The Next Digital Decade

THE NEXT DIGITAL DECADE
ESSAYS ON THE FUTURE OF THE INTERNET

Edited by Berin Szoka & Adam Marcus

THE NEXT DIGITAL DECADE
ESSAYS ON THE FUTURE OF THE INTERNET

Edited by Berin Szoka & Adam Marcus

NextDigitalDecade.com

TechFreedom
techfreedom.org
Washington, D.C.

!"#$ &'() &*$ +,-.#$"/0 -1 !/2"3(//0'4 5!"#$%&""'()*(&+67 * 8'89+(':#;
+,-.#2 +'.#21 ;"#8) ;*8) -*$/0 #8 <*$"#8=;'87 >?@? !/2"3(//0'4A$ 4#$$#'8 #$
;' ,8./*$" ;"/ +('=(/$$ ': ;/2"8'.'=1 ;"*; #4+('B/$ ;"/ ",4*8 2'80#;#'8 *80
/C+*80$ #80#B#0,*. 2*+*2#;1 ;' 2"''$/? </ =(*;/:,..1 *2)8'&./0=/ ;"/ =/8/(',$
*80 ,82'80#;#'8*. $,++'(; :'( ;"#$ +('D/2; +('B#0/0 -1 E/(#F#=87 G82?

H'(/ #8:'(4*;#'8 *-',; ;"#$ -'') #$ *B*#.*-./ *; ,"-./0+0.12/"#1'"*#()

GFIJ KLM9N9OPQL9RLMR9L

S TUNU -1 !/2"3(//0'47 <*$"#8=;'87 >?@?

!"#$ &'() #$ .#2/8$/0 ,80/( ;"/ @(/*;#B/ @'44'8$ V;;(#-,;#'89
J'8@'44/(2#*.9F"*(/V.#)/ P?U W8+'(;/0 X#2/8$/? !' B#/& * 2'+1 ': ;"#$
.#2/8$/7 B#$#; $..3455#&"1.06"#())(78*(&+520#"78"859:;7#;815<*=5 '( $/80
* ./;;/( ;' @(/*;#B/ @'44'8$7 NLN F/2'80 F;(//;7 F,#;/ PUU7 F*8 3(*82#$2'7
@*.#:'(8#*7 KONUQ7 WFV?

@'B/( >/$#=8/0 -1 Y/:: 3#/.0#8=?
THE NEXT DIGITAL DECADE: ESSAYS ON THE FUTURE OF THE INTERNET 3

TABLE OF CONTENTS
Foreword 7
!"#$% '()*+
25 Years After .COM: Ten Questions 9
!"#$% '()*+
Contributors 29
Part I: The Big Picture & New Frameworks
CHAPTER 1: The Internets Impact on
Culture & Society: Good or Bad? 49
Why We Must Resist the Temptation of Web 2.0 51
,%-#". /""%
The Case for Internet Optimism, Part 1:
Saving the Net from Its Detractors 57
,-+0 12$"#"#
CHAPTER 2: Is the Generative Internet at Risk? 89
Protecting the Internet Without Wrecking It:
How to Meet the Security Threat 91
3)%+42+% 5$44#+$%
A Portrait of the Internet as a Young Man 113
,%% !+#4).
Saving the Net from Its Supporters 139
,-+0 12$"#"#
CHAPTER 3: Is Internet Exceptionalism Dead? 163
The Third Wave of Internet Exceptionalism 165
6#$7 8)9-0+%
A Declaration of the Dependence of Cyberspace 169
,9": /)($%;*$ +%- 3);2 8)9-<))4
Is Internet Exceptionalism Dead? 179
1$0 =>

4 TABLE OF CONTENTS

Section 230 of the CDA:
Internet Exceptionalism as a Statutory Construct 189
l. Brian lolland
Internet Exceptionalism Revisited 209
Mark MacCarthy
CHAPTER 4: Has the Internet Fundamentally
Changed Economics? 237
Computer-Mediated Transactions 239
lal R. Varian
Decentralization, Freedom to Operate & Human Sociality 257
\ochai Benkler
The Economics of Information:
From Dismal Science to Strange Tales 273
Larry Downes
The Regulation of Reputational Information 293
Lric Goldman
CHAPTER 5: Who Will Govern the Net in 2020? 305
Imagining the Future of Global Internet Governance 307
Milton Mueller
Democracy in Cyberspace: Self-Governing Netizens
& a New, Global Form of Civic Virtue, Online 315
Daid R. Johnson
Whos Who in Internet Politics:
A Taxonomy of Information Technology Policy & Politics 327
Robert D. Atkinson
Part II: Issues & Applications
CHAPTER 6: Should Online Intermediaries
Be Required to Police More? 345
Trusting (and Verifying) Online Intermediaries Policing 347
lrank Pasquale
Online Liability for Payment Systems 365
Mark MacCarthy

Fuzzy Boundaries: The Potential Impact of Vague
Secondary Liability Doctrines on Technology Innovation 393
Paul Szynol
CHAPTER 7: Is Search Now an Essential Facility? 399
Dominant Search Engines:
An Essential Cultural & Political Facility 401
lrank Pasquale
The Problem of Search Engines as Essential Facilities:
An Economic & Legal Assessment 419
Georey A. Manne
Some Skepticism About Search Neutrality 435
James Grimmelmann
Search Engine Bias &
the Demise of Search Engine Utopianism 461
Lric Goldman
CHAPTER 8: What Future for Privacy? 475
Privacy Protection in the Next Digital Decade:
Trading Up or a Race to the Bottom? 477
Michael Zimmer
The Privacy Problem: Whats Wrong with Privacy? 483
Stewart Baker
A Market Approach to Privacy Policy 509
Larry Downes
CHAPTER 9: Can Speech Be Policed
in a Borderless World? 529
The Global Problem of State Censorship
& the Need to Confront It 531
John G. Palrey, Jr.
The Role of the Internet Community
in Combating Hate Speech 547
Christopher \ol

6 TABLE OF CONTENTS

CHAPTER 10: Will the Net Liberate the World? 555
Can the Internet Liberate the World? 557
Lgeny Morozo
Internet Freedom: Beyond Circumvention 565
Lthan Zuckerman


Foreword
Berin Szoka
1his book is both a beginning and an end. Its publication marks the beginning
o 1echlreedom, a new non-proit think tank that will launch alongside this
book in January 2011. Our mission is simple: to unleash the progress o
technology that improes the human condition and expands indiidual capacity
to choose. 1his book also marks an end, haing been conceied while I was
Director o the Center or Internet lreedom at 1he Progress & lreedom
loundation-beore Pll ceased operations in October 2010, ater seenteen
years.
\et this book is just as much a continuation o the theme behind both Pll and
1echlreedom: progress !" reedom.` As the historian Robert Nisbet so
elegantly put it: the condition as well as the ultimate purpose o progress is the
greatest possible degree o reedom o the indiidual.`
1
1his book`s twenty-six
contributors explore this theme and its interaction with relentless technological
change rom a wide ariety o perspecties.
Personally, this book is the perect synthesis o the themes and topics that set
me down the path o studying Internet policy in the late 1990s, and weaes
together most o the major books and authors that hae inluenced the
eolution o my own thinking on cyberlaw and policy. I hope this collection o
essays will oer students o the ield the kind o authoritatie surey that would
hae greatly accelerated my own studies. Len more, I hope this olume excites
and inspires those who may someday produce similar scholarship o their
own-perhaps to be collected in a similar olume celebrating another major
Internet milestone.
I am deeply grateul to Shane 1ews, Vice President or Global Public Policy and
Goernment Relations at VeriSign, who irst suggested publishing this sort o a
collection to commemorate the 25th anniersary o the irst .COM domain
name ,registered in 1985, by asking what the uture might bring or the
Internet. Just as I hope readers o this book will be, she had been inspired by
reading $%& '()*" +%* ,*+- ./+*0/*+ 1&2*0/!/3* 4 5(06"763+6&/, a collection o
cyberlaw essays edited by Adam 1hierer and Clyde \ayne Crews, and published
by the Cato Institute in 2003. 1his book would not exist without the
unconditional and generous support o VeriSign, the company that currently
operates the .COM registry.

1
ROBLR1 NISBL1, lIS1OR\ Ol 1lL IDLA Ol PROGRLSS 215 ,1980,.
8 FOREWORD

Nor would the book exist without the superb intellectual contributions and
patience o our twenty-six authors, and all those who assisted them. I must also
thank Pll Summer lellows Alexis Zayas, Je Ley and Zach Brieg or their
inaluable assistance with editing and organization, and Je lielding or the
book`s stunning coer artwork and design.
Most o all, I must thank Adam 1hierer and co-editor Adam Marcus. 1he two
and a hal years I spent working closely with them on a wide range o
technology policy topics at Pll were the highlight o my career thus ar.
I look orward to helping, in some small way, to discoer the uncertain uture o
progress, reedom, and technology in the next digital decade-and beyond.

Berin Szoka
December 16, 2010


25 Years After .COM: Ten Questions
Berin Szoka
\hile historians quibble oer the Internet`s birth date, one date stands out as
the day the Internet ceased being a niche or a limited number o uniersities,
goernments and military organizations, and began its transormation into a
medium that would connect billions: On March 15, 1985, Symbolics, a
Massachusetts computer company, registered symbolics.com, the Internet`s irst
commercial domain name.
2
1his book celebrates that highly symbolic`
anniersary by looking not to the Internet`s past, but to its uture. \e hae
asked twenty-six thought leaders on Internet law, philosophy, policy and
economics to consider what the next digital decade might bring or the Internet
and digital policy.
Our ten questions are all essentially ariations on the theme at the heart o
1echlreedom`s mission: \ill the Internet, on its own, improe the human
condition and expand indiidual capacity to choose` I not, what is required to
assure that technological change aoe. sere mankind Do the beneits o
goernment interention outweigh the risks Or will digital technology itsel
make digital markets work better Indeed, what would better` mean Can
\e the Netizens,` acting through the digital equialent o what Alexis de
1ocqueille called the intermediate institutions` o ciic society,` discipline
both the Internet`s corporate intermediaries ,access proiders, hosting
proiders, payment systems, social networking sites, search engines, and een
the Domain Name System operators, and our goernments
Part I ocuses on ie Big Picture & New lrameworks` questions:
1. las the Internet been good or our culture and society
2. Is the open Internet at risk rom the drie to build more secure, but less
generatie` systems and deices \ill the Internet ultimately hinder
innoation absent goernment interention
3. Is the Internet really so exceptional ater all, or will-and should-the
Internet be regulated more like traditional communications media
4. 1o ocus on one aspect o the Internet exceptionalism, has the Internet
undamentally changed economics \hat beneits and risks does this
change create
5. \ho-and what ideas-will goern the Net in 2020-at the end o the
next digital decade

2
John C Abell, DotCov Rerotvtiov tart. !itb a !biver, \IRLD MAGAZINL, March 15, 2010,
!""#$%%&&&'&()*+',-.%"!(/+01(2"*,!%3454%46%46578/1.9-:(,/8;()/"8+-",-.%
10 25 YEARS AFTER .COM: TEN QUESTIONS

Part II tackles ie Issues & Applications` questions:
6. Should intermediaries be required to police more-or be disciplined in bor
they police their networks, systems and serices \hether one thinks the
Internet is truly exceptional, and whether it has changed economics largely
determines one`s answer to these questions.
. \hile debates about the role o online intermediaries and the adequacy o
their sel-regulation ocused on net neutrality in the last digital decade, the
battle oer search neutrality` may be just as heated in the next digital
decade. Are search engines now the essential acilities` o the speech
industry that can be tamed only by regulation Or are they engines o
empowerment that will address the ery concerns they raise by ongoing
innoation
8. As the Internet accelerates the low o inormation, what uture is there or
priacy, both rom goernments and priate companies Is priacy a right
low should it be protected-rom both goernment and priate
companies
9. 1he book concludes with two Chapters regarding the Internet in a
borderless world. 1he irst ocuses on goernments` regulation o speech.
10. 1he second ocuses on the potential or goernments` disruption` b,
speech-by unettered communication and collaboration among the
citizenry. In both cases, our authors explore the consequences-and
limits-o the Internet`s empowerment o users or democracy, dissent and
pluralism.
Part I: Big Picture & New Frameworks
The Internet's Impact on Culture & Society:
Good or Bad?
Andrew Keen, the sel-declared Anti-Christ o Silicon Valley`
3
is scathing in
his criticism o the Internet, especially \eb 2.0.` Keen declares we must aoid
the siren song o democratized media,` citizen journalism, and, as the title o
his irst book puts it, the Cvtt of tbe .vatevr. le laments the technology that
arms eery citizen with the means to be an opinionated artist or writer` as
producing a techno-utopian delusion little dierent rom Karl Marx`s antasies
o a communist society-where nobody has one exclusie sphere o actiity
but each can become accomplished in any branch he wishes.`
Keen recognizes the reality o Moore`s Law-the doubling o computing
capability eery two years-but reuses to accept the idea that each adance in

3
1im Dowling, aovt tbiv/ btogger. reaa, 1lL GUARDIAN, July 20, 200,
!""#$%%&&&'<=0)+(02',-'=>%"*,!2-:-<1%344?%@=:%34%,-.#="(2<02+"!*2*"'9-->/

technology is accompanied by an equialent improement in the condition o
man.` Inormation technology is leading us into an obliion o cultural
amnesia, narcissism, and a childish rejection o the expertise, wisdom and
quality o creatie elites. lor Keen, a latter` world is one in which genius can
no longer rise aboe a sea o mediocrity, noise and triiality. lis message on
the erge o the next digital decade might as well be: Abandon all hope, ye
who enter here!` Keen`s pessimism is as strident as a certain Pollyannaish
utopianism on the other side.
Is there a middle ground Adam 1hierer, Senior Research lellow at George
Mason Uniersity`s Mercatus Center, insists there must be. In two related
essays, 1hierer describes two schools o Internet pessimism: net skeptics
generally pessimistic about technology and net loers` who think the good ol`
days` o the Internet were truly great but are nonetheless pessimistic about the
uture. 1his irst essay responds to Net skeptics like Keen-putting him in the
context o centuries o techno-pessimism, beginning with the tale rom Plato`s
Pbaearv. o 1heuth and 1hamus. 1hierer`s response is Pragmatic Optimism:
\e should embrace the amazing technological changes at work in today`s
Inormation Age but with a healthy dose o humility and appreciation or the
disruptie impact and pace o that change. \e need to think about how to
mitigate the negatie impacts associated with technological change without
adopting the paranoid tone or Luddite-ish recommendations o the pessimists.`
Is the Generative Internet at Risk?
larard Law Proessor Jonathan Zittrain summarizes the themes rom his
inluential 2008 book, 1be vtvre of tbe vtervet-.va or to to t. Zittrain is
1hierer`s prototypical Net-loing pessimist who worries how technology will
eole absent interention by those capable o steering technology in better
directions. Zittrain worries that consumer demand or security will drie the
deelopers and operators o computer networks, serices and deices to reduce
what he calls the generatiity` o their oerings. 1hus, unregulated markets
will tend to produce closed systems that limit experimentation, creatiity and
innoation. In particular, Zittrain decries the trend towards appliancized`
deices and serices-which, unlike the traditional personal computer, can load
only those applications or media authorized by the deeloper. Not only does
this diminish user control in the immediate sense, greater regulability` also
creates the potential or the Internet`s gatekeepers` to abuse their power.
1hus, Zittrain echoes the prediction made by Larry Lessig in Coae-without a
doubt the most inluential Internet policy book eer-that Let to itsel,
cyberspace will become a perect tool o control.`
4

4
Lawrence Lessig, CODL AND O1lLR LA\S Ol C\BLRSPACL 5-6 ,1999,.

In the end, he proposes essentially two kinds o solutions or Protecting the
Internet without \recking It.` 1he irst is essentially an appeal to the ciic
irtues o netizenship.` Second, regulation may be required to orce
companies to proide basic tools o transparency that empower users to
understand exactly what their machines are doing,` as well as data portability
policies.` More radically, he proposes to impose liability on deice
manuacturers who do not respond to takedown requests regarding
ulnerabilities in their code that could harm users. And, returning to his core
ear o appliancized deices, he proposes that network neutrality-style
mandates` be imposed on that subset o appliancized systems that seeks to
gain the generatie beneits o third-party contribution at one point in time
while resering the right to exclude it later.`
Ann Bartow, Proessor at the Uniersity o South Carolina School o Law,
oers a stinging rebuke o Zittrain`s 1be vtvre of tbe vtervet. She summarizes the book as
ollows: \e hae to regulate the Internet to presere its open, unregulated
nature.` ler essay draws an analogy to James Joyce`s 1916 noel, . Portrait of
tbe .rti.t a. a Yovvg Mav-emphasizing Zittrain`s desire or the independence o
his digital homeland, much as Joyce wrote about Ireland. But as a leading
cyber-eminist, she is especially critical o what she characterizes as Zittrain`s
call or an elite circle o people with computer skills and ree time who share
his policy perspectie` to rule his preerred uture ,which she calls the
Zittrainet`, as Oerlords o Good laith.`
As Bartow characterizes Zittrain`s philosophy, 1he technologies should be
generatie, but also monitored to ensure that generatiity is not abused by either
the goernment or by scoundrels, elite Internet users with, as one might say
today, mad programming skilz` should be the superisors o the Internet,
scrutinizing new technological deelopments and establishing and modeling
productie social norms online, and aerage, non-technically proicient Internet
users should ollow these norms, and should not demand security measures that
unduly burden generatiity.` In the end, she inds Zittrain`s book lacking in
clear deinitions o generatiity` and in speciic proposals or how to aoid a bad
uture or people whose interests may not be recognized or addressed by what
is likely to be a ery homogeneous group o elites` composed primary by male
elites like Zittrain.
Like Bartow, Adam 1hierer rejects Zittrain`s call or rule by a Platonic elite o
philosopher,programmer kings in the Case or Internet Optimism, Part 2:
Saing the Net rom Its vorter..` 1hierer connects the work o Larry Lessig,
Jonathan Zittrain and 1im \u as the dominant orces in cyberlaw, all united by
an oer-riding ear: 1he wide-open Internet experience o the past decade is
giing way to a new regime o corporate control, closed platorms, and walled
gardens.` 1hierer argues that they oerstate the threats to openness and

generatiity. Because companies hae strong incenties to strike the right
openness,closedness balance.. things are getting more open all the time
anyway, een though the Internet was neer quite so open or generatie as the
Openness Langelicals` imagine. In the end, he concludes it is signiicantly
more likely that the |regulated| openness` they adocate will deole into
expanded goernment control o cyberspace and digital systems than that
unregulated systems will, as the Openness Langelicals ear, become subject to
perect control` by the priate sector.` 1hus, 1hierer rejects what Virginia
Postrel called, in her 1998 book 1be vtvre ava it. vevie., the stasis
mentality.`
5
Instead, he embraces Postrel`s eolutionary dynamism: the
continuum |between openness and closedness| is constantly eoling and .
this eolution is taking place at a much aster clip in this arena than it does in
other markets.` In the end, he argues or the reedom to experiment-a
recurring theme o this collection.
Is Internet Exceptionalism Dead?
Lric Goldman, proessor at Santa Clara Uniersity School o Law, proides a
three-part historical ramework or understanding the Internet Lxceptionalism
debate. In the mid-1990s, Internet Utopianism reigned triumphant, exempliied
in the 1996 Declaration o the Independence o Cyberspace` by John Perry
Barlow, lyricist or the Grateul Dead.
6
Despite its radicalism, this lirst \ae
o Internet Lxceptionalism succeeded in getting Congress to add the only
section o the Communications Decency Act that would surie when the
Supreme Court struck down the rest o the Act on lirst Amendment grounds:
Section 230, which categorically immunizes online proiders rom liability or
publishing most types o third party content` and thus is clearly exceptionalist
because it treats online proiders more aorably than oline publishers-een
when they publish identical content.` 1hat law lies at the heart o the
philosophical debate in this Chapter and Chapter 6: Should Online
Intermediaries Be Required to Police More` 1he Second \ae ,Internet
Paranoia`, led regulators to treat the Internet more harshly than analogous
oline actiity. 1he 1hird \ae ,Lxceptionalism Prolieration`, proposed
laws treating speciic sites and serices dierently, especially social networks.
1he Deadhead Barlow was dead wrong, declare-essentially-the lon. Alex
Kozinski, Chie Judge o the Ninth Circuit Court o Appeals, and Josh
Goldoot, Department o Justice litigator-each writing only in their priate
capacity-in A Declaration o the Deevaevce o Cyberspace.` \hile they agree

5
VIRGINIA POS1RLL, 1lL lU1URL AND I1S LNLMILS ,1998,.
6
Declaration o John P. Barlow, Cognitie Dissident, Co-lounder, Llec. lrontier lound., .
Dectaratiov of tbe vaeevaevce of C,ber.ace ,leb. 8, 1996,, araitabte at
!""#$%%&3'*;;'-)<%A*2/-)/!(#%B2"*)2*"C,*2/-)/!(#C9(::/%90):-&C43DE'+*,:0)0"(-2.

that online anonymity and long-distance communications indeed make it harder
or goernments to punish law-breakers, goernments are not helpless: By
placing pressure on |intermediaries like hosting companies, banks and credit
card companies| to cut o serice to customers who break the law, we can
indirectly place pressure on Internet wrong-doers.` 1hey illustrate their point
with the examples o secondary liability or copyright inringement and Judge
Kozinski`s Roovvate..cov decision. Indeed. they reject the conceit that
|cyberspace| exists at all` as a distinct, let alone exceptional place, as well as
arguments that the costs to Internet companies o handling traditional
regulations are too high.
Columbia Law Proessor 1im \u concurs that goernments can, and do,
regulate the Internet because o what he and Jack Goldsmith called, in their
2006 book !bo Covtrot. tbe vtervet., the persistence o physicality.` 1his is not
necessarily something to be celebrated, as he notes, pointing to China`s ery
innoatieness in inding ways to repress its citizens online-a subject
addressed in this collection`s inal Chapter. Another o 1hierer`s Net-Loing
Pessimists,` \u proesses Internet optimism but insists we must be realistic
about the role o goernment.`
\u summarizes the lengthy account in his 2010 book 1be Ma.ter ritcb o how
goernment is both responsible or creating inormation monopolists and yet
also the only orce ultimately capable o dethroning them. lor \u, the Internet
is vot exceptional-rom 1he Cycle` o alternation between
centralization,closedness and decentralization,openness. \et \u agrees the
Internet is indeed an exception to the general trend o traditional media:
|t|echnologically, and in its eects on business, culture and politics.` 1hus, he
compares the ideology as expressed in its technology` and the American
exceptionalism o Alexis de 1ocqueille. \et such exceptionalism, \u warns,
cannot be assumed, but must be deended.` \u closes with a ery useul
bibliography o leading works in this ongoing debate.
l. Brian lolland, Proessor at 1exas \esleyan School o Law, responds with a
ull-bore deense o what he calls the modiied Internet Lxceptionalism`
encapsulated in Section 230-modiied` to be less audacious than Goldman`s
lirst \ae ,the Internet is inherently unregulable`,, but still bold in its
insistence that granting broad immunity to online intermediaries or the conduct
o their users is ital to the lourishing o cyber-libertarian` \eb 2.0
communities-such as wikis and social networks, capable o eoling their own
norms and enorcement mechanisms or policing behaior. lolland proides a
history o Section 230 and the debate oer Internet exceptionalism that rames
the discussion o intermediary deputization in Chapter 6. le explains how
Larry Lessig`s coniction that priate power leads to perect control, as
mentioned aboe, ultimately split the Internet Lxceptionalist consensus against

regulation o the 1990s into two camps. Both camps carried the banner o
Internet reedom but reached opposite conclusions about whether the real
threat comes rom goernment or the priate sector-most notably, regarding
Net Neutrality. Despite this racturing, lolland notes that the exceptional
deregulation made possible by Section 230 has grown, not contracted, in its
interpretation by the courts since 1996.
Similarly, Mark MacCarthy, Adjunct Proessor in the Communications Culture
and 1echnology Program at Georgetown Uniersity, explains how |t|he initial
demand rom Internet exceptionalists that the online world be let alone by
goernments has morphed into the idea that goernments should create a
global ramework to protect and spur the growth o the Internet.` Once the
exaggerated claims about the impossibility o regulating the Net made by lirst
\ae Internet Lxceptionalists proed alse, the question became not whether
|i|ntermediaries can control illegal behaior on the Internet and goernments
can control intermediaries, but .bovta they`
Based on his irst-hand experience at Visa ,described in Chapter 6,, MacCarthy
seems willing to accept more intermediary deputization than lolland but insists
that |t|he establishment o these laws needs to ollow all the rules o good
policymaking, including imposing an obligation only when the social beneits
exceed the social costs.` lurthermore, he warns that a bordered Internet in
which each country attempts to use global intermediaries to enorce its local
laws will not scale. 1his is the undamentally correct insight o the Internet
exceptionalists.` 1hus, MacCarthy concludes, I goernments are going to use
intermediaries to enorce local laws, they are going to hae to harmonize the
local laws they want intermediaries to enorce.`
Has the Internet Fundamentally
Changed Economics?
Google`s chie economist lal Varian proides a coda to the 1998 book
vforvatiov Rvte.: . trategic Cviae to tbe êtror/ covov, ,with Carl Shapiro o
the Uniersity o Caliornia at Berkeley,. 1hat book pioneered the exploration
o the unique aspects o inormation economics, and their implications or both
business and policy. lere, Varian argues that the Internet`s most
underappreciated impact on our economy lies in the obious yet under-
appreciated ubiquity o computers in our economic transactions, acilitating
our broad categories o combinatorial innoation`: new orms o contract,
data extraction and analysis, controlled experimentation, and personalization
and customization. Varian celebrates the transormatie potential o cloud
computing technology to allow een tiny companies working internationally to
launch innoatie new applications and serices that, in turn, can sere as

building blocks or new sorts o combinatorial innoation in business processes
that will oer a huge boost to knowledge worker productiity in the uture.`
larard Law Proessor \ochai Benkler is best known or his book 1be !eattb of
êtror/.-a clear allusion to Adam Smith`s 16 classic 1be !eattb of âtiov..

1hose amiliar with this part o Smith`s work iew him narrowly as an
economist ocused solely on what has traditionally been characterized as
economic exchange. But Smith in act was equal parts economist, moral
philosopher, and jurisprudentialist-and so is Benkler. Benkler`s essay,
Decentralization, lreedom to Operate, and luman Sociality,` harkens back to
Smith`s other key work, 1be 1beor, of Morat evtivevt. ,159,. lor both Smith
and Benkler, man`s natural sociability means that our distributed interactions
tend to beneit society rom the bottom-up-as i by Smith`s inisible hand.`
lor Benkler, the Internet is a global network o communications and exchange
that allows much greater low and conersation, so that many new connections
are possible on scales neer beore seen.` Like Varian, Benkler celebrates the
potential or cloud computing to acilitate accelerating and unprecedented
collaboration.
But the keys to Benkler`s uture are sociality, oluntarism, widespread
experimentation, and the reedom to experiment. 1he latter insistence makes
him highly critical o is intellectual property-copyright, patent, etc. \et he does
not address the dangers o propertizing personal data as another orm o
intellectual property. \hat does priacy-property mean or data-drien
experimentation and the reedom to experiment 1his question, unanswered
here, oers perhaps the most tantalizing organizing theme or a uture
successor to this collection o essays.
Larry Downes closes this Chapter with an expanded ersion o the discussion
o digital economics rom his 2009 book 1be ar. of Di.rvtiov-a book in the
same tradition as Varian and Shapiro`s vforvatiov Rvte. ,1998,, Postrel`s 1be
vtvre ava it. vevie. ,1998,, and Clayton Christensen`s 1be vvorator`. Ditevva
,199,. lere, Downes proposes ie principles o inormation economics that
make the digital economy dierent: ,1, Renewability: inormation cannot be
used up`, ,2, Uniersality: eeryone has the ability to use the same inormation
simultaneously,` ,3, Magnetism: Inormation alue grows exponentially as new
users absorb it,` ,4, lriction-ree: the more easily inormation lows, the more
quickly its alue increases,` and ,5, Vulnerability: 1he alue o inormation can
be destroyed through misuse or een its own success-inormation oerload.

ADAM SMI1l, AN INQUIR\ IN1O 1lL NA1URL AND CAUSLS Ol 1lL \LAL1l Ol NA1IONS 18-
21 ,Ldwin Cannan, ed., Methuen & Co., Ltd. 1904, ,16,,
!""#$%%&&&'*,-2:(9'-)<%:(9)0)1%F.("!%/.GH'!".:.

lor Downes, the Internet has changed economics in a second sense: by
relentlessly and ruthlessly cutting transaction costs-e.g., the costs o search,
inormation, bargaining, decision, policing and enorcement. 1hus, Varian`s
computer-mediation promises to dramatically latten our economy: As
transaction costs in the open market approach zero, so does the size o the
irm-i transaction costs are nonexistent, then there is no reason to hae large
companies`-what Downes calls 1he Law o Diminishing lirms.`
Downes echoes Postrel`s critique o the stasis mentality: the old rules do little
more than hold back innoation or the beneit o those who cannot or do not
know how to adapt to the economics o digital lie.` Like Benkler, Downes
particularly worries about copyright law`s ability to keep pace, but also explores
the implications o lower transactions costs or priacy, asking: \hat happens
when the cost o deleting inormation is higher than the cost o retaining it
1he answer is that nothing gets deleted.` In Chapter , both Downes and
Stewart Baker explore the costs and beneits o priacy regulation.
linally, Lric Goldman oers another three-part conceptual ramework-this
time, or understanding how the Internet has reolutionized markets or
reputational inormation. Goldman argues that well-unctioning marketplaces
depend on the ibrant low o accurate reputational inormation.` 1he Internet
may allow markets to regulate themseles better: I reputational inormation
that was preiously locked in consumers` heads` can low reely, it can play
an essential role in rewarding good producers and punishing poor ones.`
Smith`s inisible hand alone is not enough, but reputational inormation acts
like an inisible hand guiding the inisible hand`-the secondary inisible
hand.` A tertiary inisible hand` allows the reputation system to earn
consumer trust as a credible source. or to be drummed out o the market or
lack o credibility..
`
Goldman cautions against interentions that suppress reputational inormation,
but also highlights the potential unintended consequences o interentions
intended to make reputation markets work better-like anti-gaming rules and a
right-o-reply. Like lolland, Goldman emphasizes the central importance o
Section 230`s immunity in allowing reputation systems to lourish without being
crushed by intermediary liability or policing obligations.
Who Will Govern the Net in 2020?
Lach o the three authors in this Chapter wisely resists the temptation to make
oerly speciic prophesies and instead considers the broad themes likely to
shape the policy debate oer the Internet`s uture. New \ork School o Law
Proessor Daid Johnson and Syracuse Inormation Studies Proessor Milton
Mueller ocus on who .bovta goern the Net in 2020-and could just as easily
hae responded to our question about Internet Lxceptionalism-while Rob

Atkinson, President o the Inormation 1echnology and Innoation
loundation, proides a ield guide` to the eight major camps in Internet
policy.
Lchoing Postrel`s dynamist,stasist theme, like 1hierer, Mueller predicts 1he
uture o Internet goernance will be drien by the clash between its raw
technical potential and the desire o arious incumbent interests-most notably
nation-states-to assert control oer that potential|.` le hopes the Internet
will be goerned by a denationalized liberalism` based on a uniersal right to
receie and impart inormation regardless o rontiers, and sees reedom to
communicate and exchange inormation as undamental, primary elements o
human choice and political and social actiity.` 1his will require the authority
o national and subnational goernments must be contained to domains o law
and policy suited to localized or territorialized authority,` while Internet
goernance institutions must be completely detached rom nation-state
institutions. Deenders o ree speech will ultimately hae to use global ree
trade institutions to strike down censorship.
Mueller inds strong grounds or optimism in the Internet`s empowering and
democratizing nature, and in the rise o new access technologies like unlicensed
wireless broadband capable o disrupting existing Internet access bottlenecks.
But he worries about the growing technological capabilities o broadband
proiders to manage and potentially censor traic on their networks, and admits
a darker uture o strie, industrial consolidation, censorship and cyber-warare
is possible. Like Zittrain, Mueller ears a splintering o the Internet drien by
conlicts oer the Internet`s Root Serer,` and that such conlicts are bound to
intensiy as the drie to secure the Internet against cyber-threats and cyber-
warare intensiies.
Like \u, Daid Johnson, reaches back to 1ocqueille`s Devocrac, iv .verica
,1835,. \hile Mueller proposes a new liberalism, Johnson proposes
Democracy in Cyberspace: Sel-Goerning Netizens and a New, Global lorm
o Ciic Virtue, Online.` Paraphrasing 1ocqueille, Johnson argues: 1he
Internet establishes a new equality o condition and enables us to exercise
liberty to orm associations to pursue new ciic, social, and cultural goals.`
1hus, the Internet is inherently democratic`-in ways well beyond politics.
But the Internet`s nature as an engine o democratic ciic irtue` must be
deended daily by netizens-the global polity o those who collaborate online,
seek to use the new aordances o the Internet to improe the world, and care
about protecting an Internet architecture that acilitates new orms o ciic
irtue.` Johnson argues against \u`s apparent resignation to some degree o
goernment meddling online: A world in which eery local soereign seeks to
control the actiities o netizens beyond its borders iolates the true meaning o
sel-goernance and democratic soereignty.` Johnson predicts that technology

will empower users to sidestep the traditional controls imposed by
goernments-not perectly, but well evovgb. 1hus, the Internet can ulill the
more modest ambitions o lirst \ae Internet exceptionalists: by making the
Internet ecetiovatt, democratic and pluralistic.
Johnson`s approach resembles 1hierer`s Pragmatic Optimism staked out by
Adam 1hierer: "the trajectory o reedom and een ciic irtue has been, in
broad terms, oer time, constantly upward-because eeryone who gets a
chance to experience an increased leel o democratic sel-goernment-a new
equality o condition.`` Like Varian, Benkler and Downes, Johnson sees the
Internet`s acilitation o collaboration and communication as the keys to
democratic empowerment.
As a think tank eteran, Rob Atkinson oers a 1axonomy o Inormation
1echnology Policy and Politics,` describing eight camps and their positions
along our key issues. lirst is perhaps the strongest, yet also the hardest to
deine: the B2"*)2*" JK,*#"(-20:(/"/, the Netizens` who beliee that they
launched the Internet reolution,` preer inormal Internet goernance, and
generally oppose goernment interention online-especially copyright. By
contrast, F-,(0: J2<(2**)/ distrust large corporations een more than
goernment, thus leading them to adocate regulatory solutions. 1hough
Atkinson doesn`t draw the connection, this camp might well be uniied by
Lessig`s concept o code as law`-updated as choice architecture,` in the
highly inluential 2008 book ^vage: vrorivg Deci.iov. abovt eattb, !eattb, ava
aive.. by Cass Sunstein and Richard 1haler. L)** M0)>*"*)/ are those who
beliee the Internet empowers people, liberates entrepreneurs, and enables
markets`-especially by reducing transactions costs. Atkinson`s proposed tent
may be rather too large, potentially encompassing some who adocate
regulations like net neutrality or antitrust interention they beliee are the key to
reeing markets. 1he term cyber-libertarian, seems both narrower and broader
than Atkinson`s conception o ree-marketeers.`
8
Indeed, it was originally the
term Atkinson used or the Internet Lxceptionalist` camp, ocused primarily
on cyber-tibertivi.v and a anatic rejection o copyright.
M-)0: A-2/*)N0"(N*/, on the other hand, hae no qualms about enlisting
goernments to regulate the Internet` to stamp out sin and sedition. O:+
J,-2-.1 P*<=:0"-)/ reject Internet exceptionalism absolutely and insist on
continuing to regulate the Internet like all media in the public interest.` Q*,!
A-.#02(*/ R Q)0+* S//-,(0"(-2/ are united not by philosophical approach
but by their ultimate duty to shareholders, while T)(,>/802+8M-)"0)/

8
ee Adam 1hierer & Berin Szoka, C,beribertariavi.v: 1be Ca.e for Reat vtervet reeaov, 1lL
1LClNOLOG\ LIBLRA1ION lRON1, Aug. 12, 2009, !""#$%%"*,!:(9*)0"(-2',-.%344D%4U%
53%,19*)8:(9*)"0)(02(/.8"!*8,0/*8;-)8)*0:8(2"*)2*"8;)**+-.%

companies, proessional groups, and unions generally work to thwart the
Internet`s disruption o their business models-exempliying Virginia Postrel`s
stasis mindset.` Atkinson`s own camp is that o the M-+*)0"*/, who want
goernment to do no harm` to inormation technology innoations, but also
to actiely do good` by adopting policies to promote digital transormation` o
the economy.
Part II: Issues & Applications
Should Online Intermediaries
Be Required to Police More?
Seton lall Law Proessor lrank Pasquale argues that the Internet allows
intermediaries to shroud their operations in what might be called perect
opaqueness`-to extend Larry Lessig`s eared model o perect control.`
Pasquale uses the example o Google to illustrate the many ways in which
online intermediaries choose to police the Internet, een when not required to
do by goernments. Gien the critical policing role played by intermediaries,
Pasquale proposes an Internet Intermediary Regulatory Council` to help
courts and agencies adjudicate controersies concerning intermediary practice`
and assure adequate monitoring-a prerequisite or assuring a leel playing
ield online.` 1he IIRC could include a search engine diision, an ISP
diision ocusing on carriers, and eentually diisions related to social networks
or auction sites i their practices begin to raise commensurate concerns.`
\hile leaing open the possibility that the IIRC could be a priate entity,
Pasquale is unabashed in citing Robert lale, theoretician o the New Deal`s
regulatory renzy: lale`s crucial insight was that many o the leading businesses
o his day were not extraordinary innoators that desered` all the proits they
made, rather, their success was dependent on a network o laws and regulation
that could easily shit aor rom one corporate player to another.` But rather
than repealing these laws and regulation to allow the eolutionary dynamism`
o competition to play out, as Adam 1hierer proposes, Pasquale is willing to
rely on competition-promotion ia markets and antitrust only to the extent
that ,a, the intermediary in question is an economic ,as opposed to cultural or
political, orce, ,b, the oice` o the intermediary`s user community is strong,
and ,c, competition is likely to be genuine and not contried.` Otherwise,
competition is inadequate. 1he bottom line,` Pasquale concludes, is that
someone needs to be able to look under the hood` o culturally signiicant
automated ranking systems.` 1hus, the Internet is vot exceptional: Pasquale
beliees only careul regulatory oersight can protect us rom shadowy
corporations, just as in lranklin Delano Rooseelt`s telephone-and-radio era.

\hile Pasquale seems not to object to intermediaries acting as arms o the
police state so long as they are properly transparent and regulated, Mark
MacCarthy cautions against the practical problems raised by intermediary
policing and oers an analytical model or deciding when intermediary
deputization is appropriate. Based on his experience as Senior Vice President
or Public Policy at Visa Inc., MacCarthy explores how payment systems hae
handled Internet gambling and copyright inringement as exemplary case studies
in intermediary deputization because, unlike most online intermediaries,
payment systems are subject neither to Section 230`s absolute immunity or
third-party content or actiities nor to the notice-and-take-down conditional
immunity o the Digital Millennium Copyright Act.
MacCarthy inds cause or optimism about sel-regulation: regardless o the
precise legal liabilities, intermediaries hae a general responsibility to keep their
systems ree o illegal transactions and they are taking steps to satisy that
obligation.` But he insists intermediary liability should be imposed only where
real market ailures exists, where supported by an analysis o costs, beneits
and equities,` where spelled out clearly, and to the extent local laws are
harmonized internationally.
1he most troubling orm o intermediary deputization comes rom uncertain
secondary copyright liability, writes independent writer, lawyer and programmer
Paul Szynol in an expanded ersion o an essay originally written or the
Llectronic lrontier loundation. le challenges the anti-exceptionalist
arguments made by Judge Kozinski and Josh Goldoot. Szynol argues that the
ailure to clearly deine such liability chills innoation and inestment in
innoatie start-ups-and that that this problem i. unique to the Internet, gien
the astly larger scale o competition acilitated by digital markets.
Most intriguingly, Szynol argues that Kozinski and Goldoot contradict their
argument against Internet Lxceptionalism by insisting on a standard or
secondary liability online that is not actually applied oline. Szynol asks,
should a car company be held liable or driers who speed Ater all, it would
be easy enough to add a speed limit compliance chip.` \et auto manuacturers
are not orced to pay any portion o a speeding drier's ticket. Oline, in other
words, bad actors-the v.er. o technology-are punished or their own
transgressions. Online, howeer, the law chases the manuacturers-and
applies ad-hoc, ambiguous standards |o secondary liability| to their products.`
1hus, or all their denunciation o lirst \ae Lxceptionalists like John Perry
Barlow, Szynol essentially insists Kozinski and Goldoot are actually Goldman`s
Second \ae` Internet Lxceptionalists who want to impose vore punitie
regulations online than oline.

Is Search Now an "Essential Facility?"
lrank Pasquale brings his theory o intermediary regulation to ull ruition with
his sweeping call or search neutrality.` Like 1im \u in 1be Ma.ter ritcb,
Pasquale worries that antitrust law is incapable o protecting innoation and
adequately addressing the the cultural and political concerns that dominant
search engines raise." 1hus, he aims to point the way toward a new concept o
essential cultural and political acility,` which can help policymakers realize the
situations where a bottleneck has become important enough that special
scrutiny is warranted.` In particular, Pasquale sees taming search as inextricably
intertwined with protecting priacy-Lngaging in a cost-beneit analysis |as in
antitrust law| diminishes priacy's status as a right`-and Google`s potential
chokehold on inormation through the Google Books Settlement.
1he existence o competition in search, especially rom Microsot`s Bing, and
the potential or competition rom lacebook and other serices yet to be
inented, are essentially irreleant to Pasquale, while the lirst Amendment`s
protection o search engine operators are a complication to be addressed down
the road. le concludes by insisting that regulation should be supplemented by
a publicly unded alternatie to the dominant priate sector search engine-
something the lrench goernment has heaily subsidized a Luropean Quaero`
search engine. Similarly, in Chapter 6, Pasquale proposed to model his Internet
Intermediary Regulatory Council on the lrench Data Protection Authority.
1hus, Pasquale`s oer-arching ision seems to be that o a Digital New Deal-a
ta fravai.e.
Georey Manne, Proessor at Lewis & Clark Law and Lxecutie Director o
the International Center or Law & Lconomics, explains that search engines are
not the bottlenecks Pasquale suggests-and thus why een the traditional
essential acilities doctrine, which he says has been relegated by most antitrust
experts to the dustbin o history,` should not apply to them. In essence, he
argues that search neutrality` would protect only competitors, not consumers,
because een a popular search engine like Google cannot oreclose adertisers`
access to consumers` attention. Google, like any company, has no legal duty to
help its rials. More to the point, een i Google entirely dominated search, it
could not block cov.vver.` access to its competitors. 1his, argues Manne, is the
releant market to analyze-quoting Supreme Court Justice Abe lortas`s
amous admonition about excessiely narrow market deinitions: 1his Court
now approes this strange red-haired, bearded, one-eyed man-with-a-limp
classiication.`
Like Manne, New \ork Law School Proessor James Grimmelmann expresses
Skepticism about Search Neutrality,` and the signiicant practical problems it

would create. As the author o the deinitie law reiew article, 1be trvctvre of
earcb vgive ar,
9
Grimmelmann is keenly aware o the concerns raised by
search, yet he concludes that the case or search neutrality is a muddle`
because its ends and means don`t match.` Lchoing Johnson, Mueller,
lolland, and 1hierer`s iew o the Internet as a liberating, democratizing orce,
Grimmelmann is clear that the lodestar o search is user autonomy: I search
did not exist, then or the sake o human reedom it would be necessary to
inent it.` le deconstructs eight search neutrality principles-equality, object-
iity, bias, traic, releance, sel-interest, transparency and manipulation-and
inds each lacking, but cautions that it doesn`t ollow that search engines
desere a ree pass under antitrust, intellectual property, priacy, or other well-
established bodies o law,` and that some otber orm o search-speciic legal
oersight` might be appropriate.
Lric Goldman once again puts the debate in the context o its intellectual
history. Always ocused on questions o exceptionalism, Goldman concludes
search engines are neutral only in theory ,Search Lngine Utopianism`, but
must make editorial judgments just like any other media company.` le
explains that, while search engine bia. sounds scary, . such bias is both
necessary and desirable`-and the remedy o search neutrality` is probably
worse than whateer aderse consequences come with search engine bias.
Ultimately, he predicts that emerging personalization technology will soon
ameliorate many concerns about search engine bias.`
What Future for Privacy Online?
Michael Zimmer, Proessor o Inormation Studies at School o the Uniersity
o \isconsin-Milwaukee, concedes that the Internet has become a platorm
or the open low o personal inormation-lows that are largely oluntarily
proided by users.` \et Zimmer discusses lingering reasons or concern about
the Internet as a potent inrastructure or the low and capture o personal
inormation.`
Zimmer explores the conlicts among priacy laws in the U.S., Lurope, Canada
and elsewhere, but concludes that Companies are, on the whole, not moing
around in order to aoid strict priacy regulations. instead, there has been a
gradual increase in awareness and action on the issue o priacy.` Still, Zimmer
worries that the trading up` to an increased leel o protection o personal
inormation lows on our transnational digital networks has not materialized as
quickly or clearly as one might expect.` Zimmer`s answer is to demand a
renewed commitment to the rights o data subjects embodied in the Canadian
and Luropean Union approach to data protection.`

9
James Grimmelmann, 1be trvctvre of earcb vgive ar, 93 IO\A L. RLV. 1 ,200,.

Zimmer writes rom the perspectie that iews priacy as a right.` 1his is, to
put it mildly, not a perspectie shared by the other two authors in this Chapter:
Stewart Baker, a Partner at Steptoe & Johnson LLP and ormer Assistant
Secretary or Policy at the Department o lomeland Security ,DlS,, and Larry
Downes, who has expanded his essay rom his 2009 book 1be ar. of Di.rvtiov.
Baker spent his time at DlS battling priacy adocates oer programs he elt
justiied to protect Americans against terrorism-leading him to ask, \hat`s
\rong with Priacy` le traces the answer back to the 1890 law reiew article,
1he Right to Priacy` by Supreme Court Justice Louis Brandeis and Samuel
\arren that gae birth to modern priacy law. Baker rejects their reactionary
deense o the status quo` as Boston elites who didn`t much like the news
media reporting on the details o their rirate parties. In essence, Baker inds in
the priacy moement` the same stasis mentality` deined by Virginia Postrel.
Like Postrel, Baker argues or dynamism: Lach new priacy kerule inspires
strong eelings precisely because we are reacting against the eects o a new
technology. \et as time goes on, the new technology becomes commonplace.
Our reaction dwindles away. 1he raw spot grows a callous. And once the initial
reaction has passed, so does the sense that our priacy has been inaded. In
short, we get used to it.`
Baker rejects the concept o predicates` or goernment access to data ,e.g.,
requiring probable cause` or a warrant,, the Brandeisian notion that we
should all own` our personal data,` and attempting to limit uses o inormation.
Baker has little to say about the priate sector`s use o data but proposes a
system o auditing goernment employees to rigorously monitor their use o
priate inormation.
Larry Downes, too, rejects the concept o intellectual property in personal
inormation-but is willing to concede that \arren and Brandeis weren`t
entirely wrong` in that priate` inormation can also be used destructiely.`
le thus leaes open the possibility o narrow laws tailored to limiting speciic,
destructie uses o inormation-such as anti-discrimination laws. But Downes
is highly skeptical about goernmental enorcement o priacy rights,` and
ultimately echoes John Perry Barlow`s optimism about the potential or
Netizens to sole their own problems: \here there are real conlicts, where
there are wrongs, we will identiy them and address them by our means.`
10

Speciically, Downes argues that the same technologies that create the priacy
problem are also proing to be the source o its solution. Len without
goernment interention, consumers increasingly hae the ability to organize,
identiy their common demands, and enorce their will on enterprises`-
detailing examples o how reputational pressure can discipline corporate priacy

10
Barlow, .vra note 6.

practices. 1hree cheers or the sound o Lric Goldman`s three inisible hand
clapping, perhaps Ultimately, Downes ests his greatest hope in the Internet`s
potential to create new markets by lowering transactions costs-this time, a
market or priate data in which an explicit qvia ro qvo rewards consumers or
sharing their personal data or beneicial, rather than destructie uses.
Can Speech Be Policed
in a Borderless World?
John Palrey, larard Law Proessor and co-director o larard`s inluential
Berkman Center or Internet & Society, speaks with unique authority on
censorship as one o the co-authors o exhaustie sureys o global censorship
conducted by himsel, Jonathan Zittrain and others at Berkman. 1hese studies
conirm 1im \u`s conclusion that goernments can and do censor speech
eectiely, contrary to the hopes o lirst \ae Internet Lxceptionalists.
Palrey proides a beginner`s guide to the techniques used in, goals o, and
practical problems created by content iltering. Most disturbingly, he notes the
growing use o sot controls` through goernmental pressure and goernment-
ostered social norms intended to squelch dissent.
Like Zittrain, Mueller and Johnson, Palrey ears we may be headed toward a
localized ersion o the Internet, goerned in each instance by local laws.` le
thus demands a greater international debate about speech controls that orces
states to discuss whether they actually ravt their citizens to hae ull access to
the Internet or not.` In particular, he echoes Mueller`s call or international ree
trade institutions to strike down censorship barriers to ree speech.
Christopher \ol, Partner at logan lartson LLP, ocuses not on speech that
goernments hate, but on hate speech` we all-or nearly all-would ind
objectionable. \et he notes how diicult it can be to distinguish these two
categories o censorship. lurthermore, he concludes, ater much crusading
against hate speech, that laws against hate speech hae not demonstrably
reduced hate speech or deterred haters.` 1hus, he concludes that late speech
can be policed` in a borderless world, but not principally by the traditional
police o law enorcement. 1he Internet community must continue to sere as
a neighborhood watch` against hate speech online, saying something when it
sees something,` and working with online proiders to enorce community
standards.` 1hus, like Johnson, Mueller and Barlow, \ol looks to Netizens to
combat hate speech.


Can the Net Liberate the World?
1he book closes by discussing the most tragic disappointment o the lirst \ae
Internet Lxceptionalists` ision. \here John Perry Barlow insisted, deiantly,
that goernments those weary giants o lesh and steel. |did not| possess any
methods o enorcement we hae true reason to ear,` the reality is that
oppressie goernments continue to reign, sometimes een using the Internet
to sere their agenda. Can the Net liberate the world-or will it, too, become
another tool o perect control,` as Larry Lessig eared Or will imperect
controls work well enough to allow tyrants to hang on to power
Lgeny Morozo is a leading commentator on oreign aairs, a isiting scholar
at Stanord Uniersity and a Schwartz ellow at the New America loundation.
le praises the Internet`s ability to quickly disseminate inormation and allow
dissidents to organize. \et, haing grown up in the Soiet Union, he is deeply
skeptical about the much-hyped potential or \eb media to lie up to the hype
about democratization. le rejects two critical assumptions underlying this
hype. lirst, he concludes that the legitimacy o undemocratic regimes is deried
less rom brainwashing` that can be cured by exposure to the alternatie iews
online and more rom popular support or authoritarian regimes that promise to
delier economic growth or play eectiely on other concerns, such as
nationalism or religion. Second, he suggests the Internet can actually acilitate
sureillance, uel genuine support or existing regimes, allow goernment to
subtly manipulate public opinion, or simply make authoritarianism more
eicient.
John Palrey`s acid obseration in the preious Chapter bolsters Morozo`s
suggestion that much o the world may not actually ravt to be liberated: In
China and in parts o the ormer Soiet Union, ery oten the most earsome
enorcer o the state's will is the old woman on one's block, who may or may
not be on the state's payroll.`
Optimists like Johnson, Mueller, 1hierer and lolland would likely dier rom
Morozo-and the U.S. State Department has tended in this direction, too. In
January 2010, Secretary o State lillary Clinton gae a bold speech embracing
this optimism about the liberating potential o the Internet, and announcing a
commitment to supporting the deelopment o new tools that enable citizens
to exercise their rights o ree expression by circumenting politically motiated
censorship.`
11

11
lillary Rodham Clinton, Revar/. ov vtervet reeaov, Jan. 21, 2010,
!""#$%%&&&'/"0"*'<-N%/*,)*"0)1%).%3454%45%56775D'!".

Internet entrepreneur Lthan Zuckerman is a senior researcher at the Berkman
Center and ounder o Geekcorps, a non-proit dedicated to building computer
inrastructure in deeloping countries. le joined John Palrey in the study o
censorship circumention tools mentioned aboe.
12
Despite his passionate
commitment to promoting such tools, as Secretary Clinton proposed, he
concludes that \e can`t circument our way around Internet censorship`
because o the costs and practical challenges o attempting to circument
censorship on a scale suicient to make a real dierence. 1hus, he iews
circumention as just one o many tools required to thwart sot censorship,
website blocking, and attacks on dissident sites. But ultimately, what is most
required is building the right theory o change` to inorm the multi-pronged
strategy necessary or the Internet to achiee its democratizing potential.
Conclusion: Discovering the Future
of the Internet & Digital Policy

In these thirty-one essays, our authors paint a complex picture o the uture o
the Internet and digital policy: 1echnological change ineitably creates new
problems, een as it soles old ones. In the end, one`s perspectie ultimately
depends on whether one thinks the net` eect o that change is positie or
negatie-depending on how much, and in what ways, goernment interenes
online.
Personally, this collection brings me back to where I started my study o
Internet policy-reading John Perry Barlow`s Declaration o the Independence
o Cyberspace` in 1996, and Virginia Postrel`s 1be vtvre ava t. vevie. in 1998.
Despite its now-obiously excessie utopian naet about the Internet`s
crippling o the State, Barlow`s poetry still resonates deeply with many,
including mysel, as a powerul synthesis o Internet exceptionalism and cyber-
libertarianism, a ision o progress as empowerment and upliting o the user.
\et like my ormer colleague Adam 1hierer, it is Postrel`s eolutionary
dynamism that most guides me, with its emphasis not on a careully outlined
uture` or build|ing| a single bridge rom here to there, or neither here nor
there is a single point,` but on the roce.. o discoery by which the uture
eoles.
13
Like Postrel, I do not imagine that the disruption and transormation
wrought by the Digital Reolution will always be rosy or easy. But we cannot-

12
lAL ROBLR1S, L1lAN ZUCKLRMAN & JOlN PALlRL\, 200 CIRCUMVLN1ION LANDSCAPL
RLPOR1: ML1lODS, USLS, AND 1OOLS ,March 2009,, !""#$%%+0/!'!0)N0)+'*+=%
9("/")*0.%!02+:*%5%3?DVD66%344?CA(),=.N*2"(-2CW02+/,0#*'#+;X/*Y=*2,*Z3.
13
Postrel, svra note 5 at 218.

as the legendary King Canute once tried with the Lnglish Channel-command
the tides o technological change to halt.
1hierer`s Pragmatic Optimism` demands much more than a resignation to the
ineitability o change. At its heart, it is requires a cheery conidence in what
Daid Johnson dubs the 1rajectory o lreedom`-in broad terms, oer time,
constantly upward`-but also a commitment to the process by which that
trajectory is discoered. 1bi. is progress-progress a. reedom.
14
But progress
also reqvire. reedom, the reedom to discoer, innoate and experiment, i
technology is to achiee its ull potential to improe the human condition and
expand indiidual capacity to choose.
I leae it to you, the reader, to choose-to discoer-your own answers to the
many questions o law, economics, philosophy and policy explored in this
unique book.

14
ROBLR1 NISBL1, lIS1OR\ Ol 1lL IDLA Ol PROGRLSS 215 ,1980,.
29
CONTRIBUTORS
Robert D. Atkinson 31
Stewart Baker 31
Ann Bartow 32
Yochai Benkler 32
Larry Downes 33
Josh Goldfoot 34
Eric Goldman 34
James Grimmelmann 35
H. Brian Holland 35
David R. Johnson 36
Andrew Keen 36
Hon. Alex Kozinski 37
Mark MacCarthy 37
Geoffrey Manne 38
Evgeny Morozov 39
Milton Mueller 39
30 CONTRIBUTORS
John Palfrey 40
Frank Pasquale 40
Berin Szoka 41
Paul Szynol 41
Adam Thierer 42
Hal Varian 42
Christopher Wolf 43
Tim Wu 44
Michael Zimmer 44
Jonathan Zittrain 45
Ethan Zuckerman 46

Robert D. Atkinson
Robert Atkinson is the ounder and president o the Inormation 1echnology
and Innoation loundation, a \ashington, D.C.-based technology policy think
tank. le is also author o the tate êr covov, vae series and the book, 1be
Pa.t .va vtvre Of .verica`. covov,: ovg !are. Of vvoratiov 1bat Porer C,cte.
Of Crortb ,Ldward Llgar, 2005,. le has an extensie background in technology
policy, he has conducted ground-breaking research projects on technology and
innoation, is a alued adiser to state and national policy makers, and a popular
speaker on innoation policy nationally and internationally.
Beore coming to I1Il, Dr. Atkinson was Vice President o the Progressie
Policy Institute and Director o PPI`s 1echnology & New Lconomy Project.
\hile at PPI he wrote numerous research reports on technology and innoation
policy, including on issues such as broadband telecommunications, Internet
telephony, uniersal serice, e-commerce, e-goernment, middleman opposition
to e-commerce, priacy, copyright, RlID and smart cards, the role o I1 in
homeland security, the R&D tax credit, oshoring, and growth economics.
Preiously Dr. Atkinson sered as the irst Lxecutie Director o the Rhode
Island Lconomic Policy Council, a public-priate partnership including as
members the Goernor, legislatie leaders, and corporate and labor leaders. As
head o RILPC, he was responsible or drating a comprehensie economic
strategic deelopment plan or the state, deeloping a ten-point economic
deelopment plan, and working to successully implement all ten proposals
through the legislatie and administratie branches. Prior to that he was Project
Director at the ormer Congressional Oice o 1echnology Assessment. \hile
at O1A, he directed 1be 1ecbvotogicat Re.baivg of Metrootitav .verica, a seminal
report examining the impact o the inormation technology reolution on
America`s urban areas.
Stewart Baker
!"#$%&" () *%+#& is a partner in the \ashington oice o Steptoe & Johnson
LLP. le returned to the irm ollowing 3' years at the Department o
lomeland Security as its irst Assistant Secretary or Policy.
At lomeland Security, Mr. Baker created and staed the 250-person DlS
Policy Directorate. le was responsible or policy analysis across the
Department, as well as or the Department`s international aairs, strategic
planning and relationships with law enorcement and public adisory
committees. \hile at DlS, Mr. Baker led successul negotiations with
Luropean and Middle Lastern goernments oer trael data, priacy, isa
waier and related issues. le deised a new approach to isa-ree trael, orged
a congressional and interagency consensus on the plan and negotiated
acceptance with key goernments.
32 CONTRIBUTORS
Mr. Baker manages one o the nation`s premier technology and security
practices at Steptoe. Mr. Baker`s practice coers national security, electronic
sureillance, law enorcement, export control encryption, and related
technology issues. le has been a key adisor on U.S. export controls and on
oreign import controls on technology. le has also adised companies on the
requirements imposed by ClIUS.
Mr. Baker`s practice includes issues relating to goernment regulation o
international trade in high-technology products, and adice and practice under
the antidumping and counterailing duty laws o United States, Luropean
Community, Canada, and Australia. le also counsels clients on issues inoling
oreign soereign immunity, and compliance with the loreign Corrupt Practices
Act.
Mr. Baker has handled the arbitration o claims exceeding a billion dollars, is a
member o national and international rosters o arbitrators, and is the author o
articles and a book on the United Nations Commission on International 1rade
Law arbitration rules.
Mr. Baker has had a number o signiicant successes in appellate litigation and
appearances beore the United States Supreme Court. le deeloped-and
persuaded the Court to adopt-a new theory o constitutional ederalism that
remains the most ibrant 10th Amendment doctrine o the past 30 years.
Ann Bartow
(,, *%&"-$ is a Proessor o Law at the Uniersity o South Carolina School
o Law in Columbia, South Carolina. She holds a Bachelor o Science rom
Cornell Uniersity and a Juris Doctor rom the Uniersity o Pennsylania Law
School. She currently teaches an Intellectual Property Surey, Copyright Law,
1rademarks and Unair Competition Law, Patent Law and a seminar entitled
Pornography, Prostitution, Sex 1raicking and the Law. ler scholarship
ocuses on the intersection between intellectual property laws and public policy
concerns, priacy and technology law, and eminist legal theory. She also co-
administers the .#/0,01" 2%$ 3&-4#11-&1 blog, is a regular blogger at
Madisonian.net and a contributing editor at Jotwell.com.
Yochai Benkler
5-67%0 *#,+8#& is the Berkman Proessor o Lntrepreneurial Legal Studies at
larard, and aculty co-director o the Berkman Center or Internet and
Society. Beore joining the aculty at larard Law School, he was Joseph M.
lield 55 Proessor o Law at \ale. le writes about the Internet and the
emergence o networked economy and society, as well as the organization o
inrastructure, such as wireless communications.
In the 1990s he played a role in characterizing the centrality o inormation
commons to innoation, inormation production, and reedom in both its
autonomy and democracy senses. In the 2000s, he worked more on the sources
and economic and political signiicance o radically decentralized indiidual
action and collaboration in the production o inormation, knowledge and
culture. lis books include 1lL \LAL1l Ol NL1\ORKS: lO\ SOCIAL
PRODUC1ION 1RANSlORMS MARKL1S AND lRLLDOM ,2006,, which receied
the Don K. Price award rom the American Political Science Association or
best book on science, technology, and politics.
In ciil society, Benkler`s work was recognized by the Llectronic lrontier
loundation`s Pioneer Award in 200, and the Public Knowledge IP3 Award in
2006. lis articles include Orercovivg .gorabobia ,199,98, initiating the debate
oer spectrum commons,, Covvov. a. êgtectea actor of vforvatiov Proavctiov
,1998, and ree a. tbe .ir to Covvov |.e ,1998, characterizing the role o the
commons in inormation production and its relation to reedom,, rov
Cov.vver. to |.er. ,2000, characterizing the need to presere commons as a core
policy goal, across all layers o the inormation enironment,, Coa.e`. Pevgviv, or
ivv ava tbe âtvre of tbe irv ,characterizing peer production as a basic
phenomenon o the networked economy, and barivg îcet, ,2002,
characterizing shareable goods and explaining sharing o material resources
online,. lis work can be reely accessed at 9#,+8#&)-&:.
Larry Downes
2%&&; <-$,#1 is an Internet analyst and consultant, helping clients deelop
business strategies in an age o constant disruption caused by inormation
technology.
Downes is author o |vtea.bivg tbe Kitter .: Digitat trategie. for Mar/et
Dovivavce ,larard Business School Press, 1998,, which was named by the \all
Street Journal as one o the ie most important books eer published on
business and technology.
lis new book, 1be ar. of Di.rvtiov: arve..ivg tbe êr orce. tbat Corerv ie ava
v.ive.. iv tbe Digitat .ge ,Basic Books 2009, oers nine strategies or success in
naigating the accident-prone intersection o innoation and the law.
lrom 2006-2010, Downes was a nonresident lellow at the Stanord Law School
Center or Internet and Society. Beore that, he held aculty positions at the
Uniersity o Caliornia-Berkeley, Northwestern Uniersity School o Law, and
the Uniersity o Chicago Graduate School o Business. Downes is a Partner
with the Bell-Mason Group, which works with Global 1000 corporations,
proiding corporate enturing methodologies, tools, techniques and support
that accelerate corporate innoation and enturing programs.
34 CONTRIBUTORS
le has written or a ariety o publications, including U.S.A 1oday, larard
Business Reiew, Inc., \ired, CNet, Strategy & Leadership, CIO, 1he
American Schola& ava tbe larard Journal o Law and 1echnology. le was a
columnist or both 1he Industry Standard and CIO Insight. le blogs or the
=#67,-8-:; 209#&%"0-, .&-,".
Josh Goldfoot
Josh Goldoot is Senior Counsel with the Computer Crime & Intellectual
Property Section o the U.S. Department o Justice. le prosecutes hackers and
other computer criminals, and adises inestigators and other prosecutors about
priacy statutes, the lourth Amendment, and implications o emerging
technologies on law enorcement. In 2010, he was awarded the Assistant
Attorney General`s Meritorious Serice Award. le is an accomplished sotware
deeloper and computer technician, and receied a United States patent in 2008
or shape recognition technology. le is a graduate o \ale Uniersity and
earned his law degree rom the Uniersity o Virginia School o Law. le has
worked in technology law since 1999, when he adised Internet startups in
Silicon Valley on intellectual property issues. Prior to joining the Department
o Justice in 2005, he did appellate and ciil litigation, and clerked or judge Alex
Kozinski on the Ninth Circuit U.S. Court o Appeals. le was a Special
Assistant United States Attorney in the Lastern District o Virginia or six
months in 200 and 2008.
Eric Goldman
>&06 ?-8@/%, is an Associate Proessor o Law at Santa Clara Uniersity
School o Law. le also directs the school`s ligh 1ech Law Institute. Beore
joining the SCU aculty in 2006, he was an Assistant Proessor at Marquette
Uniersity Law School, General Counsel o Lpinions.com, and an Internet
transactional attorney at Cooley Godward LLP.
Lric teaches Cyberlaw and Intellectual Property and preiously has taught
courses in Copyrights, Contracts, Sotware Licensing and Proessional
Responsibility.
Lric`s research ocuses on Internet law, intellectual property, marketing, and the
legal and social implications o new communication technologies. Recent
papers hae addressed topics such as search engines and online marketing
practices.
Lric receied his BA, .vvva cvv tavae and Phi Beta Kappa, in
Lconomics,Business rom UCLA in 1988. le receied his JD rom UCLA in
1994, where he was a member o the UCLA Law Reiew, and concurrently
receied his MBA rom the Anderson School at UCLA.
James Grimmelmann
A%/#1 ?&0//#8/%,, is Associate Proessor at B#$ 5-&+ 2%$ !67--8 and a
member o its C,1"0"D"# 4-& C,4-&/%"0-, 2%$ %,@ 3-806;. le receied his
J.D. rom 5%8# 2%$ !67--8, where he was Lditor-in-Chie o LawMeme and a
member o the 5%8# 2%$ A-D&,%8. Prior to law school, he receied an A.B. in
computer science rom E%&F%&@ G-88#:# and worked as a programmer or
H06&-1-4". le has sered as a Resident lellow o the C,4-&/%"0-, !-60#";
3&-I#6" at \ale, as a legal intern or G&#%"0F# G-//-,1 and the >8#6"&-,06
.&-,"0#& .-D,@%"0-,, and as a law clerk to the lonorable Maryanne 1rump
Barry o the United States Court o Appeals or the 1hird Circuit.
le studies how the law goerning the creation and use o computer sotware
aects indiidual reedom and the distribution o wealth and power in society.
As a lawyer and technologist, he aims to help these two groups speak intelligibly
to each other. le writes about intellectual property, irtual worlds, search
engines, online priacy, and other topics in computer and Internet law. Recent
publications include 1be vtervet . a evicovvov., 8 lordham L. Re. 299
,2010,, arivg aceboo/, 94 Iowa L. Re. 113 ,2009,, 1be tbicat 1i.iov. of
Co,rigbt ar, lordham L. Re. 2005 ,2009,.
le has been blogging since 2000 at the Laboratorium: $$$)8%9-&%"-&0D/),#".
H. Brian Holland
Proessor E) *&0%, E-88%,@ joined the aculty o 1exas \esleyan School o
Law in 2009. Prior to his arrial, Proessor lolland was a Visiting Associate
Proessor at Penn State Uniersity`s Dickinson School o Law.
Ater graduating rom law school, Proessor lolland spent two years as a
judicial clerk in the U.S. Court o Appeals or the Second Circuit in New \ork.
le then joined the \ashington, D.C. oice o Jones, Day, Reais & Pogue.
lis work with the irm consisted primarily o appellate work beore the U.S.
Supreme Court and ederal courts o appeals, as well as international arbitration
beore the \orld Bank. Among the signiicant cases litigated during this period
were issues o intellectual property and constitutional law ,Lldred .
Reno,Ashcrot and Luck`s Music Library, Inc. . Reno,Ashcrot,, priacy and
identity thet ,1R\ . Andrews,, and ederal bankruptcy jurisdiction and enue.
Proessor lolland`s scholarship relects his interest in technology, goernance
and social change, with a particular emphasis on issues o authority within the
online enironment and the deelopment o social norms in mediated
communities. le is currently writing on priacy in social networks. lis most
recent work, ociat Di.tortiov: Regvtativg Prirac, iv ociat êtror/., has been a
eatured presentation at priacy conerences both in the United States and
Lurope.
36 CONTRIBUTORS
Proessor lolland receied a LL.M., with honors, rom Columbia Uniersity
School o Law, completing a sel-designed program in technology law. le
holds a J.D., summa cum laude, rom American Uniersity`s \ashington
College o Law, and a B.A. rom 1uts Uniersity. Proessor lolland is
currently pursuing his Ph.D. in Digital Media and Mass Communications at
Penn State Uniersity. lis dissertation, now in progress, applies social semiotic
theories to the concept o air use in intellectual property law.
David R. Johnson
Daid Johnson joined New \ork Law School`s aculty in spring 2004 as a
isiting proessor o law. le is a aculty member o the Institute or
Inormation Law and Policy.

Proessor Johnson joined \ilmer, Cutler & Pickering in 193 and became a
partner in 1980. lis practice ocused primarily on the emerging area o
electronic commerce, including counseling on issues relating to priacy, domain
names and Internet goernance issues, jurisdiction, copyright, taxation,
electronic contracting, encryption, deamation, ISP and OSP liability, regulation,
and other intellectual property matters.
Proessor Johnson helped to write the Llectronic Communications Priacy Act,
was inoled in discussions leading to the lramework or Global Llectronic
Commerce, and has been actie in the introduction o personal computers in
law practice. le co-authored, with Daid Post, Law and Borders: 1he Rise o
Law in Cyberspace, 48 Stanord Law Reiew 136 ,1996,. le is currently
deeloping online legal games and law practice simulations.
Proessor Johnson graduated rom \ale College with a B.A. summa cum laude
in 196. le completed a year o postgraduate study at Uniersity College,
Oxord in 1968, and earned a J.D. rom \ale Law School in 192. lollowing
graduation rom law school, Proessor Johnson clerked or a year or the
lonorable Malcolm R. \ilkey o the United States Court o Appeals or the
District o Columbia.
Andrew Keen
(,@&#$ J##, is the author o the international hit CUL1 Ol 1lL AMA1LUR:
lO\ 1lL IN1LRNL1 IS KILLING OUR CUL1URL ,2008,. Acclaimed by 1lL
NL\ \ORK 1IMLS` Michiko Kakutani as haing been written with acuity and
passion` and by A.N. \ilson in the DAIL\ MAIL as staggering`, CUL1 Ol
1lL AMA1LUR has been published in 404"##, @044#&#," 8%,:D%:# #@0"0-,1 and
was short-listed or the 2008 ligham`s Business 1echnology Book o the \ear
award.
As a pioneering Internet entrepreneur, Andrew ounded Audiocae.com in 1995
and built it into a popular irst generation Internet music company. le was the
executie producer o the new media show MB5 2000` and, between 2001 and
200, worked as a senior sales and marketing executie at seeral Silicon Valley-
based technology start-ups including Pulse, Santa Cruz Networks and Pure
Depth. le is currently the host o the popular Keen On` show on
=#676&D,67)"F as well as the host o the ideo interiew series 1he luture o
Creatiity` on the larard Business Reiew website.
Andrew was educated at London Uniersity where he was awarded a lirst Class
lonors Degree in Modern listory, as a British Council lellow at the Uniersity
o Sarajeo and at the Uniersity o Caliornia at Berkeley where he earned a
Masters Degree in Political Science.
le is currently writing a second book entitled DIGI1AL VLR1IGO: ANXIL1\,
LONLLINLSS AND INLQUALI1\ IN 1lL SOCIAL MLDIA AGL which will be
published by St Martins Press.
Hon. Alex Kozinski
G70#4 AD@:# (8#K J-L0,1+0 was appointed United States Circuit Judge or the
Ninth Circuit on Noember , 1985. le graduated rom UCLA, receiing an
A.B. degree in 192, and rom UCLA Law School, receiing a J.D. degree in
195. Prior to his appointment to the appellate bench, Judge Kozinski sered
as Chie Judge o the United States Claims Court, 1982-85, Special Counsel,
Merit Systems Protection Board, 1981-1982, Assistant Counsel, Oice o
Counsel to the President, 1981, Deputy Legal Counsel, Oice o President-
Llect Reagan, 1980-81, Attorney, Coington & Burling, 199-81, Attorney,
lorry Golbert Singer & Gelles, 19-9, Law Clerk to Chie Justice \arren L.
Burger, 196-, and Law Clerk to Circuit Judge Anthony M. Kennedy, 195-
6.
Mark MacCarthy
H%&+ H%6G%&"7; teaches and conducts research at Georgetown Uniersity`s
Communication, Culture, and 1echnology Program. le teaches courses on the
deelopment o the electronic media, technology policy and Internet reedom.
le is also an adjunct member o the Department o Philosophy where he
teaches courses in political philosophy and philosophy and priacy. le does
research and consults in the areas o inormation priacy and security, Internet
intermediary liability, global Internet reedom, the uture o the online media,
consumer inancial protection, open standards, electronic and mobile commerce
and other technology policy issues. le is an Ailiate o Georgetown
Uniersity`s McDonough School o Business Center or Business and Public
Policy, an inestigator with the RlID Consortium or Security and Priacy, and
the appointed expert o the American National Standards Institute on the
38 CONTRIBUTORS
International Organization lor Standardization ,ISO, 1echnical Management
Board ,1MB, Priacy Steering Committee
lrom 2000 to 2008, he was Senior Vice President or Global Public Policy at
Visa Inc, responsible or policy initiaties aecting electronic commerce, new
technology and inormation security and priacy. le regularly represented Visa
beore the U.S. Congress, the U.S. Administration, the U.S. lederal 1rade
Commission, the U.S ederal inancial regulators and multi-goernmental
groups such as the Organization or Lconomic Cooperation and Deelopment
and Asia Paciic Lconomic Cooperation group.
Prior to joining Visa, he spent six years as a principal and senior director with
the \exler-\alker Group, a \ashington public policy consulting irm, where
he worked with a ariety o clients on electronic commerce, inancial serices,
priacy and telecommunications. le was Vice President in charge o Capital
Cities,ABC`s \ashington oice rom 1988 to 1994, representing the
company`s interests beore Congress, the lederal Communications Commission
and other administratie agencies. lrom 1981 to 1988 he was a proessional
sta member o the louse Committee on Lnergy and Commerce, where he
handled communications policy issues. lrom 198 to 1981, he worked as an
economist perorming regulatory analyses o saety and health regulations at the
U.S. Occupational Saety and lealth Administration.
Mr. MacCarthy has a Ph.D in philosophy rom Indiana Uniersity and an MA in
economics rom the Uniersity o Notre Dame.
Geoffrey Manne
Currently the Lxecutie Director o the International Center or Law &
Lconomics ,ICLL,, a global think tank, Proessor Manne also seres as
Lecturer in Law or Lewis & Clark Law School. In this capacity he lends his
expertise to arious law school endeaors and teaches the school`s Law and
Lconomics course. 1he ICLL`s website is at $$$)8%$#6-,6#,"#&)-&:.
Manne was an Assistant Proessor o Law at Lewis & Clark rom 2003 to 2008.
lrom 2006 to 2008 he took a leae o absence rom the school to direct a law
and economics academic outreach program at Microsot, and was Director,
Global Public Policy at LLCG, an economic consulting irm, until ounding the
ICLL at the end o 2009. Prior to joining the Lewis & Clark aculty he practiced
law at Latham & \atkins in \ashington, D.C., where he specialized in antitrust
litigation and counseling. Beore priate practice Manne was a Bigelow lellow
at the Uniersity o Chicago Law School, an Olin lellow at the Uniersity o
Virginia School o Law and a law clerk to Judge Morris S. Arnold o the U.S.
Court o Appeals or the Lighth Circuit. \hile clerking he taught a seminar on
Law & Literature at the Uniersity o Arkansas at Little Rock.
During law school Manne was a research assistant to Judge Richard Posner,
Comment Lditor o the Uniersity o Chicago Law School Roundtable and a
Sta Member o the Uniersity o Chicago Legal lorum. Among his other
ocational pursuits was a brie stint at the U.S. lederal 1rade Commission. lis
research has ocused broadly on the economic implications o legal constraints
on business organizations, particularly in the contexts o antitrust, nonproit
organizations and international law. Manne is a member o the Virginia bar, as
well as the Bar o the U.S. Bankruptcy Court or the Lastern District o
Virginia. le is also a member o the American Law and Lconomics
Association, the Canadian Law and Lconomics Association and the
International Society or New Institutional Lconomics.
le blogs or the =#67,-8-:; 209#&%"0-, .&-,".
Evgeny Morozov
Lgeny Morozo is the author o 1lL NL1 DLLUSION: 1lL DARK SIDL Ol
IN1LRNL1 lRLLDOM ,Public Aairs, 2011,. le is also a isiting scholar at
Stanord Uniersity, a ellow at the New America loundation and a
contributing editor to loreign Policy magazine.
Milton Mueller
Milton Mueller teaches and does research on the political economy o
communication and inormation. le uses the theoretical tools o property
rights analysis, institutional economics and both historical and quantitatie
social science methods. le has a longstanding interest in the history o
communication technologies and global goernance institutions. Mueller
receied the Ph.D. rom the Uniersity o Pennsylania in 1989.
Mueller`s most recent research projects explore the problem o goerning the
Internet. lis new book NL1\ORKS AND S1A1LS: 1lL GLOBAL POLI1ICS Ol
IN1LRNL1 GOVLRNANCL ,MI1 Press, 2010, proides a comprehensie
oeriew o the political and economic driers o a new global politics. lis
acclaimed book RULING 1lL ROO1: IN1LRNL1 GOVLRNANCL AND 1lL
1AMING Ol C\BLRSPACL ,MI1 Press, 2002, was the irst scholarly account o
the debates oer the goernance o the domain name system. lis book,
UNIVLRSAL SLRVICL: COMPL1I1ION, IN1LRCONNLC1ION AND MONOPOL\
IN 1lL MAKING Ol 1lL AMLRICAN 1LLLPlONL S\S1LM ,MI1 Press, 199,
set out a dramatic reision o our understanding o the origins o uniersal
telephone serice and the role o interconnection in industry deelopment. le
is on the international editorial boards o the journals 1LLLCOMMUNICA1IONS
POLIC\, 1lL INlORMA1ION SOCIL1\, and INlO: 1lL JOURNAL Ol POLIC\,
RLGULA1ION AND S1RA1LG\ lOR 1LLLCOMMUNICA1ION, INlORMA1ION
AND MLDIA.
40 CONTRIBUTORS
John Palfrey
A-7, 3%84&#; is lenry N. Lss Proessor o Law and Vice Dean or Library and
Inormation Resources at larard Law School. le is the co-author o orv
Digitat: |vaer.tavaivg tbe ir.t Ceveratiov of Digitat âtire. ,Basic Books, 2008, and
.cce.. Deviea: 1be Practice ava Potitic. of vtervet itterivg ,MI1 Press, 2008,. lis
research and teaching is ocused on Internet law, intellectual property, and
international law. le practiced intellectual property and corporate law at the
law irm o Ropes & Gray. le is a aculty co-director o the *#&+/%, G#,"#&
4-& C,"#&,#" M !-60#"; at larard Uniersity. Outside o larard Law
School, he is a Venture Lxecutie at lighland Capital Partners and seres on
the board o seeral technology companies and non-proits. John sered as a
special assistant at the U.S. LPA during the Clinton Administration. le is a
graduate o larard College, the Uniersity o Cambridge, and larard Law
School.
Frank Pasquale
lrank Pasquale is a proessor o law at Seton lall Law School and a isiting
ellow at the Princeton Uniersity`s Center or Inormation 1echnology Policy.
le has a JD rom \ale, was a Marshall Scholar at Oxord, and graduated rom
larard summa cum laude. le has been a isiting proessor at \ale and
Cardozo Law Schools. le has published widely on inormation law and policy.
In 2010, he was twice inited by the National Academy o Sciences`s Committee
on Law, Science, and 1echnology and its Goernment-Uniersity-Industry
Roundtable to present on the priacy and security implications o data sensor
networks. le also was inited by the Department o lealth and luman
Serices` Oice o the National Coordinator or lealth Inormation
1echnology to present at a roundtable organized to inorm ONC`s
Congressionally mandated report on priacy and security requirements or
entities not coered by lIPAA ,relating to Section 13424 o the lI1LCl Act,.
In 2008, he presented Internet Nondiscrimination Principles or Competition
Policy Online beore the 1ask lorce on Competition Policy and Antitrust Laws
o the louse Committee on the Judiciary, appearing with the General Counsels
o Google, Microsot, and \ahoo. le is the Chair o the Priacy & Deamation
section o the American Association o Law Schools or 2010.
Pasquale has been quoted in the New \ork 1imes, San lrancisco Chronicle,
Los Angeles 1imes, Boston Globe, linancial 1imes, and many other
publications. le has appeared on CNN to comment on Google`s China policy.
le has been interiewed on internet regulation on Daid Leine`s E#%&1%;
GD8"D&# poD.C.ast, \N\C`s *&0%, 2#7&#& !7-$, the Canadian
BroaD.C.asting Corporation`s documentary Lngineering Search,` and on
National Public Radio`s =%8+ -4 "7# B%"0-,. lis recent publications include
Beyond Innoation and Competition,` Network Accountability or the
Domestic Intelligence Apparatus` ,with Danielle Citron,, Restoring
1ransparency to Automated Authority,` and Data and Power.` le is
presently working on a book titled 1he Black Box Society` which examines
and critiques the rise o secret technology in the internet and inance sectors.
Berin Szoka
Berin Szoka is ounder o =#67.&##@-/, a non-proit think tank dedicated
to unleashing the progress o technology that improes the human condition
and expands indiidual capacity to choose.
Preiously, he was a Senior lellow and the Director o the Center or Internet
lreedom at =7# 3&-:&#11 M .&##@-/ .-D,@%"0-,. Beore joining Pll, he
was an Associate in the Communications Practice Group at Latham & \atkins
LLP. Beore joining Latham, Szoka practiced at Lawler Metzger Milkman &
Keeney, LLC in \ashington and clerked or the lon. l. Dale Cook, Senior
U.S. District Judge or the Northern District o Oklahoma.
Szoka receied his Bachelor's degree in economics rom Duke Uniersity and
his juris doctor rom the Uniersity o Virginia School o Law, where he sered
as Submissions Lditor o the Virginia Journal o Law and 1echnology. le is
admitted to practice law in the District o Columbia and Caliornia ,inactie,.
le seres on the Steering Committee or the D.C. Bar's G-/ND"#& M
=#8#6-//D,06%"0-,1 2%$ !#6"0-,, and on the lAA's G-//#&60%8 !N%6#
=&%,1N-&"%"0-, (@F01-&; G-//0""## ,COMS1AC,. Szoka has chaired, and
currently seres on, the Board o Directors o the !N%6# .&-,"0#& .-D,@%"0-,,
a non-proit citizens' adocacy group ounded in 1988 and dedicated to
adancing commercial opportunity and expansion o human ciilization in
space.
Paul Szynol
3%D8 !L;,-8 was born in \arsaw, Poland, and moed to the United States in
1984, the year that New \ork City`s transit are rose rom 5 cents to 90 cents,
33 preiously unknown Bach pieces were ound in an academic library, and
Canon demoed its irst digital still camera. le has lied in New \ork City, San
lrancisco, Los Angeles, New laen, Philadelphia, New Jersey, and \arsaw,
and, during his six dries across the U.S., isited the ast majority o the
contiguous states. le graduated rom Columbia Uniersity, where he studied
history and philosophy, and \ale Uniersity, where he studied intellectual
property law. le has also taken courses at the International Center o
Photography. In the past, Paul played drums and was a computer programmer,
and he still tinkers with Pearl drums and Jaa libraries. le likes dogs,
documentary photography, music, San lrancisco, Linux, and depressing moies.
le is currently based in New \ork City.
42 CONTRIBUTORS
Adam Thierer
(@%/ =70#&#& is a senior research ellow at the Mercatus Center at George
Mason Uniersity where he works with the 1echnology Policy Program.
1hierer coers technology, media, Internet, and ree speech policy issues with a
particular ocus in online child saety and digital priacy policy issues.
1hierer has spent almost two decades in the public policy research community.
le preiously sered as the President o 1he Progress & lreedom loundation,
the Director o 1elecommunications Studies at the Cato Institute, a Senior
lellow at 1he leritage loundation as a lellow in Lconomic Policy, and a
researcher at the Adam Smith Institute in London.
1hierer is the author or editor o seen books on dierse topics such as media
regulation and child saety issues, mass media regulation, Internet goernance
and jurisdiction, intellectual property, regulation o network industries, and the
role o ederalism within high-technology markets. le earned his B.A. in
journalism and political science at Indiana Uniersity, and receied his M.A. in
international business management and trade theory at the Uniersity o
Maryland.
1hierer has sered on seeral distinguished online saety task orces, including
larard Law School`s Internet Saety 1echnical 1ask lorce, a Blue Ribbon
\orking Group` on child saety organized by Common Sense Media, the
iKeepSae Coalition, and the National Cable & 1elecommunications
Association, and the National 1elecommunications and Inormation
Administration`s Online Saety and 1echnology \orking Group.` le is also
an adisor to the American Legislatie Lxchange Council`s 1elecom & I1 1ask
lorce. In 2008, 1hierer receied the lamily Online Saety Institute`s Award
or Outstanding Achieement.`
Hal Varian
E%8 O) P%&0%, is the Chie Lconomist at Google. le started in May 2002 as a
consultant and has been inoled in many aspects o the company, including
auction design, econometric analysis, inance, corporate strategy and public
policy.
le is also an emeritus proessor at the Uniersity o Caliornia, Berkeley in
three departments: business, economics, and inormation management.
le receied his SB degree rom MI1 in 1969 and his MA in mathematics and
Ph.D. in economics rom UC Berkeley in 193. le has also taught at MI1,
Stanord, Oxord, Michigan and other uniersities around the world.
Dr. Varian is a ellow o the Guggenheim loundation, the Lconometric Society,
and the American Academy o Arts and Sciences. le was Co-Lditor o the
American Lconomic Reiew rom 198-1990 and holds honorary doctorates
rom the Uniersity o Oulu, linland and the Uniersity o Karlsruhe,
Germany.
Proessor Varian has published numerous papers in economic theory, industrial
organization, inancial economics, econometrics and inormation economics.
le is the author o two major economics textbooks which hae been translated
into 22 languages. le is the co-author o a bestselling book on business
strategy, CB.QOH(=CQB OR2>!S ( !=O(=>?CG ?RC<> =Q =E> B>=TQOJ
>GQBQH5 and wrote a /-,"78; 6-8D/, or the NL\ \ORK 1IMLS rom
2000 to 200.
Christopher Wolf
G7&01"-N7#& T-84 is a director o logan Loells` Priacy and Inormation
Management practice group. Chris is widely recognized as one o the leading
American practitioners in the ield o priacy and data security law. 1he
prestigious Practising Law Institute ,PLI, tapped Chris to sere as editor and
lead author o its irst-eer treatise on the subject, and to sere as co-editor o
its guide to the lAC1A Red llags identity thet regulations. Chris recently was
heralded or his lielong experience as a litigator` by Chambers U.S.A by
ranking him as one o the nation`s top priacy lawyers. le also was asked to
orm and co-chair 1he luture o Priacy lorum, a think tank that ocuses on
modern priacy issues with a business practical-consumer riendly perspectie,
collaborating with industry, goernment, academia, and priacy adocates.
\hen MSNBC labeled Chris a pioneer in Internet law,` it was relecting on his
participation in many o the precedent-setting matters that orm the ramework
o modern priacy law.
Chris has deep experience in the entire range o international, ederal, and state
priacy and data security laws as well as the many sectoral and geographic
regulations. Chris also counsels clients on compliance with sel-regulatory
regimes.
Chris has appeared as a speaker or the International Association o Priacy
Proessionals and or the Canadian Association o Chie Priacy Oicers. le
appears annually at the PLI Institute on Priacy and Security Law. le also has
spoken at colleges and uniersities including larard, Stanord, Berkeley, the
Uniersity o Chicago, George \ashington Uniersity, Georgetown Uniersity,
and the \ashington & Lee Uniersity School o Law. le is a requent
teleision guest on priacy and related issues, appearing on PBS, NBC,
MSNBC, CNN, lox News, and others.
44 CONTRIBUTORS
Chris is a ourth-generation \ashingtonian who started his career in
\ashington, D.C. as law clerk to 1he lonorable Aubrey L. Robinson, Jr., o
the U.S. District Court or the District o Columbia. \hile in law school, he
was a member o the !a.bivgtov c ee ar Rerier.
Tim Wu
=0/ TD is a proessor at Columbia Law School. le teaches copyright and
communications.
le is the chair o media reorm organization .&## 3&#11, and writes or !8%"#
/%:%L0,# on law, media, culture, trael, and dumplings. le has also written
or some other publications as a pure reelancer, including the NL\ \ORKLR,
the NL\ \ORK 1IMLS, \ASlING1ON POS1 \LLKLND, and lORBLS.
le is also inoled in arious other projects, usually related to alternatie
channels o content distribution. Many are run through the G-8D/90%
3&-:&%/ -, 2%$ M =#67,-8-:;) One example is 3&-I#6" 3-1,#&, another is
(8"2%$, and another is J##N 5-D& G-N;&0:7"1.
lis irst book was TEQ GQB=OQ2! =E> CB=>OB>= with Jack Goldsmith.
le is writing a new book on the long patterns o media centralization and
decentralization, the publisher is Knop , Random louse.
lis "-N061 o study are: B#"$-&+ ,#D"&%80";, the history and structure o the
media and communications industries ,the book he is currently working on,,
international problems aced by the Internet ,see TEQ GQB=OQ2! =E>
CB=>OB>=, and copyright and innoation policy ,G-N;&0:7"U1
G-//D,06%"0-,1 3-806;,.
lis brother is <%F0@ TD, author o the Xbox 360 game .D88 (D"-, and his
mother is ?0880%, TDV a scientist. le is married to Kate Judge. lis best
riends are the .%/-D1 .0F#.
Michael Zimmer
Michael Zimmer, PhD, is an %1101"%," N&-4#11-& in the !67--8 -4 C,4-&/%"0-,
!"D@0#1 at the R,0F#&10"; -4 T016-,10,WH08$%D+##, and an associate at the
G#,"#& 4-& C,4-&/%"0-, 3-806; O#1#%&67.
\ith a background in new media and Internet studies, the philosophy o
technology, and inormation policy, Zimmer studies the social, political, and
ethical dimensions o new media and inormation technologies. lis research
and teaching ocuses on:

Lthics and Inormation 1echnology
Inormation Policy
\eb Search Lngines
\eb 2.0 and Library 2.0
Priacy and Sureillance 1heory
Inormation and \eb Literacy
Access to Knowledge
Internet Research Lthics

Zimmer receied his PhD in 200 rom the <#N%&"/#," -4 H#@0%V GD8"D&#V
%,@ G-//D,06%"0-, at New \ork Uniersity under the guidance o Pros.
E#8#, B011#,9%D/, (8#K ?%88-$%;, and !0F% P%0@7;%,%"7%,. le was a
Student lellow at the C,4-&/%"0-, 2%$ C,1"0"D"# at B5R 2%$ rom 2004-
200, and was the Microsot Resident lellow at the C,4-&/%"0-, !-60#";
3&-I#6" at 5%8# 2%$ !67--8 or 200-2008. Zimmer I-0,#@ U\-Milwaukee`s
School o Inormation Studies in 2008.
Zimmer earned a B.B.A. in Marketing rom the Uniersity o Notre Dame in
1994 and worked or an electronic payment processing company in Milwaukee,
\isconsin or seeral years beore moing to New \ork City to pursue a new
career in academia. le earned an M.A. in Media Lcology rom N\U in 2002,
and his doctoral studies were supported by the Phyllis and Gerald LeBo
Doctoral lellowship in Media Lcology rom the Steinhardt School o
Lducation at New \ork Uniersity. lis dissertation research was supported by
an B!. !>! <011#&"%"0-, C/N&-F#/#," ?&%,".
Zimmer has published in international journals and deliered talks across North
America and Lurope. le has been interiewed in =7# B#$ 5-&+ =0/#1, on
âtiovat Pvbtic Raaio`s H-&,0,: >@0"0-, and !60#,6# .&0@%; programs, 1be
vffivgtov Po.t, H!B*G)6-/, CQ Magaive, 1be Movtreat Caette, =7# *-1"-,
?8-9#, HC= =#67,-8-:; O#F0#$, 1be Mitrav/ee ]ovrvat evtivet, and in arious
other national and local print and radio outlets.
Zimmer was also eatured in the Is My Cellphone Spying on Me`
commentary accompanying the XW@016 1N#60%8 #@0"0-, <P< or the hit
action,thriller moie >%:8# >;#.
Jonathan Zittrain
Jonathan Zittrain is Proessor o Law at larard Law School and the larard
Kennedy School o Goernment, co-ounder o the Berkman Center or
Internet & Society, and Proessor o Computer Science in the larard School
o Lngineering and Applied Sciences. le is a member o the Board o 1rustees
o the Internet Society and is on the board o adisors or cievtific .vericav.
Preiously he was Proessor o Internet Goernance and Regulation at Oxord
Uniersity.
46 CONTRIBUTORS
lis research interests include battles or control o digital property and content,
cryptography, electronic priacy, the roles o intermediaries within Internet
architecture, and the useul and unobtrusie deployment o technology in
education.
le perormed the irst large-scale tests o Internet iltering in China and Saudi
Arabia in 2002, and now as part o the OpenNet Initiatie he has co-edited a
study o Internet iltering by national goernments, ACCLSS DLNILD: 1lL
PRAC1ICL AND POLIC\ Ol GLOBAL IN1LRNL1 lIL1LRING, and its sequel,
ACCLSS CON1ROLLLD: 1lL SlAPING Ol PO\LR, RIGl1S, AND RULL IN
C\BLRSPACL.
lis book 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1 is aailable
rom \ale Uniersity Press and Penguin UK-and under a Creatie Commons
license. lis papers may be ound at $$$)IL)-&:.
Ethan Zuckerman
>"7%, YD6+#&/%, sered as a ellow o the Berkman Center or Internet and
Society at larard Uniersity rom 2003 through 2009. Since 2009, he`s been a
senior researcher at the center, working on projects that ocus on the impact o
technology and media on the deeloping world and on quantitatie analysis o
media. \ith lal Roberts, he is working on comparatie studies o tools or
censorship circumention, techniques or blocking-resistant publishing or
human rights sites and on the Media Cloud ramework or quantitatie study o
digital media.
Lthan and Berkman ellow Rebecca MacKinnon ounded Global Voices, a
global citizen media network. Beginning at a Berkman conerence in 2004,
Global Voices has grown into an independent Netherlands-based nonproit
with oer 200 employees and olunteers in oer 100 countries. Global Voices
/%0,"%0,1 %, 0,"#&,%"0-,%8 60"0L#, /#@0% ,#$1&--/, "&%6+1 6#,1-&170N
%,@ %@F-6%"#1 4-& 4&##@-/ -4 1N##67 -,80,#, 1DNN-&"1 :&%11&--"1 60"0L#,
/#@0% #44-&"1 and 01 % N0-,##& 0, "7# 1N%6# -4 1-60%8 "&%,18%"0-,. Global
Voices` work has been supported by the MacArthur loundation, lord
loundation, Knight loundation, lios, Open Society Institute as well as
Google, Reuters and priate donors. Lthan chairs Global Voices` global board
o directors.
In 2000, Lthan ounded Geekcorps, a non-proit technology olunteer corps.
Geekcorps pairs skilled olunteers rom U.S. and Luropean high tech
companies with businesses in emerging nations or one to our month olunteer
tours. Volunteers hae sered in 14 nations, including Ghana, Senegal, Mali,
Rwanda, Armenia and Jordan, completing oer a hundred projects. Geekcorps
became a diision o the International Lxecutie Serice Corps in 2001, where
Lthan sered as a ice president rom 2001 to 2004.
47
PART I
THE BIG PICTURE & NEW FRAMEWORKS

49

CHAPTER 1
THE INTERNETS IMPACT ON
CULTURE & SOCIETY: GOOD OR BAD?
Why We Must Resist the Temptation of Web 2.0 51
!"#$%& (%%"
Saving the Net from Its Detractors 57
!#)* +,-%$%$

50 CHAPTER 1: THE INTERNETS IMPACT ON CULTURE & SOCIETY: GOOD OR BAD?


Why We Must Resist the
Temptation of Web 2.0
By Andrew Keen
*

1he ancients were good at resisting seduction. Odysseus ought the seductie
song o the Sirens by haing his men tie him to the mast o his ship as it sailed
past the Siren`s Isle. Socrates was so intent on protecting citizens rom the
seductie opinions o artists and writers, that he outlawed them rom his
imaginary republic.
\e moderns are less nimble at resisting great seductions, particularly those
utopian isions that promise grand political or cultural salation. lrom the
lrench and Russian reolutions to the counter-cultural upheaals o the 60s
and the digital reolution o the 90s, we hae been seduced, time ater time and
text ater text, by the ision o a political or economic utopia.
Rather than Paris, Moscow, or Berkeley, the grand utopian moement o our
contemporary age is headquartered in Silicon Valley, whose great seduction is
actually a usion o two historical moements: the counter-cultural utopianism
o the 60s and the techno-economic utopianism o the 90s. lere in Silicon
Valley, this seduction has announced itsel to the world as the \eb 2.0`
moement.
On one occasion, I was treated to lunch at a ashionable Japanese restaurant in
Palo Alto by a serial Silicon Valley entrepreneur who, back in the dot.com
boom, had inested in my start-up Audiocae.com. 1he entrepreneur, a Silicon
Valley eteran like me, was pitching me his latest start-up: a technology
platorm that creates easy-to-use sotware tools or online communities to
publish weblogs, digital moies, and music. It is technology that enables anyone
with a computer to become an author, a ilm director, or a musician. 1his \eb
2.0 dream is Socrates`s nightmare: technology that arms eery citizen with the
means to be an opinionated artist or writer.
1his is historic,` my riend promised me. \e are enabling Internet users to
author their own content. 1hink o it as empowering citizen media. \e can
help smash the elitism o the lollywood studios and the big record labels. Our
technology platorm will radically democratize culture, build authentic
community, and create citizen media.` \elcome to \eb 2.0.

Andrew Keen is a eteran Silicon Valley entrepreneur and digital media critic. le blogs at
!"#$%#&'(#)*+',-./+-0 and has recently launched 12'#%!3, a podcast chat show about
media, culture, and technology. le is the author o 1lL CUL1 Ol 1lL AMA1LUR: lO\
1ODA\`S IN1LRNL1 IS KILLING OUR CUL1URL ,Crown 200,.

Buzzwords rom the old dot.com era-like cool,` eyeballs,` or burn-rate`-
hae been replaced in \eb 2.0 by language which is simultaneously more
militant and absurd: empowering citizen media, radically democratize, smash
elitism, content redistribution, authentic community. 1his sociological jargon,
once the presere o the hippie counterculture, has now become the lexicon o
new media capitalism.
\et this entrepreneur owns a >4 million house a ew blocks rom Stee Jobs`s
house. le acations in the South Paciic. lis children attend the most
exclusie priate academy on the peninsula. But or all o this he sounds more
like a cultural Marxist-a disciple o Antonio Gramsci or lerbert Marcuse-
than a capitalist with an MBA rom Stanord.
In his mind, big media`-the lollywood studios, the major record labels and
international publishing houses-really did represent the enemy. 1he promised
land was user-generated online content. In Marxist terms, the traditional media
had become the exploitatie bourgeoisie,` and citizen media, those heroic
bloggers and podcasters, were the proletariat.`
1his outlook is typical o the \eb 2.0 moement, which uses 60s radicalism
with the utopian eschatology o digital technology. 1he ideological outcome
may be trouble or all o us.
So what, exactly, is the \eb 2.0 moement As an ideology, it is based upon a
series o ethical assumptions about media, culture, and technology. It worships
the creatie amateur: the sel-taught ilmmaker, the dorm-room musician, the
unpublished writer. It suggests that eeryone-een the most poorly educated
and inarticulate amongst us-can and should use digital media to express and
realize themseles. \eb 2.0 empowers` our creatiity, it democratizes`
media, it leels the playing ield` between experts and amateurs. 1he enemy o
\eb 2.0 is elitist` traditional media.
Lmpowered by \eb 2.0 technology, we can all become citizen journalists,
citizen ideographers, or citizen musicians. Lmpowered by this technology, we
will be able to write in the morning, direct moies in the aternoon, and make
music in the eening.
Sounds amiliar It`s eerily similar to Marx`s seductie promise about indiidual
sel-realization in his German Ideology:
\hereas in communist society, where nobody has one
exclusie sphere o actiity but each can become accomplished
in any branch he wishes, society regulates the general
production and thus makes it possible or me to do one thing
today and another tomorrow, to hunt in the morning, ish in
the aternoon, rear cattle in the eening, criticise ater dinner,

just as I hae a mind, without eer becoming hunter,
isherman, shepherd or critic.
1

Just as Marx seduced a generation o Luropean idealists with his antasy o sel-
realization in a communist utopia, so the \eb 2.0 cult o creatie sel-realization
has seduced eeryone in Silicon Valley. 1he moement bridges counter-cultural
radicals o the 60s such as Stee Jobs with the contemporary geek culture o
Google`s Larry Page. Between the book-ends o Jobs and Page lies the rest o
Silicon Valley including radical communitarians like Craig Newmark ,o
Craigslist.com,, intellectual property communists such as Stanord Law
Proessor Larry Lessig, economic cornucopians like !irea magazine editor Chris
Long 1ail` Anderson, journalism proessor Je Jaris, and new media moguls
1im O`Reilly and John Battelle.
1he ideology o the \eb 2.0 moement was perectly summarized at the
1echnology Lducation and Design ,1LD, show in Monterey in 2005 when
Kein Kelly, Silicon Valley`s ber-idealist and author o the \eb 1.0 Internet
utopia 1ev Rvte. for 1be êr covov,, said:
Imagine Mozart beore the technology o the piano. Imagine
Van Gogh beore the technology o aordable oil paints.
Imagine litchcock beore the technology o ilm. \e hae a
moral obligation to deelop technology.
2

But where Kelly sees a moral obligation to aereto technology, we should actually
hae-i we really care about Mozart, Van Gogh and litchcock-a moral
obligation to qve.tiov the deelopment o technology.
1he consequences o \eb 2.0 are inherently dangerous or the itality o
culture and the arts. Its empowering promises play upon that legacy o the
60s-the creeping narcissism that Christopher Lasch described so presciently,
with its obsessie ocus on the realization o the sel.
3

Another word or narcissism is personalization.` \eb 2.0 technology
personalizes culture so that it relects ourseles rather than the world around us.
Blogs personalize media content so that all we read are our own thoughts.

1
KARL MARX & lRILDRICl LNGLLS, 1lL GLRMAN IDLOLOG\ ,1845,, text aailable at
Marxist Internet Archie,
"''4566777/0&%8,9'9/-%:6&%+",;#60&%867-%<96=>?@6:#%0&.A,)#-B-:C6+"D=&/"'0.
2
ee Dan lrost, Meetivg of Miva. iv Movtere,, SAN lRANCISCO ClRONICLL, leb. 2, 2005,
"''4566&%',+B#9/92:&'#/+-06EDD@ADEAEF6G*9,.#996=FHI=H=EJ=J),:,'&BA7-%B)A#)7&%)A
G*%'C.9<CA%-G#%'A2,9+"#BB6E ,quoting Kein Kelly,.
3
ee ClRIS1OPlLR LASCl, 1lL CUL1URL Ol NARCISSISM: AMLRICAN LIlL IN AN AGL Ol
DIMINISlING LXPLC1A1IONS ,198,.

Online stores personalize our preerences, thus eeding back to us our own
taste. Google personalizes searches so that all we see are adertisements or
products and serices we already use.
Instead o Mozart, Van Gogh, or litchcock, all we get with the \eb 2.0
reolution is more o ourseles.
Still, the idea o ineitable technological progress has become so seductie that
it has been transormed into laws.` In Silicon Valley, the most quoted o these
laws, Moore`s Law, states that the number o transistors on a chip doubles eery
two years, thus doubling the memory capacity o the personal computer eery
two years. On one leel, o course, Moore`s Law is real and it has drien the
Silicon Valley economy. But there is an unspoken ethical dimension to Moore`s
Law. It presumes that each adance in technology is accompanied by an
equialent improement in the condition o man.
But as Max \eber so conincingly demonstrated, the only really reliable law o
history is the Law o Unintended Consequences.
\e know what happened the irst time around, in the dot.com boom o the
90s. At irst there was irrational exuberance. 1hen the dot.com bubble
popped, some people lost a lot o money and a lot o people lost some money.
But nothing really changed. Big media remained big media and almost
eerything else-with the exception o Amazon.com and eBay-withered away.
1his time, howeer, the consequences o the digital media reolution are much
more proound. Apple, Google and Craigslist really are reolutionizing our
cultural habits, our ways o entertaining ourseles, our ways o deining who we
are. 1raditional elitist` media is being destroyed by digital technologies.
Newspapers are in ree-all. Network teleision, the modern equialent o the
dinosaur, is being shaken by 1iVo`s oernight annihilation o the 30-second
commercial and competition rom Internet-deliered teleision and amateur
ideo. 1he iPod is undermining the multibillion dollar music industry.
Meanwhile, digital piracy, enabled by Silicon Valley hardware and justiied by
intellectual property communists such as Larry Lessig, is draining reenue rom
established artists, moie studios, newspapers, record labels, and song writers.
Is this a bad thing 1he purpose o our media and culture industries-beyond
the obious need to make money and entertain people-is to discoer, nurture,
and reward elite talent. Our traditional mainstream media has done this with
great success oer the last century. Consider Alred litchcock`s masterpiece,
1ertigo and a couple o other brilliantly talented works o the same name: the
1999 book by Anglo-German writer \.G. Sebald, and the 2004 song by Irish
rock star Bono. litchcock could neer hae made his expensie, complex
moies outside the lollywood studio system. Bono would neer hae become
Bono without the music industry`s super-heayweight marketing muscle. And

\.G. Sebald, the most obscure o this trinity o talent, would hae remained an
unknown uniersity proessor, had a high-end publishing house not had the
good taste to discoer and distribute his work. Llite artists and an elite media
industry are symbiotic. I you democratize media, then you end up
democratizing talent. 1he unintended consequence o all this democratization,
to misquote \eb 2.0 apologist 1homas lriedman, is cultural lattening.`
4
No
more litchcocks, Bonos, or Sebalds. Just the lat noise o opinion-Socrates`s
nightmare.
\hile Socrates correctly gae warning about the dangers o a society inatuated
by opinion in Plato`s Republic, more modern dystopian writers-luxley,
Bradbury, and Orwell-got the \eb 2.0 uture exactly wrong. Much has been
made, or example, o the associations between the all-seeing, all-knowing
qualities o Google`s search engine and the Big Brother in Nineteen Lighty-
lour.
5
But Orwell`s ear was the disappearance o the indiidual right to sel-
expression. 1hus \inston Smith`s great act o rebellion in Nineteen Light-
lour was his decision to pick up a rusty pen and express his own thoughts:
1he thing that he was about to do was open a diary. 1his was
not illegal, but i detected it was reasonably certain that it
would be punished by death. \inston itted a nib into the
penholder and sucked it to get the grease o.. le dipped the
pen into the ink and then altered or just a second. A tremor
had gone through his bowels. 1o mark the paper was the
decisie act.
6

In the \eb 2.0 world, howeer, the nightmare is not the scarcity, but the oer-
abundance o authors. Since eeryone will use digital media to express
themseles, the only decisie act will be to not mark the paper. Not writing as
rebellion sounds bizarre-like a piece o iction authored by lranz Kaka. But
one o the unintended consequences o the \eb 2.0 uture may well be that
eeryone is an author, while there is no longer any audience.
Speaking o Kaka, on the back coer o the January 2006 issue o Poet. c
!riter. magazine, there is a seductie \eb 2.0 style adertisement which reads:
Kaka toiled in obscurity and died penniless. I only he`d had a
website . .

4
ee 1lOMAS lRILDMAN, 1lL \ORLD IS lLA1: A BRILl lIS1OR\ Ol 1lL 1\LN1\-lIRS1
CLN1UR\ ,2005,.
5
ee GLORGL OR\LLL, NINL1LLN LIGl1\-lOUR ,1949,.
6
a. at 6.

Presumably, i Kaka had had a website, it would be located at kaka.com-
which is today an address owned by a mad let-wing blog called 1be i.cvit
Reort. 1he ront page o this site quotes some words written by Kaka in his
diary:
I hae no memory or things I hae learned, nor things I hae
read, nor things experienced or heard, neither or people nor
eents, I eel that I hae experienced nothing, learned nothing,
that I actually know less than the aerage schoolboy, and that
what I do know is supericial, and that eery second question is
beyond me. I am incapable o thinking deliberately, my
thoughts run into a wall. I can grasp the essence o things in
isolation, but I am quite incapable o coherent, unbroken
thinking. I can`t een tell a story properly, in act, I can
scarcely talk .

One o the unintended consequences o the \eb 2.0 moement may well be
that we all, collectiely, into the amnesia that Kaka describes. \ithout an elite
mainstream media, we will lose our memory or things learnt, read, experienced,
or heard. 1he cultural consequences o this are dire, requiring the authoritatie
oice o at least an Allan Bloom,
8
i not an Oswald Spengler.
9
But here in
Silicon Valley, on the brink o the \eb 2.0 epoch, there no longer are any
Blooms or Spenglers. All we hae is the great seduction o citizen media,
democratized content and authentic online communities. And blogs, o course.
Millions and millions o blogs.

ee 1be i.cvit Reort,

"''45667#G/&%+",;#/-%:67#G6EDD>DEE@D=@F=I6"''4566777/<&2<&/+-06.
8
ee ALLAN BLOOM, 1lL CLOSING Ol 1lL AMLRICAN MIND ,198,.
9
ee OS\ALD SPLNGLLR, 1lL DLCLINL Ol 1lL \LS1 ,1918,.

The Case for Internet Optimism,
Part 1: Saving the Net
from Its Detractors
By Adam Thierer
*

Introduction: Two Schools
of Internet Pessimism
Sureying the preailing mood surrounding cyberlaw and Internet policy circa
2010, one is struck by the oerwhelming sense o pessimism regarding the long-
term prospects or a better uture. Internet pessimism,` howeer, comes in
two ery distinct laors:
1. K#' (<#4',+9M N#99,0,9',+ &G-*' '"# O.'#%.#' O04%-;,.: '"# P-' -2
Q&.<,.): 1he irst ariant o Internet pessimism is rooted in general
skepticism about the supposed beneits o cyberspace, digital technologies,
and inormation abundance. 1he proponents o this pessimistic iew oten
wax nostalgic about some supposed good ol days` when lie was much
better ,although they can`t seem to agree when those were,. At a minimum,
they want us to slow down and think twice about lie in the Inormation
Age and how it`s personally aecting each o us. Occasionally, howeer,
this pessimism borders on neo-Ludditism, with some proponents
recommending steps to curtail what they eel is the destructie impact o
the Net or digital technologies on culture or the economy. Leading
proponents o this ariant o Internet pessimism include: Neil Postman
,1ecbvoot,: 1be vrrevaer of Cvttvre to 1ecbvotog,,, Andrew Keen, ,1be Cvtt of
tbe .vatevr: or 1oaa,`. vtervet i. Kittivg ovr Cvttvre,, Lee Siegel, ,.gaiv.t tbe
Macbive: eivg vvav iv tbe .ge of tbe tectrovic Mob,, Mark lelprin, ,Digitat
arbari.v, and, to a lesser degree, Jaron Lanier ,Yov .re ôt a Caaget, and
Nicholas Carr ,1be ig ritcb and 1be battor.,.
2. K#' P-;#%9M N#99,0,9',+ &G-*' '"# R*'*%# -2 S4#..#99: A dierent
type o Internet pessimism is on display in the work o many leading
cyberlaw scholars today. Noted academics such as Lawrence Lessig, ,Coae
ava Otber ar. of C,ber.ace,, Jonathan Zittrain ,1be vtvre of tbe vtervet-
.va or to to t,, and 1im \u ,1be Ma.ter ritcb: 1be Ri.e ava att of
vforvatiov vire.,, embrace the Internet and digital technologies, but argue
that they are dying` due to a lack o suicient care or collectie oersight.

Adam 1hierer is a senior research ellow at the Q#%+&'*9 T#.'#% &' $#-%:# Q&9-.
U.,;#%9,'C where he works with the !#+".-B-:C N-B,+C N%-:%&0.

In particular, they ear that the open` Internet and generatie` digital
systems are giing way to closed, proprietary systems, typically run by
illainous corporations out to erect walled gardens and quash our digital
liberties. 1hus, they are pessimistic about the long-term surial o the
Internet that we currently know and loe.
Despite their dierent concerns, two things unite these two schools o techno-
pessimism. lirst, there is an elitist air to their pronouncements, a eritable the
rest o you just don`t get it` attitude perades much o their work. In the case
o the Net skeptics, it`s the supposed decline o culture, tradition, and economy
that the rest o us are supposedly blind to, but which they see perectly-and
know how to rectiy. lor the Net Loers, by contrast, we see this attitude on
display when they imply that a Digital Dark Age o Closed Systems is unolding
since nearious schemers in high-tech corporate America are out to suocate
Internet innoation and digital reedom more generally. 1he Net Loers
apparently see this plot unolding, but paint the rest o us out to be robotic
sheep being led to the cyber-slaughter: \e are unwittingly using serices ,AOL
in the old days, lacebook today, or deices ,the iPhone and iPad, that play right
into the hands o the ery corporate schemers determined to trap us in high and
tight walled gardens.
Unsurprisingly, this elitist attitude leads to the second belie uniting these two
ariants o Net pessimism: oveove or .ovetbivg must interene to set us on a
better course or protect those things that they regard as sacred. 1he critics
either ancy themseles as the philosopher kings who can set things back on a
better course, or imagine that such creatures exist in goernment today and can
be tapped to sae us rom our impending digital doom-whateer it may be.
Dynamism vs. the Stasis Mentality
In both cases, these two schools o Internet pessimism hae ,a, oer-stated the
seerity o the respectie problems theyè identiied and ,b, ailed to appreciate
the beneits o erotvtiovar, a,vavi.v. I borrow the term dynamism` rom
Virginia Postrel, who contrasted the conlicting worldiews o a,vavi.v and
.ta.i. so eloquently in her 1998 book, 1be vtvre ava t. vevie.. Postrel argued
that:
1he uture we ace at the dawn o the twenty-irst century is,
like all utures let to themseles, emergent, complex
messiness.` Its messiness` lies not in disorder, but in an order
that is unpredictable, spontaneous, and eer shiting, a pattern
created by millions o uncoordinated, independent decisions.
1

1
VIRGINIA POS1RLL, 1lL lU1URL AND I1S LNLMILS, at x ,1998,.

|1|hese actions shape a uture no one can see, a uture that is dynamic and
inherently unstable,` Postrel noted.
2
But that inherent instability and the
uncomortable realization that the uture is, by its ery nature, unknowable,
leads to exactly the sort o anxieties we see on display in the works o botb
arieties o Internet pessimists today. Postrel made the case or embracing
dynamism as ollows:
low we eel about the eoling uture tells us who we are as
indiiduals and as a ciilization: Do we search or .ta.i.-a
regulated, engineered world Or do we embrace a,vavi.v-a
world o constant creation, discoery, and competition Do we
alue stability and control, or eolution and learning Do we
declare with |1im| Appelo that we`re scared o the uture`
and join |Judith| Adams in decrying technology as a killing
thing` Or do we see technology as an expression o human
creatiity and the uture as initing Do we think that progress
requires a central blueprint, or do we see it as a decentralized,
eolutionary process Do we consider mistakes permanent
disasters, or the correctable by-products o experimentation
Do we crae predictability, or relish surprise 1hese two poles,
stasis and dynamism, increasingly deine our political,
intellectual, and cultural landscape. 1he central question o our
time is what to do about the uture. And that question creates
a deep diide.
3

Indeed it does, and that diide is growing deeper as the two schools o Internet
pessimism-unwittingly, o course-work together to concoct a lugubrious
narratie o impending techno-apocalypse. It makes little dierence whether
the two schools disagree on the root cause,s, o all our problems, in the end, it`s
their common call or a more regulated, engineered world` that makes them
both embrace the same stasis mindset. Again, the air o elitism rears its ugly
head, Postrel notes:
Stasist social criticism. brings up the speciics o lie only to
sneer at or bash them. Critics assume that readers will share
their attitudes and will see contemporary lie as a problem
demanding immediate action by the powerul and wise. 1his
relentlessly hostile iew o how we lie, and how we may come
to lie, is distorted and dangerous. It oeralues the tastes o
an articulate elite, compares the real world o trade-os to
antasies o utopia, omits important details and connections,

2
a.
3
a. at xi.

and conuses temporary growing pains with permanent
catastrophes. It demoralizes and dealues the creatie minds
on whom our uture depends. And it encourages the coercie
use o political power to wipe out choice, orbid
experimentation, short-circuit eedback, and trammel
progress.
4

In this essay, I ocus on the irst ariant o Internet pessimism ,the Net
skeptics, and discuss their clash with Internet optimists. I orm this narratie
using the words and themes deeloped in arious books published by Net
optimists and pessimists in recent years. I make the dynamist case or what I
call pragmatic optimism` in that I argue that the Internet and digital
technologies are reshaping our culture, economy and society-in most ways or
the better ,as the optimists argue,, but not without some serious heartburn
along the way ,as the pessimists claim,. My bottom line comes down to a
simple cost-beneit calculus: !ere re reatt, better off iv tbe .carcit, era rbev re rere
cottectiret, .vfferivg frov ivforvatiov orert, Generally speaking, I`ll take inormation
abundance oer inormation poerty any day! But we should not underestimate
or belittle the disruptie impacts associated with the Inormation Reolution.
\e need to ind ways to better cope with turbulent change in a dynamist
ashion instead o embracing the stasis notion that we can roll back the clock on
progress or recapture the good ol days`-which actually weren`t all that good.
In another essay in this book, I address the second ariant o Internet
pessimism ,the Net loers, and argue that reports o the Internet`s death hae
been greatly exaggerated. Although the Net loers will likely recoil at the
suggestion that they are not dynamists, closer examination reeals their attitudes
and recommendations to be deeply stasist. 1hey ret about a cyber-uture in
which the Internet might not as closely resemble its opening epoch. \orse yet,
many o them agree with what Lawrence Lessig said in his seminal-by highly
pessimistic-1999 book, Coae ava Otber ar. of C,ber.ace, that we hae eery
reason to beliee that cyberspace, let to itsel, will not ulill the promise o
reedom. Let to itsel, cyberspace will become a perect tool o control.`
5

Lessig and his intellectual disciples-especially Zittrain and \u-hae
continued to orecast a gloomy digital uture unless .ovetbivg i. aove to address
the Great Digital Closing we are supposedly experiencing. I will argue that,
while many o us share their appreciation o the Internet`s current nature and its
early history, their embrace o the stasis mentality is unortunate since it
orecloses the spontaneous eolution o cyberspace and inites goernment

4
a. at xii-xiii.
5
LA\RLNCL LLSSIG, CODL AND O1lLR LA\S Ol C\BLRSPACL 5-6 ,1999,.
But irst let us turn to the Net skeptics, who don`t share such an appreciation o
the potential beneits o cyberspace. Rather, their pessimism cuts deep and is
rooted in oert hostility to all things digital.
The Familiar Cycle of
Technological Revolutions
1he impact o technological change on culture, learning, and morality has long
been the subject o intense debate, and eery technological reolution brings
out a resh crop o both pessimists and Pollyannas. Indeed, a amiliar cycle has
repeat itsel throughout history wheneer new modes o production ,rom
mechanized agriculture to assembly-line production,, means o transportation
,water, rail, road, or air,, energy production processes ,steam, electric, nuclear,,
medical breakthroughs ,accination, surgery, cloning,, or communications
techniques ,telegraph, telephone, radio, teleision, hae emerged.
1he cycle goes something like this: A new technology appears. 1hose who ear
the sweeping changes brought about by this technology see a sky that is about
to all. 1hese techno-pessimists` predict the death o the old order ,which,
ironically, is oten a preious generation`s hotly-debated technology that others
wanted slowed or stopped,. Lmbracing this new technology, they ear, will
result in the oerthrow o traditions, belies, alues, institutions, business
models, and much else they hold sacred. As Dennis Baron, author o ! #$%%$&
'$()*+, has noted, the shock o the new oten brings out critics eager to warn us
away.`
6

1he Pollyannas, by contrast, look out at the unolding landscape and see mostly
rainbows in the air. 1heirs is a rose-colored world in which the technological
reolution ,- ./-& improes the general lot o mankind. I something must gie,
then the old ways be damned! lor such techno-optimists,` progress means
some norms and institutions must adapt-perhaps een disappear-or society
to continue its march orward.
Our current Inormation Reolution is no dierent. It too has its share o
techno-pessimists and techno-optimists who continue to debate the impact o
technology on human existence.
Indeed, beore most o us had een heard o

6
DLNNIS BARON, A BL11LR PLNCIL 12 ,2009,.
\illiam Powers, author o lamlet`s BlackBerry: A Practical Philosophy or Building a Good

Lie in the Digital Age, reminds us that:
wheneer new deices hae emerged, theyè presented the kinds o
challenges we ace today-busyness, inormation oerload, that sense o lie
being out o control. 1hese challenges were as real two millennia ago as they
are today, and throughout history, people hae been grappling with them and
looking or creatie ways to manage lie in the crowd.

the Internet, people were already ighting about it-or at least debating what the
rise o the Inormation Age meant or our culture, society, and economy.
Web 1.0 Fight: Postman vs. Negroponte
In his 1992 anti-technology maniesto 1ecbvoot,: 1be vrrevaer of Cvttvre to
1ecbvotog,, the late social critic Neil Postman greeted the unolding Inormation
Age with a combination o skepticism and scorn.
8
Indeed, Postman`s book was
a near-perect articulation o the techno-pessimist`s creed. Ìnormation has
become a orm o garbage,` he claimed, not only incapable o answering the
most undamental human questions but barely useul in proiding coherent
direction to the solution o een mundane problems.`
9
I let unchecked,
Postman argued, America`s new technopoly-`the submission o all orms o
cultural lie to the soereignty o technique and technology`-would destroy
the ital sources o our humanity` and lead to a culture without a moral
oundation` by undermining certain mental processes and social relations that
make human lie worth liing.`
10

Postman opened his polemic with the well-known allegorical tale ound in
Plato`s Pbaearv. about the dangers o the written word. Postman reminded us
how King 1hamus responded to the god 1heuth, who boasted that his
inention o writing would improe the wisdom and memory o the masses
relatie to the oral tradition o learning. King 1hamus shot back, the
discoerer o an art is not the best judge o the good or harm which will accrue
to those who practice it.` King 1hamus then passed judgment himsel about
the impact o writing on society, saying he eared that the people will receie a
quantity o inormation without proper instruction, and in consequence be
thought ery knowledgeable when they are or the most part quite ignorant.`
And so Postman-ancying himsel a modern 1hamus-cast judgment on
today`s comparable technological adances and those who would gloriy them:

being out o control. 1hese challenges were as real two millennia ago as they
are today, and throughout history, people hae been grappling with them and
looking or creatie ways to manage lie in the crowd.
\ILLIAM PO\LRS, lAMLL1`S BLACKBLRR\: A PRAC1ICAL PlILOSOPl\ lOR BUILDING A
GOOD LIlL IN 1lL DIGI1AL AGL 5 ,2010,. Similarly, Baron notes that rom the irst days
o writing to the present, each time a new communication technology appeared, people had
to learn all oer again how to use it, how to respond to it, how to trust the documents it
produced.` DLNNIS BARON, A BL11LR PLNCIL 5 ,2009,.
8
NLIL POS1MAN, 1LClNOPOL\: 1lL SURRLNDLR Ol CUL1URL 1O 1LClNOLOG\ ,1992,.
9
a. at 69-0.
10
a. at 52, xii.

we are currently surrounded by throngs o zealous 1heuths,
one-eyed prophets who see only what new technologies can do
and are incapable o imagining what they will vvao. \e might
call such people 1echnophiles. 1hey gaze on technology as a
loer does on his beloed, seeing it as without blemish and
entertaining no apprehension or the uture. 1hey are thereore
dangerous and to be approached cautiously. . I one is to err,
it is better to err on the side o 1hamusian skepticism.
11

Nicholas Negroponte begged to dier. An unapologetic 1heuthian technophile,
the ormer director o the MI1 Media Lab responded on behal o the techno-
optimists in 1995 with his prescient polemic, eivg Digitat.
12
It was a paean to
the Inormation Age, or which he sered as one o the irst high prophets-
with !irea magazine`s back page sering as his pulpit during the many years he
sered as a regular columnist.
Appropriately enough, the epilogue o Negroponte`s eivg Digitat was entitled
An Age o Optimism` and, like the rest o the book, it stood in stark contrast
to Postman`s pessimistic worldiew. Although Negroponte conceded that
technology indeed had a dark side` in that it could destroy much o the old
order, he belieed that destruction was both ineitable and not cause or much
concern. Like a orce o nature, the digital age cannot be denied or stopped,`
he insisted, and we must learn to appreciate the ways digital technology can be
a natural orce drawing people into greater world harmony.`
13
,1his sort o
techno-determinism is a theme ound in many o the Internet optimist works
that ollowed Negroponte.,
1o Postman`s persistent claim that America`s technopoly lacked a moral
compass, Negroponte again conceded the point but took the glass-is-hal-ull
iew: Computers are not moral, they cannot resole complex issues like the
rights to lie and to death. But being digital, neertheless, does gie much cause
or optimism.`
14
lis deense o the digital age rested on the our ery
powerul qualities that will result in its ultimate triumph: decentralizing,
globalizing, harmonizing, and empowering.`
15
Gazing into his techno-crystal
ball in 1995, Negroponte orecast the ways in which those qualities would
reolutionize society:

11
a. at 5.
12
NIClOLAS NLGROPON1L, BLING DIGI1AL ,1995,.
13
a. at 229, 230.
14
a. at 228-9.
15
a. at 229.

1he access, the mobility, and the ability to eect change are
what will make the uture so dierent rom the present. 1he
inormation superhighway may be mostly hype today, but it is
an understatement about tomorrow. It will exist beyond
people`s wildest predictions. As children appropriate a global
inormation resource, and as they discoer that only adults
need learner`s permits, we are bound to ind new hope and
dignity in places where ery little existed beore.
16

In many ways, that`s the world we occupy today: one o unprecedented media
abundance and unlimited communications and connectiity opportunities.
But the great debate about the impact o digitization and inormation
abundance did not end with Postman and Negroponte. 1heirs was but Act I in
a drama that continues to unold, and grows more heated and complex with
each new character on the stage. 1his conlict between stability and progress,
security and prosperity, dynamism and stasis, has led to the creation o a major
political ault line in American politics,` argues Robert D. Atkinson: On one
side are those who welcome the uture and look at the New Lconomy as largely
positie. On the other are those who resist change and see only the risks o new
technologies and the New Lconomy.`
1
Atkinson expands on this theme in
another essay in this collection.
18

Web War II
1he disciples o Postman and Negroponte are a colorul, dierse lot. 1he
players in Act II o this drama occupy many dierse proessions: journalists,
technologists, business consultants, sociologists, economists, lawyers, etc. 1he
two camps disagree with each other een more ehemently and ocierously
about the impact o the Internet and digital technologies than Postman and
Negroponte did.
In Lxhibit 1, I hae listed the Internet optimists and pessimists alongside their
key works. 1his ery binary treatment obiously cannot do justice to the
arying shades o optimism or pessimism in in each, but is nonetheless helpul.

16
a. at 231.
1
ROBLR1 D. A1KINSON, 1lL PAS1 AND lU1URL Ol AMLRICA`S LCONOM\ 201 ,2004,. As a
result,` he says, a political diide is emerging between preserationists who want to hold
onto the past and modernizers who recognize that new times require new means.`
18
Robert D. Atkinson, !bo`. !bo iv vtervet Potitic.: . 1aovov, of vforvatiov 1ecbvotog, Potic,
c Potitic., inra at 162.
!"#$%$& (
)#*+&#$,- )*.#-/0#$1*2
3 4)#* 5-&*6-*& 70&$8$2&29:
)#,8+2$,- )*.#-/0#/%*2
3 4)#* 5-&*6-*& ;*22$8$2&29:
Nicholas Negroponte, <*$-= >$=$&,1
,1995,
Kein Kelly, 7+& /? @/-&6/1A )#* B*C
<$/1/=D /? E,.#$-*2F G/.$,1 GD2&*82F
,-H &#* !./-/8$. I/61H ,1995,
Virginia Postrel, )#* J+&+6* ,-H
5&2 !-*8$*2 ,1998,
James Surowiecki, )#* I$2H/8 /?
@6/CH2 ,2004,
Chris Anderson, )#* K/-= ),$1A I#D &#*
J+&+6* /? <+2$-*22 $2 G*11$-= K*22 /?
E/6* ,2006,
Steen Johnson, !L*6D&#$-= <,H $2 M//H
J/6 N/+ ,2006,
Glenn Reynolds, O- O68D /? >,L$H2A
P/C E,6Q*&2 ,-H )*.#-/1/=D
!80/C*6 76H$-,6D ;*/01* &/ <*,& <$=
E*H$,F <$= M/L*6-8*-&F ,-H 7&#*6
M/1$,&#2 ,2006,
\ochai Benkler, )#* I*,1&# /?
B*&C/6Q2A P/C G/.$,1 ;6/H+.&$/-
)6,-2?/682 E,6Q*&2 ,-H J6**H/8
,2006,
Clay Shirky, P*6* @/8*2 !L*6D%/HDA
)#* ;/C*6 /? 76=,-$R$-= C$&#/+&
76=,-$R,&$/-2 ,2008,
Don 1apscott & Anthony D. \illiams,
I$Q$-/8$.2A P/C E,22 @/11,%/6,&$/-
@#,-=*2 !L*6D&#$-= ,2008,
Neil Postman, )*.#-/0/1DA )#*
G+66*-H*6 /? @+1&+6* &/
)*.#-/1/=D ,1993,
Sen Birkerts, )#* M+&*-%*6=
!1*=$*2A )#* J,&* /? S*,H$-=
$- ,- !1*.&6/-$. O=* ,1994,
Cliord Stoll, P$=#T)*.#
P*6*&$.A S*?1*.&$/-2 /? ,
@/80+&*6 @/-&6,6$,- ,1999,
Cass Sunstein, S*0+%1$.U./8
,2001,
1odd Gitlin, E*H$, V-1$8$&*HA
P/C &#* )/68*-& /? 58,=*2
,-H G/+-H2 7L*6C#*182 7+6
K$L*2 ,2002,
1odd Oppenheimer, )#*
J1$.Q*6$-= E$-HA G,L$-=
!H+.,&$/- ?6/8 &#* J,12*
;6/8$2* /? )*.#-/1/=D ,2003,
Andrew Keen, )#* @+1& /? &#*
O8,&*+6A P/C )/H,DW2
5-&*6-*& $2 X$11$-= /+6 @+1&+6*
,200,
Stee 1albott, >*L$.*2 /? &#*
G/+1A <,&&1$-= ?/6 7+6 G*1L*2 $-
,- O=* /? E,.#$-*2 ,200,
"#$% &'(() )#* <$= GC$&.#A
S*C$6$-= &#* I/61HF ?6/8
!H$2/- &/ M//=1* ,2008,

!"#$%$& ( @/-&$-+*H
)#*+&#$,- )*.#-/0#$1*2
3 4)#* 5-&*6-*& 70&$8$2&29:
)#,8+2$,- )*.#-/0#/%*2
3 4)#* 5-&*6-*& ;*22$8$2&29:
Je lowe, @6/CH2/+6.$-=A I#D
&#* ;/C*6 /? &#* @6/CH 52
>6$L$-= &#* J+&+6* /? <+2$-*22
,2008,
1yler Cowen, @6*,&* N/+6 7C-
!./-/8DA )#* ;,&# &/
;6/20*6$&D $- , >$2/6H*6*H I/61H
,2009,
Dennis Baron, O <*&&*6 ;*-.$1A
S*,H*62F I6$&*62F ,-H &#* >$=$&,1
S*L/1+&$/- ,2009,
Je Jaris, I#,& I/+1H M//=1*
>/Y ,2009,
Clay Shirky, @/=-$&$L* G+601+2A
@6*,&$L$&D ,-H M*-*6/2$&D $- ,
@/--*.&*H O=* ,2010,
Nick Bilton, 5 K$L* $- &#* J+&+6*
Z P*6*W2 P/C 5& I/6Q2 ,2010,
*+,#- *+../) I#,& )*.#-/1/=D
I,-&2 ,2010,
Lee Siegel, O=,$-2& &#* E,.#$-*A <*$-=
P+8,- $- &#* O=* /? &#* !1*.&6/-$.
E/% ,2008,
Mark Bauerlein, )#* >+8%*2&
M*-*6,&$/-A P/C &#* >$=$&,1 O=*
G&+0*?$*2 N/+-= O8*6$.,-2 ,-H
[*/0,6H$R*2 7+6 J+&+6* ,2008,
Mark lelprin, >$=$&,1 <,6%,6$28A O
I6$&*6W2 E,-$?*2&/ ,2009,
Maggie Jackson, >$2&6,.&*HA )#*
!6/2$/- /? O&&*-&$/- ,-H &#* @/8$-=
>,6Q O=* ,2009,
John lreeman, )#* )D6,--D /? !TE,$1A
)#* J/+6T)#/+2,-HTN*,6 [/+6-*D &/
N/+6 5-%/" ,2009,
Jaron Lanier, N/+ O6* B/& , M,H=*&
,2010,
Nick Carr, )#* G#,11/C2A I#,& &#*
5-&*6-*& 52 >/$-= &/ 7+6 <6,$-2 ,2010,
\illiam Powers, P,81*&W2 <1,.Q<*66DA
O ;6,.&$.,1 ;#$1/2/0#D ?/6 <+$1H$-= ,
M//H K$?* $- &#* >$=$&,1 O=* ,2010,

In Lxhibit 2, I hae sketched out the major lines o disagreement between these
two camps and diided those disagreements into ,1, @+1&+6,1 \ G/.$,1 %*1$*?2
s. ,2, !./-/8$. \ <+2$-*22 %*1$*?2.

!"#$%$& ]
70&$8$2&2 ;*22$8$2&2
@+1&+6,1 \ G/.$,1 %*1$*?2
Net is 0,6&$.$0,&/6D Net is 0/1,6$R$-=
Net acilitates 0*62/-,1$R,&$/-
,welcome o Daily Me`
that digital tech allows,
Net acilitates ?6,=8*-&,&$/-
,ear o the Daily Me`,
a =1/%,1 L$11,=*` %,1Q,-$R,&$/- and
ears o 8/% 6+1*`
#*&*6/=*-*$&D , encourages diersity
o thought and expression
#/8/=*-*$&D , Net
leads to close-mindedness
allows 2*1?T,.&+,1$R,&$/- H$8$-$2#*2 0*62/-#//H
Net a tool o 1$%*6,&$/-
Z *80/C*68*-&
Net a tool o requent
8$2+2* Z ,%+2*
Net can help *H+.,&* the masses H+8%2 H/C- the masses
anonymous communication
encourages ibrant debate -
whistleblowing ,a net good,
anonymity debases culture &
leads to lack o accountability
welcome $-?/68,&$/- ,%+-H,-.*,
beliee it will create new
opportunities or learning
concern about $-?/68,&$/- /L*61/,H,
esp. impact on learning & reading
!./-/8$. \ <+2$-*22 %*1$*?2
%*-*?$&2 /? 4J6**9 ,increasing
importance o =$?& *./-/8D`,
./2&2 /? 4J6**9 ,ree` ~ threat to
quality & business models,
8,22 ./11,%/6,&$/- is
generally more important
$-H$L$H+,1 *??/6& is
generally more important
embrace o ,8,&*+6` creatiity superiority o 406/?*22$/-,1$28`
stress importance o /0*-
2D2&*82` o production
stress importance o 06/06$*&,6D`
models o production
wiki` model ~ C$2H/8 /? .6/CH2^
beneits o .6/CH2/+6.$-=
wiki` model ~ 2&+0$H$&D /? .6/CH2^
collectie intelligence is oxymoron, -
2#,6*.6/00*6` concern about
exploitation o ree labor

\hen you boil it all down, there are two major points o contention between
the Internet optimists and pessimists:
1. 1he impact o technology on 1*,6-$-= Z .+1&+6* and the role o *"0*6&2
L2U ,8,&*+62 in that process.
2. 1he promise-or perils-o 0*62/-,1$R,&$/-, or both indiiduals and
society.
Lach dispute is discussed in more detail below.
Differences Over Learning,
Culture & Truth
As with 1heuth and 1hamus, today`s optimists and skeptics dier about who is
the best judge o what constitutes progress, authority, and truth` and how
technological change will impact these things.
The Pessimists Critique
Consider the heated debates oer the role o amateur` creations, user-
generation content, and peer-based orms o production. Pessimists tend to
ear the impact o the Net and the rise o what Andrew Keen has called the
cult o the amateur.`
19
1hey worry that proessional` media or more
enlightened oices and iewpoints might be drowned out by a cacophony o
competing-but less compelling or enlightened-oices and iewpoints.
\ithout enorceable scarcity` and protection or the enlightened class,` the
pessimists wonder how high quality` news or high art` will be unded and
disseminated. Some, like Keen, een suggest the need to re-create media
scarcity` to sae culture or proessional content creators.
20

Some o these pessimists clearly think in zero-sum terms: More amateur`
production seems to mean less proessional` content creation will be possible.
lor example, Lee Siegel, author o 01'#-23 34+ 5'$4#-+6 7+#-1 89:'- #- 34+ 01+ ;<
34+ =.+$3(;-#$ 5;>, says that by empowering the masses to hae more o a oice,
unbiased, rational, intelligent, and comprehensie news . will become less

19
ANDRL\ KLLN, 1lL CUL1 Ol 1lL AMA1LUR: lO\ 1ODA\`S IN1LRNL1 IS KILLING OUR
CUL1URL ,200,.
20
Andrew Keen, 0(3 ? &;::+($+6 @+'34 >/ A;9B9>+, AD\LLK, Oct. 15, 200,
#&&0A\\C*%U,6.#$L*U/6=\C*%\]__`_(_a_]bcc]\#&&0A\CCCU,HC**QU./8\,C\8,=,R$-
*\,6&$.1*dH$201,DUe20YL-+d./-&*-&d$Hf(__ghc`]_b. lor a response, see Adam 1hierer,
B4;91432 ;- 0-C(+D *++-) E'(3 F6 B4+ @'-1+(2 ;< 34+ G3'2#2 5+-3'.#3/, 1LClNOLOG\ LIBLRA1ION
lRON1, Oct. 18, 200, #&&0A\\&*.#1$%*6,&$/-U./8\]__a\(_\(`\&#/+=#&2T/-T,-H6*CT
Q**-T0,6&T]T&#*TH,-=*62T/?T&#*T2&,2$2T8*-&,1$&D.

and less aailable.`
21
|G|iing eeryone a oice,` he argues, can also be a way
to keep the most creatie, intelligent, and original oices rom being heard.`
22

1he centrality o \ikipedia, the collaboratie online encyclopedia, to this
discussion seres as a microcosm o the broader debate between the optimists
and the pessimists. Almost eery major optimist and pessimist tract includes a
discussion o \ikipedia, it generally seres as a hero in the works o the ormer
and a illain in the latter. lor the pessimists, \ikipedia marks the decline o
authority, the death o objectiity, and the rise o mobocracy` since it allows
anyone with opposable thumbs and a ith-grade education |to| publish
anything on any topic.`
23
1hey ear that truth` becomes more relatiistic
under models o peer collaboration or crowd-sourced initiaties.
24

1he pessimists also hae ery little good to say about \ou1ube, blogs, social
networks, and almost all user-generated content. 1hey treat them with a
combination o conusion and contempt. |S|el-expression is not the same
thing as imagination,` or art, Siegel argues.
25
Instead, he regards the explosion
o online expression as the narcissistic` bloiation o the masses and argues it
is destroying true culture and knowledge. Lchoing Postman`s assertion that
inormation has become a orm o garbage,` Siegel says that the Under the
inluence o the Internet, knowledge is withering away into inormation.`
26
Our
new age o inormation abundance is not worth celebrating, he says, because
inormation is powerlessness.`
2

Some pessimists argue that all the new inormation and media choices are
largely alse choices that don`t beneit society. lor example, Siegel disputes
what he regards as oerly-romanticized notions o online participation` and
personal democracy.` Keen goes urther reerring to them as the great
seduction.` le says the \eb 2.0 reolution has peddled the promise o

21
LLL SILGLL, AGAINS1 1lL MAClINL: BLING lUMAN IN 1lL AGL Ol 1lL LLLC1RONIC MOB
165 ,2008,. lor a reiew o the book, .ee Adam 1hierer, oo/ Rerier: ee ieget`. Against the
Machine, 1LClNOLOG\ LIBLRA1ION lRON1, Oct. 20, 2008,
"''4566'#+"B,G#%&',-./+-06EDD>6=D6ED6G--<A%#;,#7AB##A9,#:#BVWEV>DVXX9A
&:&,.9'A'"#A0&+",.#.
22
a. at 5.
23
Keen, .vra note 19, at 4.
24
\ikipedia, with its ideo-game like mode o participation, and with its mountains o triial
actoids, o shiting mounds o gossip, o inane personal details, is knowledge in the process
o becoming inormation.` Siegel, .vra note 21, at 152.
25
a. at 52.
26
a. at 152.
2
a. at 148.

bringing more truth to more people . but this is all a smokescreen.`
28
\hat
the \eb 2.0 reolution is really deliering,` he argues, is supericial
obserations o the world around us rather than deep analysis, shrill opinion
rather than considered judgment.`
29

Occasionally, the pessimists resort to some airly immature name-calling tactics
while critiquing Inormation Age culture. It would be one thing i such a
|digital| reolution produced Mozarts, Linsteins, or Raphaels,` says noelist
Mark lelprin, but it doesn`t... It produces mouth-breathing morons in
backward baseball caps and pants that all down, Slurpee-sucking geeks who
seldom see daylight, pretentious and earnest hipsters who want you to wear
bamboo socks so the world won`t end . beer-drinking duuses who pay to
watch noisy cars driing around in a circle or eight hours at a stretch.`
30

Some pessimists also claim that prolierating new media choices are merely
orce-ed commercial propaganda or that digital technologies are spawning
needless consumerism. New technologies unquestionably make purchases
easier and more conenient or consumers. 1o this extent, they do help,` says
the proliic Uniersity o Chicago law proessor Cass Sunstein. But they help
ar less than we usually think, because they accelerate the consumption treadmill
without making lie much better or consumers o most goods.`
31

In Siegel`s opinion, eeryone is just in it or the money. \eb 2.0 is the
brainchild o businessmen,` and the producer public` is really just a totalized
consumerist` society.`
32
Countless unpaid bloggers-in it or the loe o the
conersation and debate-are merely brainwashed sheep whom Siegel argues
just don`t realize the harm they are doing. |1|he bloggers are playing into the
hands o political and inancial orces that want nothing more than to see the
critical, scrutinizing media disappear.`
33
le reseres special scorn or Net
eangelists who beliee that something truly exciting is happening with the new
online conersation. According to Siegel, they are simply in a mad rush to earn
proits or push a erent idealism.`
34

1he pessimists also ear that these new technologies and trends could hae
proound ramiications not just or entertainment culture, but also or the

28
29
a.
30
MARK lLLPRIN, DIGI1AL BARBARISM: A \RI1LR`S MANIlLS1O 5 ,2009,.
31
CASS SUNS1LIN, RLPUBLIC.COM 121 ,2010,.
32
Siegel, .vra note 21, at 128.
33
a. at 141.
34
a. at 25-6.

uture o news and proessional journalism. 1hey worry about the loss o
trusted intermediaries and traditional authorities. lor example, Keen ears that
\ikipedia, is almost single-handedly killing the traditional inormation
business.`
35
1hey also argue that ree culture` isn`t ree at all, it`s oten just
parasitic copying or blatant piracy.
Similarly, Nick Carr and Jaron Lanier worry about the rise o digital
sharecropping,` where a small group o elites make money o the back o ree
labor. 1o Carr, many new \eb 2.0 sites and serices are essentially
agglomerations o the creatie, unpaid contributions o their members. In a
twist on the old agricultural practice o sharecropping, the site owners proide
the digital real estate and tools, let the members do all the work, and then
harest the economic riches.`
36
And in opening his book, Lanier says
Ultimately these words will contribute to the ortunes o those ew who hae
been able to position themseles as lords o the computing clouds.`
3

linally, some pessimists worry deeply about the impact o computers and digital
technologies on learning. 1hey ear these trends will ineitably result in a
general dumbing down` o the masses or een the disappearance o reading,
writing, and other arts. 1ypiying this iew is Mark Bauerlein`s 1be Dvvbe.t
Ceveratiov: or tbe Digitat .ge tvefie. Yovvg .vericav. ava ]eoaraie. Ovr vtvre
,2008,, but similar concerns are on display in the works o Sen Birkerts,
38

Cliord Stoll,
39
1odd Gitlin,
40
and 1odd Oppenheimer.
41

The Optimists Response
1he optimists` response is rooted in the belie that, despite their highly
disruptie nature, the Internet and new digital technologies empower and
enlighten indiiduals and, thereore, generally beneit society.

35
36
NIClOLAS CARR, 1lL BIG S\I1Cl: RL\IRING 1lL \ORLD, lROM LDISON 1O GOOGLL 13-
8 ,2008,.
3
LANILR, \OU ARL NO1 A GADGL1 at 1 ,2010,.
38
SVLN BIRKLR1S, 1lL GU1LNBLRG LLLGILS: 1lL lA1L Ol RLADING IN AN LLLC1RONIC AGL
,1994,.
39
CLIllORD S1OLL, lIGl-1LCl lLRL1IC: RLlLLC1IONS Ol A COMPU1LR CON1RARIAN
,1999,.
40
1ODD GI1LIN, MLDIA UNLIMI1LD: lO\ 1lL 1ORMLN1 Ol IMAGLS AND SOUNDS
OVLR\lLLMS OUR LIVLS ,2002,.
41
1ODD OPPLNlLIMLR, 1lL lLICKLRING MIND: SAVING LDUCA1ION lROM 1lL lALSL
PROMISL Ol 1LClNOLOG\ ,2003,.

1he optimists tend to argue that new modes o production ,especially peer-
based production, will oer an adequate-i not superior-alternatie to
traditional modalities o cultural or artistic production. Despite displacing some
institutions and cultural norms, they claim digital technologies create more
opportunities. 1hey speak o collectie intelligence,`
42
the wisdom o
crowds,`
43
the importance o peer production,
44
and the rise o what uturist
Alin 1oler irst reerred to as prosumers.`
45
1here has been a undamental
shit in the balance o power between consumers and salesmen oer the last
generation and it points in the direction o consumers,` 1yler Cowen argues in
his book, Create Yovr Orv covov,: 1be Patb to Pro.erit, iv a Di.oraerea !orta.
46

1he peer production trend is stressed in works such as 1be !eattb of êtror/.:
or ociat Proavctiov 1rav.forv. Mar/et. ava reeaov, by \ochai Benkler,
4
and
!i/ivovic.: or Ma.. Cottaboratiov Cbavge. rer,tbivg, by Don 1apscott and
Anthony D. \illiams.
48
A new economic democracy is emerging in which we
all hae a lead role,` claim 1apscott and \illiams,
49
because the economics o
production hae changed signiicantly.`
50

Most optimists also argue that new business models will eole to support what
had preiously been proided by proessional content creators or news
proiders. Glenn Reynolds ,.v .rv, of Daria., and Dan Gillmor ,!e tbe Meaia,
reer o the rise o we-dia` ,user-generated content and citizen journalism, that
is an increasingly important part o the modern media landscape. Gillmor, a
ormer av ]o.e Mercvr, êr. columnist, speaks o a modern reolution .
because technology has gien us a communications toolkit that allows anyone to
become a journalist at little cost and, in theory, with global reach. Nothing like
this has eer been remotely possible beore,` he argues.
51
And the optimists
generally don`t spend much time lamenting the obliteration o large media

42
lLNR\ JLNKINS, CONVLRGLNCL CUL1URL: \lLRL OLD AND NL\ MLDIA COLLIDL 4
,2006,.
43
JAMLS SURO\ILCKI, 1lL \ISDOM Ol CRO\DS ,2004,.
44
DON 1APSCO11 & AN1lON\ D. \ILLIAMS, \IKINOMICS: lO\ MASS COLLABORA1ION
ClANGLS LVLR\1lING 1, 6 ,2008,.
45
ALVIN 1OllLLR, 1lL 1lIRD \AVL 265 ,1980,.
46
1\LLR CO\LN, CRLA1L \OUR O\N LCONOM\: 1lL PA1l 1O PROSPLRI1\ IN A DISORDLRLD
\ORLD 11 ,2009,.
4
\OClAI BLNKLLR, 1lL \LAL1l Ol NL1\ORKS: lO\ SOCIAL PRODUC1ION 1RANSlORMS
MARKL1S AND lRLLDOM ,2006,.
48
1apscott & \illiams, .vra note 44, at 15.
49
a. at 15.
50
a. at 68.
51
DAN GILLMOR, \L 1lL MLDIA at xii ,2004,.

institutions, either because they think little o their past perormance or,
alternatiely, beliee that whateer watchdog` role they played can be illed by
others. \e are seeing the emergence o new, decentralized approaches to
ulilling the watchdog unction and to engaging in political debate and
organization,` Benkler claims.
52

Optimists also beliee that the Inormation Age oers real choices and genuine
oices, and they ocierously dispute charges o diminished quality by
prosumers, amateur creators, new media outlets, and citizen journalists.
Moreoer, they do not ear the impact o these new trends and technologies on
learning or culture. Surely the technophobes who romanticize the pencil don`t
want to return us to the low literacy rates that characterized the good old days
o writing with pencils and quills,` Baron asks. Still, a ew critics object to the
new technologies because they enable too many people to join the guild o
writers, and they might paraphrase 1horeau`s objection to the telegraph: these
new computer writers, it may be, hae nothing to say to one another.`
53

linally, in addressing the sharecropper concern raised by Carr and Lanier, the
optimists insist most people aren`t in it or the money. Shirky notes that
lumans intrinsically alue a sense o connectedness,` and much o what they
do in the social media world is a true labor o loe.
54
Amateurs aren`t just pint-
sized proessionals, people are sometimes happy to do things or reasons that
are incompatible with getting paid,` he says.
55
Mostly they do it or loe o
knowledge or a belie in the importance o ree culture,` the optimists claim.
The Debate Over the Promise
or Perilsof Personalization
Optimists and pessimists tend to agree that the Internet and \eb 2.0` is
leading to more personalized` media and inormation experiences. 1hey
disagree ehemently, howeer, on whether this is good or bad. 1hey
particularly disagree on what increased inormation customization means or
participatory democracy and the uture o relations among people o dierse
backgrounds and ideologies. linally, they dier on how serious o a problem
inormation oerload` is or society and indiiduals.

52
Benkler, .vra note 4, at 11.
53
54
CLA\ SlIRK\, COGNI1IVL SURPLUS: CRLA1IVI1\ AND GLNLROSI1\ IN A CONNLC1LD AGL
58-9 ,2010,.
55
a.

The Optimists Case
Let`s take the optimists irst this time.
1he optimists tend to embrace what Nicholas Negroponte irst labeled 1he
Daily Me` ,i.e., hyper-personalized news, culture, and inormation,. In 1995,
Negroponte asked us to:
Imagine a uture in which your interace agent can read eery
newswire and newspaper and catch eery 1V and radio
broadcast on the planet, and then construct a personalized
summary. 1his kind o newspaper is printed in an edition o
one..
Imagine a computer display o news stories with a knob that,
like a olume control, allows you to crank personalization up
or down. \ou could hae many o these controls, including a
slider that moes both literally and politically rom let to right
to modiy stories about public aairs. 1hese controls change
your window onto the news, both in terms o size and its
editorial tone. In the distant uture, interace agents will read,
listen to, and look at each story in its entirety. In the near
uture, the iltering process will happen by using headers, those
bits about bits.
56

1hat uture came about sooner than een Negroponte could hae predicted.
\e all hae a Daily Me` at our disposal today thanks to RSS eeds, lacebook,
Google Alerts, 1witter, email newsletters, instant messaging, and so on. 1hese
tools, among others, can proide tailored, automated search results sered up
instantaneously. 1he optimists argue that this increased tailoring and
personalization o our media experiences empowers heretoore silenced masses.
1his worldiew is typiied by the title o Glenn Reynolds` book: .v .rv, of
Daria.: or Mar/et. ava 1ecbvotog, vorer Oraivar, Peote to eat ig Meaia, ig
Corervvevt ava Otber Cotiatb..
5
1he optimists argue that our participatory
culture` promotes greater cultural heterogeneity and gies eeryone a better
chance to be heard. In a world o media conergence, eery important story
gets told, eery brand gets sold, and eery consumer gets courted across
multiple media platorms,` says lenry Jenkins, author o Covrergevce Cvttvre.
58

56
Negroponte, .vra note 12, at 153-54.
5
GLLNN RL\NOLDS, AN ARM\ Ol DAVIDS: lO\ MARKL1S AND 1LClNOLOG\ LMPO\LR
ORDINAR\ PLOPLL 1O BLA1 BIG MLDIA, BIG GOVLRNMLN1 AND O1lLR GOLIA1lS ,2006,.
58
lLNR\ JLNKINS, CONVLRGLNCL CUL1URL: \lLRL OLD AND NL\ MLDIA COLLIDL 3
,2006,. 1apscott & \illiams, .vra note 44, at 41.

Again, they stress the empowering nature o digital technology as a good in and
o itsel. 1he mass amateurization o publishing undoes the limitations
inherent in haing a small number o traditional press outlets,` Shirky claims.
59

1his leads to greater openness, transparency, exposure to new thinking and
opinions, and a diersity o thought and societal participation. Shirky speaks o
the cognitie surplus` unleashed by these changes and its myriad beneits or
society and culture:
1he harnessing o our cognitie surplus allows people to
behae in increasingly generous, public, and social ways,
relatie to their old status as consumers and couch potatoes.
1he raw material o this change is the ree time aailable to us,
time we can commit to projects that range rom the amusing to
the culturally transormatie. . llexible, cheap, and inclusie
media now oers us opportunities to do all sorts o things we
once didn`t do. In the world o the media,` we were like
children, sitting quietly at the edge o a circle and consuming
whateer the grown-ups in the center o the circle produced.
1hat has gien way to a world in which most orms o
communication, public and priate, are aailable to eeryone in
some orm.
60

Shirky een suggests that 1he world`s cognitie surplus is so large that small
changes can hae huge ramiications in aggregate,` and hae beneicial impacts
on politics, adocacy, and generosity.`
\hen it comes to concerns about inormation oerload,` most optimists see
little reason or concern. 1yler Cowen argues that using search tools like
Google and other inormation gathering and processing technologies actually
lengthen our attention spans in another way, namely by allowing greater
specialization o knowledge:`
61

\e don`t hae to spend as much time looking up arious acts
and we can ocus on the particular areas o interest, i only
because general knowledge is so readily aailable. It`s neer
been easier to wrap yoursel up in a long-term intellectual
project, yet without losing touch with the world around you.

59
CLA\ SlIRK\, lLRL COMLS LVLR\BOD\: 1lL PO\LR Ol ORGANIZING \I1lOU1
ORGANIZA1IONS 65 ,2008,.
60
CLA\ SlIRK\, COGNI1IVL SURPLUS, .vra note 54, at 63.
61
1\LLR CO\LN, CRLA1L \OUR O\N LCONOM\: 1lL PA1l 1O PROSPLRI1\ IN A DISORDLRLD
\ORLD 55 ,2009,.

As or inormation oerload, it is you who chooses how much
stu` you want to experience and how many small bits you
want to put together . . 1he quantity o inormation coming
our way has exploded, but so has the quality o our ilters.
62

Chris Anderson preiously made this point in his book, 1be ovg 1ait.
Anderson deined ilters as the catch-all phrase or recommendations and all
the other tools that help you ind quality in the Long 1ail` and noted that
these technologies and serices sit through a ast array o choices to present
you with the ones that are most right or you.`
63
1he job o ilters is to screen
out |the| noise` or inormation clutter, Anderson says.
64
Cowen argues that the
iltering technologies are getting better at this siting and processing process, bvt
.o too are bvvav., he says. 1he key to this, he argues, is that we are getting better
at ordering` inormation.
On balance, thereore, the optimists argue that personalization beneits our
culture and humanity. Dennis Baron concludes, Lnglish suries, conersation
thries online as well as o, and on balance, digital communications seems to
be enhancing human interaction, not detracting rom it.`
65

The Pessimists Response
1he pessimists argue that all this Pollyannaish talk about a new age o
participatory democracy is bunk. Instead o welcoming increased inormation
and media personalization, they lament it. 1hey ear that 1he Daily Me` that
the optimists laud will lead to homogenization, close-mindedness, an online
echo-chamber, inormation oerload, corporate brainwashing, etc. \orst,
hyper-customization o websites and online technologies will cause extreme
social ragmentation,` polarization,` balkanization,` extremism` and een
the decline o deliberatie democracy.
66

Siegel and Keen are probably the most searing in this critique. 1o Siegel, or
example, the Daily Me` is little more that the creation o a narcissistic
culture` in which exaggeration` and the loudest, most outrageous, or most

62
a.
63
ClRIS ANDLRSON, 1lL LONG 1AIL 108 ,2006,.
64
a. at 115.
65
66
Carr worries that eery little choice moes us close toward such social isolation: Lery time
we subscribe to a blog, add a riend to our social network, categorize an email message as
spam, or een choose a site rom a list o search results, we are making a decision that
deines, in some small way, whom we associate with and what inormation we pay attention
to.` NIClOLAS CARR, 1lL BIG S\I1Cl: RL\IRING 1lL \ORLD, lROM LDISON 1O GOOGLL
160 ,2008,.

extreme oices sway the crowd o oices this way, the cutest, most sel-eacing,
most ridiculous, or most transparently raudulent o oices sway the crowd o
oices that way.`
6
le calls \eb 2.0 democracy`s atal turn` in that, instead o
allowing indiiduals to create their own cultural and commercial choices,` it
has instead created a more potent orm o homogenization.`
68
Keen ears the
rise o a dangerous orm o digital narcissism` and the degeneration o
democracy into the rule o the mob and the rumor mill.`
69

1his echoes concerns irst raised by Cass Sunstein in his 2001 book
Revbtic.cov.
0
In that book, Sunstein reerred to Negroponte`s Daily Me` in
contemptuous terms, saying that the hyper-customization o websites and
online technologies was causing extreme social ragmentation and isolation that
could lead to political extremism. A system o limitless indiidual choices, with
respect to communications, is not necessarily in the interest o citizenship and
sel-goernment,` he wrote.
1
Sunstein was essentially claiming that the
Internet is breeding a dangerous new creature: Anti-Democratic Man.
2

Group polarization is unquestionably occurring on the Internet,` he
proclaimed, and it is weakening what he called the social glue` that binds
society together and proides citizens with a common group identity.`
3
I that
continues unabated, Sunstein argued, the potential result could be nothing short
o the death o deliberatie democracy and the breakdown o the American
system o goernment.
Some o the pessimists, like Keen, go urther and claim that the moral abric o
our society is being unraeled by \eb 2.0. It seduces us into acting on our
most deiant instincts and allows us to succumb to our most destructie ices.
And it is corroding the alues we share as a nation.`
4
Nick Carr summarizes the
iews o the pessimists when he says: it`s clear that two o the hopes most dear
to the Internet optimists-that the \eb will create a more bountiul culture and
that it will promote greater harmony and understanding-should be treated

6
Siegel, .vra note 21, at 9.
68
a. at 6.
69
Keen, .vra note 19, at 54-5.
0
CASS SUNS1LIN, RLPUBLIC.COM ,2001,.
1
a. at 123.
2
ee Adam 1hierer, arivg Devocrac, frov tbe vtervet, RLGULA1ION ,lall 2001, 8-9,
"''4566777/+&'-/-%:64*G96%#:*B&',-.6%#:;E?.H6,.%#;,#7/4)2.
3
Sunstein, .vra, at 1, 89.
4

with skepticism. Cultural impoerishment and social ragmentation seem
equally likely outcomes.`
5

Another common theme in the works o the pessimists is summarized by the
title o Siegel`s book ,.gaiv.t tbe Macbive,. 1hey ear the mechanization o the
soul`
6
or humanity`s surrender` to the machine reolution.`
In opening o
Yov .re ôt a Caaget, Lanier ears that these words will mostly be read by
nonpersons-automatons or numb mobs composed o people who are no
longer acting as indiiduals.`
8
1he trick is not to subject man and nature to
the laws o the machine,` says lelprin, but rather to control the machine
according to the laws and suggestions o nature and human nature. 1o
subscribe to this does not make one a Luddite.`
9

linally, the pessimists are also concerned about the impact o online anonymity
on human conduct and language. 1hey argue anonymity leads to less
accountability or, more simply, just plain bad manners. I our national
conersation is carried out by anonymous, sel-obsessed people unwilling to
reeal their real identities, then,` Keen argues, community denigrates into
anarchy.`
80

So Whos Right?
On balance, the optimists generally hae the better o the argument today. \e
really are better o in an age o inormation abundance than we were in the
scarcity era we just exited. Nonetheless, the pessimists make many air points
that desere to be taken seriously. But they need a more reasonable articulation
o those concerns and a constructie plan or how to moe orward without a
call or extreme reactionary solutions.
A hybrid approach here might be thought o as pragmatic optimism,` which
attempts to rid the optimist paradigm o its kookier, pollyannish thinking while
also taking into account some o the ery legitimate concerns raised by the
pessimists, but rejecting its caustic, neo-Luddite ringe elements and stasis
mentality in the process.

5
Carr, .vra note 36, at 16.
6
lelprin, .vra note 30, at 100.
a. at 9, 100.
8
Lanier, .vra note 3, at 1.
9
80

Thoughts on the Pessimists
lirst and oremost, i they hope to be taken more seriously, Net skeptics need
better spokespersons. Or, they at least need a more moderated, less hysterical
tone when addressing alid concerns raised by technological progress. It`s oten
diicult to take the pessimists seriously when they exude outright hostility to
most orms o technological progress. Most o them deny being high-tech
troglodytes, but the tone o some o their writing, and the thrust o some o
their recommendations, exhibit occasional Luddite tendencies-een i they
don`t always come out and call or extreme measures to counteract dynamism.
Moreoer, the name-calling they sometimes engage in, and their derision or the
digital generation can be just as insulting and immature as the online mob`
they repeatedly castigate in their works. 1oo oten, their criticism deoles into
philosophical snobbery and blatant elitism, as in the works o lelprin, Siegel,
and Keen. Constantly looking down their noses at digital naties and all
amateur` production isn`t going to help them win any conerts or respect or
their positions. Moreoer, one wonders i they hae ingered the right culprit
or ciilization`s supposed decline, since most o the ills they identiy predate
the rise o the Internet.
1he pessimists are oten too quick to proclaim the decline o modern
ciilization by looking only to the baser elements o the blogosphere or the
more caustic oices o cyberspace. 1he Internet is a cultural and intellectual
bazaar where one can ind both the best and the worst o humanity on display
at any gien moment. 1rue, brutishness and barbarism,` as lelprin calls it,
81

can be ound on many cyber-corners, but not att o its corners. And, contrary
to lelprin`s assertion that blogging begins the mad race to the bottom,`
82
one
could just as easily cite countless instances o the healthy, unprecedented
conersations that blogs hae enabled about a dierse array o topics.
1heir claim that the Daily Me` and inormation specialization will lead to a
ariety o ills is also somewhat oerblown. It`s particularly hard to accept
Sunstein and Carr`s claims that increased personalization is breeding
extremism,` anaticism` and radicalization.` A recent study by Matthew
Gentzkow and Jesse M. Shapiro o the Uniersity o Chicago Booth School o
Business lent credibility to this, inding no eidence that the Internet is
becoming more segregated oer time` or leading to increased polarization as
Sunstein and other pessimists ear.
83
Instead, their indings show that the Net

81
82
a. at 42.
83
Matthew Gentzkow & Jesse M. Shapiro, aeotogicat egregatiov Ovtive ava Offtive, ClICAGO
BOO1l \ORKING PAPLR No. 10-19, April 5, 2010,
"''45664&4#%9/99%./+-069-BH64&4#%9/+20Y&G9'%&+'J,)Z=@>>XED.

has encouraged more ideological integration and is actually driing us to
experience new, unanticipated iewpoints.
84

\hile it`s true the Internet has gien some extremists a new soapbox to stand
on and spew their hatred and stupidity, the act is that such oices and
iewpoints hae always existed. 1he dierence today is that the Internet and
digital platorms hae gien us a platorm to counter such societal extremism.
As the old saying goes, the answer to bad speech is more speech-not a
crackdown on the underlying technologies used to coney speech. It should
not be orgotten that, throughout history, most extremist, totalitarian
moements rose to power by taking oer the scarce, centralized media
platorms that existed in their countries. 1he decentralization o media makes
such a take-oer ar less plausible to imagine.
Sometimes the pessimists seem to just be suering rom a bit o old-ogeyism.
Lanier, or example, dismisses most modern culture as retro` and a petty
mashup o preweb culture.`
85
It`s as i culture roze just beore it became
digitally open, and all we can do now is mine the past like salagers picking oer
a garbage dump.`
86
Many pessimists are guilty o such hyper-nostalgia about
those mythical good ol days` when all was supposedly much better. It`s a
common rerain weè heard rom many social and cultural critics beore. But
such cultural critiques are prooundly subjectie. Many pessimists simply seem
to be well passed the adenture window.`
8
1he willingness o humans to try
new things and experiment with new orms o culture-our adenture
window`-ades rapidly ater certain key points in lie, as we gradually settle in
our ways. Many cultural critics and aerage olk alike seem coninced the best
days are behind us and the current good-or-nothing generation and their new-
angled gadgets and culture are garbage. At times this deoles into a ull-blown
moral panic.
88
It`s perectly normal and probably healthy to examine whether
these changes are good or bad,` says êr Yor/ 1ive. blogger Nick Bilton,
author o ire iv tbe vtvre c ere`. or t !or/.. But we`ll also no doubt

84
1his study suggests that Internet users are a bunch o ideological Jack Kerouacs. 1hey`re
not burrowing down into comorting nests. 1hey`re cruising ar and wide looking or
adenture, inormation, combat and arousal.` Daid Brooks, Riaer. ov tbe torv, NL\ \ORK
1IMLS, April 19, 2010, "''4566777/.C',0#9/+-06ED=D6D?6ED6-4,.,-.6EDG%--<9/"'0B.
85
Lanier, .vra note 3, at 131.
86
a. at 133.
8
Adam 1hierer, 1be .arevtvre !ivaor, Raaio orvat. ava Meaia Orver.bi Rvte.,
1LClNOLOG\ LIBLRA1ION lRON1, Aug. 16, 2006,
"''4566'#+"B,G#%&',-./+-06EDDI6D>6=I6'"#A&);#.'*%#A7,.)-7A%&),-A2-%0&'9A&.)A
0#),&A-7.#%9",4A%*B#9.
88
ee Adam 1hierer, Parevt., Kia. c Potic,va/er. iv tbe Digitat .ge: afegvaraivg .gaiv.t 1ecbvo
Pavic.,` INSIDL ALLC ,July 2009, at 16-,
"''4566777/&B#+/-%:6&064)26O.9,)#J[*BCDX/4)2.

look back at many o the debates a generation rom now and see that a lot o
these ears were inlated and maybe a bit ridiculous, too.`
89

1he sharecropper` concern raised by Carr and Lanier is also oer-stated. 1his
logic ignores the non-monetary beneits that many o us eel we extract rom
today`s online business models and social production processes. Most o us eel
we get a lot back as part o this new alue exchange. Carr and Lanier are
certainly correct that Google, lacebook, MySpace, and a lot o other Net
middlemen are getting big and rich based on all the user-generated content
lowing across their sites and systems. On the other hand, most cyber-citizens
extract enormous beneits rom the existence o those ,mostly ree and
constantly improing, platorms and serices. It`s a ery dierent sort o alue
exchange and business model than in the past, but we are adjusting to it.
\et or all o \ikipedia`s alue as a reerence o irst ,but certainly not inal,
resort, the pessimists hae almost nothing good to say about it. Much the same
goes or open source and other collaboratie eorts. 1hey don`t appear willing
to accept the possibility o any beneits coming rom collectie eorts. And
they wrongly treat the rise o collectie , collaboratie eorts as a zero-sum
game, imagining it represents a net loss o indiidual eort & personhood.`
1hat simply doesn`t ollow. 1he masses hae been gien more o a oice
thanks to the rise o \eb 2.0 collaboratie technologies and platorms, but that
doesn`t mean that media proessionals` don`t still exist. Most bloggers, or
example, build their narraties around acts and stories ound in respected
mainstream media` outlets. It`s true that those outlets must now compete in a
broad sense with many new orms o competition or human attention, but it
doesn`t mean they still won`t play a lead role in the new inormation ecosystem.
Most o all, the pessimists can and must come to terms with the Inormation
Reolution while oering more constructie ava racticat solutions to
legitimately diicult transitional problems created by disintermediating
inluences o the digital technologies and Net. Ater all, practically speaking,
what would the pessimists hae us do i we can`t mitigate the problems they
identiy \hateer the mix o good and bad,` Notes !att treet ]ovrvat
columnist Gordon Croitz, technology only adances and cannot be put back
in the bottle.`
90
\ould the pessimists hae us attempt to put the digital genie
back in bottle with burdensome restrictions on technology or the creation o a
permissions-based system o innoation |\|hether it`s good or society or

89
NICK BIL1ON, I LIVL IN 1lL lU1URL & lLRL`S lO\ I1 \ORKS 63 ,2010,.
90
L. Gordon Croitz, . 1ecbvotog, Cooa or aa. Ye.. \ALL S1RLL1 JOURNAL, Aug. 23, 2010,
"''4566-.B,.#/79\/+-06&%',+B#6(]=DDD=?E?D@EF?>FDH@FX>D?@F@??=?I==X=?H>HHD/"'0B.

bad . is somewhat irreleant at this point,` argues Nick Bilton.
91
1here`s no
turning back the clock.` Similarly, Ben Casnocha has correctly noted that:
the wind at the backs o att techno-optimists . |is| the
orward momentum o technological deelopment. \ou
cannot turn back the clock. It is impossible to enision a uture
where there is te.. inormation and ewer people on social
networks. It is ery possible to enision increasing abundance
along with better ilters to manage it. 1he most constructie
contributions to the debate, then, heed Moore`s Law in the
broadest sense and oer speciic suggestions or how to
harness the change or the better.
92

Regrettably, most o the leading Net pessimists hae ailed to do this in their
work. loweer, good templates or how to accomplish this can be ound in
recent books by \illiam Powers ,avtet`. tac/err,: . Practicat Pbito.ob, for
vitaivg a Cooa ife iv tbe Digitat .ge,
93
and John lreeman ,1be 1,ravv, of Mait:
1be ovr1bov.avaYear ]ovrve, to Yovr vbo,.
94
1hese authors, although
somewhat pessimistic in their iew o technology`s impact on lie and learning,
oer outstanding sel-help tips and plans o action about how to reasonably
assimilate new inormation technologies into our lies. 1heir key insight: the
Internet and digital technologies aren`t going away, so we must igure out how
to deal with them in a responsible manner-both indiidually and collectiely.
It`s essential other pessimists come to grips with that act.
1he pessimists are at their best when highlighting the ery legitimate concerns
about the challenges that accompany technological change, including the impact
o the digital reolution on proessional` media, the decline o authority

91
Bilton, .vra note 89, at 216.
92
Ben Casnocha, Rtea Deretovevt, 1lL AMLRICAN, July 1, 2009,
"''4566777/&0#%,+&./+-06&%+",;#6EDDX6\*.#6%99'#)A)#;#B-40#.'. Clay Shirky has
also noted that 1here is neer going to be a moment when we as a society ask ourseles,
Do we want this Do we want the changes that the new lood o production and access and
spread o inormation is going to bring about`` Clay Shirky, lLRL COMLS LVLR\BOD\:
1lL PO\LR Ol ORGANIZING \I1lOU1 ORGANIZA1IONS 3 ,2008,.
93
\ILLIAM PO\LRS, lAMLL1`S BLACKBLRR\: A PRAC1ICAL PlILOSOPl\ lOR BUILDING A
GOOD LIlL IN 1lL DIGI1AL AGL ,2010,. ee at.o Adam 1hierer, Coivg ritb vforvatiov
Orertoaa: 1bovgbt. ov lamlet`s BlackBerry b, !ittiav Porer., 1LClNOLOG\ LIBLRA1ION
lRON1, Sept. 6, 2010, "''4566'#+"B,G#%&',-./+-06ED=D6DX6DI6+-4,.:A7,'"A
,.2-%0&',-.A-;#%B-&)A'"-*:"'9A-.A"&0B#'9AGB&+<G#%%CAGCA7,BB,&0A4-7#%9.
94
JOlN lRLLMAN, 1lL 1\RANN\ Ol L-MAIL: 1lL lOUR-1lOUSAND-\LAR JOURNL\ 1O
\OUR INBOX ,2009,. lor a reiew o the book, .ee Adam 1hierer, Cav vvav. Coe ritb
vforvatiov Orertoaa. 1,ter Corev c ]obv reevav ]oiv tbe Debate, 1LClNOLOG\ LIBLRA1ION
lRON1, Aug. 23, 2009, "''4566'#+"B,G#%&',-./+-06EDDX6D>6EH6+&.A"*0&.9A+-4#A7,'"A
,.2-%0&',-.A-;#%B-&)A'CB#%A+-7#.A\-".A2%##0&.A\-,.A'"#A)#G&'#.

among trusted experts and intermediaries, and the challenge o inding creatie
ways to und proessional` media and art going orward.
Thoughts on the Optimists
Again, the optimists currently hae the better o this debate: \eb 2.0 is
generally beneiting culture and society. It is almost impossible to accept that
society has not beneited rom the Internet and new digital technologies
compared to the past era o inormation scarcity. 1he Digital Reolution has
greatly empowered the masses and oered them more inormational inputs.
But the optimists need to be less pollyannaish and aoid becoming the
technopolists` ,or digital utopians, that Postman eared were taking oer our
society. 1here`s oten too much Rousseauian romanticism at work in some
optimist writings. Just as the pessimists are oten guilty assuming the Net and
digital technologies are responsible or ar too many ills, the optimists
occasionally do the opposite by engaging in what Nick Carr labels the
Internet`s liberation mythology.` 1he Internet isn`t remaking man or changing
human nature in any undamental way. Nor can it liberate us rom all earthly
constraints or magically sole all o ciilization`s problems. Moreoer, when it
comes to economics, all this talk about the Long 1ail being the uture o
business` ,Chris Anderson, and o \ikinomics . changing eerything
through mass collaboration,` ,1apscott and \illiams, erges on irrational
techno-exuberance.
In particular, optimists oten oerplay the beneits o collectie intelligence,
collaboration, and the role o amateur production. 1hey are occasionally guilty
o the eleation o inormation to metaphysical status` as Postman lamented.
95

lor example, the optimists could rame \iki` and peer-production models as
a covtevevt to proessional media, not a retacevevt or it. Could the equialent
o 1be êr Yor/ 1ive. really be cobbled together by amateurs daily It seems
highly unlikely. And why aren`t there any compelling open source ideo games
Similarly, ree and open source sotware ,lOSS, has produced enormous social
, economic beneits, but it would be oolish to beliee that lOSS ,or wiki`
models, will replace att proprietary business models. Lach model or mode o
production has its place and purpose and they will continue to co-exist and
compete.
\e wouldn`t necessarily be better o i all the proessional` media producers
and old intermediaries disappeared, een i it is no doubt true that many o
them will. Some optimists play the old media just doesn`t get it` card ar too
oten and snobbishly dismiss many producers` alid concerns and eorts to
reinent themseles.

95
Postman, .vra note 8, at 61.

1here`s also a big dierence between remix culture` and rip-o culture.`
Many optimists turn a blind eye to blatant copyright piracy, or example, or
een deend it as either a positie deelopment or simply ineitable. Remix
culture generally enhances and extends culture and creatiity. But blatant
content piracy depries many o society`s most gited creators o the incentie
to produce culturally beneicial works. Likewise, hacking, circumention, and
reerse-engineering all play an important and legitimate role in our new digital
economy, but one need not accept the legitimacy o those actiities when
conducted or nearious purposes ,think identity thet or chip-modding to
acilitate ideo game piracy.,
1he optimists should be cautious about predicting sweeping positie changes
rom the Internet or \eb 2.0 technologies. Consider Shirky`s generally upbeat
assessment o the impact o cognitie surplus.` 1here`s a lot o luy talk and
anecdotal examples in Shirky`s book about how the cognitie surplus spawned
by cyber-lie has aected politics, adocacy, and generosity,` but I think it`s a
stretch to imply that the Net is going to upend political systems. In another
essay in this collection, Lgeny Morozo challenges Shirky on some o these
points, arguing that the Internet will not automatically presere-neer mind
improe-the health o democratic politics.`
96
le`s right. 1hat digital
technology and the Internet will help reshape society and politics to some
degree is indisputable. But that doesn`t mean the Net will radically reshape
political systems or human nature anytime soon.
linally, the optimists would be wise to separate themseles rom those extreme
oices in their community who speak o the noosphere` and global
consciousness` and long or the eentual singularity. \hile he doesn`t go quite
so ar, !irea editor Kein Kelly oten pushes techno-optimism to its extreme.
In his latest book, !bat 1ecbvotog, !avt., Kelly speaks o what he calls the
technium` as a orce` or een a liing organism that has a ital spirit` and
which has its own wants` and a noticeable measure o autonomy.`
9

1reating technology as an autonomous orce is silly, een dangerous, thinking.
It is to imbue it with attributes and eelings that simply do not exist and would
probably not be desirable i they did. \et, some optimists speak in atalistic
terms and make such an outcome seem desirable. 1hey sound like they long or
lie in 1be Matri-`Bring on sentient robot masters and the Singularity!` 1hus
does an optimist cross oer into the realm o quixotic techno-utopianism.
Optimists need to place technological progress in context and appreciate that, as
Postman argued, there are some moral dimensions to technological progress
that desere attention. Not all change is good change. 1he optimists need to be

96
Lgeny Morozo, !itt tbe êt iberate tbe !orta., inra at 443.
9
KLVIN KLLL\, \lA1 1LClNOLOG\ \AN1S 198, 41, 15, 13 ,2010,.

mature enough to understand and address the downsides o digital lie without
dismissing its critics. On the other hand, some o those moral consequences are
prooundly o.itire, which the pessimists usually ail to appreciate or een
acknowledge.
Conclusion: Toward Pragmatic Optimism
Again, I beliee the optimists currently hae the better o this debate. It`s
impossible or me to beliee we were better o in an era o inormation poerty
and un-empowered masses. I`ll take inormation oerload oer inormation
poerty any day! As Dennis Baron puts it: 1he Internet is a true electronic
rontier where eeryone is on his or her own: all manuscripts are accepted or
publication, they remain in irtual print oreer, and no one can tell writers what
to do.`
98

1he rise o the Internet and digital technologies has empowered the masses and
gien eeryone a soapbox on which to speak to the world. O course, that
doesn`t necessarily mean all o them will hae something interesting to say! \e
shouldn`t exalt user-generated content as a good in and o itsel. It`s quality, not
olume, that counts. But such human empowerment is worth celebrating,
despite its occasional downsides.
99
Abundance is better than the old analog
world o ew choices and ewer oices.
loweer, the pessimists hae some ery legitimate concerns regarding how the
passing o the old order might leae society without some important things. lor
example, one need not endorse bailouts or a dying newspaper industry to
nonetheless worry about the important public serice proided by inestigatie
journalists: \ho will take up those eorts i large media institutions go under
because o digital disintermediation
1he skeptics are also certainly correct that each o us should think about how to
better balance new technologies and assimilate them into our lies and the lies
o our amilies and communities. lor example, children need to learn new
digital literacy` and cyber-citizenship` skills to be say Netizens.
1o be clear, I am vot suggesting that these questions should be answered by
goernment. 1here exist many other ways that society can work to presere

98
99
Just as well-meaning scientists and consumers eared that trains and comic books and
teleision would rot our brains and spoil our minds, I beliee many o the skeptics and
worrywarts today are missing the bigger picture, the greater alue that access to new and
aster inormation is bringing us.` Nick Bilton, I LIVL IN 1lL lU1URL & lLRL`S lO\ I1
\ORKS 136 ,2010,.

important alues and institutions without embracing the stasis mentality and
using coercion to accomplish that which should be pursued oluntarily.
As noted, the nostalgia the pessimists typically espouse or the past is a
common rerain o cultural and technological critics who ear our best days are
behind us. 1he truth typically proes less cataclysmic, o course. 1he great
thing about humans is that we adapt better than other creatures. \hen it comes
to technological change, resiliency is hard-wired into our genes. 1he techno-
apocalypse neer comes,` notes tate`. Jack Shaer, because cultures tend to
assimilate and normalize new technology in ways the retul neer anticipate.`
100

\e learn how to use the new tools gien to us and make them part o our lies
and culture. Indeed, we hae lied through reolutions more radical than the
Inormation Reolution. \e cav adapt and learn to lie with some o the
legitimate diiculties and downsides o the Inormation Age.
Generally speaking, the sensible middle ground position is pragmatic
optimism`: \e should embrace the amazing technological changes at work in
today`s Inormation Age but with a healthy dose o humility and appreciation
or the disruptie impact and pace o that change. \e need to think about how
to mitigate the negatie impacts associated with technological change without
adopting the paranoid tone or Luddite-ish recommendations o the pessimists.
I`m particularly persuaded by the skeptics` call or all o us to exercise some
restraint in terms o the role technology plays in our own lies. \hile pessimists
rom Plato and Postman certainly went too ar at times, there is more than just
a kernel o truth to their claim that, taken to an extreme, technology can hae a
deleterious impact on lie and learning. \e need to ocus on the Aristotelian
mean. \e must aoid neo-Luddite calls or a return to the good ol days` on
the one hand, while also rejecting techno-utopian Pollyannaism on the other.
\e need not go to all or nothing` extremes.
In the end, howeer, I return to the importance o eolutionary dynamism and
the importance o leaing a broad sphere or continued experimentation by
indiiduals and organizations alike. lreedom broaat, cov.trvea is aluable in its
own right-een i not all o the outcomes are optimal. As Clay Shirky rightly
notes:
1his does not mean there will be no diiculties associated with
our new capabilities-the deenders o reedom hae long
noted that ree societies hae problems peculiar to them.
Instead, it assumes that the alue o reedom outweighs the

100
Jack Shaer, Digitat âtire Catv. tbe .viov. Ma..e., SLA1L, Sept. 13, 2010,
"''4566777/9B&'#/+-06,)6EEIF=I=.

problems, not based on calculation o net alue but because
reedom is the right thing to want or society.
101

linally, we cannot ignore the practical diiculties o halting or een slowing
progress-assuming we somehow collectiely decided we wanted to do so.
1urning back the clock seems almost unathomable at this point absent extreme
measures that would sacriice so many o the beneits the Inormation Age has
brought us-not to mention the curtailment o reedom that it would demand.
Regardless, the old 1heuth-1hamus debate about the impact o technological
change on culture and society will continue to rage. 1here is no chance this
debate will die down anytime soon. ,Just wait till new technologies like irtual
reality go mainstream!, Despite real challenges in adapting to technological
change, I remain generally optimistic about the prospects or technology to
improe the human condition.

101
Shirky, .vra note 59, at 298.

89

CHAPTER 2
IS THE GENERATIVE INTERNET AT RISK?
Protecting the Internet Without Wrecking It:
How to Meet the Security Threat 91
!"#$%&$# ()%%*$)#
A Portrait of the Internet as a Young Man 113
+## ,$*%"-
Saving the Net from Its Supporters 139
+.$/ 0&)1*1*

90 CHAPTER 2: IS THE GENERATIVE INTERNET AT RISK?


Protecting the Internet Without
Wrecking It: How to Meet the
Security Threat
By Jonathan Zittrain
*

On Noember 2, 1988, 5-10 o the 60,000 computers hooked up to the
Internet started acting strangely. Inentories o aected computers reealed
that rogue programs were demanding processor time. \hen concerned
administrators terminated these programs, they reappeared and multiplied.
1hey then discoered that renegade code was spreading through the Internet
rom one machine to another. 1he sotware-now commonly thought o as
the irst Internet worm-was traced to a twenty-three-year-old Cornell
Uniersity graduate student, Robert 1appan Morris, Jr., who had launched it by
inecting a machine at MI1 rom his terminal in Ithaca, New \ork.
Morris said he unleashed the worm to count how many machines were
connected to the Internet, and analysis o his program conirmed his benign
intentions. But his code turned out to be buggy. I Morris had done it right, his
program would not hae drawn attention to itsel. It could hae remained
installed or days or months, and quietly perormed a wide array o actiities
other than Morris`s digital headcount.
1he mainstream media had an intense but brie ascination with the incident. A
goernment inquiry led to the creation o the Deense Department-unded
Computer Lmergency Response 1eam Coordination Center at Carnegie Mellon
Uniersity, which seres as a clearinghouse or inormation about iruses and
other network threats. A Cornell report on what had gone wrong placed the
blame solely on Morris, who had engaged in a juenile act` that was selish
and inconsiderate.` It rebuked elements o the media that had branded Morris
a hero or dramatically exposing security laws, noting that it was well known
that the computers` Unix operating systems were imperect. 1he report called
or uniersity-wide committees to proide adice on security and acceptable
use. It described consensus among computer scientists that Morris`s acts
warranted some orm o punishment, but not so stern as to damage
permanently the perpetrator`s career.`

Proessor o Law, larard Law School and larard Kennedy School, Proessor o
Computer Science, larard School o Lngineering and Applied Sciences, Co-lounder,
Berkman Center or Internet & Society. 1his chapter originally appeared in the March,April
2008 BOS1ON RLVIL\.

In the end, Morris apologized, earned three years o criminal probation,
perormed our hundred hours o community serice, and was ined >10,050.
le transerred rom Cornell to larard, ounded a dot-com startup with some
riends in 1995, and sold it to \ahoo! in 1998 or >49 million. le is now a
respected, tenured proessor at MI1.
In retrospect, the commission`s recommendations-urging users to patch their
systems and hackers to grow up-might seem nae. But there were ew
plausible alternaties. Computing architectures, both then and now, are
designed or lexibility rather than security. 1he decentralized ownership and
non-proprietary nature o the Internet and the computers connected to it made
it diicult to implement structural improements. More importantly, it was
hard to imagine cures that would not entail drastic, wholesale, purpose-altering
changes to the ery abric o the Internet. Such changes would hae been
wildly out o proportion to the perceied threat, and there is no record o their
haing een been considered.
Generatie systems are powerul-they enable extraordinary
numbers o people to deise new ways to express themseles
in speech, art, or code, perhaps because they lack central
coordination and control.
By design, the uniersity workstations o 1988 were generatie: 1heir users
could write new code or them or install code written by others. 1his generatie
design lies on in today`s personal computers. Networked PCs are able to
retriee and install code rom each other. \e need merely click on an icon or
link to install new code rom aar, whether to watch a ideo newscast embedded
within a \eb page, update our word processing or spreadsheet sotware, or
browse satellite images.
Generatie systems are powerul and aluable, not only because they oster the
production o useul things like \eb browsers, auction sites, and ree
encyclopedias, but also because they enable extraordinary numbers o people to
deise new ways to express themseles in speech, art, or code and to work with
other people. 1hese characteristics can make generatie systems ery successul
een though-perhaps especially because-they lack central coordination and
control. 1hat success attracts new participants to the generatie system.
1he lexibility and power that make generatie systems so attractie are,
howeer, not without risks. Such systems are built on the notion that they are
neer ully complete, that they hae many uses yet to be conceied o, and that
the public can be trusted to inent good uses and share them. But multiplying
breaches o that trust threaten the ery oundations o the system.
\hether through a sneaky ector like the one Morris used, or through the ront
door, when a trusting user elects to install something that looks interesting

without ully understanding it, opportunities or accidents and mischie abound.
A hobbyist computer that crashes might be a curiosity, but when a home or
oice PC with years` worth o ital correspondence and papers is compromised,
it can be a crisis. And when thousands or millions o indiidual, business,
research, and goernment computers are subject to attack, we may ind
ourseles aced with a undamentally new and harrowing scenario. As the
unsustainable nature o the current state o aairs becomes more apparent, we
are let with a dilemma that cannot be ignored: low do we presere the
extraordinary beneits o generatiity, while addressing the growing
ulnerabilities that are innate to it
! ! !
low proound is today`s security threat Since 1988, the Internet has suered
ew truly disruptie security incidents. A network designed or communication
among academic and goernment researchers appeared to scale beautiully as
hundreds o millions o new users signed on during the 1990s, and three types
o controls seemed adequate to address emerging dangers.
lirst, the hacker ethos rowns upon destructie hacking. Most iruses that
ollowed Morris`s worm had completely innocuous payloads: In 2004, Mydoom
spread like wildire and reputedly cost billions in lost productiity, but the worm
did not tamper with data, and it was programmed to stop spreading at a set
time. \ith rare exceptions like the inamous Loebug worm, which oerwrote
iles with copies o itsel, the ew highly malicious iruses that run contrary to
the hacker ethos were so poorly coded that they ailed to spread ery ar.
Second, network operations centers at uniersities and other institutions
became more proessionalized between 1988 and the adent o the mainstream
Internet. lor a while, most Internet-connected computers were staed by
proessionals, administrators who generally heeded admonitions to patch
regularly and scout or security breaches. Less adept mainstream consumers
began connecting unsecured PCs to the Internet in earnest only in the mid-
1990s. 1hen, transient dial-up connections greatly limited both the amount o
time during which they were exposed to security threats, and the amount o
time that, i compromised and hijacked, they would contribute to the problem.
linally, bad code lacked a business model. Programs to trick users into
installing them, or to sneak onto the machines, were written or amusement.
Bad code was more like graiti than illegal drugs: 1here were no economic
incenties or its creation.
1oday each o these controls has weakened. \ith the expansion o the
community o users, the idea o a set o ethics goerning actiity on the Internet
has eaporated. Anyone is allowed online i he or she can ind a way to a

computer and a connection, and mainstream users are transitioning rapidly to
always-on broadband connections.
Moreoer, PC user awareness o security issues has not kept pace with
broadband growth. A December 2005 online saety study ound 81 o home
computers to be lacking irst-order protection measures such as current
antiirus sotware, spyware protection, and eectie irewalls.
1

Perhaps most signiicantly, bad code is now a business. \hat seemed genuinely
remarkable when irst discoered is now commonplace: Viruses that
compromise PCs to create large zombie botnets` open to later instructions.
Such instructions hae included directing PCs to become remotely-controlled e-
mail serers, sending spam by the thousands or millions to e-mail addresses
harested rom the hard disk o the machines themseles or gleaned rom
Internet searches, with the entire process typically proceeding behind the back
o the PCs` owners. At one point, a single botnet occupied iteen percent o
\ahoo!`s search capacity, running random searches on \ahoo! to ind text that
could be inserted into spam e-mails to throw o spam ilters.
2
Dae Dagon,
who recently let Georgia 1ech Uniersity to start a bot-ighting company
named Damballa, pegs the number o botnet-inected computers at close to 30
million.
3
Dagon said, lad you told me ie years ago that organized crime
would control one out o eery ten home machines on the Internet, I would not
hae belieed that.`
4
So long as spam remains proitable, that crime will persist.
Botnets can also be used to launch coordinated attacks on a particular Internet
endpoint. lor example, a criminal can attack an Internet gambling \eb site and
then extort payment to make the attacks stop. 1he going rate or a botnet to
launch such an attack is reputed to be about >5,000 per day.
5

Viruses are thus aluable properties. \ell-crated worms and iruses routinely
inect ast swaths o Internet-connected personal computers. Antiirus endor
Lugene Kaspersky o Kaspersky Labs told an industry conerence that they
may not be able to withstand the onslaught.`
6
IBM`s Internet Security Systems

1
AOL,NCSA ONLINL SAlL1\ S1UD\ 2 ,Dec. 2005,,
#$$%&''((()*+),-.'+/0$,0$'-12'345,6'/334+,6'#,5%'%-3'613,$786$.-789::;)%-3
2
1im \eber, Crivivat. Ma, Orerrbetv tbe !eb,` BBC NL\S, Jan. 25, 200,
#$$%&''0,(6)**+)+/).<'9'#4'*.640,66'=9>?=@A)6$2.
3
Bob Sullian, . Yovr Covvter a Crivivat., RLD 1APL ClRONICLLS, Mar. 2, 200,
#$$%&''B,-$1%,)260*+)+/2'9::C':D'*/$686$/B7)#$25.
4
a.
5
a.
6
a.

reported a 40 increase in sotware ulnerabilities reported by manuacturers
and white hat` hackers between 2005 and 2006.
Nearly all o those

ulnerabilities could be exploited remotely, and oer hal allowed attackers to
gain ull access to the machine and its contents.
As the supply o troubles has increased, the capacity to address them has
steadily diminished. Patch deelopment time increased throughout 2006 or all
o the top operating system proiders.
8
1imes shortened modestly across the
board in the irst hal o 200, but, on aerage, enterprise endors were still
exposed to ulnerabilities or 55 days-plenty o time or hazardous code to
make itsel elt.
9
,1he patch interals or browsers tend to be shorter than
those or operating systems., \hat is more, antiirus researchers and irms
require extensie coordination eorts simply to agree on a naming scheme or
iruses as they emerge.
10
1his is a ar cry rom a common strategy or battling
them.
In addition, the idea o casually cleaning a irus o a PC is gone. \hen
computers are compromised, users are now typically adised to reinstall
eerything on them. lor example, in 200, some PCs at the U.S. National
Deense Uniersity ell ictim to a irus. 1he institution shut down its network
serers or two weeks and distributed new laptops to instructors.
11
In the
absence o such drastic measures, a truly mal` piece o malware could be
programmed to, say, erase hard dries, transpose numbers inside spreadsheets
randomly, or intersperse nonsense text at arbitrary interals in \ord documents
ound on inected computers-and nothing would stand in the way.
Recognition o these basic security problems has been slowly growing in
Internet research communities. Nearly two-thirds o academics, social analysts,
and industry leaders sureyed by the Pew Internet & American Lie Project in
2004 predicted serious attacks on network inrastructure or the power grid in

IBM IN1LRNL1 SLCURI1\ S\S1LMS, X-lORCL 2006 1RLND S1A1IS1ICS ,Jan. 200,,
#$$%&''((()466)0,$'-/+.2,0$6'(#4$,%1%,B6'E8F/B+,8GH,+8IB4,3)%-3.
8
S\MAN1LC, GLOBAL IN1LRNL1 SLCURI1\ 1lRLA1 RLPOR1, 1RLNDS lOR JUL\-DLCLMBLR
200 at 24-28 ,April 2008,,
#$$%&'',J15)67210$,+)+/2'2<$K403/',0$,B%B46,'(#4$,8%1%,B6'*L
(#4$,%1%,B840$,B0,$86,+.B4$78$#B,1$8B,%/B$8H4448:@L9::?),0L.6)%-3.
9
a. at 6.
10
ee, e.g., Common Malware Lnumeration: Reducing Public Conusion During Malware
Outbreaks, #$$%&''+2,)24$B,)/BK' ,last isited June 1, 200,.
11
Bill Gertz & Rowan Scarborough, v.iae tbe Rivg-ôte. frov tbe Pevtagov, \ASl. 1IMLS, Jan. 5,
200, at A5, aailable at #$$%&''((()K,B$M345,)+/2'K,B$M345,'B40K:AA9:C)#$25.

the coming decade.
12
Security concerns will lead to a undamental shit in our
tolerance o the status quo, either by a catastrophic episode, or, more likely, a
glacial death o a thousand cuts.
Consider, in the latter scenario, the burgeoning realm o badware` ,or
malware`, beyond iruses and worms: Sotware that is oten installed at the
user`s initation. 1he popular ile-sharing program KaZaA, though adertised
as spyware-ree,` contains code that users likely do not want. It adds icons to
the desktop, modiies Microsot Internet Lxplorer, and installs a program that
cannot be closed by clicking Quit.` Uninstalling the program does not
uninstall all these extras, and the aerage user does not know how to get rid o
the code itsel. \hat makes such badware bad` has to do with the leel o
disclosure made to a consumer beore he or she installs it. 1he most common
responses to the security problem cannot easily address this gray zone o
sotware.
Many technologically say people think that bad code is simply a Microsot
\indows issue. 1hey beliee that the \indows OS and the Internet Lxplorer
browser are particularly poorly designed, and that better` counterparts
,GNU,Linux and Mac OS, or the lireox and Opera browsers, can help shield
a user. But the added protection does not get to the undamental problem,
which is that the point o a PC-regardless o its OS-is to enable its users to
easily reconigure it to run new sotware rom anywhere. \hen users make
poor decisions about what sotware to run, the results can be deastating to
their machines and, i they are connected to the Internet, to countless others`
machines as well.
1he cybersecurity problem deies easy solution because any o its most obious
ixes will undermine the generatie essence o the Internet and PC. Bad code is
an ineitable side eect o generatiity, and as PC users are increasingly
ictimized by bad code, consumers are likely to reject generatie PCs in aor o
sae inormation appliances-digital ideo recorders, mobile phones, iPods,
BlackBerrys, and ideo game consoles-that optimize a particular application
and cannot be modiied by users or third-parties. It is entirely reasonable or
consumers to actor security and stability into their choice. But it is an
undesirable choice to hae to make.
! ! !
On January 9, 200, Stee Jobs introduced the iPhone to an eager audience
crammed into San lrancisco`s Moscone Center. A beautiul and brilliantly

12
Susannah lox et al., 1be vtvre of tbe vtervet: v a vrre,, 1ecbvotog, ert. ava cbotar. ratvate
!bere tbe êtror/ . eaaea iv tbe êt er Year., Jan. 9, 2005, at i,
#$$%&''((()%,(40$,B0,$)/BK'NNF'B'A@;'B,%/B$8-46%517)16%.

engineered deice, the iPhone blended three products into one: an iPod, with
the highest-quality screen Apple had eer produced, a phone, with cleerly
integrated unctionality, such as oicemail that came wrapped as separately
accessible messages, and a deice to access the Internet, with a smart and
elegant browser, and built-in map, weather, stock, and e-mail capabilities.
Stee Jobs had no clue how the Apple II would be used. 1he iPhone-or all
its startling inentieness-is precisely the opposite.
1his was Stee Jobs`s second reolution. 1hirty years earlier, at the lirst \est
Coast Computer laire in nearly the same spot, the twenty-one-year-old Jobs,
wearing his irst suit, exhibited the Apple II personal computer to great buzz
amidst ten thousand walking, talking computer reaks.`
13
1he Apple II was a
machine or hobbyists who did not want to uss with soldering irons: all the
ingredients or a unctioning PC were proided in a conenient molded plastic
case. Instead o puzzling oer bits o hardware or typing up punch cards to
eed into someone else`s mainrame, Apple owners aced only the hurdle o a
cryptic blinking cursor in the upper let corner o the screen: the PC awaited
instructions. But the hurdle was not high. Some owners were inspired to
program the machines themseles, but beginners, too, could load sotware
written and then shared or sold by their more skilled counterparts. 1he Apple
II was a blank slate, a bold departure rom preious technology that had been
deeloped and marketed to perorm speciic tasks.
1he Apple II quickly became popular. And when programmer and
entrepreneur Dan Bricklin introduced the irst killer application or the Apple II
in 199-VisiCalc, the world`s irst spreadsheet program-sales o the ungainly
but ery cool machine took o. An Apple running VisiCalc helped to conince
a skeptical world that there was a place or the PC on eeryone`s desk.
1he Apple II was quintessentially generatie technology. It was a platorm. It
inited people to tinker with it. lobbyists wrote programs. Businesses began
to plan on selling sotware. Jobs ,and Apple, had no clue how the machine
would be used. 1hey had their hunches, but, ortunately or them ,and the rest
o us,, nothing constrained the PC to the hunches o the ounders.
1he iPhone-or all its startling inentieness-is precisely the opposite.
Rather than a platorm that inites innoation, the iPhone comes
preprogrammed. In its irst ersion, you were not allowed to add programs to
the all-in-one deice that Stee Jobs sells you except ia the Siberia o its \eb

13
Daid l. Ahl, 1be ir.t !e.t Coa.t Covvter aire, in 3 1lL BLS1 Ol CRLA1IVL COMPU1ING
98 ,Daid Ahl & Burchenal Green eds., 1980,, aailable at
#$$%&''((()1$1B41B+#4J,6)/BK'*++D'6#/(%1K,)%#%O%1K,8>?.

browser. Its unctionality was locked in, though Apple could change it through
remote updates. Indeed, those who managed to tinker with the code and enable
iPhone-support o more or dierent applications, were on the receiing end o
Apple`s threat to transorm the iPhone into an iBrick.
14
A threat, to be sure,
that Apple later at least partially disaowed. 1he machine was not to be
generatie beyond the innoations that Apple ,and its exclusie carrier, A1&1,
wanted. In its second ersion a year later, the iPhone boasted the App Store.
Sotware deelopers could code or the phone - but the deelopers, and then
each piece o sotware, would require approal rom Apple beore it could be
made aailable to iPhone users. Apple would receie a 30 cut o sales,
including in-app` sales o upgrades, and an app could be banned retroactiely
ater initial approal. 1his made the iPhone contingently generatie,` a hybrid
status that, depending on how you look at it, is either the best or the worst o
both worlds: a melding o the sterile and the generatie.
Jobs was not shy about these restrictions. As he said at the iPhone launch: \e
deine eerything that is on the phone .. \ou don`t want your phone to be
like a PC. 1he last thing you want is to hae loaded three apps on your phone
and then you go to make a call and it doesn`t work anymore.`
15

In the arc rom the Apple II to the iPhone, we learn something important about
where the Internet has been, and something een more important about where
it is going. 1he PC reolution was launched with PCs that inited innoation
by others. So, too, with the Internet. Both were designed to accept any
contribution that ollowed a basic set o rules ,either coded or a particular
operating system, or respecting the protocols o the Internet,. Both
oerwhelmed their respectie proprietary, non-generatie competitors: PCs
crushed stand-alone word processors and the Internet displaced such
proprietary online serices as CompuSere and AOL.
But the uture is looking ery dierent because o the security situation-not
generatie PCs attached to a generatie network, but appliances tethered to a
network o control. 1hese appliances take the innoations already created by
Internet users and package them neatly and compellingly, which is good-but
only i the Internet and PC can remain suiciently central in the digital
ecosystem to compete with locked-down appliances and acilitate the next
round o innoations. 1he balance between the two spheres is precarious, and
it is slipping toward the saer appliance. lor example, Microsot`s Xbox 360

14
Michael, !""#$ &'() *+ ,'( -./0123 45#612$7 08965$) :0+9 ;$<+ &6=+>'/$ 4"7'+$, Apple Gazette,
Sep. 24, 200, !""#$%%&&&'(##)*+(,*""*'-./%0#!.1*%(##)*23(4320"2/(42560-72
81).-7*920#!.1*32&0"!21*:"23.;"&(6*28#9("*%.
15
&$$ John Marko, &+$?$ @6A) :'#2) +9$ B0C9+/6"$ !C'05, N.\. 1IMLS, Jan. 12, 200, aailable at
!""#$%%&&&'14"0/*3'-./%<==>%=?%?<%"*-!1.).+4%?<(##)*'!"/).

ideo game console is a powerul computer, but, unlike Microsot`s \indows
operating system or PCs, it does not allow just anyone to write sotware that
can run on it - games must be licensed by Microsot. Bill Gates sees the Xbox
at the center o the uture digital ecosystem, rather than its periphery: It is a
general purpose computer . . . |\|e wouldn`t hae done it i it was just a
gaming deice. \e wouldn`t hae gotten into the category at all. It was about
strategically being in the liing room.`
16

Deices like iPhones and Xbox 360s may be saer to use, and they may seem
capacious in eatures so long as they oer a simple \eb browser. But by
ocusing on security and limiting the damage that users can do through their
own ignorance or carelessness, these appliances also limit the beneicial tools
that users can create or receie rom others-enhancements they may be
clueless about when they are purchasing the deice.
I the PC ceases to be at the center o the inormation
technology ecosystem, the most restrictie aspects o
inormation appliances will come to the ore.
Security problems related to generatie PC platorms may propel people away
rom PCs and toward inormation appliances controlled by their makers. I we
eliminate the PC rom many dens or liing rooms, we eliminate the test bed and
distribution point o new, useul sotware rom any corner o the globe. \e
also eliminate the saety ale that keeps those inormation appliances honest.
I 1iVo makes a digital ideo recorder that has too many limits on what people
can do with the ideo they record, people will discoer DVR sotware like
Myth1V that records and plays 1V shows on their PCs. I mobile phones are
too expensie, people will use Skype. But people do not buy PCs as insurance
policies against appliances that limit their reedoms, een though PCs sere
exactly this ital unction. People buy them to perorm certain tasks at the
moment o acquisition. I PCs cannot reliably perorm these tasks, most
consumers will not see their merit, and the saety ale will be lost. I the PC
ceases to be at the center o the inormation technology ecosystem, the most
restrictie aspects o inormation appliances will come to the ore.
In act, the dangers may be more subtly packaged. PCs need not entirely
disappear as people buy inormation appliances in their stead. PCs can
themseles be made less generatie. Users tired o making the wrong choices
about installing code on their PCs might choose to let someone else decide
what code should be run. lirewalls can protect against some bad code, but they
also complicate the installation o new good code. As antiirus, antispyware,

16
Ryan Block, . vvcbtive Cbat ritb itt Cate., LNGADGL1, Jan. 8, 200,
#$$%&''(((),0K1-K,$)+/2'9::C':A':?'1L5.0+#$42,L+#1$L(4$#L*455LK1$,6L1$L+,6'.

and anti-badware barriers prolierate, there are new barriers to the deployment
o new good code rom unpriileged sources. And in order to guarantee
eectieness, these barriers are becoming increasingly paternalistic, reusing to
allow users easily to oerrule them. Lspecially in enironments where the user
o the PC does not own it-oices, schools, libraries, and cyber-cas-barriers
are being put in place to preent the running o any code not speciically
approed by the releant gatekeeper. Users may ind themseles limited to
using a \eb browser. And while \eb 2.0` promises many more uses or a
browser-consumers can now write papers and use spreadsheets through a
browser, and sotware deelopers now write or \eb platorms like lacebook
instead o PC operating systems -these \eb platorms are themseles tethered
to their makers, their generatiity contingent on the continued permission o the
platorm endors.
Short o completely banning unamiliar sotware, code might be diided into
irst- and second-class status, with second-class, unapproed sotware allowed
to perorm only certain minimal tasks on the machine, operating within a digital
sandbox. 1his technical solution is saer than the status quo but imposes
serious limits. It places the operating system creator or installer in the position
o deciding what sotware will and will not run. 1he PC will itsel hae become
an inormation appliance, not easily reconigured or extended by its users.
1he key to aoiding such a uture is to gie the market a reason not to abandon
or lock down the PCs that hae sered it so well, also giing most goernments
reason to rerain rom major interention into Internet architecture in the name
o public saety. 1he solutions to the generatie dilemma will rest on social and
legal as much as technical innoation, and the best guideposts can be ound in
other generatie successes in those arenas. Mitigating abuses o openness
without resorting to lockdown will depend on a community ethos embodied in
responsible groups with shared norms and a sense o public purpose, rather
than in the hands o a single gatekeeper, whether public or priate.
In the medium term, the battle between generatie and sterile will be played out
between the iPhone and Android, which despite its own ersion o an App
Store, also allows outside code to run that doesn`t come rom the store, and
with projects like Boxee and Google 1V, which are seeking to bridge the gap
between the PC and the liing room. Lach deice sets the dial set at a dierent
point between complete open` and completely closed.` And those dials can
shit: ater a security spill,` Android could be reprogrammed oernight to be
more restrictie in the code it runs, and by the same token, Apple could decide
to loosen its restrictions on iPhone code.
! ! !
\e need a strategy that addresses the emerging security troubles o today`s
Internet and PCs without killing their openness to innoation. 1his is easier

said than done, because our amiliar legal tools are not particularly attuned to
maintaining generatiity. A simple regulatory interention-say, banning the
creation or distribution o deceptie or harmul code-will not work because it
is hard to track the identities o sophisticated wrongdoers, and, een i ound,
many may not be in cooperatie jurisdictions. Moreoer, such interention may
hae a badly chilling eect: Much o the good code we hae seen has come
rom unaccredited people sharing what they hae made or un, collaborating in
ways that would make business-like regulation o their actiities burdensome or
them. 1hey might be dissuaded rom sharing at all.
\e can ind a balance between needed change and undue restriction i we think
about how to moe generatie approaches and solutions that work at one
layer` o the Internet-content, code, or technical-to another. Consider
\ikipedia, the ree encyclopedia whose content-the entries and their
modiications-is ully generated by the \eb community. 1he origins o
\ikipedia lie in the open architecture o the Internet and \eb. 1his allowed
\ard Cunningham to inent the wiki, generic sotware that oers a way o
editing or organizing inormation within an article, and spreading this
inormation to other articles. Unrelated non-techies then used \ikis to orm
\eb sites at the content layer, including \ikipedia. People are ree not only to
edit \ikipedia, but to take all o its contents and experiment with dierent ways
o presenting or changing the material, perhaps by placing the inormation on
otherwise unrelated \eb sites in dierent ormats. \hen abuses o this
openness beset \ikipedia with andalism, copyright inringement, and lies, it
turned to its community-aided by some important technical tools-as the
primary line o deense, rather than copyright or deamation law. Most recently,
this eort has been aided by the introduction o Virgil Griith`s \ikiscanner, a
simple tool that uses \ikipedia`s page histories to expose past instances o
article whitewashing by interested parties.
Unlike a orm o direct regulation that would hae locked down the site, the
\ikipedian response so ar appears to hae held many o \ikipedia`s problems
at bay. \hy does it work so well Generatie solutions at the content layer
seem to hae two characteristics that suggest broad approaches to lowering the
risks o the generatie Internet while presering its openness. lirst, much
participation in generating \eb content-editing \ikipedia entries, blogging, or
een engaging in transactions on eBay and Amazon that ask or reiews and
ratings to establish reputations-is understood to be an innately social actiity.
1hese serices solicit and depend upon participation rom the public, and their
participation mechanisms are easily mastered. 1he same possibility or broad
participation exists one leel down at the technical layer, but it has not yet been
as ully exploited: Mainstream users hae thus ar been eager to hae someone
else sole underlying problems, which they perceie as technical rather than
social. Second, many content-layer enterprises hae deeloped technical tools
to support collectie participation, augmenting an indiidualistic ethos with

community-acilitating structures. In the Internet and PC security space, on the
other hand, there hae been ew tools aailable to tap the power o groups o
users to, say, distinguish good code rom bad.
1he eectieness o the social layer in \eb successes points to two approaches
that might sae the generatie spirit o the Net, or at least keep it alie or
another interal. 1he irst is to reconigure and strengthen the Net`s
experimentalist architecture to make it it better with the ast expansion in the
number and types o users. 1he second is to deelop new tools and practices
that will enable releant people and institutions to help secure the Net
themseles instead o waiting or someone else to do it.
Generative PCs with Easy Reversion
\ikis are designed so that anyone can edit them. 1his creates a genuine and
ongoing risk o bad edits, through either incompetence or malice. 1he damage
that can be done, howeer, is minimized by the wiki technology, because it
allows bad changes to be quickly reerted. All preious ersions o a page are
kept, and a ew clicks by another user can restore a page to the way it was
beore later changes were made. So long as there are more users ,and
automated tools they create, detecting and reerting andalism than there are
users andalizing, the community wins. ,1ruly, the price o reedom is eternal
igilance.,
Our PCs can be similarly equipped. lor years \indows XP ,and now Vista,
has had a system restore eature, where snapshots are taken o the machine at a
moment in time, allowing later bad changes to be rolled back. 1he process o
restoring is tedious, restoration choices can be rustratingly all-or-nothing, and
the system restoration iles themseles can become corrupted, but it represents
progress. Len better would be the introduction o eatures that are
commonplace on wikis: A quick chart o the history o each document, with an
ability to see date-stamped sets o changes going back to its creation. Because
our standard PC applications assume a saer enironment than really exists,
these eatures hae neer been demanded or implemented. Because wikis are
deployed in enironments prone to andalism, their contents are designed to be
easily recoered ater a problem.
1he next stage o this technology lies in new irtual machines, which would
obiate the need or cyber cas and corporate I1 departments to lock down
their PCs.
In an eort to satisy the desire or saety without ull lockdown, PCs can be
designed to pretend to be more than one machine, capable o cycling rom one
personality to the next. In its simplest implementation, we could diide a PC
into two irtual machines: Red` and Green.` 1he Green PC would house
reliable sotware and important data-a stable, mature OS platorm and tax

returns, term papers, and business documents. 1he Red PC would hae
eerything else. In this setup, nothing that happens on one PC can easily aect
the other, and the Red PC could hae a simple reset button that restores a
predetermined sae state. Someone could conidently store important data on
the Green PC and still use the Red PC or experimentation. 1his isn`t rocket
science - there`s already sotware out there to amount to a Green,Red diide
on a \indows machine - but it`s not so easy or the aerage user to deploy and
use.
Lasy, wiki-style reersion, coupled with irtual PCs, would accommodate the
experimentalist spirit o the early Internet while acknowledging the important
uses or those PCs that we do not want to disrupt. Still, this is not a complete
solution. 1he Red PC, despite its experimental purpose, might end up
accumulating data that the user wants to keep, occasioning the need or what
Internet architect Daid D. Clark calls a checkpoint Charlie` to moe sensitie
data rom Red to Green without also carrying a irus or anything else
undesirable. 1here is also the question o what sotware can be deemed sae or
Green-which is just another ersion o the question o what sotware to run
on today`s single-identity PCs.
lor these and related reasons, irtual machines will not be panaceas, but they
might buy us some more time. And they implement a guiding principle rom
the Net`s history: an experimentalist spirit is best maintained when ailures can
be contained as learning experiences rather than expanding to catastrophes.
A Generative Solution to Bad Code
1he Internet`s original design relied on ew mechanisms o central control.
1his lack o control has the generatie beneit o allowing new serices to be
introduced, and new destinations to come online, without any up-ront etting
or blocking by either priate incumbents or public authorities. \ith this
absence o central control comes an absence o measurement. 1he Internet
itsel cannot say how many users it has, because it does not maintain user
inormation. 1here is no awareness at the network leel o how much
bandwidth is being used by whom. lrom a generatie point o iew this is
good because it allows initially whimsical but data-intensie uses o the network
to thrie ,remember goldish cams,-and perhaps to become ital ,now-
routine ideoconerencing through Skype, rom, unsettlingly, the makers o
KaZaA,.
Because we cannot easily measure the network and the
character o the actiity on it, we cannot easily assess and deal
with threats rom bad code without laborious and imperect
cooperation among a limited group o security sotware
endors.

But limited measurement is starting to hae generatie drawbacks. Because we
cannot easily measure the network and the character o the actiity on it, we
cannot easily assess and deal with threats rom bad code without laborious and
imperect cooperation among a limited group o security sotware endors. 1he
uture o the generatie Net depends on a wider circle o users able to grasp the
basics o what is going on within their machines and between their machines
and the network.
\hat might this system look like Roughly, it would take the orm o toolkits
to oercome the digital solipsism that each o our PCs experiences when it
attaches to the Internet at large, unaware o the size and dimension o the
network to which it connects. 1hese toolkits would run unobtrusiely on the
PCs o participating users, reporting back-to a central source, or perhaps only
to each other-inormation about the ital signs and running code o that PC,
which could help other PCs determine the leel o risk posed by new code.
\hen someone is deciding whether to run new sotware, the toolkit`s
connections to other machines could tell the person how many other machines
on the Internet are running the code, what proportion o machines belonging to
sel-described experts are running it, whether those experts hae ouched or it,
and how long the code has been in the wild.
Building on these ideas about measurement and code assessment, larard
Uniersity`s Berkman Center and the Oxord Internet Institute-
multidisciplinary academic enterprises dedicated to charting the uture o the
Net and improing it-hae begun a project called StopBadware
,www.stopbadware.org,, designed to assist rank-and-ile Internet users in
identiying and aoiding bad code. 1he idea is not to replicate the work o
security endors like Symantec and McAee, which, or a ee, seek to bail new
iruses out o our PCs aster than they pour in. Rather, these academic groups
are deeloping a common technical and institutional ramework that enables
users to deote some bandwidth and processing power or better measurement
o the eect o new code. A irst step in the toolkit was deeloped as lerdict
PC.` lerdict PC was a small piece o sotware that assembles ital signs like
number o pop-up windows or crashes per hour. |It incorporates that data into
a dashboard usable by mainstream PC owners. Lorts like lerdict - including
such entures as Soluto ,((()6/5.$/)+/2, - will test the idea that solutions
that hae worked or generating content might also be applicable to the
technical layer. Such a system might also illuminate Internet iltering by
goernments around the world, as people participate in a system where they can
report when they cannot access a \eb site, and such reports can be collated by
geography.
A ull adoption o the lessons o \ikipedia would gie PC users the
opportunity to hae some ownership, some shared stake, in the process o
ealuating code, especially because they hae a stake in getting it right or their

own machines. Sharing useul data rom their PCs is one step, but this may
work best when the data goes to an entity committed to the public interest o
soling PC security problems and willing to share that data with others. 1he
notion o a ciic institution here does not necessarily mean cumbersome
goernance structures and ormal lines o authority so much as it means a sense
o shared responsibility and participation. 1hink o the olunteer ire
department or neighborhood watch: \hile not eeryone is able to ight ires or
is interested in watching, a critical mass o people are prepared to contribute,
and such contributions are known to the community more broadly.
1he success o tools drawing on group generatiity depends on participation,
which helps establish the legitimacy o the project both to those participating
and those not. Internet users might see themseles only as consumers whose
purchasing decisions add up to a market orce, but, with the right tools, users
can also see themseles as participants in the shaping o generatie space-as
netizens.
Along with netizens, hardware and sotware makers could also get inoled.
OS makers could be asked or required to proide basic tools o transparency
that empower users to understand exactly what their machines are doing. 1hese
need not be as sophisticated as lerdict. 1hey could proide basic inormation
on what data is going in and out o the box and to whom. Insisting on getting
better inormation to users could be as important as proiding a speedometer or
uel gauge on an automobile-een i users do not think they need one.
Internet Serice Proiders ,ISPs, can also reasonably be asked or required to
help. 1hus ar, ISPs hae been on the sidelines regarding network security.
1he justiication is that the Internet was rightly designed to be a dumb network,
with most o its eatures and complications pushed to the endpoints. 1he
Internet`s engineers embraced the simplicity o the end-to-end principle or
good reasons. It makes the network more lexible, and it puts designers in a
mindset o making the system work rather than designing against eery possible
thing that could go wrong. Since this early architectural decision, keep the
Internet ree` adocates hae adanced the notion o end-to-end neutrality as
an ethical ideal, one that leaes the Internet without iltering by any o its
intermediaries, routing packets o inormation between sender and recipient
without anyone looking along the way to see what they contain. Cyberlaw
scholars hae taken up end-to-end as a battle cry or Internet reedom, inoking
it to buttress arguments about the ideological impropriety o iltering Internet
traic or aoring some types or sources o traic oer others.
Lnd-to-end neutrality has indeed been a crucial touchstone or Internet
deelopment. But it has limits. Lnd-to-end design preseres users` reedom
only because the users can conigure their own machines howeer they like.
But this depends on the increasingly unreliable presumption that whoeer runs

a machine at a gien network endpoint can readily choose how the machine
should work. Consider that in response to a network teeming with iruses and
spam, network engineers recommend more bandwidth ,so the transmission o
deadweights` like iruses and spam does not slow down the much smaller
proportion o legitimate mail being carried by the network, and better
protection at user endpoints. But users are not well positioned to painstakingly
maintain their machines against attack, and intentional inaction at the network
leel may be sel-deeating, because consumers may demand locked-down
endpoint enironments that promise security and stability with minimum user
upkeep.
Strict loyalty to end-to-end neutrality should gie way to a new principle asking
that any modiications to the Internet`s design or the behaior o ISPs be made
in such a way that they will do the least harm to generatie possibilities. 1hus, it
may be preerable in the medium-term to screen-out iruses through ISP-
operated network gateways rather than through constantly updated PCs. 1o be
sure, such network screening theoretically opens the door to undesirable
iltering. But we need to balance this speculatie risk against the growing threat
to generatiity. ISPs are in a good position to help in a way that alls short o
undesirable perect enorcement acilitated through endpoint lockdown, by
proiding a stopgap while we deelop the kinds o community-based tools that
can promote salutary endpoint screening.
Len search engines can help create a community process that has impact. In
2006, in cooperation with the larard and Oxord StopBadware initiatie,
Google began automatically identiying \eb sites that had malicious code
hidden in them, ready to inect browsers. Some o these sites were set up or
the purpose o spreading iruses, but many more were otherwise-legitimate
\eb sites that had been hacked. lor example, isitors to chuckroast.com can
browse leece jackets and other oerings and place and pay or orders.
loweer, Google ound that hackers had subtly changed the chuckroast.com
code: 1he basic unctionalities were untouched, but code injected on the home
page would inect many isitors` browsers. Google tagged the problem, and
appended to the Google search result: \arning: 1his site may harm your
computer.` 1hose who clicked on the results link anyway would get an
additional warning rom Google and the suggestion to isit StopBadware or
pick another page.
1he site`s traic plummeted, and the owner ,along with the thousands o others
whose sites were listed, was understandably anxious to ix it. But cleaning a
hacked site takes more than an amateur \eb designer. Requests or specialist
reiew inundated StopBadware researchers. Until StopBadware could check
each site and eriy it had been cleaned o bad code, the warning pages stayed
up. Prior to the Google,StopBadware project, no one took responsibility or
this kind o security. Ad hoc alerts to the hacked sites` webmasters-and their

ISPs-garnered little reaction. 1he sites were ulilling their intended purposes
een as they were spreading iruses to isitors. \ith Google,StopBadware,
\eb site owners hae experienced a major shit in incenties or keeping their
sites clean.
1he result is perhaps more powerul than a law that would hae directly
regulated them, and it could in turn generate a market or irms that help
alidate, clean, and secure \eb sites. Still, the justice o Google,StopBadware
and similar eorts remains rough, and market orces alone might not direct the
desirable leel o attention to those wrongly labeled as people or \eb sites to be
aoided, or properly labeled but with no place to seek help.
1he touchstone or judging such eorts is whether they relect the generatie
principle: Do the solutions arise rom and reinorce a system o
experimentation Are the users o the system able, so ar as they are interested,
to ind out how the resources they control-such as a PC-are participating in
the enironment Done well, these interentions can encourage een casual
users to hae some part in directing what their machines will do, while securing
those users` machines against outsiders who hae not been gien permission by
the users to make use o them. Automatic accessibility by outsiders-whether
by endors, malware authors, or goernments-can deprie a system o its
generatie character as its users are limited in their own control.
Data Portability
1he generatie Internet was ounded and cultiated by people and institutions
acting outside traditional markets, and later carried orward by commercial
orces. Its success requires an ongoing blend o expertise and contribution
rom multiple models and motiations. Ultimately, a moe by the law to
allocate responsibility to commercial technology players in a position to help
but without economic incentie to do so, and to those among us, commercially
inclined or not, who step orward to sole the pressing problems that elude
simpler solutions may also be in order. low can the law be shaped i one wants
to reconcile generatie experimentation with other policy goals beyond
continued technical stability 1he next ew proposals are ocused on this
question about the constructie role o law.
One important step is making locked-down appliances and \eb 2.0 sotware-
as-a-serice more palatable. Ater all, they are here to stay, een i the PC and
Internet are saed. 1he crucial issue here is that a moe to tethered appliances
and \eb serices means that more and more o our experiences in the
inormation space will be contingent: A serice or product we use at one
moment could act completely dierently the next, since it can be so quickly
reprogrammed by the proider without our assent. Lach time we power up a
mobile phone, ideo game console, or BlackBerry, it might hae gained some

eatures and lost others. Lach time we isit a \eb site oering an ongoing
serice like e-mail access or photo storage, the same is true.
As arious serices and applications become more sel-contained within
particular deices, there is a minor interention the law could make to aoid
undue lock-in. Online consumer protection law has included attention to
priacy policies. A \eb site without a priacy policy, or one that does not lie
up to whateer policy it posts, is open to charges o unair or deceptie trade
practices. Similarly, makers o tethered appliances and \eb sites keeping
customer data ought to be asked to oer portability policies. 1hese policies
would declare whether users will be allowed to extract their data should they
wish to moe their actiities rom one appliance or \eb site to another. In
some cases, the law could create a right o data portability, in addition to merely
insisting on a clear statement o a site`s policies.
A requirement o data portability is a generatie insurance policy applying to
indiidual data whereer it might be stored. And the requirement need not be
onerous. It could apply only to uniquely proided personal data such as photos
and documents, and mandate only that such data ought to readily be extractable
by the user in some standardized orm. Maintaining data portability will help
people pass back and orth between the generatie and the non-generatie, and,
by permitting third-party backup, it will also help preent a situation in which a
non-generatie serice suddenly goes oline, with no recourse or those who
hae used the serice to store their data.
Appliance Neutrality
Reasonable people disagree on the alue o deining and legally mandating
network neutrality. But i there is a present worldwide threat to neutrality in the
moement o bits, it comes rom enhancements to traditional and emerging
appliancized` serices like Google mash-ups and lacebook apps, in which the
serice proider can be pressured to modiy or kill others` applications on the
ly. Surprisingly, parties to the network neutrality debate-who hae ocused
on ISPs-hae yet to weigh in on this phenomenon.
In the late 1990`s, Microsot was ound to possess a monopoly in the market or
PC operating systems.
1
Indeed, it was ound to be abusing that monopoly to
aor its own applications-such as its Internet Lxplorer browser-oer third-
party sotware, against the wishes o PC makers who wanted to sell their
hardware with \indows preinstalled but adjusted to suit the makers` tastes.
Microsot was orced by the law to meet ongoing requirements to maintain a

1
United States . Microsot Corp., 84 l. Supp. 2d 9, 19 ,D.D.C. 1999,.

leel playing ield between third-party sotware and its own by allowing third-
party sotware to be pre-installed on new \indows computers.
\e hae not seen the same requirements arising or appliances that do not
allow, or strictly control, the ability o third parties to contribute rom the start.
So long as the market`s aorite ideo game console maker neer opens the
door to generatie third-party code, it is hard to see how the irm could be
ound to be iolating competition law. A manuacturer is entitled to make an
appliance and to try to bolt down its inner workings so that they cannot be
modiied by others. So when should we consider network neutrality-style
mandates or appliancized systems 1he answer lies in that subset o
appliancized systems that seeks to gain the generatie beneits o third-party
contribution at one point in time while resering the right to exclude it later.
1he common law recognizes ested expectations. lor example, the law o
aderse possession dictates that people who openly occupy another`s priate
property without the owner`s explicit objection ,or, or that matter, permission,
can, ater a lengthy period o time, come to legitimately acquire it. More
commonly, property law can ind prescriptie easements-rights-o-way across
territory that deelop by orce o habit-i the owner o the territory ails to
object in a timely ashion as people go back and orth across it. 1hese and
related doctrines point to a deeply held norm: Certain consistent behaiors can
gie rise to obligations, sometimes despite ine print that tries to preent those
obligations rom coming about.
Applied to the idea o application neutrality, this norm o protecting settled
expectations might suggest the ollowing: I Microsot wants to make the Xbox
a general purpose deice but still not open to third-party improement, no
regulation should preent it. But i Microsot does welcome third-party
contribution, it should not be able to subsequently impose barriers to outside
sotware continuing to work. Such behaior is a bait-and-switch that is not easy
or the market to anticipate and that stands to allow a platorm maker to exploit
habits o generatiity to reach a certain plateau, dominate the market, and then
make the result proprietary-exactly what the Microsot \eb browser case
rightly was brought to preent.
1he ree sotware moement has produced some great works,
but under preailing copyright law een the slightest bit o
poison,` in the orm o code rom a proprietary source, could
amount to legal liability or anyone who copies or een uses
the sotware.
Generative Software
At the code layer, it is not easy or the law to maintain neutrality between the
two models o sotware production that hae emerged with the Net: Proprietary

sotware whose source code recipe is nearly always hidden, and ree sotware-
ree not in terms o the price, but the openness o its code to public reiew and
modiication. 1he ree sotware moement has produced some great works,
but under preailing copyright law een the slightest bit o poison,` in the
orm o code rom a proprietary source, could amount to legal liability or
anyone who copies or een uses the sotware. 1hese standards threaten the
long-term lourishing o the ree sotware moement: 1he risks are more
burdensome than need be.
But there are some changes to the law that would help. 1he kind o law that
shields \ikipedia and \eb site hosting companies rom liability or
unauthorized copyrighted material contributed by outsiders, at least so long as
the organization acts expeditiously to remoe inringing material once it is
notiied, ought to be extended to the production o code itsel. Code that
incorporates inringing material ought not be gien a ree pass, but those who
hae promulgated it without knowledge o the inringement would hae a
chance to repair the code or cease copying it beore becoming liable.
Modest changes in patent law could help as well. I those who see alue in
sotware patents are correct, inringement is rampant. And to those who think
patents chill innoation, the present regime needs reorm. 1o be sure, amateurs
who do not hae houses to lose to litigation can still contribute to ree sotware
projects-they are judgment proo. Others can contribute anonymously,
eading any claims o patent inringement since they simply cannot be ound.
But this turns coding into a gray market actiity, eliminating what otherwise
could be a thriing middle class o contributing irms should patent warare
ratchet into high gear.
1he law can help leel the playing ield. lor patent inringement in the United
States, the statute o limitations is six years, or ciil copyright inringement it is
three. Unortunately, this limit has little meaning or computer code because
the statute o limitations starts rom the time o the last inringement. Lery
time someone copies ,or perhaps een runs, the code, the clock starts ticking
again on a claim o inringement. 1his should be changed. 1he statute o
limitations could be clariied or sotware, requiring that anyone who suspects
or should suspect his or her work is being inringed sue within, or instance, one
year o becoming aware o the suspect code. lor example, the acts o those
who contribute to ree sotware projects-namely, releasing their code into a
publicly accessible database like Sourcelorge-could be enough to start the
clock ticking on that statute o limitations. In the absence o such a rule,
lawyers who think their employers` proprietary interests hae been
compromised can wait to sue until a gien piece o code has become wildly
popular-essentially sandbagging the process in order to let damages rack up.

Generative Licenses
1here is a parallel to how we think about balancing generatie and sterile code
at the content layer: Legal scholars Lawrence Lessig and \ochai Benkler, as well
as others, hae stressed that een the most rudimentary mixing o cultural icons
and elements, including snippets o songs and ideo, can accrue thousands o
dollars in legal liability or copyright inringement without harming the market
or the original proprietary goods.
18
Benkler beliees that the explosion o
amateur creatiity online has occurred despite this system. 1he high costs o
copyright enorcement and the widespread aailability o tools to produce and
disseminate what he calls creatie cultural bricolage` currently allow or a
ariety o oices to be heard een when what they are saying is theoretically
sanctionable by ines up to >30,000 per copy made, >150,000 i the inringement
is done willully.`
19
As with code, the status quo shoehorns otherwise laudable
actiity into a sub-rosa gray zone.
As tethered appliances begin to take up more o the inormation space, making
inormation that much more regulable, we hae to guard against the possibility
that content produced by citizens who cannot easily clear permissions or all its
ingredients will be squeezed out. Len the gray zone will constrict.
! ! !
Regimes o legal liability can be helpul when there is a problem and no one has
taken ownership o it. No one ully owns today`s problems o copyright
inringement and deamation online, just as no one ully owns security problems
on the Net. But the solution is not to conscript intermediaries to become the
Net police.
Under preailing law, \ikipedia could get away with much less stringent
monitoring o its articles or plagiarized work, and it could leae plainly
deamatory material in an article but be shielded in the United States by the
Communications Decency Act proision exempting those hosting material rom
responsibility or what others hae proided. \et \ikipedia polices itsel
according to an ethical code that encourages contributors to do the right thing
rather than the required thing or the proitable thing.
1o harness \ikipedia`s ethical instinct across the layers o the generatie
Internet, we must igure out how to inspire people to act humanely in digital
enironments. 1his can be accomplished with tools-some discussed aboe,
others yet to be inented. lor the generatie Internet to come ully into its

18
LA\RLNCL LLSSIG, RLMIX: MAKING AR1 AND COMMLRCL 1lRIVL IN A l\BRID LCONOM\
,2008,, \OClAI BLNKLLR, 1lL \LAL1l Ol NL1\ORKS ,2006,.
19
\OClAI BLNKLLR, 1lL \LAL1l Ol NL1\ORKS 25 ,2006,.

own, it must allow us to exploit the connections we hae with each other. Such
tools allow us to express and lie our ciic instincts online, trusting that the
expression o our collectie character will be one at least as good as that
imposed by outside soereigns-soereigns who, ater all, are only people
themseles.
Our generatie technologies need technically skilled people o good will to keep
them going, and the ledgling generatie actiities-blogging, wikis, social
networks-need artistically and intellectually skilled people o goodwill to sere
as true alternaties to a centralized, industrialized inormation economy that
asks us to identiy only as consumers o meaning rather than as makers o it.
1he deciding actor in whether our current inrastructure can endure will be the
sum o the perceptions and actions o its users. 1raditional state soereigns,
pan-state organizations, and ormal multi-stakeholder regimes hae roles to
play. 1hey can reinorce conditions necessary or generatie blossoming, and
they can also step in when mere generosity o spirit cannot resole conlict. But
that generosity o spirit is a society`s crucial irst line o moderation.
Our ortuitous starting point is a generatie deice on a neutral Net in tens o
millions o hands. Against the trend o sterile deices and serices that will
replace the PC and Net stand new architectures like those o Boxee and
Android. 1o maintain that openness, the users o those deices must
experience the Net as something with which they identiy and belong. \e must
use the generatiity o the Net to engage a constituency that will protect and
nurture it.

A Portrait of the Internet
as a Young Man
By Ann Bartow
*

Introduction
1he core theory o Jonathan Zittrain`s 2008 book 1be vtvre of tbe vtervet-.va
or to to t is this: Good laws, norms, and code are needed to regulate the
Internet, to preent bad laws, norms, and code rom compromising its creatie
capabilities and ettering its ecund lexibility. A ar snarkier, i less alliteratie,
summary would be \e hae to regulate the Internet to presere its open,
unregulated nature.`
Zittrain uses brie, inormal accounts o past eents to build two main theories
that dominate the book. lirst, he claims that open access, which he calls
generatiity, is under threat by a trend toward closure, which he reers to as
tetheredness, which is counterproductiely aored by proprietary entities.
1hough consumers preer openness and the autonomy it coners, ew take
adantage o the opportunities it proides, and thereore underalue it and too
readily cede it in aor o the promise o security that tetheredness brings.
Second, he argues that i the Internet is to ind salation it will be by the grace
o true netizens,` olunteers acting collectiely in good aith to cultiate
positie social norms online.
One o the themes o the James Joyce noel irst published in 1916, . Portrait of
tbe .rti.t a. a Yovvg Mav
1
is the Irish quest or autonomous rule. Jonathan
Zittrain`s 1be vtvre of tbe vtervet-.va or to to t is similarly inused with
the author`s desire or principled, legitimate goernance-only o the place
called cyberspace, rather than the author`s meatspace homeland.
Portrait`s protagonist, Stephen Dedalus, internally deines himsel as an artist
through a nonlinear process o experiences and epiphanies. le consciously
decides that it should be his mission to proide a oice or his amily, riends,
and community through his writing. 1hough Dedalus opts out o the

Proessor o Law, Uniersity o South Carolina School o Law. 1his essay was adapted
rom . Portrait of tbe vtervet a. a Yovvg Mav, 108 MICl. L. RLV. 109 ,2010,, aailable at
#$$%&''((()24+#4K1051(B,J4,()/BK'1B$4+5,6'1L%/B$B14$L/3L$#,L40$,B0,$L16L1L7/.0KL
210. 1he author dedicates this essay to her son Casey, and to the memory o C. Ldwin
Baker.
1
JAMLS JO\CL, A POR1RAI1 Ol 1lL AR1IS1 AS A \OUNG MAN ,1916,.

traditional orms o participation in society, he enisions his writing as a way to
productiely inluence society. Jonathan Zittrain charts the deelopment o the
Internet as a nonlinear process wrought by both conscious hard work and
sweeping serendipity. le also stries to proide a oice or technologically elite
Internet users, and to inluence the deelopment o online culture. le paints a
portrait o the uture Internet as chock ull o so many enigmas and puzzles that
it will keep the cyberlaw proessors busy or decades, een though according to
Zittrain, law as traditionally conceptualized will not be important.
In addition to inoking Joyce, I chose the title o this essay or its decisie
inocation o maleness. Lmbedded within Zittrain`s theories o generatiity,
there is also a perplexing gender story, in which men are ertile, crediting
themseles with helping to birth` the ield o cyberlaw,
2
and engaging in
stereotypically domestic pursuits such as baking` restrictions into gadgetry.
3

Non-generatie appliances are deemed sterile`
4
by Zittrain, sterility being the
conceptual opposite o generatiity. lis deployment o reproductie imagery is
odd. A metaphor equating an author`s creatie output to a child is oten
inoked in the context o copyright law by people arguing that authors should
hae extensie control oer the works they create.
5
Zittrain`s ariation
characterizes controlled technological innoations as unable to produce progeny
at all. 1he metaphor works better i tetheredness is instead enisaged as a orm
o birth control, preenting unwanted ospring only. Certainly the producers
o closed deices or locked sotware are able to proide, and generally
enthusiastic about proiding, new and improed ersions o their goods and
serices to paying customers.

2
ee, e.g., Lawrence Lessig, Amazon.com Customer Reiew o 1lL lU1URL Ol 1lL
IN1LRNL1-AND lO\ 1O S1OP I1, C,bertar 2.0,
#$$%&''((()121M/0)+/2'B,J4,('PADAPCAQRDSTUV'B,3W+28+B8B-%8%,B2 ,Dec. 4,
2008, ,1he ield o cyberlaw, or the law o the Internet-a fieta betea birtb . has suered
because people like me hae spent too much time cheerleading, and not enough time
ocusing the world on the real problems and threats that the Internet has produced.`,
,emphasis added,, .ee at.o Daniel J. Soloe, Prirac, ava Porer: Covvter Databa.e. ava Metabor.
for vforvatiov Prirac,, 53 S1AN. L. RLV. 1393, 1416-1 ,2001, ,noting that Roger Clarke is
credited with coining the term dataeillance`,. Roger Clarke published suggestions or
Internet regulations as early as 1988. ee Roger A. Clarke, vforvatiov 1ecbvotog, ava
Datareittavce, 31 COMM. ACM 498, 508-11 ,1988,.
3
JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1 2 ,2008,.
,Jobs was not shy about these restrictions baked into the iPhone.`,. |bereivafter ZI11RAIN,
1lL lU1URL Ol 1lL IN1LRNL1|.
4
ee, e.g., ia. at 2 ,1he iPhone is the opposite. It is sterile.`,, 3 ,Generatie tools are not
inherently better than their non-generatie ,sterile`, counterparts.`,.
5
See Malla Pollack, 1orara. a evivi.t 1beor, of tbe Pvbtic Dovaiv, or Re;ectivg tbe Cevaerea coe of
tbe |vitea tate. Co,rigbtabte ava Patevtabte vb;ect Matter, 12 \M. & MAR\ J. \OMLN & L. 603,
606-0 ,2006,, see \illiam Patry, Cevaer ava Co,rigbt, 1lL PA1R\ COP\RIGl1 BLOG, Jun. 20,
2008, #$$%&''(455412%1$B7)*5/K6%/$)+/2'9::?':='K,0-,BL10-L+/%7B4K#$)#$25.

Zittrain oers a well-executed collection o stories that are intended to anchor
his global theories about how the Internet should optimally unction, and how
two classes o Internet users should behae: 1he technologies should be
generatie, but also monitored to ensure that generatiity is not abused by either
the goernment or by scoundrels, elite Internet users with, as one might say
today, mad programming skilz` should be the superisors o the Internet,
scrutinizing new technological deelopments and establishing and modeling
productie social norms online, and aerage, non-technically proicient Internet
users should ollow these norms, and should not demand security measures that
unduly burden generatiity.
1he anecdotes are entertaining and educational, but they do not constructiely
cohere into an instruction manual on how to aoid a bad uture or people
whose interests may not be recognized or addressed by what is likely to be a
ery homogeneous group o elites manning ,and I do mean man-ning, gien the
masculine dominance o the ield, the irtual battlements they oluntarily design
to deend against online orces o eil. And some o the conclusions Zittrain
draws rom his stories are questionable. So, I question them below.
Generativity Versus Tetheredness
Is a False Binary
Pitting generatiity against tetheredness creates a alse binary that dries a lot o
Zittrain`s theorizing. 1he book was published in May o 2008, but its origins
can be ound in his earlier legal scholarship and mainstream media writings. In
2006, Jonathan Zittrain published an article entitled 1be Ceveratire vtervet.
6
In it,
he asserted the ollowing:
Cyberlaw`s challenge ought to be to ind ways o regulating-
though not necessarily through direct state action-which code
can and cannot be readily disseminated and run upon the
generatie grid o Internet and PCs, lest consumer sentiment
and preexisting regulatory pressures prematurely and tragically
terminate the grand experiment that is the Internet today.

Like the article, the book is useul or prooking thought and discussion, and it
teaches the reader many disparate acts about the eolution o a number o
dierent technologies. But it does not proide much direction or actiists,
especially not those who aor using laws to promote order. Zittrain has come

6
Jonathan L. Zittrain, 1be Ceveratire vtervet, 119 lARV. L. RLV. 194 ,2006, |hereinater
Zittrain, 1be Ceveratire vtervet|.
a. at 199.

to bury cyberspace law as promulgated by goernments, not to praise it.
Cyberlaw` as redeined by Zittrain is no longer the science o adapting existing
real-space legal constructs to the online enironment. Instead it is a collection
o best practices chosen by people with the technological proiciency to impose
them, top down, on the ignorant olks who are selishly drien by their shallow
consumer sentiments ,ri., a desire or simplicity and security oer openness and
generatiity,.
An abstract or the book, eatured at its dedicated website, states:
1he Internet`s current trajectory is one o lost opportunity. Its
salation, Zittrain argues, lies in the hands o its millions o
users. Drawing on generatie technologies like \ikipedia that
hae so ar suried their own successes, this book shows how
to deelop new technologies and social structures that allow
users to work creatiely and collaboratiely, participate in
solutions, and become true netizens.`
8

I will bluntly state ,splitting an ininitie in the process, that I did not learn how
to deelop new technologies or new social structures rom reading this book. It
coninced me that new technologies and new social structures could contribute
productiely to the Internet i they deelop appropriately, but Zittrain does not
proide road maps or an instruction manual or deeloping them. le calls or
|c|iic technologies |that| seek to integrate a respect or indiidual reedom and
action with the power o cooperation,`

but doesn`t paint a clear picture o
which precise qualities these technologies or social structures would hae,
beyond cultiating generatiity.
9

Zittrain relentlessly inorms the reader that generatiity is a ery good thing-
except when it is abused by maleactors. But what, exactly, is generatiity
Zittrain inokes the terms generatie, non-generatie, and generatiity
constantly throughout the book ,oer 500 times,, but the deinition o
generatie doesn`t remain constant. Sometimes it means creatie or innoatie,
while other times it connotes openness, accessibility, or reedom.
10

8
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3.
9
Jonathan Zittrain, or to Cet !bat !e .tt !avt, CA1O UNBOUND, May 6, 2009,
#$$%&''((()+1$/L.0*/.0-)/BK'9::>':;':='X/01$#10LM4$$B140'#/(L$/LK,$L(#1$L(,L
155L(10$'.
10
Covare ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 84 ,Generatie systems
allow users at large to try their hands at implementing and distributing new uses, and to ill a
crucial gap when innoation is undertaken only in a proit-making model .`,, ritb ia. at 113
,|1|he PC telephone program Skype is not amenable to third-party changes and is tethered
to Skype or its updates. Skype`s distribution partner in China has agreed to censor words


Zittrain had written preiously that Generatiity denotes a technology`s oerall
capacity to produce unprompted change drien by large, aried, and
uncoordinated audiences.`
11
Similarly, in the book he says, Ceveratirit, i. a
.,.tev`. caacit, to roavce vvavticiatea cbavge tbrovgb vvfitterea covtribvtiov. frov broaa
ava rariea avaievce..`
12
le lists ie elements o generatiity:
,1, how extensiely a system or technology leerages a set o
possible tasks, ,2, how well it can be adapted to a range o
tasks, ,3, how easily new contributors can master it, ,4, how
accessible it is to those ready and able to build on it, and ,5,
how transerable any changes are to others- including ,and
perhaps especially, non-experts.
13

Generatie also seems to mean idiot-resistant. In his article 1be Ceveratire
vtervet he explains that PCs are highly adaptable machines that are connected to
a network with little centralized control, resulting in a grid that is nearly
completely open to the creation and rapid distribution o the innoations o
technology-say users to a mass audience that can enjoy those innoations
without haing to know how they work.`
14
In the book, he makes the same
point repeatedly-that most mainstream` or rank-and-ile` computer users
are either passie beneiciaries or ictims o generatiity, rather than generatie
actors.
15
1here is a highly inluential generatie class o indiiduals who use
generatiity in socially productie ways. 1here is a nearious group o
reprobates who abuse generatiity to create online haoc. And then there are
the rest o the people online, sending and receiing emails, reading and writing
blogs, participating on social-networking sites, renewing antiirus subscriptions,
banking, shopping, and reading newspapers online. 1hese users are blithely
unaware o the generatiity that proided this ast electronic bounty and
complacently beliee that, as long as they continue to pay an Internet serice
proider ,ISP`, or Internet access, its deliery will remain relatiely smooth

like lalun Gong` and Dalai Lama` in its text messaging or the Chinese ersion o the
program. Other serices that are not generatie at the technical layer hae been similarly
modiied .`,.
11
Zittrain, 1be Ceveratire vtervet, .vra note 6, at 1980.
12
JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1, .vra note 3
at 0 ,emphasis in original,.
13
a. p. 1.
14
Zittrain, 1be Ceveratire vtervet, .vra note 6.
15
ee, e.g., ia. at 3, .ee at.o ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 4, 8, 43,
44-45, 51, 56, 59, 8, 100, 102, 130, 151-52, 155, 59-60, 198, 243, 245.

and uninterrupted. \hen they call or more security or electronic deices, they
themseles are the damage` that generatiity has to route around.`
16

1he anti-generatie concept o tetheredness also does some deinitional shape-
shiting throughout the tome. Sometimes it means unmodiiable, while other
times it means controlled by proprietary entities, who may or may not acilitate,
or een tolerate, alterations o their wares by end users. According to Zittrain,
the dangers o tethers are twoold: Priate companies can regulate how
consumers use their products, and serices and goernments can use them to
censor or spy on their citizens.
1

1ethers can be good things i you are a mountain climber, or i you don`t want
your horse to run o without you. And ar more pertinently, tethers acilitate
sotware updating or law-ixing and hole-patching purposes. Untethered
sotware would require manual updates, a labor-intensie prospect that would
require a degree o technical proiciency that many Internet users may lack.
low many people are prepared to gie up the adantages o tetheredness in the
interest o presering generatiity is unclear. \ithout tethered appliances, the
unctionality o the Internet will be compromised. 1ry using a program that is
no longer updated or supported by its endor. Its obsolescence may render it
untethered, but unless you hae some pretty good programming chops, its
useulness will decline rapidly. Zittrain ears people will exchange generatiity
or security in binary ashion, but the relationship between tetheredness and
conenience needs to be taken into account, as these ariables will also aect
consumer preerences and behaiors.
1he undamental security most people seek is probably operability. Any threat
to sericeability, whether rom too much generatiity or too many tethers, will
prooke a call or action rom users. I couldn`t hae accessed the downloadable
ersion o Zittrain`s book without a host o tethered utilities, including my
computer`s operating system, my Internet browser, and Adobe Acrobat, which
all update automatically with great requency, as I consented to allow them to
do when I agreed to the terms o use laid out in the associatie end user license
agreements ,LULAs`,. 1he same with my printer sotware, my antiirus
program, my online media players, the online games I play, and eery other
Internet-related utility I use. In a sense, this proes Zittrain`s assertion that we
hae ceded control oer the mechanisms o online interace to electronic leash-

16
1his is a sideways reerence to the John Gilmore quote, 1he Net interprets censorship as
damage and routes around it.` ee Philip Llmer-De\itt, ir.t âtiov iv C,ber.ace, 1IML,
Dec. 6, 1993, at 62, 64, araitabte at
#$$%&''((()$42,)+/2'$42,'21K1M40,'1B$4+5,':Y>ACAY>C>C=?Y::)#$25.
1
ee, e.g., ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 56-5, 113 ,discussing
Skype,, 109-10, 113 ,discussing OnStar,, 113 ,discussing China`s use o Google.cn,, 210-14
,discussing mobile phones,.

!"#$%"&' )*+,&)-. /0)1 2# 3,* 2,4# )2# )"3"&' ,- !#$$ ,- 35)"4,)"5& !+5&'. 6
-0-7#8) 35-) 59 0- %#9#++#% )5 )#)2#+"&' 8533#+8",$ #&)#+7+"-#- 4#+* #,+$* "& )2#
#45$0)"5& 59 )2# 3,"&-)+#,3 6&)#+&#)1 +,)2#+ )2,& +#8#&)$*. :"))+,"& +#9#+#&8#-
7"5&##+"&' 6;<- =5370;#+4# ,&% >?@ ,- 7+57+"#),+* -#+4"8#- )2,) !#+#
54#+!2#$3#% A* )2# '#&#+,)"4")* 59 <=- ,&% )2# 6&)#+&#).
BC
D* "&")",$
&5&,8,%#3"8 #E7#+"#&8#- !")2 )2# 6&)#+&#) 8537+"-#% !,")"&' ,&E"50-$* 95+
=5370;#+4# ,&% )2#& >?@ )5 9"&"-2 "&-),$$"&' 07%,)#- !2#& 6 &##%#% )5 82#8F
3* #G3,"$1 ,&% 6 2,% )5 7,* 95+ 3* 6&)#+&#) )"3# A* )2# 3"&0)#. H2"&'- 5&$*
!#&) %5!&2"$$ !2#& >?@ !#&) )5 ,& I,$$ *50 8,& #,)J 7,*3#&) -)+08)0+#1
7+54"%"&' 0&$"3")#% 6&)#+&#) 95+ , 9"E#% 35&)2$* 9##. K-#+- -0+'#% A0) >?@L-
8,7,8")* 850$% &5) 3##) )2# %#3,&%.
BM
K-#+- %"%&L) !,&) -#80+")*1 )2#* !,&)#%
7#+95+3,&8#. H#)2#+#%&#--1 5+ -53#)2"&' -"3"$,+1 3,* 2,4# A##& $"&F#% "&
-53# !,* )5 >?@L- %"99"80$)"#- 3##)"&' ")- 80-)53#+-L %#3,&%1 A0) 54#+-#$$"&'
,&% "&-099"8"#&) -#+4#+ 8,7,8")* !#+# )2# )+0# 80$7+")- "& )#+3- 59 "&2"A")"&'
57#+,A"$")*. 6& ,%%")"5&1 "9 :"))+,"& "- 85++#8) )2,) =5370;#+4# ,&% >?@
#E#37$"9* )2# #4"$- 59 )#)2#+"&'1 ")L- 7+#))* 8$#,+ )2# 3,+F#) 70&"-2#% )25-#
#&)")"#- 7+#))* 2,+-2$* !")250) 6&)#+&#) '54#+&,&8#G-)*$# "&)#+4#&)"5&-.
;59)!,+# ,&% #$#8)+5&"8 %#4"8#- 8,& A# -"30$),&#50-$* '#&#+,)"4# ,&% )#)2#+#%.
>&% ") "- 0&9,"+ )5 8+")"8"N# 7#57$# !25 O0")# +#,-5&,A$* +#$* 5& )#)2#+#%&#-- )5
F##7 )2#"+ 85370)#+- ,&% #$#8)+5&"8 #O0"73#&) 07%,)#% ,&% 90$$* 90&8)"5&,$.
D,&* ,4#+,'# 6&)#+&#) 0-#+- 3"'2) $"F# 35+# )+,&-7,+#&8* ,A50) )2# &,)0+# ,&%
#E)#&) 59 )2# )#)2#+- )2,) 85&&#8) )2#"+ 85370)#+- )5 $,+'# 30$)"&,)"5&,$
85+75+,)"5&-1 A0) -25+) 59 2,4"&' ,8)0,$ $,!- )2,) +#O0"+# +#$#4,&) %"-8$5-0+#-1
)2"- 85&-03#+ %#-"+# "- 0&$"F#$* )5 A# 3#). P5+ )2#31 '#&#+,)"4")* "- 0&$"F#$* )5
A# 2#$790$ 5+ #&$"'2)#&"&'1 ,- :"))+,"& 85++#8)$* &5)#-1 A#8,0-# )2#* ,+# &5)
-F"$$#% #&50'2 )5 ),F# ,%4,&),'# 59 "). 6& )2# ,A-#&8# 59 2#$790$ $,!-1 )2#* ,+#
,) )2# 3#+8* 59 A0-"&#-- 35%#$-.
Generativity: The Good,
the Bad & the Ugly
:"))+,"&L- -)5+"#- ,+# "&)#&%#% )5 -25! )2,) '#&#+,)"4# )#82&5$5'"#- ,+# A#))#+
)2,& )#)2#+#% 5&#-. /0) ,&5)2#+ -)+,&% 59 2"- &,++,)"4# "$$0-)+,)#- )2,)
'#&#+,)"4")* 8,& A# 0-#% %#-)+08)"4#$*1 )5 -0775+) )2# 85&)#&)"5& )2,) ") 8,&&5)

BC
H2# <= +#45$0)"5& !,- $,0&82#% !")2 <=- )2,) "&4")#% "&&54,)"5& A* 5)2#+-. ;5 )55 !")2 )2#
6&)#+&#). /5)2 !#+# '#&#+,)"4#Q )2#* !#+# %#-"'&#% )5 ,88#7) ,&* 85&)+"A0)"5& )2,) 95$$5!#%
, A,-"8 -#) 59 +0$#- R#")2#+ 85%#% 95+ , 7,+)"80$,+ 57#+,)"&' -*-)#31 5+ +#-7#8)"&' )2#
7+5)585$- 59 )2# 6&)#+&#)S. /5)2 54#+!2#$3#% )2#"+ +#-7#8)"4# 7+57+"#),+*1 &5&G'#&#+,)"4#
8537#)")5+-1 -082 ,- )2# 3,F#+- 59 -),&%G,$5&# !5+% 7+58#--5+- ,&% 7+57+"#),+* 5&$"&#
-#+4"8#- $"F# =5370;#+4# ,&% >?@. :6HHT>6U1 HVW PKHKTW ?P HVW 6UHWTUWH1 !"#$% &5)#
X1 ,) YXZY[.
BM
&''( '*+*1 H"35)2* =. /,+3,&&1 ,"-+' ./ $"0' .12! 3''4 /5 678 !'$92:'1 =\/WTH>@]1 ?8). Y^1 BMM_1
!""#$%%&&&'()*+,"-./'(01%234567*'!"1.

be unettered. At its worst, he warns, generatiity will enable bad actors to
exploit tethers or nearious purposes, while tethers will simultaneously restrain
positie generatie responses to these challenges. lis accounts o degenerate
generatiity rest uneasily with his exhortation that acilitating generatiity should
be the guiding principle o Internet goernance.
le also suggests deploying the generatie principle to determine whether and
when it makes sense to iolate the end-to-end principle` in the context o
debates about network neutrality.
20
And the quantum o generatiity that is
promoted becomes the measure or assessing the legitimacy and eectieness o
what he characterizes as the intrusions o cyberlaw. le writes:
1he touchstone or judging such eorts should be according
to the generatie principle: do the solutions encourage a system
o experimentation Are the users o the system able, so ar as
they are interested, to ind out how the resources they
control-such as a PC-are participating in the eniron-
ment
21

lostering generatiity thus becomes the Prime Directie o Internet
goernance.
22
But there are problems he raises elsewhere in the book that
generatiity may not address, or may in act exacerbate. lor example, Zittrain
reerences OnStar a number o times, warning that it can be used by law
enorcement or sureillance purposes because it is tethered, and can be
accessed remotely.
23
Putting aside questions about whether OnStar is accurately
described as part o the Internet, one wonders o what practical use OnStar
would be to its clients i it wasn`t tethered. OnStar seems to be a serice that
caters to people who want higher leels o proactie inormation and security
when they are driing than the combination o a GPS unit and mobile phone
can proide. OnStar customers don`t want generatiity, they want someone to
call the police and an ambulance or tow truck i they hae an accident so they

20
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 185.
21
a. at 13.
22
1he Prime Directie is a plot deice cooked up by a patently optimistic 1V writer ,either
1re/ producer Gene L. Coon or writer 1heodore Sturgeon, depending on who you ask, in
the mid-1960s. It`s a reshmen-year philosophy student`s reaction to the Cold \ar, when
America and the Soiets were playing out their hostilities by proxy third-world conlicts.
Lectiely, they were interering in the deelopment` o underpriileged countries to
urther their own ends with some awul immediate and long-term results. In Roddenberry`s
ision, humanity had eoled beyond such puppeteering and become an adanced` race.`
ee Jay Garmon, !b, tar 1re/`. Prive Directire i. .tvia`, 1LClRLPUBLIC.COM, leb. 12, 200,
#$$%&''*5/K6)$,+#B,%.*54+)+/2)+/2'K,,<,0-'O%W;DD.
23
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 109-10, 113, 11-18, 18.

don`t hae to, or to track down the location o their ehicle i it is stolen.
Security means more to them than priacy, and i they don`t consciously realize
they are exchanging one or the other when they sign up with OnStar, it seems
to me the best solution is to require OnStar to inorm them o this trade-o in
simple and unambiguous terms. 1he law could also require OnStar to proide
urther inormation, perhaps including a primer on the search and seizure
jurisprudence o lourth Amendment law. Making OnStar generatie, so that
priate citizens can readily discern incursions by goernment actors, would not
gie OnStar customers any more o what they appear to want-a high leel o
security oertly linked to constant, dedicated superision. Lnhanced
generatiity might also proide opportunities or priate spying or intentional
serice disruptions by the ery illains Zittrain spills so much ink warning
against.
Many o his examples o useul online-goernance initiaties rely on extensie
amounts o olunteer labor. But the important technological innoations
related to the Internet were motiated by some orm o sel-interest. 1he U.S.
Deense Department deeloped the Internet as a decentralized communications
system that would be diicult to disrupt during wartime.
24
1im Berners-Lee
inented the \orld \ide \eb as a way to acilitate communications with other
physicists.
25
Pornographers hae long used spam, browser hijacking, and
search-engine manipulation to reach the eyeballs o potential customers.
26
All
may hae relied on generatiity ,though one might question how open and
accessible the Deense Department was, but not all are socially beneicial.
2

Sometimes Internet users may donate their labor inoluntarily. 1heir online
actiities are harested and bundled into what Zittrain applauds as the mediated
wisdom o the masses. lor example, he notes as ollows:

24
ee Joseph D. Schleimer, Protectivg Co,rigbt. at tbe ac/bove eret of tbe vtervet, 15 UCLA
LN1. L. RLV. 139, 149 ,2008,, .ee at.o JANL1 ABBA1L, INVLN1ING 1lL IN1LRNL1 -41 ,1999,.
25
ABBA1L, .vra, .ee at.o Dick Kaser, 1be Cv, !bo Dia tbe !!! 1bivg at tbe Ptace !bere e Dia
t, INlO. 1ODA\, leb. 2004, at 30.
26
ee, e.g., Porvograber. Cav oot Yov !itb i1ecb, lIL1LRGUIDL.COM,
#$$%&''((()345$,BK.4-,)+/2'%/B063//5)#$2 ,setting orth arious ways in which
pornographers use technology to ool children, ,last isited Oct 21, 2009,, PL\ IN1LRNL1 &
AMLRICAN LIlL PROJLC1, SPAM IS S1AR1ING 1O lUR1 LMAIL ,2003,,
#$$%&''((()%,(40$,B0,$)/BK'NB,66LP,5,16,6'9::D'R%12L46L6$1B$40KL$/L#.B$L
,2145)16%H ,accounting or pornography-related spams` impact on email,.
2
ee geveratt, Ann Bartow, Porvograb,, Coerciov, ava Co,rigbt ar 2.0, 10 VAND. J. LN1. & 1LCl.
L. 99, 800 ,2008, ,Pornography is a dominant industrial orce that has drien the eolution
o the Internet.`,.

1he alue o aggregating data rom indiidual sources is well
known. \ochai Benkler approingly cites Google Pagerank
algorithms oer search engines whose results are auctioned,
because Google draws on the indiidual linking decisions o
millions o \eb sites to calculate how to rank its search results.
I more people are inking to a \eb site criticizing Barbie dolls
than to one selling them, the critical site will, all else equal,
appear higher in the rankings when a user searches or
Barbie.`
28

But all else is unlikely to be equal. Mattel can hire reputation-deense
companies like ReputationDeender
29
to bury the critical sites about Barbie
using search engine-optimization techniques and to surreptitiously edit
\ikipedia entries.
30
lor-proit entities don`t just want to spy on and control
their customers with tethers. 1hey also want to manipulate as much o the
Internet as possible to their beneit, and this logically includes taking steps to
highlight positie inormation and minimize the isibility o disparagement by
third parties.
Additionally, collectie actions by the online masses can be oppressie. I more
people link to websites gloriying sexual iolence against women than to
websites where women are treated as i they are ully human, those sites appear
higher in the rankings when a user searches or a wide ariety o things related
to sex. 1he same is potentially true or racist and homophobic sites and other
content that depict discrete groups in derogatory ways. In this way, negatie
stereotypes can be reinorced and spread irally.
31

linally, in the Google PageRank example, the power and input o the masses is
being harnessed, or proit, by a large corporation. Google is doubtlessly happy
to use generatie tools when they are eectie. But contrast the Google search

28
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3 at 160 ,ootnote omitted,.
29
ee ia. at 230 ,asserting that ReputationDeender uses moral suasion` as its primary
technique or manipulating search-engine results,. I oer a ery dierent perspectie on
this. ee Ann Bartow, vtervet Defavatiov a. Profit Cevter: 1be Movetiatiov of Ovtive ara..vevt,
32 lARV. J.L. & GLNDLR 383 ,2009,.
30
Zittrain himsel noted something similar, writing, I the \ikipedia entry on \al-Mart is
one o the irst hits in a search or the store, it will be important to \al-Mart to make sure
the entry is air-or een more than air, omitting true and releant acts that nonetheless
relect poorly on the company.` ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3
at 139.
31
ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3 at 14. Zittrain tacitly
acknowledges this: 1here are plenty o online serices whose choices can aect our lies.
lor example, Google`s choices about how to rank and calculate its search results can
determine which ideas hae prominence and which do not.`

engine with Google`s Gmail, and it becomes apparent that the same company
will keep a serice tethered and proprietary when doing so best suits its
purposes.
32

1he idiosyncratic online juggernaut that is \ikipedia, to which Zittrain deotes
irtually an entire chapter, also illustrates some o the downsides o excessie
generatiity.
33
\ikipedia is an online encyclopedia that, at least in theory,
anyone can edit. Zittrain is clearly enamored o it, writing, \ikipedia stands at
the apex o amateur endeaor: an undertaking done out o sheer interest in or
loe o a topic, built on collaboratie sotware that enables a breathtakingly
comprehensie result that is the sum o indiidual contributions, and one that is
extraordinarily trusting o them.`
34
Zittrain proides a lot o inormation about
\ikipedia, and the ast majority o it skews positie. le writes, \ikipedia has
charted a path rom crazy idea to stunning worldwide success`,
35
and
\ikipedia is the canonical bee that lies despite scientists` skepticism that the
aerodynamics add up`,
36
and asserts that the manner in which \ikipedia
operates is the essence o law.`
3
Perhaps echoing Zittrain`s enthusiasm, one
researcher determined \ikipedia has been cited in oer 400 U.S. court
opinions.
38

Among myriad other acts and anecdotes, Zittrain notes that \ikipedia co-
ounder Larry Sanger is controersial because possibly he is gien too much
credit or his limited contributions to \ikipedia.
39
le also notes that another
person inoled with \ikipedia, ormer \ikimedia loundation member Angela

32
ee geveratt, Paul Boutin, Reaa M, Mait, Ptea.e, SLA1L, Apr. 15, 2004,
#$$%&''651$,)260)+/2'4-'9:>?>@=, Deane, Critic. Retea.e tbe ovva. ov CMait,
GADGL1OPIA, Apr. 10, 2004, #$$%&''K1-K,$/%41)+/2'%/6$'99;@, Google \atch,
#$$%&''((()K//K5,L(1$+#)/BK'K2145)#$25, Brian Morrissey, .v art, oo/ at or Cvait
!or/., DMNL\S, Apr. 19, 2004, #$$%&''((()-20,(6)+/2'10L,1B57L5//<L1$L#/(L
K2145L(/B<6'1B$4+5,'?D>@=.
33
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, chapter six.
34
a. at 96.
35
a. at 136.
36
a. at 148.
3
a. at 144.
38
Lee l. Peoples, 1be Citatiov of !i/ieaia iv ]vaiciat Oiviov., 12 \ALL J.L. & 1LCl.
,orthcoming 2009,, araitabte at
#$$%&''%1%,B6)66B0)+/2'6/5D'%1%,B6)+32O1*6$B1+$84-WA9C9@DC.
39
ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 143 ,At times-they are
constantly in lux-\ikipedia`s articles about \ikipedia note that there is controersy oer
the co-ounder` label or Sanger.`,, .ee at.o ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra
note 3 at 142, 145.

Beesley Starling, unsuccessully ought to hae her \ikipedia entry deleted.
40

1hat a man who wants undesered credit and a woman who wants no attention
at all hae likely both been thwarted by \ikipedians is something Zittrain seems
to iew as a positie indicator. Angela Beesley Starling probably eels ery
dierently, especially i her reasons or wanting her \ikipedia entry deleted
included pressing personal saety concerns. 1he talk` page o her \ikipedia
biography quotes her as saying, I`m sick o this article being trolled. It`s ull o
lies and nonsense.`
41
1he orced publicity o \ikipedia entries is something all
women may encounter under \ikipedia`s system o sel-goernance that has
many indicia o the rule o law without heay reliance on outside authority or
boundary.`
42
Research suggests that women, though 51 o the population,
comprise a mere 13 o \ikipedia contributors,
43
or reasons that probably
hae to do with the culture o this entity, which women may experience more
negatiely than men do.
Certainly notable liing eminists hae been on the receiing end o a campaign
o nasty and untruthul edits to \ikipedia entries they would probably preer
not to hae. Many entries on eminism hae been written or edited by people
who are actiely hostile toward eminists, but they preail because they seem to
hae a lot o ree time and the ew eminists who enter the wikiray seem to get
drien out or edited into obliion. 1o take just one example, the entries about
Melissa larley,
44
Catharine MacKinnon,
45
and Sheila Jeries
46
hae all been

40
a. at 143.
41
ee .vgeta ee.te, tartivg 1at/age,
\IKIPLDIA,#$$%&'',0)(4<4%,-41)/BK'(4<4'Z15<&[0K,518I,,65,78R$1B540K ,last isited
Sept. 4, 2009, ,Angela Beesley has tried to hae her biography on \ikipedia deleted, saying
I`m sick o this article being trolled. It`s ull o lies and nonsense.` 1he Register and
\ikitruth claim that her objections are ironic in light o the generally liberal policy o
\ikipedia administrators to the accuracy and notability o biographies in \ikipedia o liing
people. Seth linkelstein, who tried to hae his own entry rom \ikipedia remoed, called it
a pretty stunning ote o no-conidence. Len at least some high-ups can`t eat the dog
ood.``, ,ootnotes omitted,.
42
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3 at 143.
43
ee, e.g., Andrew LaVallee, Ovt, 1 of !i/ieaia Covtribvtor. .re !ovev, tva, a,., \ALL S1.
J., Aug. 31, 2009, #$$%&''*5/K6)(6X)+/2'-4K4$6'9::>':?'DA'/057LADL/3L(4<4%,-41L
+/0$B4*.$/B6L1B,L(/2,0L6$.-7L6176, Jennier Van Groe, tva,: !ovev ava !i/ieaia Dov`t
Mi, MASlABLL, Sept. 1, 2009, #$$%&''216#1*5,)+/2'9::>':>':A'(/2,0L(4<4%,-41,
Cathy Daidson, !i/ieaia ava !ovev, lAS1AC, Sept. 2, 2009,
#$$%&''((()#16$1+)/BK'*5/K6'+1$#7L-1J4-6/0'(4<4%,-41L10-L(/2,0.
44
ee Meti..a arte,, \IKIPLDIA, #$$%&'',0)(4<4%,-41)/BK'(4<4'\,546618F1B5,7 ,last isited
July28, 2009,.
45
ee Catbarive MacKivvov, \IKIPLDIA,
#$$%&'',0)(4<4%,-41)/BK'(4<4']1$#1B40,8\1+^400/0 ,last isited July 28, 2009,.

heaily edited
4
by a rabid pornography proponent named Peter G. \erner
48

who sometimes also uses the pseudonym Iamcuriousblue.
49
Lach entry is the
irst result returned ater a Google search o their names. le has deleted or
attempted to hae deleted entries about other eminists.
50
le shows up under
one identity or another in irtually eery entry in which eminism is mentioned.
And he successully coninced the \ikipedia community to ban a eminist
actiist who igorously contested his edits.
51
Any group that is not well
represented within the \ikipedia editing community is likely to experience
similar marginalization.
Recently, \ikipedia announced that the entries o liing people will receie a
mandatory layer o intermediation. A new eature called lagged reisions` will
require that an experienced olunteer editor sign o on any changes beore they
become permanent and publicly accessible.
52
A New \ork 1imes report noted
that this would diide \ikipedia`s contributors into two classes-experienced,
trusted editors, and eeryone else-altering \ikipedia`s implicit notion that
eeryone has an equal right to edit entries.`
53
1his seems to be one realization
o what Zittrain broadly desires-control oer the ignorant wikimasses by a
designated elite. But the project became signiicantly less collaboratie and
open when this change was made.
\ikipedia entries are generated by a massie assemblage o olunteers with
unknown motiations and agendas. Group behaior is always unpredictable, a
act that Zittrain acknowledges but under-appreciates. One somewhat
organized assemblage that calls itsel Anonymous launches cyber-attacks that

46
ee beita ]effre,., \IKIPLDIA, #$$%&'',0)(4<4%,-41)/BK'(4<4'R#,4518T,33B,76 ,last isited
July 28, 2009,.
4
ee, e.g., Catbarive MacKivvov 1at/age, \IKIPLDIA,
#$$%&'',0)(4<4%,-41)/BK'(4<4'Z15<&]1$#1B40,8\1+^400/0 ,last isited July 28, 2009,.
48
ee Peter C !erver |.erage, \IKIPLDIA,
#$$%&'',0)(4<4%,-41)/BK'(4<4'_6,B&N,$,B8V8`,B0,B ,last isited July 28, 2009,.
49
ee avcvriov.btve |.erage, \IKIPLDIA,
#$$%&'',0)(4<4%,-41)/BK'(4<4'_6,B&a12+.B4/.6*5., ,last isited July 28, 2009,.
50
ee, e.g., .rticte. for aetetiov,Cber,t iva.e, eetboff, \IKIPLDIA,
#$$%&'',0)(4<4%,-41)/BK'('40-,H)%#%O$4$5,W`4<4%,-41&[B$4+5,683/B8-,5,$4/0']#,B75
8b40-6,78R,,5#/33c/5-4-WA;:AA:?A; ,last isited Sept. 25, 2009,, .ee at.o î//i Craft
1at/age, \IKIPLDIA, #$$%&'',0)(4<4%,-41)/BK'(4<4'Z15<&d4<<48]B13$ ,last isited Sept.
25, 2009,.
51
1elephone interiew with Nikki Crat, .ee at.o î//i Craft 1at/age, .vra ,containing
conersation in which user Iamcuriousblue discredits Nikki Crat`s \ikipedia article,.
52
Noam Cohen, !i/ieaia to ivit Cbavge. to .rticte. ov Peote, N.\. 1IMLS, Aug. 25, 2009, at B1.
53
a.

online norms do not seem to hae any cognizable role in addressing.
54
As with
\ikipedians, Anonymous is hostile to others and outsiders. One blogger noted:
Interestingly . Anon neer seems to take down the big sites.
\almart.com and the Pentagon are sae rom his attentions.
It`s not that Anon is a big an o \almart or the goernment.
It`s just so much easier to attack the ulnerable. Big business
and big goernment aren`t ulnerable on the Internet. 1hey
can aord not to be.
Small discussion boards and blogs, particularly ones that
adocate unpopular points o iew, are oten run by indiiduals
who put up their own unds, i they can scrape them together,
and who must be their own I1 departments. 1hey can`t aord
the type o security that requires the big bucks. And since they
hae jobs ,unlike Anon, apparently,, they hae to put their
desire to maintain an Internet presence in the balance with
supporting themseles and their amilies. \hen the crunch
comes and time pressures set in, it`s not the Internet presence
that wins out.
So the actions o these apolitical` hackers do hae a political
end: 1hey remoe unpopular, radical, ringe iewpoints rom

54
ee e.g., Shaun Daies, ô Cv..ivg` 1eev ace. êt ate Cavaigv, NINLMSN NL\S, Jan. 18,
2009, #$$%&''0,(6)040,260)+/2)1.'$,+#0/5/K7'C9:AA;'0/L+.6640KL$,,0L31+,6L0,$L
#1$,L+12%14K0 ,stating McKay latch`s No Cussing Club, which encourages teens to chill
on the proanity`, claims to hae oer 20,000 members worldwide. latch, a 15-year-old
rom South Pasadena in Caliornia, garnered wide media coerage or his anti-swearing
campaign, including an appearance on Dr Phil. But at the beginning o the year, latch`s
email inbox began clogging up with hate mail rom an unknown source. Pizza and porn
delieries became commonplace or his amily, who eentually called in the lBI ater
numerous receiing|sic| death threats and obscene phone calls. Anonymous appears to be
behind the attacks, with threads on sites such as 4chan.org and 11chan.org identiying their
members as the culprits. And the pain may not yet be oer or the latch amily-
Anonymous appears to be planning uture raids and has threatened to wipe this cancer |the
No Cussing Club| rom the ace o the internet`.|sic| In one 4chan thread, a number o
users boasted about sending bogus pizza delieries and een prostitutes to the latchs`
house, although it was impossible to eriy i these claims were genuine. 1he same thread
also contained a credit card number purported to be stolen rom latch`s ather, phone
numbers, the amily`s home address and latch`s instant messenger address.`,, .ee at.o ebiva
tbe aaae of tbe .vov,vov. ate Crov, RLLIGIOUS lRLLDOM \A1Cl, July 6, 2009,
#$$%&''((()B,54K4/.63B,,-/2(1$+#)/BK'2,-41L0,(6B//2'*,#40-L$#,L31+1-,L/3L
$#,eG9e?:e>]10/072/.6eG9e?:e>fL#1$,LKB/.%', .ee also .te !vori,
LNC\CLOPALDIA DRAMA1ICA, #$$%&'',0+7+5/%,-41-B121$4+1)+/2'[5,H8`./B4 ,last
isited July 28, 2009,.

the web. Big goernment doesn`t hae to eliminate the
subersie websites, Anon will do it.
55

1he actiities o Anonymous hae been characterized as domestic terrorism.
56

And Anonymous certainly takes adantage o generatie technologies, just as
\ikipedians with reprehensible agendas do. Zittrain asserts that bad actors like
Anonymous are driing the demand or increased security,
5
but he doesn`t
proide any targeted mechanisms or hindering them, or explain why increasing
security necessarily compromises productie generatiity.
The Zittrainnets Netizens:
Overlords of Good Faith
As with a James Joyce noel, there are a ariety o transactions that the careul
reader negotiates with the author. Lach section has to be read independently o
the others, because while it may cohere internally, it may not combine with
other delineated portions to paint a consistent picture o Zittrain`s preerred
uture or the Internet, which will hereater be called the Zittrainnet.`
Some o the recommendations he makes inite broad democratic participation
in Zittrainnet goernance, while other times he warns against it and suggests
ways to decrease the threats posed by outsiders-whether by endors, malware
authors, or goernments.`
58
One wonders how something as disaggregated as
the Internet can hae outsiders, until recognition dawns about what Zittrain is
truly suggesting, at least part o the time, in terms o who should control the
Internet to best ensure its eolution into the Zittrainnet: an elite circle o people
with computer skills and ree time who share his policy perspectie.
Technologists Rule
Zittrain doesn`t contemplate anyone` deeloping sericeable code. Zittrain`s
iew is that only a select ew can take productie adantage o generatiity, and
within this elite group are bad actors as well as good. le thinks that cyberlaw is
the appropriate mechanism to encourage positie uses o generatiity while

55
VeraCity, Dovivator 1evtacte., #$$%&''J,B1)(/B-%B,66)+/2'9::C':?'9@'-/2401$/BL
$,0$1+5,6' ,Aug. 24, 200,.
56
VA. lUSION C1R., VA. DLP`1 Ol S1A1L POLICL, 2009 VIRGINIA 1LRRORISM 1lRLA1
ASSLSSMLN1 48 ,2009,, araitabte at #$$%&''((()403/(1B6)+/2'J4BK4041L3.64/0L+,0$,BL
B,5,16,6#/2,KB/(0L$,BB/B462L-/+.2,0$'.
5
ee geveratt, ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at chapter 3. 1his is one
o the central claims o the book.
58

thwarting the troublesome ones, cyberlaw being computer-code construction
and norm entrepreneurship within Internet communities, as well as more
traditionally recognized modes o law ormation such as statutes and
regulations.
59
As ar as who exactly will diine good generatiity rom bad, and
wield the mighty sword o cyberlaw to deend the ormer and deeat the latter,
Zittrain is decidedly ague. In the Solutions` section o the tome Zittrain lists
two approaches that might sae the generatie spirit o the Net`:
1he irst is to reconigure and strengthen the Net`s
experimentalist architecture to make it it better with its now-
mainstream home. 1he second is to create and demonstrate
the tools and practices by which releant people and
institutions can help secure the Net themseles instead o
waiting or someone else to do it.
60

By releant people and institutions` Zittrain seems to mean technologically
skilled, Internet users o good will.
61
But as ar as who it is that will
reconigure and strengthen the Net`s experimentalist architecture` or who will
create and demonstrate the tools and practices` on behal o these releant
people and institutions ,shall we call them generatiators`,, Zittrain oers ew
speciics. le mentions uniersities generally,
62
and two organizations he is
ailiated with speciically, larard Uniersity`s Berkman Center ,where he is
one o 13 Directors-all male, o course
63
, and the Oxord Internet Institute
,where he is a Research Associate
64
,, which he describes as multidisciplinary
academic enterprises dedicated to charting the uture o the Net and improing
it.`
65
1hose who share his isions or the Zittrainnet are supposed to unction
as norm entrepreneurs, guiding lights or the undereducated, inadequately
skilled online masses to ollow, sheep-like.
Less-releant people are described as |r|ank-and-ile Internet users |who| enjoy
its beneits while seeing its operation as a mystery, something they could not

59
a. chapter 5.
60
a. at 152.
61
a. at 246.
62
a. at 198, 245.
63
ee Peote, Berkman Center or Internet and Society at larard Uniersity,
#$$%&''+7*,B)51()#1BJ1B-),-.'%,/%5,.
64
ee Peote, OXlORD IN1LRNL1 INS1I1U1L UNIVLRSI1\ Ol OXlORD,
#$$%&''((()/44)/H)1+).<'%,/%5,'O6$1$.6W+.BB,0$c$7%,Wc<,7(/B-6WM4$$B140.
65

possibly hope to aect.`
66
1hese ignorant non-generatiators righten Zittrain,
because when he ears that, a crisis comes, they will pressure the goernment to
enhance Internet security at the expense o Internet generatiity, out o short-
sighted, ill-inormed perceptions o their own sel-interest.
6
le knows better
than they do what`s best or them.
In a related article he published in egat .ffair. to promote the book, Zittrain
explains:
I the Internet does hae a September 11 moment, a scared
and rustrated public is apt to demand sweeping measures to
protect home and business computers-a metaphorical USA
Patriot Act or cyberspace. Politicians and endors will likely
hasten to respond to these pressures, and the end result will be
a radical change in the technology landscape. 1he biggest
casualty will likely be a undamental characteristic that has
made both the Internet and the PC such powerul phenomena:
their generatiity.`
68

Many o the stories Zittrain tells in the book are intended to persuade readers
that unless somebody does something, the Internet will do what the book`s
coer suggests: derail and drie oer a cli. But ater ominously warning his
audience repeatedly that Steps Must Be 1aken Immediately,` the particulars o
whom that somebody is and the details o what s,he should be doing are neer
made explicit.
In addition, the law component o cyberlaw gets surprisingly little attention in
the book, gien that Zittrain is a law proessor. According to Larry Lessig,
1his book will redeine the ield we call the law o cyberspace.`
69
1his is

66
a. at 245.
6
a.
68
ee Jonathan Zittrain, !itbovt a êt, LLGAL AllAIRS, Jan.,leb. 2006, at 34, aailable at
#$$%&''((()5,K1513314B6)/BK'466.,6'T10.1B7LF,*B.1B7L
9::='3,1$.B,8M4$$B1408X103,*:=)26%, .ee at.o Lawrence Lessig, Z`. oo/ . Ovt, LLSSIG 2.0,
May 1, 2008, #$$%&''5,664K)/BK'*)/K'X.6$8%51408*B455410$' |hereinater Lessig, Z`. oo/ .
Ovt|, Lawrence Lessig, 1be .tate of C,bertar, 200:, LLSSIG 2.0, Dec. 30, 2005,
#$$%&''5,664K)/BK'*)/K'B,1-8$#46' ,stating Legal Aairs has a antastic collection o
essays about arious cyberspace related legal issues by some o my aorite writers about the
subject. Zittrain`s piece outlines the beginning o his soon to be completed book. It shall be
called Z-theory.`,.
69
ee Lessig, Z`. oo/ . Ovt, .vra. Lessig explains his thoughts regarding the importance o
Zittrain`s book in his blog:
1his book will redeine the ield we call the law o cyberspace. 1hat sounds
like a hokey blurb no doubt. But hokeness |sic| does not mean it is not true.


worrisome to anyone still struggling to ascertain the parameters o cyberlaw in
the irst instance, beyond the macro concerns about top-down ersus bottom-
up approaches to goernance identiied by the scholars mentioned aboe. 1he
role o law in Zittrainnet`s rule o law is extremely limited. Laws concerning
jurisdiction, priacy, ree speech, copyrights, and trademarks oten transmogriy
into cyberlaw when they are inoked in an Internet context, but they exist and
eole oline too, which preents their total capture by cyberlaw scholars.
Zittrain`s redeinition o cyberlaw compresses debates that engage complicated,
intersecting bodies o law into a much narrower conersation about the alue o
generatiity, and how best to secure the appropriate leel o it. In general
Zittrain seems quite pessimistic about whether cyberlaw can achiee anything
positie beyond somehow-he neer tells us how-ostering generatiity. At
one point in the book he een describes the enorcement o laws online as
something that could result in net social losses, and thereore a mechanism o
Internet goernance that is inerior to retention o generatie technologies.`
0

Zittrain seems to hae a lot more conidence in technologists than in attorneys.
le waxes rhapsodic about the wisdom and orethought o the ramers` o the
Internet throughout the tome.
1
One o the primary` ways he proposes to
address tetheredness and its associatie ills is a series o conersations,
arguments, and experiments whose participants span the spectrum between
network engineers and PC sotware designers, between expert users with time

It is true. 1he ield beore this book was us cheerleaders trying to conince a
skeptical ,academic, world about the importance and alue o certain central
eatures o the network. Zittrain gies these eatures a name-generatiity-
and then shows us an aspect o this generatie net that we cheerleaders
would rather you not think much about: the extraordinary explosion o
malware and the like that the generatie net has also generated.
0
ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 113-114. Zittrain states:
1echnologies that lend themseles to an easy and tightly coupled expression
o goernmental power simply will be portable rom one society to the next.
It will make irreleant the question about how irms like Google and Skype
should operate outside their home countries.
1his conclusion suggests that although some social gain may result rom
better enorcement o existing laws in ree societies, the gain might be more
than oset by better enorcement in societies that are less ree-under
repressie goernments today, or anywhere in the uture. I the gains and
losses remain coupled, it might make sense to aor retention o generatie
technologies to put what law proessor James Boyle has called the
Libertarian gotcha` to authoritarian regimes: i one wants technological
progress and the associated economic beneits, one must be prepared to
accept some measure o social liberalization made possible with that
technology.
1
ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at , 2, 31, 33, 34, 69, 99.

to spend tinkering and those who simply want the system to work-but who
appreciate the dangers o lockdown` ,p. 13,. On the Zittrainnet, with the
exception o a select ew cyberlaw proessors, academics in disciplines other
than law, particularly computer science, are going to be the true beneolent
dictators o cyberlaw, mediating disputes with technological innoations and
enorcing their judgments through code.
The Private Sector
Zittrain quite understandably doubts that or-proit entities will sellessly
prioritize the well-being o the Internet oer their own commercial gain. So,
they are unlikely to consistently adhere to pro-generatie business plans unless
they can be coninced that doing so will beneit them. One o Zittrain`s
objecties in writing the book was to educate the reader about the ways that
extensie generatiity can sere commercial goals. loweer, while corporate
actors may ind Zittrain`s book o interest, I suspect actual experiences in the
marketplace will be what dries their decisions about tethers and generatiity.
Zittrain opens his book with what is ramed as an apocryphal tale: Apple II
computers were reolutionary because they acilitated the deelopment o new
and original uses by outsiders, but thirty years later the same company launched
an anti-generatiity counterreolution o sorts by releasing its innoatie iPhone
in a locked ormat intended to discourage the use o applications that were not
deeloped or approed by Apple.
2

But how would Zittrain change this Surely when the company made this
decision, it knew een more than Zittrain about the role that generatiity played
in the success o the Apple II, but still chose a dierent strategy or the iPhone.
Airmatie curtailment o its generatiity initially lowered the risk that iPhones
would be plagued by iruses or malware, and allowed Apple to control the ways
that most consumers use them. \ould Zittrain hae orced generatiity into
the mechanics o the iPhone by law Or, would he strip Apple o its ability to
use the law to interere when others hack the iPhone and make it more
customizable Or, would he instead simply wait or the market to show Apple
the error o its degeneratie ways le neer speciies. \hat he says at the end
o his iPhone discussion is:
A lockdown on PCs and a corresponding rise o tethered
appliances will eliminate what today we take or granted: a
world where mainstream technology can be inluenced, een
reolutionized, out o let ield. Stopping this uture depends

2
ee geveratt, ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 86-8 ,summarizing
work by Lric on lippel on the subject,.

on some wisely deeloped and implemented locks, along with
new technologies and a community ethos that secures the keys
to those locks among groups with shared norms and a sense o
public purpose, rather than in the hands o a single gatekeeping
entity, whether public or priate.
3

It sounds like Zittrain wants to preent Apple rom interering when consumers
modiy their iPhones. But how he proposes to achiee this is addressed only
generally, much later in the book when he suggests ague, persuasion-based
solutions. My inner pragmatist thinks strong consumer protection laws might
be a iable option to this and many other problems he articulates in the book,
but Zittrain mentions that possibility only glancingly, in the context o
maintaining data portability.
4

In July o 2008, Apple began allowing sotware deelopers to sell sotware or
the iPhone, and tens o thousands o applications hae subsequently been
independently deeloped or the iPhone,
5
suggesting either successul
deployment o a strategic multistep product rollout Apple had planned all along,
or a midcourse marketing correction. In either eent, ater the App Store the
iPhone cannot accurately be described as non-generatie, at least as I
understand the concept,
6
and what Zittrain characterized as a problem seems
to hae been largely soled without the interention o cyberlaw. 1he iPhone is
still tethered, o course, possibly giing consumers just enough rope to hang
themseles i Apple decides to interere with the contents or operation o any
gien phone. But tethering also acilitates positie interactions, such as updates
and repairs. It is now, to use a phrase Zittrain uses in a dierent context, |a|
technology that splits the dierence between lockdown and openness.`

It is true that Apple could alter the iPhone`s balance between generatiity and
tetheredness without notice or reason. But there is eery reason to expect that
Apple will try to keep its customers happy, especially gien increased
competition by deices running Google`s Android operating system-with its

3
4
a. at 1.
5
ee, e.g., Jon lortt, iPbove a.: or fvv ava rofit., lOR1UNL 1LCl DAIL\, July 6, 2009,
#$$%&''2/0,7)+00)+/2'9::>':C':='$,+#0/5/K7'1%%5,84%#/0,81%%6)3/B$.0,'40-,H)
#$2
6
ee, e.g., Adam 1hierer iPbove 2.0 crac/ea iv bovr. . rbat ra. tbat Zittraiv tbe.i. agaiv., 1lL
1LClNOLOG\ LIBLRA1ION lRON1, July 10, 2008,
#$$%&''$,+#54*,B1$4/0)+/2'9::?':C'A:'4%#/0,L9:L+B1+<,-L40L#/.B6L(#1$L(16L$#1$L
M4$$B140L$#,646L1K140'.


een more open apps marketplace.
8
A recent short reiew o the book in 1be
Ob.errer noted:
1he problem acing books about the internet is that by the
time they hae hit the sheles, they are already dated. 1his is
clear on the second page o 1he luture o the Internet, where
Jonathan Zittrain writes that the iPhone is purposeully
resistant to applications` ,programmes allowing the phone to
do cleer things apart rom make calls,.
9

1he problem acing this book is deeper than datedness. Zittrain is wrong in his
assumptions about rigidity and ixedness.
80
In the abstract generatiity and
tetheredness may be opposites, but in reality they can exist within a single
appliance. le actually makes this point when he describes computers with dual
applications designated red` and green,` one generatie and the other
secure.
81
But he does not acknowledge that many technological deices already

8
\i-\yn \en & Michal Le-Ram, Coogte`. 1 Pbove to Covete ritb tbe iPbove, 1LClLAND,
Sept. 1, 2008, #$$%&''$,+#510-)*5/K6)3/B$.0,)+00)+/2'9::?':>'AC'K//K5,6LA>>L
%#/0,L$/L +/2%,$,L(4$#L$#,L4%#/0,'.
9
lelen Zaltzman, 1be vtvre of tbe vtervet b, ]ovatbav Zittraiv, OBSLRVLR ,London,, June 14,
2009, at 26, araitabte at #$$%&''((()K.1B-410)+/).<'*//<6'9::>'X.0'A@'3.$.B,L
40$,B0,$LM4$$B140LB,J4,(.
80
ee Adam 1hierer, Rerier of Zittraiv`. vtvre of tbe vtervet, 1lL 1LClNOLOG\ LIBLRA1ION
lRON1, Mar. 23, 2008, #$$%&''$,+#54*,B1$4/0)+/2'9::?':D'9D'B,J4,(L/3LM4$$B1406L
3.$.B,L/3L$#,L40$,B0,$'. 1hierer writes:
My primary objection to Jonathan`s thesis is that ,1, he seems to be oer-
stating things quite a bit, and in doing so, ,2, he creates a alse choice o
possible utures rom which we must choose. \hat I mean by alse choice is
that Jonathan doesn`t seem to beliee a hybrid uture is possible or desirable.
I see no reason why we can`t hae the best o both worlds-a world ull o
plenty o tethered appliances, but also plenty o generatiity and openness.
ee at.o 1imothy B. Lee, iivg | Coae !itb 20,20 iva.igbt, lRLLDOM 1O 1INKLR, May
14, 2009, #$$%&''((()3B,,-/2L$/L$40<,B)+/2'*5/K'$*5,,'64M40KL+/-,L9:9:L#40-64K#$.
Lee writes:
I think Jonathan Zittrain`s 1he luture o the Internet and low to Stop It
makes the same kind o mistake Lessig made a decade ago: oerestimating
regulators` ability to shape the eolution o new technologies and
underestimating the robustness o open platorms. 1he eolution o
technology is mostly shaped by engineering and economic constraints.
Goernment policies can sometimes orce new technologies underground,
but regulators rarely hae the kind o ine-grained control they would need
to promote generatie` technologies oer sterile ones, any more than they
could hae stopped the emergence o cookies or DPI i they`d made
dierent policy choices a decade ago.
81
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 154-5.

shit between tethered and generatie unctions, drien by the demands o their
users.
Making assumptions about consumer preerences can be hazardous, especially
or olks who tend to associate mostly with people who share common
interests, common backgrounds, a common race, a common gender. 1he
Zittrainnet`s netizens, being human, are likely to engage in all manner o
typecasting and generalizing when they redesign their Internet sectors o
interest. I the leading netizens echo the demographic pattern o the cyberlaw
scholars, white men with elite educations will be making most o the calls.
82

And Internet goernance will be exceedingly top-down.
At present companies can dramatically alter the leels o tetheredness and
generatiity in their products and serices or any reason or no reason at all, and
Zittrain neer explains what sort o regulations or market interentions he
thinks are necessary to achiee or presere the Zittrainnet. le is critical o
companies that assist totalitarian goernments with sureillance or censorship
initiaties,
83
but ails to acknowledge the reason that many technologies that can
be readily employed to spy on people are deeloped: Companies want to be able
to shadow and scrutinize their customers themseles. Consumers usually agree
to this scrutiny in nonnegotiable LULA terms and conditions. lor companies,
closely ollowing the acts and omissions o their customers or client base is
generatie behaior, een though it relies on tethers. Inormation about
consumers can lead to innoations in goods and serices as well as in marketing
them.
Governments
Zittrain expresses grae concerns about goernment interention on the
Internet. le does not seem to beliee that goernment actors can competently
saeguard users, or eectiely regulate technology. And he ears goernments
will urther harness the Internet to adance sureillance and censorship agendas
that are anathema to reedom. Zittrain writes with deep oreboding:
1he rise o tethered appliances signiicantly reduces the
number and ariety o people and institutions required to apply
the state`s power on a mass scale. It remoes a practical check
on the use o that power. It diminishes a rule`s ability to attain

82
ee Anupam Chander, !bo.e Revbtic., 69 U. ClI. L. RLV. 149, 1484-85 ,2002, ,reiewing
CASS SUNS1LIN, RLPUBLIC.COM ,2001,,.
83
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 112-13.

legitimacy as people choose to participate in its enorcement,
or at least not stand in its way.
84

So it seems strange to learn that his solution to too much tethering is a latter-
day Manhattan Project.`
85
1he Manhattan Project was, o course, the code
name or the U.S. goernment`s secret project to deelop a nuclear bomb. It
may hae been staed by scientists, many o whom were academics, but it was
organized, unded, and strictly controlled by the goernment.
86
An analogous
initiatie to ormulate the Zittrainnet would hardly be open and accessible to
the online public. Moreoer, goernments generally take some kind o
proprietary interest in the outcomes o projects they und. Len under the
Bayh-Dole Act,
8
which allows uniersities in the United States to patent
inentions deeloped with ederal unding, the U.S. goernment retains march-
in rights.
88
Zittrain seems to want the resources that goernments can proide
without any o the restrictions or obligations goernments will, as experience
suggests, ineitably impose. It`s possible that a well-crated Zittrainet Project
could receie the unconditional support o goernment actors, but I don`t think
this is terribly likely to happen.
Surprisingly, one o the success stories or generatiity that Zittrain reerences is
the Digital Millennium Copyright Act o 1998.
89
Not only did this require
goernment interention in the orm o traditional law, but it also relied on
tethering. \eb sites could not take down potentially inringing material without
retaining a leel o control that enables this.
In addition to generatiity, one o the deining principles o the Zittrainnet will
be adherence to lirst Amendment principles. Zittrain`s descriptions o online
reedom and autonomy suggest a strong belie that all the countries o the world

84
a. at 118.
85
a. at 13.
86
U.S. DLP`1 Ol LNLRG\, OllICL Ol lIS1OR\ & lLRI1AGL RLS., art, Corervvevt vort, iv
1lL MANlA11AN PROJLC1: AN IN1LRAC1IVL lIS1OR\,
#$$%&''((()+3/)-/,)K/J'2,C:'210#1$$10'A>D>LA>@9)#$2 ,last isited July 30, 2009,,
1be Mavbattav Pro;ect ;ava efore), iv 1lL NUCLLAR \LAPON ARClIVL,
#$$%&''0.+5,1B(,1%/01B+#4J,)/BK'_61'\,-'\,-)#$25 ,last isited Oct. 4, 2009,, U.S.
DLP`1 Ol LNLRG\, OllICL Ol lIS1OR\ & lLRI1AGL RLS., . 1evtatire Deci.iov to vita tbe
ovb, iv 1lL MANlA11AN PROJLC1: AN IN1LRAC1IVL lIS1OR\,
#$$%&''((()+3/)-/,)K/J'2,C:'210#1$$10'$,0$1$4J,8-,+464/08*.45-)#$2 ,last isited
July 30, 2009,.
8
35 U.S.C. 200-212 ,2006,.
88
a. 203.
89
ee Pub. L. No. 105-304, 112 Stat. 2860 ,1998,. ee at.o ZI11RAIN, 1lL lU1URL Ol 1lL
IN1LRNL1, .vra note 3, at 119-20 ,stating Zittrain`s discussion o the DMCA,.

should honor and implement the ree-speech alues o the lirst Amendment,
whether they want to or not.
90
1his raises complicated issues o state
soereignty and international law that Zittrain does not address.
Conclusion
Iè been ery hard on 1be vtvre of tbe vtervet in this reiew, but I truly did
enjoy reading it. 1he book is ery inormatie, i you can sit through the
portions contried to illustrate an unconincing macro theory o the Internet. I
wish Zittrain had written a book that set out only to describe the history and
state o the Internet, rather than one that was ormulated to support
questionable generalizations and grandiose prescriptions. le could hae told
many o the same extremely interesting stories, but with more balance and less
o a blatant big think` agenda.
1he book is woeully lacking in speciics, in terms o adancing the reorms
Zittrain asserts are necessary. Len i I were willing to buy into Zittrain`s claim
that presering and enhancing generatiity should be the organizing principle o
the Internet goernance interentions, the mechanics o how this could be
pursued holistically are neer reealed. And the technicalities by which good
generatiity could be ostered while bad generatiity was simultaneously
repressed are similarly unstated. 1he only extensiely deeloped account o a
generatie system Zittrain unabashedly admires is \ikipedia, which he admits is
undemocratic.
91
It is also a system that acilitates repression o unpopular
iewpoints, and this is likely to aect outsider groups most dramatically.
\ho will step orward to somehow cultiate the Zittrainnet is a mystery. 1he
uture o the Internet, Zittrain asserts, would be much saer in the hands o
those who can competently saeguard it. le describes these people in ery
general terms as being skilled and o good aith. 1hese hands do not belong to
people who are ailiated with dot-coms, because they use tethering to constrain
generatiity when doing so is proitable. Nor do they belong to dot-go
bureaucrats, who are at best uninormed and at worst eager to use the Internet
to enorce regimes o totalitarian rule. Readers o the book learn a lot more
about who Zittrain thinks should vot be in control o the Internet than who
should be. But there are a number o hints and suggestions scattered
throughout its pages that he beliees he and his colleagues are capable o
directing the Internet`s uture wisely and beneicently. I they are going to
attempt to do this by writing books, perhaps Zittrain`s oering makes sense as a

90
Covtra Joel R. Reidenberg, Yaboo ava Devocrac, ov tbe vtervet, 42 JURIML1RICS 261 ,2002,.
91
ee ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 141 ,And \ikipedia is
decidedly not a democracy: consensus is aored oer oting and its head counts.`,.

declaration o irst principles. Maybe his next book will describe the steps along
the path to the Zittrainnet more concretely.


The Case for Internet Optimism,
Part 2: Saving the Net from Its
Supporters
By Adam Thierer
*

v av eartier e..a,, argvea tbat tro ai.tivct .trava. of vtervet e..ivi.v ivcrea.ivgt,
aovivate vtervet otic, ai.cv..iov.. 1be e..ivi.v of êt ./etic. i. rootea iv a geverat
./etici.v of tbe .vo.ea bevefit. of c,ber.ace, aigitat tecbvotogie., ava ivforvatiov
abvvaavce. ere, re.ova to a rer, aifferevt .trava of vtervet e..ivi.v-ove ere..ea b,
fav. of tbe vtervet ava c,ber.ace rbo vovetbete.. fear tbat aar/ aa,. tie abeaa vvte.. .te.
are ta/ev to .are tbe êt frov a rariet, of itt., e.eciatt, tbe erceirea eva of oevve...
Introduction: Is the
Web Really Dying?
1he Death o the Internet` is a hot meme
in Internet policy these days. Much as a
amous 1ive magazine coer asked Is
God Dead` in 1966,
1
!irea magazine, the
magazine or the modern digerati,
proclaimed in a recent coer story that
1he \eb is Dead.`
2
A ew weeks later,
1be covovi.t magazine ran a coer story
retting about 1he \eb`s New \alls,`
wondering how the threats to the
Internet`s openness can be aerted.`
3
1he
primary concern expressed in both essays:

Adam 1hierer is a senior research ellow at the \,B+1$.6 ],0$,B 1$ V,/BK, \16/0
_04J,B64$7 where he works with the Z,+#0/5/K7 N/54+7 NB/KB12.
1
Is God Dead` 1IML, April 8, 1966,
((()$42,)+/2'$42,'+/J,B6':YA==@AYA>==:@:?Y::)#$25
2
Chris Anderson & Michael \ol, 1be !eb . Deaa. ovg ire tbe vtervet, \IRLD, Aug. 1,
2010, ((()(4B,-)+/2'21K1M40,'9:A:':?'338(,*B4%'155'A. Incidentally, there`s a long
history o pundits declaring just about eerything dead` at some point, rom email, RSS,
and blogging to eReaders, browser, and een lacebook and 1witter. ee larry McCracken,
1be 1ragic Deatb of Practicatt, rer,tbivg, 1LClNOLOGIZLR, Aug. 18, 2010,
#$$%&''$,+#0/5/K4M,B)+/2'9:A:':?'A?'$#,L$B1K4+L-,1$#L/3L%B1+$4+1557L,J,B7$#40K
3
1be !eb`. êr !att., 1lL LCONOMIS1, Sept. 2, 2010,
(((),+/0/246$)+/2'B,6,1B+#'1B$4+5,6I7R.*X,+$'-46%517R$/B7)+32O6$/B784-WA=>@D;
C>c12%g6.*X,+$afWD@?>=Dc12%g36B+W0(5

1he wide-open Internet experience o the
past decade is giing way to a new regime
o corporate control, closed platorms, and
walled gardens.
1his ear is gien uller elucidation in
recent books by two o the intellectual
godathers o modern cyberlaw: Jonathan
Zittrain`s 1be vtvre of tbe vtervet-And
or to to t,
4
and 1im \u`s 1be Ma.ter
ritcb: 1be Ri.e ava att of vforvatiov
vire..
5
1hese books are best understood
as the second and third installments in a
trilogy that began with the publication o
Lawrence Lessig`s seminal 1999 book, Coae
ava Otber ar. of C,ber.ace.

Lessig`s book ramed much o how we study and discuss cyberlaw and Internet
policy. More importantly, Coae spawned a bova fiae philosophical moement
within those circles as a polemic against both cyber-libertarianism and Internet
exceptionalism ,closely related moements,, as well as a sort o call to arms or
a new Net actiist moement. 1he book gae this moement its central
operating principle: Code and cyberspace cav be bent to the will o some
amorphous collectie or public will, and it oten vv.t be i we are to aoid any
number o impending disasters brought on by nearious-minded ,or just plain
incompetent, olks in corporate America scheming to achiee perect control`
oer users.
It`s diicult to know what to label this school o thinking about Internet policy,
and Pro. Lessig has taken oense at me calling it cyber-collectiism.`
But
the collectiism o which I speak is a more generic type, not the hard-edged
Marxist brand o collectiism o modern times. Instead, it`s the belie that
markets, property rights, and priate decision-making about the uture course o
the Net must yield to supposedly more enlightened actors and mechanisms. As
Declan McCullagh has remarked, Lessig and his students

4
JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1 ,2008,.
5
1IM \U, 1lL MAS1LR S\I1Cl: 1lL RISL AND lALL Ol INlORMA1ION LMPIRLS ,2010,.
6
LA\RLNCL LLSSIG, CODL AND O1lLR LA\S Ol C\BLRSPACL ,1999,.
Adam 1hierer, Ovr Covftict of C,ber1i.iov., CA1O UNBOUND, May 14, 2009, !!!"#$%&'
()*&()+"&,-./001.02.34.$+$5'%678,8,.&(,'#&)9:7#%'&9'#;*8,'<7=7&)=.

preer . what probably could be called technocratic
philosopher kings, o the breed that Plato`s 1be Revbtic said
would be best able to guard the laws and institutions o our
State-let them be our guardians.` 1hese technocrats would be
entrusted with making wise decisions on our behal, because,
according to Lessig, politics is that process by which we
collectiely decide how we should lie.`
8

\hat is it, exactly, that these cyber-collectiists seek to protect or accomplish
1o the extent it can be boiled down to a single term, their rallying cry is:
Openness! Openness` is almost always 1he Good, anything closed`
,restricted or proprietary, in nature is 1he Bad. 1hus, since they recoil at the
cyber-collectiist` label, we might think o adherents to this philosophy as
Openness Langelicals,` since they eangelize in aor o openness` and
seemingly make all else subserient to it.
lor example, in vtvre of tbe vtervet, Zittrain argues that, or a ariety o reasons,
we run the risk o seeing the glorious days o generatie` deices and the
open` Internet gie way to more tethered appliances` and closed networks.
le says:
1oday, the same qualities that led to |the success o the
Internet and general-purpose PCs| are causing |them| to alter.
As ubiquitous as Internet technologies are today, the pieces are
in place or a wholesale shit away rom the original chaotic
design that has gien rise to the modern inormation
reolution. 1his counterreolution would push mainstream
users away rom the generatie Internet that osters innoation
and disruption, to an appliancized network that incorporates
some o the most powerul eatures o today`s Internet while
greatly limiting its innoatie capacity-and, or better or
worse, heightening its regulability. A seductie and more
powerul generation o proprietary networks and inormation
appliances is waiting or round two. I the problems associated
with the Internet and PC are not addressed, a set o blunt
solutions will likely be applied to sole the problems at the
expense o much o what we loe about today`s inormation
ecosystem.
9

8
Declan McCullagh, !bat arr, Diav`t Cet, CA1O UNBOUND, May 4, 2009, ((()+1$/L
.0*/.0-)/BK'9::>':;':@'-,+510L2++.551K#'(#1$L51BB7L-4-0$LK,$
9
Zittrain, .vra note 4 at 8.

In other words, Zittrain ears most will lock to tethered appliances in a search
or stability or security. 1hat`s troubling, he says, because those tethered
appliances are less open` and more likely to be regulable,` either by large
corporate intermediaries or goernment oicials. 1hus, the uture o the
Internet` Zittrain is hoping to stop` is a world dominated by tethered digital
appliances and closed walled gardens because they are too easily controlled by
other actors.
My primary bee with these Openness Langelicals` is not that openness and
generatiity aren`t ine generic principles but that:
1. 1hey tend to signiicantly oerstate the seerity o this problem ,the
supposed decline o openness or generatiity, that is,,
2. I`m more willing to allow eolutionary dynamism to run its course within
digital markets, een i that means some closed` deices and platorms
remain ,or een thrie,, and,
3. It`s signiicantly more likely that the openness` adocated by Openness
Langelicals will deole into expanded goernment control o cyberspace
and digital systems than that unregulated systems will become subject to
perect control` by the priate sector, as they ear.
More generally, my problem with this moement-and Zittrain`s book, in
particular-comes down to the dour, depressing the-Net-is-about-to-die` ear
that seems to uel this worldiew. 1he message seems to be: Lnjoy the good
old days o the open Internet while you can, because any minute now it will be
crushed and closed-o by corporate marauders!` Lessig started this nerous
hand-wringing in !"#$ when he ominously predicted that Let to itsel,
cyberspace will become a perect tool o control.`
10
1oday, his many disciples
in academia ,including Zittrain and \u, and a wide ariety o regulatory
adocacy groups continue to preach this gloomy gospel o impending digital
doom and perect control` despite plenty o eidence that supports the case
or optimism.
lor example, \u warns there are orces threatening the Internet as we know
it`
11
while Zittrain worries about a handul o gated cloud communities whose
proprietors control the aailability o new code.`
12
At times, this paranoia o

10
LA\RLNCL LLSSIG, CODL AND O1lLR LA\S Ol C\BLRSPACL ,1999, at 5-6.
11
\U, &'()* note 5 at .
12
Jonathan Zittrain, +"&, -. ,/$ !0"'#, NL\ \ORK 1IMLS, July 19, 2009,
!!!"#$%&'()"*+',-../,.0,-.,+1&#&+#,-.2&%%34&#"5%'6.

some in the Openness Langelical clan borders on outright hysteria. In August
2008, a Public Knowledge analyst likened Apple`s management o applications
in its iPhone App Store to the tyranny o Orwell`s 11!
13
In other words, the
Big Brother they want us to ear is Cororate Big Brother. Someday ery soon,
we are repeatedly told, the corporate big boys will toss the proerbial master
switch,` suocating Internet innoation and digital reedom, and making us all
cyber-slaes within their commercialized walled gardens. 1he possibility o
consumers escaping rom these walled gardens or aoiding them altogether is
treated as remote-i the notion is entertained at all.
\e might think o this ear as 1he Great Closing,` or the notion that, unless
radical interentions are pursued-oten through regulation-a Digital Dark
Age o Closed Systems will soon unold, complete with myriad America Online-
like walled gardens, sterile and tethered deices,` corporate censorship, and
gouging o consumers. linally, the implicit message in the work o all these
hyper-pessimistic critics is that markets must be steered in a more sensible
direction by those technocratic philosopher kings ,although the details o their
blueprint or digital salation are oten scarce,.
Problems with The Great Closing Thesis
1here are serious problems with the Great Closing` thesis as set orth in the
high-tech threnody o Lessig, Zittrain, \u, and other Openness Langelicals, or
às 1be êr Yor/ 1ive. has called them, digital doomsayers.`
14

No Clear Definitions of Openness or Closedness;
Both Are Matters of Degree
Open` s. closed isn`t as black and white as some Openness Langelicals
make it out to be. lor example, Zittrain praises the supposedly more open
nature o PCs and the openness to innoation made possible by Microsot`s
\indows operating system. low ironic, since so many hae blasted \indows
as the Great Satan o closed code! Meanwhile, while most others think o
Apple as eeryone`s aorite example o innoation,`
15
Zittrain makes the

13
Alex Curtis, evefit. of iPbove . tore 1aivtea b, 11ti/e Covtrot, Public Knowledge Blog,
Aug. 11, 2008, ((()%.*54+<0/(5,-K,)/BK'0/-,'AC:D 1he tech gadget website
Gizmodo recently ran a similar Apple-as-Big-Brother essay: Matt Buchanan, ig rotber .te
ava tbe Deatb of tbe Prograv, GIZMODO, Oct. 22, 2010, #$$%&''K4M2/-/)+/2';=C:?A9'*4KL
*B/$#,BL1%%5,L10-L$#,L-,1$#L/3L$#,L%B/KB12.
14
Lric Panner, Proctaivea Deaa, !eb i. borivg igv. of êr ife, NL\ \ORK 1IMLS, Oct. 31,
2010, ((()07$42,6)+/2'9:A:'AA':A'$,+#0/5/K7':A(,*(1556)#$25
15
Amar Bhide, Dov`t ect Mvcb rov tbe RcD 1a Creait, \ALL S1RLL1 JOURNAL, Sept. 11,
2010,
#$$%&''/0540,)(6X)+/2'1B$4+5,'RIA:::A@9@:;9C@?C:@=@@@:@;C;@?A;D@A>DD@@:??)#$25

iPhone and iPad out to be sterile, tethered` appliances. But the company`s
App Store has oered millions o innoators the opportunity to produce almost
eery conceiable type o mobile application the human mind could imagine or
those deices.
16
Moreoer, those Apple deices don`t block completely open`
communications applications or interaces, such as \eb browsers, email and
SMS clients, or 1witter. In the abstract,` notes Uniersity o South Carolina
School o Law proessor Ann Bartow, generatiity and tetheredness may be
opposites, but in reality they can exist within a single appliance.`
1

\hile the Apple deices seem to proe that, in reality, almost att modern digital
deices and networks eature some generatie and non-generatie` attributes.
No one has eer created, and no one will eer create, a system that allows any
user to create anything he or she wants. Instead, eery system designer makes
innumerable tradeos and imposes countless constraints,` note James
Grimmelmann and Paul Ohm.
18
Lery generatie technology aces .
tradeos. Good system designers always restrict generatiity o some kinds in
order to encourage generatiity o other kinds. 1he trick is in striking the
balance,` they argue.
19
\et, Zittrain neer ully analyzes .titgeveratirit,
systems, those with generatie layers built upon non-generatie layers, or ice-
ersa.`
20

1he zero-sum ear that the ascendancy o mobile apps means less generatiity`
or the death o the \eb` is another myth. Nick Bilton o 1be êr Yor/ 1ive.
notes:
Most o these apps and \eb sites are so intertwined that it`s
diicult to know the dierence. \ith the exception o
downloadable games, most \eb apps or news and serices
require pieces o the \eb and Internet to unction properly. So
as more deices become connected to the Internet, een i
they`re built to access beautiul walled gardens, like mobile

16
Apple, .te`. . tore Dorvtoaa. 1o 1bree ittiov, Jan. 5, 2010,
((()1%%5,)+/2'%B'54*B1B7'9:A:':A':;1%%6$/B,)#$25
1
Ann Bartow, . Portrait of tbe vtervet a. a Yovvg Mav, 108 MIClIGAN LA\ RLVIL\ 6, at 1102-
03, ((()24+#4K1051(B,J4,()/BK'166,$6'%-36'A:?'='*1B$/()%-3
18
James Grimmelmann & Paul Ohm, Dr. Ceveratire or: or earvea to to !orr,ivg ava ore
tbe iPbove, MAR\LAND LA\ RLVIL\ ,2010, at 940-41.
19
a. at 941.
20
a. at 944. ,emphasis in original,.

apps or 1V-speciic interaces, they will continue to access the
\eb too, enabling each platorm to grow concurrently.
21

Ironically, it was Chris Anderson, editor o !irea and author o the apocalyptic
\eb is Dead` coer story, who best explained why ears o 1he Great
Closing` are largely oerblown:
Lcommerce continues to thrie on the \eb, and no company
is going to shut its \eb site as an inormation resource. More
important, the great irtue o today`s \eb is that so much o it
is noncommercial. 1he wide-open \eb o peer production, the
so-called generatie \eb where eeryone is ree to create what
they want, continues to thrie, drien by the nonmonetary
incenties o expression, attention, reputation, and the like.
22

And Je Bertolucci o PC !orta makes it clear generatie computing is alie and
well:
1he next big computing platorm won`t be a ersion o
Apple`s Mac OS, Google`s Android, or Microsot`s \indows.
It`s already here-and it`s the \eb. And the drie to oer the
most compelling window to the \eb possible, ia the browser,
is intense. 1he browser is spreading beyond the PC and
smartphone to new types o gadgetry, including 1V set-top
boxes and printers. 1his is a trend that will accelerate in the
coming years.
23

The Evils of Closed Systems or Digital
Appliances Are Greatly Over-Stated
Openness Langelicals oten ail to appreciate how there obiously must hae
been a need , demand or some closed` or sterile` deices or else the market
wouldn`t hae supplied them. \hy .bovtav`t people who want a simpler or more
secure digital experience be oered such options \u worries that deices like
the iPad are computers that hae been reduced to a strictly limited set o
unctions that they are designed to perorm extremely well.`
24
Needless to say,

21
Nick Bilton, . tbe !eb D,ivg. t Doe.v`t oo/ 1bat !a,, NL\ \ORK 1IMLS BI1S BLOG, Aug.
1, 2010, #$$%&''*4$6)*5/K6)07$42,6)+/2'9:A:':?'AC'$#,LKB/($#L/3L$#,L-740KL(,*
22
Anderson & \ol, .vra note 2.
23
Je Bertolucci, Yovr ror.er iv ire Year., PC \ORLD, June 16, 2010,
((()%+(/B5-)+/2'1B$4+5,'A>>:CA'7/.B8*B/(6,B840834J,87,1B6)#$25
24
\u, .vra note 5 at 292.

it will be hard or many consumers to sympathize with \u`s complaint that
products work too well!
loweer, as noted throughout this essay, it`s also not quite true that those
deices are as closed or crippled as their critics suggest. As Grimmelmann and
Ohm aptly note, restricting generatiity in one place ,or example, by building
computers with ixed circuit boards rather than a tangle o reconigurable wires,
can massiely enhance generatiity oerall ,by making computers cheap and
usable enough that eeryone can tinker with their sotware,.`
25
lor example, in
Noember 2010, Damon Albarn, lead singer o the popular band Gorillaz,`
announced that the group`s next album would be recorded entirely on an iPad.
26

Regardless, just how ar would these critics go to keep deices or platorm
perectly generatie` or open` ,assuming we can een agree on how to deine
these concepts, Do the Openness Langelicals really think consumers would
be better sered i they were orced to end or themseles with deices that
arried totally unconigured Should the iPhone or iPad, or example, be
shipped to market with no apps loaded on the main screen, orcing eeryone to
go ind them on their own Should 1iVos hae no interactie menus out-o-
the-box, orcing consumers to go online and ind some homebrew` code that
someone whipped up to gie users an open source programming guide
Some o us are able to do so, o course, and those o us who are tech geeks
sometimes ind it easy to look down our noses at those who want their hand
held through cyberspace, or who aor more simplistic deices. But there`s
nothing wrong with those indiiduals who seek simplicity, stability, or security
in their digital deices and online experiences-een i they ind those solutions
in the orm o tethered appliances` or walled gardens.` Not eeryone wants
to tinker or to experience cyberspace as geeks do. Not eeryone wants to
program their mobile phones, hack their consoles, or write their own code.
Most people lie perectly happy lies without eer doing any o these things!
Nonetheless, many o those mere mortals` ritt want to use many o the same
toys that the tech geeks use, or they may just want to take more cautious steps
into the occasionally cold pool called cyberspace-one tippy toe at a time. \hy
shouldn`t those users be accommodated with lesser` deices or a curated`
\eb experience Kein Kelly argues that there`s another way o looking at
these trends. Digital tools are becoming more specialized, he argues, and with
the adent o rapid abrication . specialization will leap ahead so that any tool
can be customized to an indiidual`s personal needs or desires.`
2
Viewed in

25
Grimmelmann & Ohm, .vra note 18, at 923.
26
Davov .tbarv Recora. êr Coritta .tbvv ov av iPaa, NML NL\S, Noember 12, 2010,
#$$%&''((()02,)+/2'0,(6'K/B4551M';D?A=
2
Kein Kelly, !bat 1ecbvotog, !avt. ,2010, at 295-6.

this light, the Openness Langelicals would hold back greater technological
specialization in the name o presering market norms or structures they preer.
1he best argument against digital appliancization is that the desire or more
stable and secure systems will lead to a more regulable` world-i.e., one that
can be more easily controlled by both corporations and goernment. As
Zittrain puts it:
\hether sotware deeloper or user, olunteering control oer
one`s digital enironment to a Manager means that the
manager can change one`s experience at any time-or worse,
be compelled to by outside pressures. . 1he amously
ungoernable Internet suddenly becomes much more
goernable, an outcome most libertarian types would be
concerned about.
28

No doubt, concerns about priacy, child saety, deamation, cybersecurity,
identity thet and so on, will continue to lead to calls or more interention. At
the corporate leel, howeer, some o that potential interention makes a great
deal o sense. lor example, i ISPs are in a position to help do something to
help alleiate some o these problems-especially spam and iruses-what`s
wrong with that Again, there`s a happy balance here that critics like Zittrain
and \u ail to appreciate. Bruce Owen, an economist and the author o 1be
vtervet Cbattevge to 1eteri.iov, discussed it in his response to Zittrain`s recent
book:
\hy does Zittrain think that oerreaction is likely, and that its
costs will be unusually large Neither prediction is sel-eident.
laced with the risk o inection or mishap, many users already
restrain their own taste or PC-mediated adenture, or install
protectie sotware with similar eect. lor the most risk-aerse
PC users, it may be reasonable to welcome tethered` PCs
whose suppliers compete to oer the most popular
combinations o reedom and saety. Such risk-aerse users are
reacting, in part, to negatie externalities rom the poor
hygiene o other users, but such users in turn create positie
externalities by limiting the population o PCs ulnerable to
contagion or hijacking. As ar as one can tell, this can as easily
produce balance or under-reaction as oerreaction-it is an
empirical question. But, as long as lexibility has alue to users,

28
Jonathan Zittrain, a. tbe vtvre of tbe vtervet aevea Sept. , 2010, CONCURRING
OPINIONS blog, ((()+/0+.BB40K/%404/06)+/2'1B+#4J,6'9:A:':>'#16L$#,L3.$.B,L/3L
$#,L40$,B0,$L+/2,L1*/.$)#$25

suppliers o hardware and interconnection serices will hae
incenties to oer it, in measured ways, or as options.
29

Indeed, we can ind happy middle-ground solutions that balance openness and
stability-and platorm operators must be ree to discoer where that happy
medium is through an ongoing process o trial and error, or only through such
discoery can the right balance be struck in a constantly changing landscape. A
world ull o hybrid solutions would oer more consumers more choices that
better it their speciic needs.
linally, to the extent something more must be done to counter the supposed
regulability o cyberspace, the solution should not be new limitations on
innoation. Instead o imposing restrictions on code or coders to limit
regulability, we should instead place more constraints on our goernment,s,.
Consider priacy and data collection concerns. \hile, as a general principle, it
is probably wise or companies to minimize the amount o data they collect
about consumers to aoid priacy concerns about data breaches, there are also
beneits to the collection o that data. So rather than legislating the right` data
retention rules, we should hold companies to the promises they make about
data security and breaches, and tightly limit the powers o goernment to access
priate inormation through intermediaries in the irst place.
Most obiously, we could begin by tightening up the Llectronic
Communications Priacy Act ,LCPA, and other laws that limit goernment
data access.
30
More subtly, we must continue to deend Section 230 o the
Communications Decency Act, which shields intermediaries rom liability or
inormation posted or published by users o their systems, because ,among
many things, such liability would make online intermediaries more susceptible
to the kind o back-room coercion that concerns Zittrain, Lessig and others. I
we`re going to be legislating the Internet, we need more laws like that, not those
o the middleman deputization` model or those that would regulate code to
achiee this goal.
Companies Have Strong Incentives to Strike
the Right Openness/Closedness Balance
Various social and economic inluences help ensure the scales won`t be tipped
completely in the closed or non-generatie direction. 1he \eb is built on

29
Bruce Owen, .. ovg a. teibitit, a. 1atve to |.er., vtier. !itt are vcevtire. to Offer t,
BOS1ON RLVIL\, March,April 2008, ((()*/6$/0B,J4,()0,$'IPDD)9'/(,0)%#%
30
A broad coalition has proposed such reorms. See ((()-4K4$15-.,%B/+,66)/BK.

!"#$%&'( &$$*+,-. /$-0,123/3 ,1* !"33$33$3 ,1 $45%,"%*21,%6 ($7$( "&
5%,13!,%$1-6 21 5$%/3 "& 253 "!$%,52"138
9"%$"7$%: 50$ +%$,.21; 1$#3 -6-($ &"% 5$-0 *$7$("!/$153 -,1 +$ /$,3'%$* 1"5
21 *,63: +'5 21 /21'5$3 "% $7$1 3$-"1*38 <7$%6 +"1$0$,*$* /"7$ /$$53
2//$*2,5$ ,1* 215$13$ 3-%'5216 +6 +(";;$%3: 5$-0 !%$33: !'1*253: ;,*;$5 325$3: etc8
=$7$% 0,3 50$ #025$>0"5 3!"5(2;05 "& !'+(2- ,55$152"1 +$$1 3" 215$13$ 21 0$(!21;
5" 3021$ , (2;05 "1 -"%!"%,5$ /2335$!3 ,1* &"%-21; 50$2% -"%%$-52"18 ?$ 3,# 5023
*61,/2- ,5 #"%. #250 50$ @,-$+"". A$,-"1 21-2*$15:
BC
D"";($E A'FF *$+,-($:
BG

H/,F"1 11 21-2*$15:
BB
H!!($E3 @(,30 %$35%2-52"13:
BI
50$ J"16 %""5.25 $!23"*$:
BK

,1* "50$% $4,/!($38
Things Are Getting More Open
All the Time Anyway
9"35 -"%!"%,5$ ,55$/!53 5" +"55($ '! 21&"%/,52"1 "% -("3$ "&& 50$2% !(,5&"%/3
$1* +,*(68 L0$ #,(($* ;,%*$13 "& 50$ !,35 &,2($* /23$%,+(68 M1 -%252N'21;
O255%,21E3 +"".: H11 A,%5"# 0,3 1"5$* 50,5 P2& O255%,21 23 -"%%$-5 50,5
Q"/!'J$%7$ ,1* H/$%2-, R1(21$ SHRTU $4$/!(2&6 50$ $72(3 "& 5$50$%21;: 25E3
!%$556 -($,% 50$ /,%.$5 !'1230$* 50"3$ $15252$3 !%$556 0,%30(6 #250"'5 M15$%1$5
;"7$%1,1-$>356($ 215$%7$152"138V
BW
M1*$$*: ($5E3 1"5 &"%;$5 50,5 HRT #,3 50$
+2;: +,* -"%!"%,5$ +"";$6/,1 "& T$332;E3 Coae ,1* 6$5: X'35 , *$-,*$ (,5$%: 25 0,3
+$$1 %$($;,5$* 5" ,1 ,(3">%,1 21 50$ M15$%1$5 $-"3635$/8

BC
ee =,1-6 D"0%21;: aceboo/ ace. Cta...ctiov vit Orer eacov: =<L?RYZ?RYT[8QR9: H';8
CB: G\\]: !""#$%%&&&'()"&*+,&*+-.'/*0%()&1%2334%345634789/):**,789/)17/-9117
9/";*(71<;"7*=)+'!"0-8
BG
ee Y6,1 ^,'(: PC ait: Coogte ace. 1C Covtaivt Orer v Prirac,: HYJ L<Q_=MQH: @$+8
C`: G\C\: !""#$%%9+1")/!(;/9'/*0%1)/<+;">%()&1%2353%32%)#;/789;-7?**?-)789/)17
/*0#-9;("7*=)+7:<@@7#+;=9/>7;11<)1'9+18
BB
ee a"01 L2//$%: .vaov ettte. 11 vit, et. ivit. ov Kivate Detetiov.: HYJ L<Q_=MQH: R-58
G: G\\b: !""#$%%9+1")/!(;/9'/*0%&):%()&1%233A%53%909@*(71";#<-9")17")+017*87
:**,7.)-)";*(7=;975A4B71)""-)0)("'9+18
BI
ee Y"+ ^$;"%,%": .te aa`. Re;ectiov of .aobe ta.b Covta igvat tbe Pta,er`. Deatb Kvett: L_<
?HJ_M=DLR= ^RJL: @$+8 `: G\C\: !""#$%%&&&'&91!;(?"*(#*1"'/*0%&#7
.>(%/*(")("%9+";/-)%2353%32%3C%DE2353323C3534A'!"0-8
BK
ee ?2.2!$*2,: ov, MC CD Co, Protectiov cavaat,
!""#$%%)('&;,;#).;9'*+?%&;,;%F*(>GHIJGKLG/*#>G#+*")/";*(G1/9(.9- S(,35
,--$33$* [$-8 b: G\C\U8
BW
A,%5"#: .vra 1"5$ C` ,5 C\]]:
&&&'0;/!;?9(-9&+)=;)&'*+?%911)"1%#.81%534%M%:9+"*&'#.8

Z#, [2,B4+1 h0540, ]16, R$.-7&
P,2,2*,B40K S,6$,B-17i6 F1+, /3 j]5/6,-k GJ45
\hen it comes to closed` systems, eil has a ace, but it seems the ace is
always changing. \hen Lessig penned Coae a decade ago, it was American
Online ,AOL, that was set to become the corporate enslaer o cyberspace. lor
a time, it was easy to see why Lessig and others might hae been worried.
1wenty ie million subscribers were willing to pay >20 per month to get a
guided tour o AOL`s walled garden ersion o the Internet. 1hen AOL and
1ime \arner announced a historic mega-merger that had some predicting the
rise o new totalitarianisms`
3
and corporate Big Brother.`
38

But the deal quickly went o the rails.
39
By April 2002, just two years ater the
deal was struck, AOL-1ime \arner had already reported a staggering >54
billion loss.
40
By January 2003, losses had grown to >99 billion.
41
By September
2003, 1ime \arner decided to drop AOL rom its name altogether and the deal
continued to slowly unrael rom there.
42
In a 2006 interiew with the !att
treet ]ovrvat, 1ime \arner President Jerey Bewkes amously declared the
death o synergy` and went so ar as to call synergy bullsht`!
43
In early 2008,
1ime \arner decided to shed AOL`s dial-up serice
44
and in 2009 spun o
AOL entirely.
45
lurther deconsolidation ollowed or 1ime \arner, which

3
Norman Soloman, .O 1ive !arver: Cattivg 1be aitbfvt 1o 1beir Kvee., Jan. 2000,
((()314B)/BK'2,-41L*,1$':::AAD)#$25
38
Robert Scheer, Covfe..iov. of av Cotvvvi.t, Jan. 14, 2000, ONLINL JOURNALISM RLVIL\,
((()/XB)/BK'/XB'(/B<%51+,'A:AC>==A:>)%#%
39
Adam 1hierer, . rief i.tor, of Meaia Merger ,.teria: rov .O1ive !arver to Covca.t
^C, Progress & lreedom loundation, PROGRLSS ON POIN1 16.25, Dec. 2, 2009,
((()%33)/BK'466.,6L%.*6'%/%6'9::>'%/%A=)9;L+/2+16$LdI]L2,BK,BL21-0,66)%-3
40
lrank Pellegrini, !bat .O 1ive !arver`. :1 ittiov o.. Meav., April 25, 2002, 1IML
ONLINL, ((()$42,)+/2'$42,'*.640,66'1B$4+5,':Y?;>>Y9DD@D=Y::)#$25
41
Jim lu, .O o.e. 1ea 1vrver ava bittiov, CNL1 NL\S.COM, Jan. 30, 2004,
#$$%&''0,(6)+0,$)+/2'[hbL5/6,6LZ,-LZ.B0,BL10-L>>L*4554/0'9A::LA:9D8DL
>?9=@?)#$25
42
a.
43
Matthew Karnitschnig, .fter Year. of Pv.bivg ,verg,, 1ive !arver vc. a,. vovgb, \ALL
S1RLL1 JOURNAL, June 2, 2006,
#$$%&''/0540,)(6X)+/2'1B$4+5,'RIAA@>9A?:A=;:>=>;C@)#$25
44
Geraldine labrikant, 1ime \arner Plans to Split O AOL`s Dial-Up Serice, NL\ \ORK
1IMLS, leb. , 2008,
((()07$42,6)+/2'9::?':9':C'*.640,66':C(1B0,B)#$25O8BWAc1-H005WAc/B,3W65/K
40c1-H005HWA9:>=;@:D:Ll%GVI'0DXR;ZVQE=DfhdQK
45
Press Release, 1ime \arner, 1ime \arner Inc. Completes Spin-o o AOL Inc. ,Dec. 10,
2009,, #$$%&''((()$42,(1B0,B)+/2'+/B%'0,(6B//2'%B':Y9:?A9YA>@=?D;Y::)#$25.

spun o its cable 1V unit and arious other properties. Looking back at the
deal, lortune magazine senior editor at large Allan Sloan called it the turkey o
the decade.`
46

In the larger scheme o things, AOL`s story has already become an aterthought
in our chaotic cyber-history. But we shouldn`t let those old critics orget about
their lugubrious lamentations. 1o recap: the big, bad corporate illain o
Lessig`s Coae attempted to construct the largest walled garden eer, and partner
with a titan o the media sector in doing so-ava tbi. aa.tarat, tot faitea vi.erabt,.
1he hysteria about AOL`s looming monopolization o instant messaging-and
with it, the rest o the \eb-seems particularly silly: 1oday, anyone can
download a ree chat client like Digsby or Adium to manage multiple IM
serices rom AOL, \ahoo!, Google, lacebook and just about anyone else, all
within a single interace, essentially making it irreleant which chat serice your
riends use.
lrom this case study one would think the Openness Langelicals would hae
gained a newound appreciation or the eolutionary and dynamic nature o
digital markets and come to understand that, in markets built upon code, the
pace and nature o change is unrelenting and utterly unpredictable. Indeed,
covtra Lessig`s lament in Coae that Let to itsel, cyberspace will become a
perect tool o control,` cyberspace has proen ar more diicult to control`
or regulate than any o us eer imagined. 1he olume and pace o technological
innoation we hae witnessed oer the past decade has been nothing short o
stunning.
Critics like Zittrain and \u, howeer, wants to keep beating the cyber-sourpuss
drum. So, the ace o corporate eil had to change. 1oday, Stee Jobs has
become the supposed apotheosis o all this closed-system eil instead o AOL.
Jobs seres as a prime illain in the books o Zittrain and \u and in many o
the essays they and other Openness Langelicals pen. It`s worth noting,
howeer, that their enemies list is growing longer and now reads like a \ho`s
\ho` o high-tech corporate America. According to Zittrain and \u`s books,
`we need to worry about just about eery major player in the high-tech
ecosystem-telcos, cable companies, wireless operators, entertainment
proiders, lacebook, and others.
Len Google-Silicon Valley`s supposed saior o Internet openness-is not
spared their scorn. Google is the Internet`s switch,` \u argues. In act, it`s

46
Allan Sloan, Ca.b for.` ava tbe Year`. Otber Ctvv/er., \ASlING1ON POS1, No. 1, 2009,
((()(16#40K$/0%/6$)+/2'(%L
-70'+/0$,0$'1B$4+5,'9::>'AA'A='[P9::>AAA=:DCC;)#$25

the world`s most popular Internet switch, and as such, it might een be
described as the current custodian o the Master Switch.` More ominously, he
warns, it is the switch that transorms mere communications into
networking-that ultimately decides who reaches what or whom.`
4

It seems, then, that the ace o closed` eil is constantly morphing. Shouldn`t
that tell us something about how dynamic these markets are

1here are ew reasons to beliee that today`s eorts to build such walled
gardens would end much dierently. Indeed, increasingly when companies or
coders erect walls o any sort, holes orm quickly. lor example, it usually
doesn`t take long or a determined group o hackers to ind ways around
copy,security protections and root` or jailbreak` phones and other deices.
48

Once hacked, users are usually then able to conigure their deices or
applications howeer they wish, eectiely thumbing their noses at the
deelopers. 1his process tends to unold in a matter o just days, een hours,
ater the release o a new deice or operating system.
d.2*,B /3 f176 I,3/B, d,( f,J4+,6 `,B, jP//$,-k /B jT145*B/<,0k
49

original iPhone 10 days
original iPod 1ouch 35 days
iPhone 3G 8 days
iPhone 3GS 1 day
iPhone 4 38 days
iPad 1 day
1-Mobile G1 ,irst Android phone, 13 days
Palm Pre 8 days

O course, not eery user will make the eort-or take the risk
50
-to hack their
deices in this ashion, een once instructions are widely aailable or doing so.

4
48
In liing proo that as long as there`s a thriing geek an culture or a deice, it will neer be
long or the new ersion to be jailbroken: behold iOS 4.1. Most people are perectly willing
to let their deices do the talking or them, accept what`s gien, and just run sanctioned
sotware. But there are those intrepid ew-who actually make up a airly notable portion o
the market-who want more out o their deices and ind ways around the handicaps built
into them by the manuacturers.` Kit Dotson, êr iO for .te 11 irvrare Retea.ea,
Provtt, Decr,tea, SiliconAngle, Sept. 28, 2010, #$$%&''6454+/010K5,)+/2'*5/K'9:A:':>'
9?'0,(L4/6L3/BL1%%5,L$JL34B2(1B,LB,5,16,-L%B/2%$57L-,+B7%$,-
49
Original research conducted by author and Adam Marcus based on news reports.

Nonetheless, een i copyright law might sometimes seek to restrict it, the
hacking option still exists or those who wish to exercise it. Moreoer, because
many manuacturers know their deices are likely to be hacked, they are
increasingly willing to make them more open` right out o the gates or oer
more unctionality,lexibility to make users happy.
Innovation Continues to Unfold Rapidly
in Both Directions along the Open
vs. Closed Continuum
As noted aboe, part o Zittrain and \u`s lament seems to be that the deices
that the boi ottoi choose might crowd out those aored by tinker-happy tech
geeks ,o which I count mysel a proud member,. But we geeks need not ear
such oreclosure. Just because there are some closed` systems or deices on
the market, it doesn`t mean innoation has been oreclosed among more
open` systems or platorms. A hybrid uture is both possible and desirable.
Again, we can hae the best o both worlds-a world ull o plenty o closed
systems or een tethered appliances,` but also plenty o generatiity and
openness. As \eb 2.0 pioneer 1im O`Reilly notes:
I`m not terribly taken in by the rhetoric that says that because
content silos are going up, and we`re seeing more paid content,
the open web is oer. Indiiduals, small companies,
entrepreneurs, artists, all hae enormous ability to share and
distribute their work and ind an audience. I don`t see that
becoming less in today`s enironment.
51

Consider the battle between the Apple iPhone and Google Android mobile
phone operating systems. Zittrain says Android is a sort o canary in the coal
mine`
52
or open platorms, but ignores the rantic pace o its growth, now
accounting or one-quarter o mobile \eb traic just three years ater its
inception
53
and stealing away Apple`s marketshare in the process.
54
Beyond

50
Rooting or jailbreaking a smartphone creates the risk o bricking` the deice-rendering it
completely inoperable ,and thus no more useul than a brick,. Additionally, hacking deices
in this ashion typically oids any manuacturer warranty.
51
1be !eb i. Deaa. . Debate, \IRLD, Aug. 1, 2010,
((()(4B,-)+/2'21K1M40,'9:A:':?'338(,*B4%8-,*1$,'155'A
52
$#,L40$,B0,$L+/2,L1*/.$)#$25
53
Sean lollister, .varoia .ccovvt. for OveQvarter of Mobite !eb 1raffic, a,. Qvavtca.t,
LNGADGL1, Sept. 4, 2010, (((),0K1-K,$)+/2'9:A:':>':@'10-B/4-L1++/.0$6L3/BL/0,L
m.1B$,BL/3L2/*45,L(,*L$B1334+L6176Lm.1, .varoia Mo.t Povtar Oerativg ,.tev iv |..


downplaying Android`s success as a marketplace triumph or openness ,and
proo o the non-goernmental orces that work to orce a balance between
openness and closedness,, Zittrain also reerts to the kill switch` boogeyman:
le warns us that any day now Google could change its mind, close the Android
platorm, and kill an app, or the entire phone` remotely.
55
But where`s the
business sense in that \hat`s the incentie or Google to pursue such a course
o action \ould Google be able to produce all those millions o apps
currently produced by independent deelopers 1hat seems both unlikely and
unpopular. Meanwhile, how many times has supposedly control-minded Apple
actually thrown the dreaded kill switch` on apps 1here are tens o millions o
apps in Apple`s App Store and hundreds o billions o downloads. I Stee Jobs
is supposed to be the great illain o independent innoation, he seems to be
doing a pretty bad job at it! 1he App Store is, by some estimates, now a multi-
billion-dollar-a-year business,` note Grimmelmann and Ohm.
56
1he iPhone is
a hotbed o creatie tinkering, people are doing amazing things with it.`
5

In act, \u admits Apple`s App Store oers a seemingly unlimited ariety o
unctions` and that Apple does allow outsiders to deelop applications on its
platorm` since the deeat o the Macintosh by \indows taught Jobs that a
platorm completely closed to outside deelopers is suicide.`
58
1hat should be
the end o the story. \et \u`s ear o that big proerbial kill switch` oerrides
all: Any day now, that switch will be thrown and Lessig`s pessimistic predictions
o perect control` will inally come to pass, he implies. As \u says, all
innoation and unctionality are ultimately subject to Apple`s eto.`
59
And
consider the lament o 1om Conlon o Povtar cievce: Once we replace the
personal computer with a closed-platorm deice such as the iPad, we replace

.vovg Recevt vartbove v,er., NILLSLN \IRL, Oct. 5, 2010,
#$$%&''*5/K)04,56,0)+/2'04,56,0(4B,'/0540,82/*45,'10-B/4-L2/6$L%/%.51BL
/%,B1$40KL676$,2L40L.L6L12/0KLB,+,0$L621B$%#/0,L*.7,B6
54
1ricia Duryee, .te Covtivvea 1o o.e |.. Mar/et.bare De.ite i/e rov iPbove 1 ate.,
MOCONL\S.NL1, Sept. 15, 2010, #$$%&''2/+/0,(6)0,$'1B$4+5,'@A>L1%%5,L+/0$40.,-L
$/L5/6,L.)6)L21B<,$6#1B,L-,6%4$,L6%4<,L3B/2L4%#/0,L@L61, Miguel lelt, 1he iPhone
las a Real light on Its lands, NL\ \ORK 1IMLS BI1S, Oct. 5, 2010,
#$$%&''*4$6)*5/K6)07$42,6)+/2'9:A:'A:':;'$#,L4%#/0,L#16L1LB,15L34K#$L/0L4$6L
#10-6'
55
$#,L40$,B0,$L+/2,L1*/.$)#$25
56
Grimmelmann & Ohm, .vra note 18 at 923.
5
a.
58
59
a.

reedom, choice, and the ree market with oppression, censorship, and
monopoly.`
60
But Apple is hardly the only game in town, and each time Apple
creates a new product category ,iPod, iPhone, iPad, etc.,, other companies are
quick to ollow with their own, usually more open systems, oten running
Google`s Android operating system.
Neither \u nor Zittrain, howeer, spend much time inestigating how oten
their proerbial kill switch is actually thrown-by Apple or anyone else. 1here
hae been a handul o examples, but those are hardly the rule. 1he ra.t
majority o all applications are immediately accepted and oered on the
platorm. Moreoer, i they rere blocked, they could quickly be ound on other
platorms. Again, there are plenty o alternaties to Apple products i you don`t
like their ,somewhat, more restrictie policies regarding application
deelopment.
Bottom line: 1oday`s supposed walled gardens` are less walled` than eer
beore, and closed` systems aren`t really so closed.
The Internet Was Never Quite
So Open or Generative
At times, Zittrain and others seem to hae created an Internet imago, an
idealized conception o a supposed better time when cyberspace was more open
and ibrant. But let`s ace it, the good ol` days` that many Openness
Langelicals seem to be longing or weren`t really so glorious. \ere you online
back in 1994 Did you enjoy 1rumpet \insock and noisy 14.4 baud modems
Did you like loading up multiple 5'-inch loppy disks just to boot your
machine Needless to say, most o us don`t miss those days.
lere`s the other orgotten actor about the Net`s early history: Until the Net
was commercialized, it was an extremely closed system. As Geert Loink
reminds us:
|In| |t|he irst decades|,| the Internet was a closed world, only
accessible to ,\estern, academics and the U.S. military. In
order to access the Internet one had to be an academic
computer scientist or a physicist. Until the early nineties it was
not possible or ordinary citizens, artists, business|es| or
actiists, in the USA or elsewhere, to obtain an email address

60
1om Conlon, 1be iPaa`. Cto.ea ,.tev: ovetive. ate eivg Rigbt, POPULAR SCILNCL, Jan. 29,
2010, ((()%/%6+4)+/2'K1-K,$6'1B$4+5,'9:A:L:A'4%1-eG9e?:e>>6L+5/6,-L676$,2L
6/2,$42,6L4L#1$,L*,40KLB4K#$

and make use o the rudimentary UNIX-based applications. .
It was a network o networks-but still a closed one.
61

Ironically, it was only because Lessig and Zittrain`s much-dreaded AOL and
CompuSere came along that many olks were een able to experience and
enjoy this strange new world called the Internet. 1he act that millions o
Americans or the irst time experienced the Internet through serices like AOL
,and continue to do so, is a reality that Zittrain simply oerlooks,` notes
Loink.
62
Could it be that those glorious good ol` days` Zittrain longs or were
really due to the way closed walled gardens` like AOL and CompuSere held
our hands to some extent and gae many new Netizens a guided tour o
cyberspace
Regardless, we need not reisit or reconsider that history. 1hat`s ancient history
now because the walls around those gardens came crumbling down.
Summary
\hen you peel away all the techno-talk and hand-wringing, what Zittrain and
other Openness Langelicals object to is the act that some people are making
choices that they don`t approe o. 1o be generous, perhaps it`s because they
beliee that the mere mortals` don`t ully understand the supposed dangers o
the choices they are making. But my contention here has been that things just
aren`t as bad as they make them out to be. More pointedly, who are these critics
to say those choices are irrational
Again, so what i some mere mortals choose more closed` deices or
platorms because they require less tinkering and just work` It isn`t the end o
the world. 1hose deices or platorms aren`t really as closed as they suggests-
in act, they are ar more open in some ways that the earlier technologies and
platorms Zittrain, et.at. gloriy. And it simply doesn`t ollow that just because
.ove consumers choose to use appliances` that it`s the end o the generatie
deices that others so cherish. General-purpose computers are so useul that
we`re not likely to abandon them,` notes Princeton Uniersity computer science
proessor Ld lelten.
63
lor example, a October 2010 NPD Group surey

61
Geert Loink, Zittraiv`. ovvaatiovat M,tb of tbe Oev vtervet, NL1 CRI1IQUL B\ GLLR1
LOVINK, Oct. 12, 2008,
#$$%&''0,$(/B<+.5$.B,6)/BK'(%2.'K,,B$'9::?'A:'A9'M4$$B1406L3/.0-1$4/015L27$#L
/3L$#,L/%,0L40$,B0,$'
62
a.
63
Ld lelten, iPaa to 1e.t Zittraiv`. vtvre of tbe vtervet 1be.i., lRLLDOM 1O 1INKLR blog, leb.
4, 2010, ((()3B,,-/2L$/L$40<,B)+/2'*5/K'3,5$,0'4%1-L$,6$LM4$$B1406L3.$.B,L40$,B0,$L
$#,646

reealed that contrary to popular belie, the iPad isn`t causing cannibalization
in the PC market because iPad owners don`t exhibit the same buying and
ownership patterns as the typical consumer electronics customer.`
64
According
to NPD, only 13 o iPad owners sureyed bought an iPad instead o a PC,
while 24 replaced a planned e-reader purchase with an iPad. 1hus, to the
extent the iPad was replacing anything, it would be other non-generatie`
deices like e-readers.
In a similar ein, James \atters, Senior Manager o Cloud Solutions
Deelopment at VMware, argues:
Innoation will be alie and well because the undamental
technologies at the core o cloud computing are designed or
massie, ibrant, explosie, awesome, and amazing application
innoation. 1here will always be a big place in the market or
companies who achiee design simplicity by limiting what can
be done on their platorms-Apple and lacebook may march
to massie market share by this principle-but as long as the
technologies underpinning the network are open,
programmable, extensible, modular, and dynamic as they are
and will be, innoation is in good hands.
65

1hus, we cav hae the best o both worlds-a world ull o plenty o tethered`
appliances, but also plenty o generatiity and openness. \e need not make a
choice between the two, and we certainly shouldn`t be demanding someone else
make it or us.
Against the Stasis Mentality
& Static Snapshots
1here are some important practical questions that the Openness Langelicals
oten ail to acknowledge in their work. Beyond the thorny question o how to
deine openness` and generatiity,` what metric should be used when existing
yardsticks become obsolete so regularly
1his points to two major ailings in the work o all the cyber-collectiists-
Lessig in Coae, Zittrain in vtvre of tbe vtervet, and \u in 1be Ma.ter ritcb:

64
êart, 0 Percevt of vitiat iPaa ate. are vcrevevtat ava vot Cavvibatiivg tbe PC Mar/et, .ccoraivg
to ^PD, NPD Group PRLSS RLLLASL, October 1, 2010,
((()0%-)+/2'%B,66'B,5,16,6'%B,668A:A::A)#$25
65
James \atters, ^Y1 Kic/. Off Ctova Paravoia erie., SILICONANGLL blog, July 21, 2009,
#$$%&''6454+/010K5,)+/2'*5/K'9::>':C'9A'07$L<4+<6L/33L+5/.-L%1B10/41L,-4$/B415L
6,B4,6

1. 1hey hae a tendency to adopt a static, snapshot iew o markets and
innoation, and,
2. 1hey oten express an oerly nostalgic iew o the past ,without
making it clear when the good old days` began and ended, while
adopting an excessiely pessimist iew o the present and the chances
or progress in the uture.
1his is what Virginia Postrel was reerring to in !"# %&'&(# )*+ ,'- .*#/0#- when
she criticized the stasis mentality because It oeralues the tastes o an
articulate elite, compares the real world o trade-os to antasies o utopia,
omits important details and connections, and conuses temporary growing pains
with permanent catastrophes.`
66
And it is what economist Israel Kirzner was
speaking o when warned o the shortsightedness o those who, not
recognizing the open-ended character o entrepreneurial discoery, repeatedly
all into the trap o orecasting the uture against the background o '1+)23-
expectations rather than against the unknowable background o tomorrow`s
discoeries.`
6

Indeed, there seems to be a complete lack o appreciation among the Openness
Langelicals or just how rapid and unpredictable the pace o change in the
digital realm has been and will likely continue to be. 1he relentlessness and
intensity o technological disruption in the digital economy is truly
unprecedented but oten under-appreciated. \eè had multiple mini-industrial
reolutions within the digital ecosystem oer the past 15 years. Again, this is
eolutionary dynamism` at work. ,Actually, it`s more like (#415&'01*)(2
dynamism!, Nothing-)6-15&'#52 *1'"0*7-that was sitting on our desks in 1995
is still there today ,in terms o digital hardware and sotware,. It`s unlikely that
much o what was on our desk in 899: is still there either-with the possible
exception o some crusty desktop computers running \indows XP. 1hus, at a
minimum, analysts o innoation in this space should . extend the time
horizon or our assessment o the generatie ecosystem`
68
to ensure they are
not guilty o the static snapshot problem.
Speaking o \indows, it perectly illustrates the complexity o deining
generatie systems. Compare the hal-lie o \indows PC operating systems-
which Zittrain indirectly gloriies in his book as generatiity nirana-to the
hal-lie o Android operating systems. Both Apple and Android-based deices

66
VIRGINIA POS1RLL, 1lL lU1URL AND I1S LNLMILS ,1998,, at xii-xiii.
6
ISRALL KIRZNLR, DISCOVLR\ AND 1lL CAPI1ALIS1 PROCLSS ,Uniersity o Chicago Press,
1985,, at xi.
68
Grimmelmann & Ohm, -&;() note 18 at 94.

hae seen multiple OS upgrades since release. Some application deelopers
actually complain about this rantic pace o mobile OS reolutions,` especially
with the Android OS, since they must deal with multiple deices and OS
ersions instead o just one Apple iPhone. 1hey`d rather see more OS
consistency among the Android deices or which they`re deeloping to
acilitate quicker and more stable rollouts. 1hey also hae to consider whether
and how to deelop the same app or seeral other competing platorms.
Meanwhile, \indows has oered a more stable` deeloping platorm or
deelopers because Microsot rolls out OS upgrades at a much slower pace.
Should we should consider an OS with a slower upgrade trajectory more
generatie` than an OS that experiences constant upgrades i, in practice, the
ormer allows or more open` ,and potentially rapid, independent innoation
by third parties O course, there other actors that play into the generatiity`
equation,
69
but it would be no small irony to place the \indows PC model on
the higher pedestal o generatiity than the more rapidly-eoling mobile OS
ecosystem.
Conclusion: Toward Evolutionary
Dynamism & Technological Agnosticism
\hether we are debating where arious deices sit on a generatiity continuum
,o open` ersus closed` systems,, or what its where on a code ailure`
continuum ,o perect code` ersus market ailure`,, the key point is that tbe
covtivvvv it.etf i. cov.tavtt, erotrivg and that this eolution is taking place at a much
aster clip in this arena than it does in other markets. Coders don`t sit still.
People innoate around ailure.` Indeed, market ailure` is really just the
glass-is-hal-empty iew o a golden opportunity or innoation. Markets
eole. New ideas, innoations, and companies are born. 1hings generally
change or the better-and do so rapidly.

69
|G|eneratiity is essential but can neer be absolute. No technological system is perectly
generatie at all leels, or all users, oreer. 1radeos are ineitable.` Grimmelmann and
Ohm, .vra note 18 at 923.

In light o the radical reolutions constantly unolding in this space and
upending existing models, it`s itally important we aoid deining down`
market ailure. 1his is not based on a blind aith in ree markets, but rather a
proound appreciation or the act that iv var/et. bvitt vov coae, tbe ace ava vatvre
of cbavge i. vvretevtivg ava vttert, vvreaictabte. Covtra Lessig`s lament in Coae that
Let to itsel, cyberspace will become a perect tool o control`-cyberspace
has proen ar more diicult to control` or regulate than any o us eer
imagined. Again, the olume and pace o technological innoation we hae
witnessed oer the past decade has been nothing short o stunning.
\e need to gie eolutionary dynamism a chance. Sometimes it`s during what
appears to be a gien sector`s darkest hour that the most exciting things are
happening within it-as the AOL case study illustrates. It`s easy to orget all the
anxiety surrounding AOL and its market power` circa 1999-2002, when
scholars like Lessig predicted that the company`s walled garden approach would
eentually spread and become the norm or cyberspace. As made clear in the
breakout aboe, howeer, the exact opposite proed to be the case. 1he critics
said the sky would all, but it most certainly did not.
Similarly, in the late 1990s, many critics-including goernments both here and
in the LU-claimed that Microsot dominated the browser market. Dour
predictions o perpetual Internet Lxplorer lock-in ollowed. lor a short time,
there was some truth to this. But innoators weren`t just sitting still, exciting
things were happening. In particular, the seeds were being planted or the rise
o lireox and Chrome as robust challengers to IL`s dominance-not to
mention mobile browsers. O course, it`s true that roughly hal o all websurers

still use a ersion o IL today. But IL`s share o the market is alling rapidly
0
as
iable, impressie alternaties now exist and innoation among these
competitors is more ibrant than eer.
1
1hat`s all that counts. 1he world
changed, and or the better, despite all the doomsday predictions we heard less
than a decade ago about Microsot`s potential dominance o cyberspace.
Moreoer, all the innoation taking place at the browser layer today certainly
undercuts the gloomy death o the Net` thesis set orth by Zittrain and others.
1hus, as O`Reilly argues, this case study again shows us the power o open
systems and eolutionary dynamism:
Just as Microsot appeared to hae eerything locked down in
the PC industry, the open Internet restarted the game, away
rom what eeryone thought was the main action. I guarantee
that i anyone gets a lock on the mobile Internet, the same
thing will happen. \e`ll be surprised by the innoation that
starts happening somewhere else, out on the ree edges. And
that ree edge will eentually become the new center, because
open is where innoation happens. |.| it`s ar too early to call
the open web dead, just because some big media companies
are excited about the app ecosystem. I predict that those same
big media companies are going to get their clocks cleaned by
small innoators, just as they did on the web.
2

In sum, history counsels patience and humility in the ace o radical uncertainty
and unprecedented change. More generally, it counsels what we might call
technological agnosticism.` \e should aoid declaring openness` a
sacrosanct principle and making eerything else subserient to it without regard
to cost or consumer desires. As Anderson notes, there are many \eb
triumphalists who still beliee that there is only One 1rue \ay, and will ight to
the death to presere the open, searchable common platorm that the \eb
represented or most o its irst two decades ,beore Apple and lacebook, to
name two, decided that there were Other \ays,.`
3
1he better position is one
based on a general agnosticism regarding the nature o technological platorms
and change. In this iew, the spontaneous eolution o markets has alue in its

0
1im Steens, vtervet torer att. etor :0 Percevt Ctobat Mar/et.bare, Cbrove |.age 1rite.,
LNGADGL1, Oct. 5, 2010, (((),0K1-K,$)+/2'9:A:'A:':;'40$,B0,$L,H%5/B,BL31556L
*,5/(L;:L%,B+,0$LK5/*15L21B<,$6#1B,L+#B
1
Nick \ingield & Don Clark, ror.er. Cet a aceift, \ALL S1RLL1 JOURNAL, Sept. 15, 2010,
#$$%&''/0540,)(6X)+/2'1B$4+5,'RIA:::A@9@:;9C@?C:@9?;A:@;C;@>9A:9;A@;?9?;=)#$25
2
1be !eb i. Deaa. . Debate, \IRLD, Aug. 1, 2010,
((()(4B,-)+/2'21K1M40,'9:A:':?'338(,*B4%8-,*1$,'155'A
3
a.

own right, and continued experimentation with new models-be they open`
or closed,` generatie` or tethered`-should be permitted.
Importantly, one need not beliee that the markets in code are perectly
competitie` to accept that they are competitie evovgb` compared to the
alternaties-especially those re-shaped by regulation. Code ailures` are
ultimately better addressed by oluntary, spontaneous, bottom-up, marketplace
responses than by coerced, top-down, goernmental solutions. Moreoer, the
decisie adantage o the market-drien, eolutionary approach to correcting
code ailure comes down to the rapidity and nimbleness o those responses.
Let`s gie those other orces-alternatie platorms, new innoators, social
norms, public pressure, etc.-a chance to work some magic. Lolution happens,
i you let it.
163

CHAPTER 3
IS INTERNET EXCEPTIONALISM DEAD?
The Third Wave of Internet Exceptionalism 165
!"#$ &'()*+,
A Declaration of the Dependence of Cyberspace 169
-(./ 0'1#,23# +,) 4'25 &'()6''7
Is Internet Exceptionalism Dead? 179
8#* 9:
Section 230 of the CDA: Internet Exceptionalism
as a Statutory Construct 189
;< ="#+, ;'((+,)
Internet Exceptionalism Revisited 209
>+"3 >+$?+"75@

164 CHAPTER 3: IS INTERNET EXCEPTIONALISM DEAD?


The Third Wave of Internet
Exceptionalism
By Eric Goldman
*

lrom the beginning, the Internet has been iewed as something special and
unique.` lor example, in 1996, a judge called the Internet a unique and
wholly new medium o worldwide human communication.`
1

1he Internet`s perceied noelty has prompted regulators to engage in Internet
exceptionalism`: crating Internet-speciic laws that dierge rom regulatory
precedents in other media. Internet exceptionalism has come in three distinct
waes:
The First Wave: Internet Utopianism
In the mid-1990s, some people antasized about an Internet utopia` that
would oercome the problems inherent in other media. Some regulators,
earing disruption o this possible utopia, sought to treat the Internet more
aorably than other media.
4 U.S.C. 230 ,Section 230`-a law still on the books, is a lagship example
o mid-1990s eorts to presere Internet utopianism. 1he statute categorically
immunizes online proiders rom liability or publishing most types o third
party content. It was enacted ,in part, to presere the ibrant and competitie
ree market that presently exists or the Internet and other interactie computer
serices, unettered by lederal or State regulation.`
2
1he statute is clearly
exceptionalist because it treats online proiders more aorably than oline
publishers-een when they publish identical content.
The Second Wave: Internet Paranoia
Later in the 1990s, the regulatory pendulum swung in the other direction.
Regulators still embraced Internet exceptionalism, but instead o aoring the
Internet, regulators treated the Internet more harshly than analogous oline
actiity.
lor example, in 2005, a 1exas website called Lie-shot.com announced that it
would oer Internet hunting.` 1he website allowed paying customers to

Associate Proessor and Director, ligh 1ech Law Institute, Santa Clara Uniersity School o
Law. Lmail: !"#$%&'()"&'*$+,#&. \ebsite: -../011222+!3*,"#$%&'(+#3".
1
American Ciil Liberties Union . Reno, 929 l. Supp. 824 ,L.D. Pa. 1996,.
2
4 U.S.C. 230,b,,2,.

control, ia the Internet, a gun on its game arm. An employee manually
monitored the gun and could oerride the customer`s instructions. 1he website
wanted to gie people who could not otherwise hunt, such as paraplegics, the
opportunity to enjoy the hunting experience.
3

1he regulatory reaction to Internet hunting was swit and seere. Oer three-
dozen states banned Internet hunting.
4
Caliornia also banned Internet ishing
or good measure.
5
loweer, regulators neer explained how Internet hunting
is more objectionable than physical space hunting.
lor example, Caliornia Sen. Debra Bowen criticized Internet hunting because it
isn`t hunting, it`s an inhumane, oer the top, pay-per-iew ideo game using
lie animals or target practice . . Shooting lie animals oer the Internet
takes absolutely zero hunting skills, and it ought to be oensie to eery
legitimate hunter.`
6

Sen. Bowen`s remarks relect numerous unexpressed assumptions about the
nature o hunting` and what constitutes air play. In the end, howeer,
hunting may just be hunting,` in which case the response to Internet hunting
may just be a typical example o aderse Internet exceptionalism.

The Third Wave:
Exceptionalism Proliferation
1he past ew years hae brought a new regulatory trend. Regulators are still
engaged in Internet exceptionalism, but each new adance in Internet
technology has prompted exceptionalist regulations towards that technology.
lor example, the emergence o blogs and irtual worlds has helped initiate a
push towards blog-speciic and irtual world-speciic regulation. In eect,
Internet exceptionalism has splintered into pockets o smaller exceptionalist
eorts.

3
Sylia Moreno, Mov.e Ctic/ rivg. ove 1britt of tbe vvt, \ASl. POS1, May 8, 2005.
4
vtervet vvtivg av., 1he lumane Society o the United States,
-../011222+-454+#3"12!678*$!419:;1*(.!3(!.-5(.*("<&'/+/%8 ,last isited Aug. 23,
2010,.
5
Zachary M. Seward, vtervet vvtivg a. Cot to to - f t rer tart., \ALL S1. J., Aug. 10,
200.
6
Michael Gardner, !eb vvt.` iv Cro.. air. of arva/er., S.D. UNION-1RIBUNL, Apr. 6,
2005.
Lric Goldman, . !eb ite for vvtivg Po.e. Qve.tiov. .bovt Kittivg, S.J. MLRCUR\ NL\S, July
25, 2005.

Regulatory responses to social networking sites like lacebook and MySpace are
a prime example o Internet exceptionalism splintering. Rather than regulating
these sites like other websites, regulators hae sought social networking site-
speciic laws, such as requirements to eriy users` age,
8
combat sexual
predators
9
and suppress content that promotes iolence.
10
1he result is that the
regulation o social networking sites diers not only rom oline enterprises but
rom other websites as well.
Implications
Internet exceptionalism-either aoring or disaoring the Internet-is not
inherently bad. In some cases, the Internet truly is unique, special or dierent
and should be regulated accordingly. Unortunately, more typically, anti-
Internet exceptionalism cannot be analytically justiied and instead relects
regulatory panic.
In these cases, anti-Internet regulatory exceptionalism can be harmul, especially
to Internet entrepreneurs and their inestors. It can distort the marketplace
between \eb enterprises and their oline competition by hindering the \eb
business` ability to compete. In extreme cases, such as Internet hunting,
unjustiied regulatory interention may put companies out o business.
Accordingly, beore enacting any exceptionalist Internet regulation ,and
especially any anti-Internet regulation,, regulators should articulate how the
Internet is unique, special or dierent and explain why these dierences justiy
exceptionalism. Unortunately, emotional oerreactions to perceied Internet
threats or harms typically trump such a rational regulatory process. Knowing
this tendency, perhaps we can better resist that temptation.

8
Nick Alexander, .ttorve,. Ceverat .vvovvce .greevevt !itb M,ace Regaraivg ociat êtror/ivg
afet,, NAA GAZL11L, Jan. 18, 2008,
-../011222+(''"+#3"1'..#3(!=4<"!(!3'$<'((#5(,!<'"3!!&!(.<2*.-<&=4/',!<3!"'
3%*("<4#,*'$<(!.2#3>*("<4'8!.=+/-/, Brad Stone, aceboo/ ettte. ritb êr Yor/, N.\.
1IMLS BI1S BLOG, Oct. 16 200, -../0116*.4+6$#"4+(=.*&!4+,#&1?@@A1B@1BC18',!6##>7
4!..$!472*.-7(!27=#3>1.
9
KIDS Act o 200 ,l.R. 19,S. 431, ,requiring sexual predators to register their email
addresses and other screen names and enabling social networking sites to access those
electronic identiiers so that the sexual predators can be blocked rom registering with the
social networking sites,.
10
l. Res. 224 ,200, ,resolution requesting that social networking sites proactiely remoe
enemy propaganda rom their sites,` such as ideos made by terrorists,.


A Declaration of the
Dependence of Cyberspace
By Hon. Alex Kozinski
*
& Josh Goldfoot
**

!"#$%&'$&() "+ (,$ -&./)(%012 3"%2.4 5"/ 6$1%5 701&() "+ +2$), 1&. )($$24
- 8"'$ +%"' 95:$%);18$4 (,$ &$6 ,"'$ "+ <0&.= >& :$,12+ "+ (,$ +/(/%$4
- 1)? 5"/ "+ (,$ ;1)( (" 2$1#$ /) 12"&$= @"/ 1%$ &"( 6$28"'$ 1'"&7 /)=
@"/ ,1#$ &" )"#$%$07&(5 6,$%$ 6$ 71(,$%=
1

1hat was the opening o A Declaration o the Independence o Cyberspace.`
1he would-be Cyber-Jeerson who wrote it was John Perry Barlow, a co-
ounder o the Llectronic lrontier loundation, a noted libertarian and a
Grateul Dead lyricist. le deliered the Declaration on lebruary 8, 1996, the
same day that President Clinton signed into law the Communications Decency
Act. 1hat Act was chiely an early eort to regulate Internet pornography.
Many had concerns about that law, and, indeed, the Supreme Court would
eentually declare most o it unconstitutional.
2

Barlow`s argument inoked what he belieed was a more decisie criticism than
anything the Supreme Court could come up with. Barlow saw the Internet as
literally untouchable by our laws. Lxtolling the power o anonymity, he taunted
that our identities hae no bodies, so, unlike you, we cannot obtain order by
physical coercion.` Unlike the Declaration o Independence, this was not a
declaration that cyberspace was newly independent, it was an obseration that
cyberspace had always been independent, and will always remain independent,
because its denizens were beyond the law`s reach.
Needless to say, the weary giants o lesh and steel did not take kindly to the
Declaration. 1hey ought back hard and won numerous battles: witness the all
o Napster, Grokster, Aimster and innumerable other ile-sharing and child-
pornography-trading sites and serices. Ironically, the Department o

Chie Judge, United States Court o Appeals or the Ninth Circuit.

B.A., \ale Uniersity, J.D., Uniersity o Virginia School o Law, 1rial Attorney, Department
o Justice, Criminal Diision, Computer Crime & Intellectual Property Section. 1his essay
was originally published in 32 COLUMBIA JOURNAL Ol LA\ & 1lL AR1S, no. 4, 2009 at 365.
1he iews expressed in this essay are the iews o the authors and do not necessarily
represent the iews o the U.S. Department o Justice or the United States.
1
John Perry Barlow, A Declaration o the Independence o Cyberspace ,leb. 8, 1996,,
!""#$%%!&'()*(++*&,-%./0,1&2%3(410,0"5&6785601*!"'1.
2
Reno . American Ciil Liberties Union, 521 U.S. 844 ,199,=

lomeland Security now has a National Strategy to Secure Cyberspace.`
3
Len
the cyber-libertarians hae shited their ocus: 1he Llectronic lrontier
loundation, which Barlow co-ounded, now accepts that there may be a place
or so-called network neutrality` regulation, een though it regulates how
subscribers access the Internet and how content reaches them.
4

In other ways, the Declaration has proed prescient. As ar back as 1996,
Barlow had identiied that the Internet poses a signiicant problem or
goernments. 1hen, as now, people used the Internet to break the law. 1he
Internet gies those people two powerul tools that help them escape the law`s
eorts to ind and punish them. lirst, the Internet makes anonymity easy.
1oday any 11-year-old can obtain a ree e-mail account, ree website and ree
ideo hosting. 1he companies that proide these things ask or your name, but
they make no eort to eriy your answer, as a result, only Boy Scouts tell them
the truth. \ou can be tracked through your Internet protocol ,IP, address, but
it is not too tough to use proxies or some neighbor`s open \i-li connection to
get around that problem. 1hus, i your online conduct eer hurts someone, it
will be diicult or the ictim to eer ind out who you are and sue you.
Second, the Internet makes long-distance international communication cheap.
1his allows the world`s miscreants, con-artists and thiees easy access to our
gullible citizens. \hen people ind out theyè been had, they oten ind that
they hae no practical recourse because o the extraordinary diiculties inoled
in pursuing someone oerseas. 1he Internet`s global nature makes it easy or
people to hide rom our courts.
1hese two adantages o Internet law-breakers pose a serious and recurring
problem. 1hat problem has been particularly painul or intellectual property
rights holders. It is common knowledge that instead o buying music or
moies, you can use the Internet to download perect copies or ree rom
indiiduals known only by their IP addresses. In some cases, wrongdoers hae
become so bold that they demand payment in exchange or the opportunity to
download inringing material.
1he situation seemed unsolable to Barlow and others in 1996. Armed with
anonymity and inulnerability, Internet actors could ignore eorts to apply law
to the Internet. Barlow concluded that the Internet`s nature posed an
insurmountable barrier to any eort at legal enorcement. Some scholars een

3
1he National Strategy to Secure Cyberspace, leb. 2003,
-../011222+%-4+"#D1E$*63'3=1'44!.41F'.*#('$<G=6!34/',!<H.3'.!"=+/%8.
4
ee -../4011222+!88+#3"18*$!418*$!(#%!1((1I;;FF,#&&!(.4+/%8.

began work on theorizing how the dierse denizens o cyberspace might join
together and go about creating their own indigenous legal system.
5

But oer time, a solution to Barlow`s problem appeared. Let us entertain, or a
moment, the conceit that there is a cyberspace,` populated by people who
communicate online. 1he denizens o cyberspace exist simultaneously in
cyberspace and in the real lesh-and-steel world. 1heir cyberspace seles can be
completely anonymous, their real-lie seles are easier to identiy. 1heir
cyberspace seles hae no physical presence, their real-lie seles both exist and
hae base material desires or PlayStations, Porsche Boxsters and attte.tar
Catactica memorabilia. 1heir physical seles can be ound in the real world and
made to pay in real dollars or sere real time behind real bars or the damage
their cyber-seles cause.
1he dilemma that online law-breakers ace is that their cyberspace crimes hae
real-lie moties and ulill real-lie needs. 1hereore, they need some way to
translate their online misdeeds into oline beneits. 1he teenager downloads a
MP3 so that he can listen to it. 1he con-artist asks or money to be wired to
him so that he can withdraw it and buy things with it. 1he ringe actiist who e-
mails a death threat to a judge does so in the hopes that the judge will change
his behaior in the real world.
1hese Internet actors usually rely on real-world institutions to get what they
want. 1hey use Internet Serice Proiders ,ISPs, and hosting companies to
communicate, and they use banks and credit card companies to turn online
gains into cash. \ithout these institutions, they either could not accomplish
their online harms, or they would not be able to beneit rom them in the real
world. Unlike anonymous cyberspace miscreants, howeer, these institutions
hae street addresses and real, physical assets that can satisy judgments in the
United States. By placing pressure on those institutions to cut o serice to
customers who break the law, we can indirectly place pressure on Internet
wrong-doers. 1hrough this pressure, we hae a powerul tool to promote
online compliance with the law.
In some cases, or some oenses, we hae the legal tools to do this already. lor
intellectual property cases, the tool or holding those institutions liable is
secondary liability: contributory and icarious inringement. 1he Ninth Circuit
has led the way in deeloping the law in this area. In Perfect 10 r. Coogte, the
court noted the cases that had applied contributory inringement to Internet
actors, and summarized their holdings as saying that a computer system
operator can be held contributorily liable i it has actual knowledge that speciic
inringing material is aailable using its system . and can take simple measures

5
ee, e.g., Daid G. Post, .varcb,, tate, ava tbe vtervet: .v ..a, ov arMa/ivg iv C,ber.ace,
JOURNAL Ol ONLINL LA\, Article 3, ,1995,, araitabte at -../011443(+,#&1'64.3',.JKLMLNC.

to preent urther damage to copyrighted works . yet continues to proide
access to inringing works.`
6
In other words, i people are using your stu to
inringe copyrights, and you know about it, and you can easily stop them, but
you do not, then you are on the hook.
1he motie behind secondary liability is simple. Leryone agrees that the direct
inringers ideally should be the ones to pay. But there might be too many o
them to sue, or, they might be anonymous, or, they might be in Nigeria. 1his
can make them apparently inulnerable to lawsuits. 1hat inulnerability has a
cause: someone is proiding the tools to inringe and looking the other way.
1he doctrine o secondary liability says that such behaior is unacceptable.
1hose who proide powerul tools that can be used or good or eil hae some
responsibility to make sure that those tools are used responsibly.
Put more directly: with some changes to the law, the institutions that enable the
anonymity and inulnerability o cyberspace denizens can be held accountable
or what their anonymous and inulnerable customers do. 1he anonymity o
cyberspace is as much a creation o men as it is a creation o computers. It is
the result o policy choices. \e hae accepted, without serious examination,
that it is perectly ine or a business to grant ree \eb space and e-mail to any
schmuck who comes o the street with an IP address, and then either keep no
record o that grant or discard the record quickly. Businesses that do this are
lending their powerul and potentially harmul capabilities and demanding little
accountability in return. 1hat arrangement has obious beneits but also
obious costs. 1he ictims o online torts and crimes bear these costs, and
those ictims are, oerwhelmingly, third parties. 1hey include big moie
studios, middle-aged Internet newbies and, unortunately in some cases, young
children.
I the legal rules change, and companies are held liable more oten or what
their users do, then the cost o anonymity would shit away rom ictims and
toward the proiders. In this world, proiders will be more careul about
identiying users. Perhaps online assertions o identity will be backed up with
oline proo, proiders will be more careul about proiding potential scam
artists in distant jurisdictions with the tools to practice their crat. All this
would be expensie or serice proiders, but not as expensie as it is or
injured parties today.
Secondary liability should not reach eery company that plays any hand in
assisting the online wrong-doer, o course. Beore secondary liability attaches,
the plainti must show that the deendant proided a crucial serice, knew o
the illegal actiity, and had a right and a cost-justiied ability to control the

6
Perect 10, Inc. . Amazon.com, Inc., 508 l.3d 1146, 112 ,9th Cir. 200, ,quotations,
citations and italics omitted,.

inringer`s actions. 1his rule will in almost eery case exclude electrical utilities,
landlords, and others whose contributions to illegal actiity are minuscule.
\hile we hae come a long way rom Barlow`s Declaration o the
Independence o Cyberspace, the central idea behind it-that the Internet is a
special place, separate somehow rom the brick and mortar world, and thus
subject to special rules and regulations, or no rules and regulations-lingers.
1he name itsel has a powerul inluence: we don`t speak o telephone-space`
or radio-space` or 1V-space`-though we do hae 1eleision City in
lollywood. Prior technological adances that aided in connecting people were
generally recognized as tools to aid lie in the real world, no one claimed that
they made up a separate dimension that is somehow dierent and separate rom
the real word. Lery time we use the term cyberspace` or the now-outmoded
Inormation Superhighway,` we buy into the idea that the world-wide network
o computers that people use or electronic commerce and communication is a
separate, organic entity that is entitled to special treatment.
1his idea o cyberspace as a separate place subject to a dierent set o rules-
one where courts ought to tread lightly lest they disturb the natural order o
things and thereby cause great harm-still arises in many court cases.

1he irst o these is Perfect 10 r. 1i.a-a case where one o the authors o this
piece was in the dissent.
8
1he acts are simple: plainti produces and owns
pictures o scantily-clad young women, which it sells online. It alleged that
unknown parties had copied the pictures and were selling them online, at a
lower price, using serers in remote locations where the legal system was not
hospitable to copyright and trademark lawsuits, and, moreoer, they could old
up their tents and open up business elsewhere i anyone really tried to pursue
them. So the plainti didn`t try to sue the primary inringers, instead, it went
ater the credit card companies that were processing the payments or what they
claimed were pirated photographs.

Some disclaimers: One o the authors o this piece ,Chie Judge Kozinski, sat on the panel
that decided some o the cases gien as examples here. le wants to make it clear that he
won`t re-argue the cases here. Both inoled split decisions, and his iews as to how those
cases should hae come out is set out in his opinions in those cases. lis colleagues on the
other side are not present to argue their positions and, in any eent, it`s unseemly to continue
a judicial debate ater the case is oer. lurthermore, despite his disagreement with his
colleagues, he respects and appreciates their iews. 1he judges that came out the other way
are some o the dearest o his colleagues, and some o the inest judges anywhere. 1he
disagreement is troubling, because they bring a wealth o intelligence, diligence, talent,
experience and objectiity to the problem, and he can`t quite igure out why they see things
so dierently.
8
Perect 10, Inc. . Visa Int`l Ser. Ass`n, 494 l.3d 88 ,9th Cir. 200,.

1his was by ar not the irst case that applied the doctrine o secondary
inringement to electronic commerce. 1he cases go back at least to the 1995
case o Retigiov. 1ecbvotog, Cevter r. êtcov,
9
a case inoling the liability o an ISP
or damage caused when it posted copyrighted Scientology documents to
USLNL1, at the direction o one o its users. And, o course, the Napster,
Aimster and Grokster cases all dealt with the secondary liability o those who
assist others in inringement.
10
Perfect 10, though, presented a noel question:
how do you apply the doctrine o secondary inringement to people who help
the transaction along, but neer hae any physical contact with the protected
work
1wo excellent and conscientious Ninth Circuit jurists, Judges Milan Smith and
Stephen Reinhardt, said there was no liability, whereas the dissenting judge
concluded that there was. Visa, the dissent argued, was no dierent rom any
other company that proided a serice to inringers, knew what it was doing,
and had the ability to withdraw its serice and stop the inringement, but did
nothing.
1his debate its within a larger context. In the majority`s rejection o
contributory liability, it cited a public policy decision that ound that the
Internet`s deelopment should be promoted by keeping it ree o legal
regulation. Relatedly, the majority distinguished some precedent by saying that
its tests were deeloped or a brick-and-mortar world` and hence do not lend
themseles well to application in an electronic commerce context.`
11

1his argument channels Barlow`s declaration that users o the Internet are
entitled to special treatment ,or, as he would hae it, entitled to no treatment,.
1he chie justiication or this argument is that the Internet is so new, exotic
and complicated that the imposition o legal rules will chill, stile, discourage or
otherwise squelch the budding geniuses who might otherwise create the next
Google, Pets.com, or lamsterDance.com. lor example, the Llectronic
lrontier loundation argued to the Supreme Court during the Grokster case that
i the Ninth Circuit`s opinion were reersed, the eect would threaten
innoation by subjecting product design to expensie and indeterminate judicial

9
Religious 1ech. Ctr. . Netcom, 90 l.Supp. 1361 ,N.D. Cal. 1995,.
10
ee A&M Records, Inc. . Napster, Inc., 284 l.3d 1091 ,9th Cir. 2002,, In re Aimster
Copyright Litig., 334 l.3d 643 ,th Cir. 2003,, Metro-Goldwyn-Mayer Studios Inc. .
Grokster, Ltd., 545 U.S. 913 ,2005,.
11
Perfect 10, 494 l.3d at 98, n.9.

second-guessing.`
12
1he Ninth Circuit was reersed, and i that decision slowed
the pace o product design, no one seems to hae noticed.
1his argument became particularly central in a second case, air ov.ivg Covvcit
of av ervavao 1atte, r. Roovvate..cov.
13
1he case inoled a claim that the
commercial website Roommates.com iolated state and ederal air housing
laws by helping to pair up roommates according to their personal preerences,
the exercise o which is allegedly prohibited by law. Again, one o the authors
o this piece was a judge on that case, and was in the majority at both the panel
and the ev bavc leel-despite the eorts o some conscientious and brilliant
dissenting judges, o whose intellectual rigor and commitment to the rule o law
no one can doubt.
1he majority mostly held that Roommates.com could be held liable, i the
plainti`s allegations were proen true. 1he court held essentially that an online
business had to be held to the same substantie law as businesses in the brick-
and-mortar world. 1he dissenters saw things quite dierently, to them, the
majority placed in jeopardy the surial o the Internet. lere is a taste o the
dissent:
On a daily basis, we rely on the tools o cyberspace to help us
make, maintain, and rekindle riendships, ind places to lie,
work, eat, and trael, exchange iews on topics ranging rom
terrorism to patriotism, and enlighten ourseles on subjects
rom aardarks to Zoroastrianism.` . 1he majority`s
unprecedented expansion o liability or Internet serice
proiders threatens to chill the robust deelopment o the
Internet that Congress enisioned . . \e should be looking
at the housing issue through the lens o the Internet, not rom
the perspectie o traditional publisher liability.
14

And inally, the unkindest cut o all: 1he majority`s decision, which sets us
apart rom ie circuits, . iolates the spirit and serendipity o the Internet.`
15

1he argument that a legal holding will bring the Internet to a standstill makes
most judges listen closely. Just think o the panic that was created when the

12
ee Brie or Respondents, Metro-Goldwyn-Mayer Studios Inc. . Grokster, Ltd., No. 04-480
,9th Cir. Mar. 1, 2005,, aailable at 2005 \L 508120 and at
-../0112?+!88+#3"1O919?91PQP<D<Q3#>4.!31?@@N@M@B<3!4/#(%!(.4<63*!8+/%8.
13
lair lousing Council o San lernando Valley . Roommates.Com, LLC, 521 l.3d 115 ,9th
Cir. 2008,.
14
a. at 116- ,ootnote omitted,.
15
a. at 11.

Blackberry serer went down or a ew hours. No one in a black robe wants to
be responsible or anything like that, and when intelligent, hard-working,
thoughtul colleagues argue that this will be the eect o one o your rulings,
you hae to think long and hard about whether you want to go that way. It
tests the courage o your conictions.
Closely related is the argument that, een i you don`t bring down the existing
structure, the threat o liability will stile innoation, so that the progress we
hae seen in recent years-and the gains in productiity and personal
satisaction-will stop because the legal structure has made innoation too risky
or expensie. 1he innoation argument is partly right but mostly wrong.
Certainly, some innoators will shy away rom legally murky areas. It`s hard to
think o a worse recipe or creatiity than haing a lawyer attend eery
engineering meeting. But promoting innoation alone cannot be a suicient
justiication or exempting innoators rom the law. An unortunate result o
our complex legal system is that almost eeryone is conused about what the
law means, and eeryone engaged in a business o any complexity at some point
has to consult a lawyer. I the need to obey the law stiles innoation, that
stiling is just another cost o haing a society ruled by law. In this sense, the
Internet is no dierent than the pharmaceutical industry or the auto industry:
1hey ace ormidable legal regulation, yet they continue to innoate.
1here is an een more undamental reason why it would be unwise to exempt
the innoators who create the technology that will shape the course o our lies:
Granting them that exemption will yield a generation o technology that
acilitates the behaior that our society has decided to prohibit. I the Internet
is still being deeloped, then we should do what we can to guide its
deelopment in a direction that promotes compliance with the law.
lor example, what use is innoation` in creating a job hunting site i the
innoators produce a site that inites employers to automatically reject any
applicant rom a particular race Perhaps the job site is a bold new innoation
that makes hiring ar easier and more eicient than it has eer been. But i this
site is used widely, it will acilitate racial discrimination in hiring-conduct that
society has already decided it must prohibit. Similarly, is a ile-sharing serice
such as Grokster worth the harm it causes by oering no built-in tools or
identiying participants or establishing they hae the right to share` the iles
they copy lar rom exempting this growing industry rom the law, we should
igorously enorce the law as the industry grows, so that when it is mature, its
serices won`t guide behaior toward conduct that society has decided to
discourage. As diicult as it might be or innoators today, it is easier than the
alternaties: orcing them to rebuild eerything ten years down the road, or
grudgingly accepting that we hae surrendered key aspects o our ability to
goern our society through law.

It is Barlow who is generally credited with taking the word cyberspace` rom
the science iction o \illiam Gibson and applying it to the Internet.
16
In doing
so, he launched the conceit that such a space` exists at all. 1his was wholly
unjustiied. It is a mistake to all into Barlow`s trap o belieing that the set o
human interactions that is conducted online can be neatly grouped together into
a discrete cyberspace` that operates under its own rules. 1echnological
innoations gie us new capabilities, but they don`t change the undamental
ways that humans deal with each other. 1he introduction o telephones and
cars did create new legal questions. 1hose questions all reoled around what
the acceptable uses o the new technologies were. low closely can you ollow
the car in ront o you on the highway Can you repeatedly dial someone`s
phone to annoy them Can you tap into a phone conersation or put a tape
recorder in a phone booth Oer time, courts and legislatures answered these
questions with new legal rules. 1hey had to, the essence o the controersy
arose rom the new technological abilities. But no one thought that telephones
and cars changed the legal rules surrounding what was said on a telephone or
where a car traeled. Can an oral contract be ormed with a telephone call O
course, it is still two people speaking. Is it trespassing to drie across my
neighbor`s ront yard O course, you are on his land.
Like cars and telephones, the Internet prompts new questions about the
acceptable uses o the new technology. Is port-scanning a orm o hacking
\hen does title to a domain name legally transer \hile analogies to settled
legal rules are helpul in answering these questions, they are not conclusie.
Answers to these questions will look like new legal rules.
But when the Internet is inoled in a controersy only because the parties
happened to use it to communicate, new legal rules will rarely be necessary.
\hen the substance o the oense is that something was communicated, then
the harm occurs regardless o the tools used to communicate. I an attorney
betrays a client`s conidence, the duty to the client is breached regardless o
whether the attorney used a telephone, a newspaper, a radio station, or the
Internet. 1he choice o communication medium might aect the magnitude o
the harm, but i it is illegal or A to communicate X to B without C`s
permission, there is no reason to ashion new rules o liability that depend on
the mode o communication used.
1here are some ways that the Internet might require courts to re-think legal
rules. 1he Internet makes long-distance communication cheaper than it was
beore. 1o the extent that existing legal rules were premised on the assumption
that communications were expensie, the Internet might require a reappraisal.
Courts are already reealuating, or example, what it means to do business

16
ee John Perry Barlow, Crive ava Pvtevevt: v .aravce of tbe ar ov tbe tectrovic rovtier,
\lOLL LAR1l RLV., Sept. 22, 1990, at 44, 45.

within a state, or purposes o the long-arm statute, when the deendant`s
business establishment` is a serer located in Uzbekistan.
\et the ast majority o Internet cases that hae reached the courts hae not
required new legal rules to sole them. It has been iteen years since America
Online unleashed its hordes o home computing modem-owners on e-mail and
the Internet and iteen years since the release o the Mosaic \eb browser.
Ater all that time, we hae today relatiely ew legal rules that apply only to the
Internet. Using the Internet, people buy stocks, adertise used goods and apply
or jobs. All those transactions are goerned by the exact same laws as would
goern them i they were done oline.
1hose who claim the Internet requires special rules to deal with these ordinary
controersies hae trouble explaining this history. Despite this dearth o
Internet-speciic law, the Internet is doing wonderully. It has suried
speculatie booms and busts, made millionaires out o many and, unortunately,
rude bloggers out o more than a ew. 1he lack o a special Internet ciil code
has not hurt its deelopment.
1he Internet, it turns out, was neer so independent or soereign as early
idealists belieed. It was an astounding social and technological achieement,
and it continues to change our lies. But it has not proen to be inulnerable to
legal regulation-at least, not unless we choose to make it inulnerable. As
intriguing as Barlow`s Declaration o Independence was, the original 16
Declaration is more proound in its understanding o the purpose and abilities
o goernment: men hae rights o Lie, Liberty and the pursuit o
lappiness,` and to secure these rights, Goernments are instituted among
Men.` 1he goernment that we hae instituted retains its purpose o securing
those rights, and it accomplishes that purpose through the law. \e hae seen
that our goernment has many tools at its disposal through which it can bring
law to the Internet`s ar reaches. 1he Internet might pose obstacles toward that
job, but those obstacles can be oercome. 1he question is whether we will do it.

Is Internet Exceptionalism Dead?
By Tim Wu
*

In 1831, Alexis de 1ocqueille released Devocrac, iv .verica, the ounding text
o American exceptionalism.` Ater long study in the ield, America, he had
concluded, was just dierent than other nations. In an oten-quoted passage, de
1ocqueille wrote:
1he position o the Americans is thereore quite exceptional,
and it may be belieed that no democratic people will eer be
placed in a similar one. 1heir strictly Puritanical origin-their
exclusiely commercial habits-een the country they inhabit,
which seems to diert their minds rom the pursuit o science,
literature, and the arts-the proximity o Lurope, which allows
them to neglect these pursuits without relapsing into
barbarism-a thousand special causes, o which I hae only
been able to point out the most important-hae singularly
concurred to ix the mind o the American upon purely
practical objects. lis passions, his wants, his education, and
eerything about him seem to unite in drawing the natie o
the United States earthward, his religion alone bids him turn,
rom time to time, a transient and distracted glance to heaen.
1

Is there such a thing as Internet exceptionalism I so, just what is the Internet
an exception to It may appear technical, but this is actually one o the big
questions o our generation, or the Internet has shaped the United States and
the world oer the last twenty years in ways people still struggle to understand.
lrom its beginnings the Internet has always been dierent rom the networks
that preceded it-the telephone, radio and teleision, and cable. But is it
dierent in a lasting way
1he question is not merely academic. 1he greatest Internet irms can be
succinctly deined as those that hae best understood what makes the Internet
dierent. 1hose that hae ailed to understand the Network o Networks`-
say, AOL, perished, while those that hae, like Google and Amazon, hae
lourished. lence the question o Internet exceptionalism is oten a multi-
billion dollar question. 1he state o the Internet has an obious eect on
national and international culture. It is also o considerable political releance,

Proessor, Columbia Law School, lellow, New America loundation

1
ALLXIS DL 1OCQULVILLL, DLMOCRAC\ IN AMLRICA 519 ,lenry Reee trans., D. Appleton
and Company 1904, ,1831,.

both or enorcement o the laws, and the rise o candidates and social
moements.
\hat makes the question so interesting is that the Internet is both obiously
exceptional and unexceptional at the same time. It depends on rbat you might
think it is an exception to. It is clear that the Internet was a dramatic reolution
and an exception to the ordinary ways o designing communications systems.
But whether it enjoys a special immunity to the longer and deeper orces that
shape human history is, shall we say, yet to be seen.
R R R
In the early 2000s, Jack Goldsmith and I wrote !bo Covtrot. tbe vtervet.
2
1he
book is an explicitly anti-exceptionalist work. It addressed one particular way
that the Internet might be an exception, namely, the susceptibility, as it were, o
the Internet to regulation by the laws o nations. lrom the mid-1990s onward it
was widely thought that the Internet would proe impossible to control or
regulate. Some legal scholars, in interesting and proocatie work, argued that
in some ways the Network might be considered to hae its own soereignty,
like a nation-state.
3
1hat was the boldest claim, but the general idea that the
Internet was diicult or impossible to regulate was, at the time, a political,
journalistic and academic commonplace, taken or granted. lor example,
relecting his times, in 1998 President Clinton gae a speech about China`s
eorts to control the Internet. Now, there`s no question China has been
trying to crack down on the Internet-good luck` he said. 1hat`s sort o like
trying to nail Jello to the wall.`
4

1hat was the conentional wisdom. In our book we suggested that despite the
wonders o the Network it did not present an existential challenge to national
legal systems, reliant, as they are, on threats o physical orce.
5
\e predicted
that nations would, and to some degree already had, reassert their power oer
the Network, at least, or matters they cared about. 1hey would assert their
power not oer the Network in an abstract sense, but the actual, physical
humans and machinery who lie underneath it. Many o the book`s chapters
ended with people in jail, unsurprisingly, China proided the strongest example
o what a State will do to try to control inormation within its borders.

2
1IM \U & JACK GOLDSMI1l, \lO CON1ROLS 1lL IN1LRNL1 ,2006,.
3
Daid Post & Daid Johnson, ar ava oraer.-1be Ri.e of ar iv C,ber.ace, 48 S1AN. L. RLV.
136 ,1996,.
4
R. MIClALL ALVARLZ & 1lAD L. lALL, POIN1, CLICK AND VO1L: 1lL lU1URL Ol
IN1LRNL1 VO1ING 3 ,2004,.
5
JOlN AUS1IN, 1lL PROVINCL Ol JURISPRUDLNCL DL1LRMINLD ,\ilrid L. Rumble, ed.,
Cambridge Uni. Press 1995, ,1832,.

Drama aside, in a deeper way, we were interested in what you might call the
persistence o physicality. Despite its irtual qualities, behind the concept o a
global network were liing human beings, blood and lesh. 1he human body`s
susceptibility to pain and imprisonment is a large part o what the nation-state
bases its rule on, and that had not changed. \e predicted that the nation`s
threat o physical orce, otherwise known as laws, would thereore shape the
Network as much as its ounding ambitions.
lere is how we put the point in the introduction to our book, written in about
2005 or so:
Our age is obsessed with the search or the newest new
thing.` Our story, by contrast, is about old things-ancient
principles o law and politics within nations, cooperation and
clashes between nations, and the enduring releance o
territory, and physical coercion. It is a story where 1homas
lobbes is as important as Bill Gates. Like it or not, these old
things are as important to the Net`s deelopment, i not more
so, than any technological or intellectual breakthrough.
In these pages we present a strong resistance to Internet
exceptionalism, or any arguments that new technologies can
only be understood using noel intellectual rameworks. Like
other reolutionary communication technologies, the Internet
has changed the way we lie, and ostering undreamt o new
orms o social organization and interaction. But also like
other reolutionary communication technologies, the Internet
has not changed the undamental roles played by territorial
goernment.
\e are optimists who loe the internet and beliee that it can
and has made the world a better place. But we are realistic
about the role o goernment and power in that uture, and
realists about the prospects or the uture.
I regret to say that it has been the Chinese goernment that has done the most
to proe our basic thesis correct. 1he Jello was, somehow, nailed to the wall.
Despite nearly a decade o \esterners ,most particularly \estern newspaper
columnists, assuming or hoping that the Net would bring down the Chinese
state, it didn`t happen, indeed it neer een came close. And so, ie years later
the basic ideas in our book seem hard to contest. Consequently, this one
particular species o Internet exceptionalism-the idea that the network has its
own soereignty in a sense, or is an exception to law-has weakened and may
be dead.

In the summer o 2010, in act, as i to hammer to point home, the Chinese
goernment released a new \hite Paper on Internet Policy.` It made its
centerpiece the phrase coined by the Internet exceptionalists o the 1990s:
Internet Soereignty.` loweer, that phrase did not mean what it did in the
1990s. Rather as the People`s Daily, the state newspaper, explained, Internet
Soereignty` means that all oreign I1 companies operating in China must
abide by China`s laws and |be| subject to Beijing`s oersight.`
6

R R R
Leaing law aside, howeer, the larger questions o Internet Lxceptionalism
remain unanswered. It is surely one thing or the Internet to be a liing
exception to the legal system, a soereign unto itsel in some way. But is the
Network an exception as an ivforvatiov vetror/, as a means or a nation or world
to communicate lere, surely, the exceptionalist is on ar stronger ground.
\hateer you might say about eorts to use the Internet to aoid law, we
cannot doubt that the Networks o Networks` has changed the way we
communicate in dramatic ashion. 1echnologically, and in its eects on
business, culture and politics, the Internet seems, by almost any account, an
exception, dierent rom the way other systems o mass communications hae
operated, whether the telephone, radio, or the teleision.
1his point seems so obious as to be commonplace to anyone who`s lied
through the 1990s. Unlike teleision, radio and newspapers, which all are
speech outlets or a priileged ew, the Internet allows anyone to be a publisher.
Unlike the priate cable networks, the Internet is public and, in its totality,
owned by no one. Unlike the telephone system, it carries ideo, graphics, the
\eb, and supports any idea anyone can come up with. It has played host to
generations o new inentions, rom email and the \orld \ide \eb to the
search engine, rom shops like eBay and Amazon to social networking and
blogging. It has challenged and changed industries, rom entertainment to
banking and trael industries. 1hese eatures and others are what hae made
the Network so interesting or so many years.
1he question is whether, howeer, the Internet is dierent in a ta.tivg way.
\hat do I mean, a lasting way` I rely on the sense that certain ideas, once
spread, seem to lodge permanently, or or centuries at least-e.g., the idea o
property, ciil rights, or accination. Lach is an idea that, once receied, has a
way o embedding itsel so deeply as to be nearly impossible to dislodge. In
contrast are ideas that, while doubtlessly important, tend, in retrospect, to orm
a rather interesting blip in history, a reolution that came and went. \ill we

6
Inormation Oice o the State Council o the People`s Republic o China, 1be vtervet iv
Cbiva, 2010, -../011222+,-*('+#3"+,(1"#D!3(&!(.12-*.!/'/!31(#%!<A@KMN@T+-.&,
!bite aer etaiv. vtervet orereigvt,`, PLOPLL`S DAIL\ ONLINL, June 9, 2010,
-../011!("$*4-+/!#/$!%'*$=+,#&+,(1K@@@B1K@AAC1K@ATN1A@BTCM@+-.&$.

think o the open age o the Internet the way we think o communism, or the
hula-hoop

I the Internet is exceptional in a lasting way, it must be or its ideology as
expressed in its technology. And in this sense its exceptionalism is similar to
American exceptionalism. Both the Nation and the Network were ounded on
unusual and distinct ideologies, ollowing a reolution ,one actual, another
technological,. In a typical account, writer Seymour Martin Lipset writes in
.vericav cetiovati.v: . Dovbteagea rora: 1he United States is exceptional
in starting rom a reolutionary eent . it has deined its rai.ov a`tre
ideologically.`
8
Or, as one-time Columbia proessor Richard lostadter wrote
in the 20th century, it has been our ate as a nation to not to hae ideologies,
but to be one.`
9
De 1ocqueille put American exceptionalism down to
particular eatures o the United States-the religiosity o its ounding, its
proximity to yet reedom rom Lurope, and, as he wrote, a thousand special
causes.`
10

Looking at the Internet, its ounding and its deelopment, we can ind the same
pattern o a reolution, an ideology, and many special causes.` \hile much o
it was purely technical, there were deeply reolutionary ideas, een by
technological standards, at the heart o the Internet, een i sometimes they
were arried at in accidental ashion or or pragmatic reasons.
O course, ully describing all that makes the Internet dierent would take
another Devocrac, iv .verica, and we hae the beneit o many writers whoè
tried to do just that, whether in Katie laner and Matthew Lyon`s !bere
!iara. ta, v ate, the oral accounts o its creators, classic works like J.l.
Saltzer et al., vatova .rgvvevt. iv ,.tev De.igv, or Jonathan Zittrain`s 1be
vtvre of tbe vtervet.
11

Iè spent some time thinking about these questions, and I want to suggest that it isn`t really
possible to answer the question in ull without understanding the story o the networks that
preceded the Internet. My ullest answer to the question Iè posed, then, is in 1lL
MAS1LR S\I1Cl ,Knop 2010,, an eort to try and ind the patterns, oer time, that
surround reolutionary technologies. 1his time, unlike in \lO CON1ROLS 1lL IN1LRNL1,
when it comes to the broader question o the Internet as a way o moing inormation, I
tend to side with the exceptionalists, though it is a close call.
8
SL\MOUR MAR1IN LIPSL1, AMLRICAN LXCLP1IONALISM 18 ,1996,.
9
JAMLS M. JASPLR, RLS1LLSS NA1ION 38 ,2000,.
10
ALLXIS DL 1OCQULVILLL, DLMOCRAC\ IN AMLRICA 519 ,lenry Reee trans., D. Appleton
and Company 1904, ,1831,.
11
KA1IL lAlNLR & MA11lL\ L\ON, \lLRL \IZARDS S1A\ UP LA1L: 1lL ORIGINS Ol 1lL
IN1LRNL1 ,1996,,J. l. Saltzer, D. P. Reed & D. D. Clark, va1ova .rgvvevt. iv ,.tev
De.igv, 2 ACM 1RANSAC1IONS ON COMPU1LR S\S1LMS ,1OCS, 2-288 ,1984,, JONA1lAN
ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1 ,2009,.

1o understand what makes the Internet dierent, the origins o the Internet
bear careul examination. lirst, the Network`s predecessors ,the telephone,
cable, etc., were all commercial enterprises irst and oremost, inented and
deployed ,in the U.S., by priate irms. 1he Internet, in contrast, was ounded
as a research network, explicitly non-commercial and public or the irst decade
o its existence. Priate companies were inoled, yes, but it was not a
commercial operation in the same sense that, say, the cable networks always
were.
Perhaps, thanks to its origins, the Internet was ounded with an ideology that
was ar more explicit than most-a kind o pragmatic libertarianism whose
inluence remains. 1he early Internet scientists had arious principles that they
were proud o. One example is Daid Clark`s memorable adage. \e reject:
kings, presidents, and oting. \e beliee in: rough consensus and running
code.` Another is ound in a amous Request lor Comments written by
Internet ounder Jon Postel, setting orth the ollowing as a principle or
network operators: Be conseratie in what you do. Be liberal in what you
accept rom others.`
12

1he Network constituted not just a technological adance, though it was that as
well, but also a rejection o dominant theories o system design and, in a deeper
sense, a reolution in inormation goernance. 1he early Internet researchers
were designing a radically decentralized network in an age-the mid-1960s-
when highly centralized systems ran nearly eery aspect o American and world
lie. In communications this was represented by A1&1, the great monopolist,
with its mighty and near-perect telephone network. But it could also be ound
in other aspects o society, rom the enlarged Deense Department that ran the
Cold \ar, the new, giant goernment agencies that ran social programs, and
enormous corporations like General Motors, IBM, and General Llectric.
So when Vint Cer and his colleagues put the Internet on the 1CP,IP protocol
in 1982 ,its eectie launch`,, most inormation networks-and I don`t mean
this is a pejoratie sense-could be described as top-down dictatorships. One
entity-usually a irm or a part o the State ,or both,, like A1&1 or the BBC,
decided what the network would be. 1he Internet, in contrast, has long been
goerned more like a ederation o networks, and in some respects, like a
Republic o Users. 1hat is implicit in the ability o anyone to own an IP
address, set up a website, and publish inormation-something neer true, and
still not true, on any other network.

12
Paulina Borsook, or .varcb, !or/., \IRLD ,Oct. 1995,,
-../011222+2*3!%+,#&12*3!%1'3,-*D!1M+B@1*!.8+-.&$, Jon Postel, Inormation Sciences
Institute o the Uniersity o Southern Caliornia, DOD Standard 1ransmission Control
Protocol 13 ,1980,, araitabte at -../011.##$4+*!.8+#3"1-.&$138,ACBU4!,.*#(7?+B@.

1hroughout its history, the uniersal Network has, true to the goernance
structure, seen a pattern o innoation that is unlike any other. 1his too is the
subject o much scholarship and popular recognition-the mode o
decentralized innoation` that had led eery seeral years or so to the next
wonder, starting with email, through the \eb, search engines, online retail, \eb
ideo, social networking, and onward. 1hese innoations arried in a highly
disorganized ashion oten led by amateurs and outsiders. 1he spread o
computer-networking itsel began with amateur geeks gloriied in 1980s ilms
like !ar Cave..
13
It is hard to think o a truly important Internet inention that
came rom a irm that predated the Internet. Society-changers like Craigslist,
eBay, \ikipedia and blogs are obiously the products o geeks.
R R R
Can it last Can the Internet remain, in this sense, exceptional \hateer the
Internet`s original ideas, it is easy to argue that all this, too, shall pass. 1he
argument rom transience suggests that all that seems reolutionary about the
Internet is actually just a phase common to speech inentions. In other words,
the Internet is ollowing a path already blazed by other reolutionary inentions
in their time, rom the telephone to radio. Such disruptie innoations usually
do arrie as an outsider o some kind, and will pass through what you might call
a utopian` or open` phase-which is where we are now. But that`s just a
phase. As time passes, een yesterday`s radical new inention becomes the
oundation and sole possession o one or more great irms, monopolists, or
sometimes, the state, particularly in totalitarian regimes like the Soiet Union or
the 1hird Reich. 1he openness ends, replaced with better production alue and
tighter controls. It is, in other words, back to normal, or at least what passed
or normal or most o human history.
\e might learn rom the ate o the broadcast radio, the darling new technology
o the 1920s.
14
In the 1920s, opening a radio station was relatiely easy, not
quite as easy as a website, but within the reach o amateurs. American radio was
once radically decentralized, open and rather utopian in its aspirations. But by
the 1930s, broadcast in the United States was increasing controlled by the
chains-most o all, the National Broadcast Company, NBC, who brought
better programming, but also much less o the amateur, open spirit. But that`s
nothing compared to countries like Germany and the Soiet Union, where radio
became the domain o the state, used to control and cajole. In Germany, eery
citizen was issued a people`s receier` tuned only to Nazi channels, and within

13
\ar Games ,Metro-Goldwyn-Mayer 1983,
14
1his story o radio can be ound in 1IM \U, 1lL MAS1LR S\I1Cl, chaps 3, 5 ,2010,.

the space o a decade, the ree radio had became what Joseph Goebbels called
the spiritual weapon o the totalitarian state.`
15

\et I ind it hard to imagine such a dramatic or immediate ate or the Internet.
It seems in so many ways too established, its alues too enmeshed in society, to
disappear in an instant.
Perhaps it would be more accurate to suggest that there are aspects o the
Internet ideology that are more and less likely to ade, to become yesterday`s
ideas. At one extreme, the Internet`s core technological ideas, protocol layering
& packet-switching, seem unlikely to go anywhere. 1he reason is that these
techniques hae become the basis o almost all inormation technology, not just
the Internet itsel. 1he telephone networks are today layered and packet-
switched, een i they don`t rely on the Internet Protocol.
More ulnerable, howeer, are the Internet`s early ideas o openness and
decentralized operation-putting the intelligence in the edges, as opposed to the
center o the network. Originally described by engineers as the L2L principle,
and popularly contained in the catch-phrase Net Neutrality,` these principles
hae suried the arrial o broadband networks. \et by its nature, Net
Neutrality seems easier to upset, or discrimination in inormation systems has
long been the rule, not the exception. 1here are, importantly, certain
commercial adantages to discriminatory networking that are impossible to
deny, temptations that een the Internet`s most open irms ind diicult to
resist. So while I may personally think open networking is important or
reasons related to innoation and ree speech, it seems obious to me that open
networking principles cav be dislodged rom their current perch.
Another open question is whether some o the means o production and
cultural creatiity that are associated with the Internet are destined or lasting
importance. \e hae recently lied through an era when it was not unusual or
an amateur ideo or blog to gain a greater iewership than ilms made or tens
o millions. But is that, Lessig`s remix culture,`
16
a noelty o our times \e
also lie in era where ree sotware is oten better than that which you pay or.
1hey are the products o open production systems, the subject o \ochai
Benkler`s 1be !eattb of êtror/., the engines behind Linux and \ikipedia and
other mass projects-as discussed in Benkler`s essay in this collection.
1
O
course such systems hae always existed, but will they retreat to secondary

15
Quoted in Garth S. Jowett, GAR1l JO\L11 & VIC1ORIA O`DONNLLL, RLADINGS IN
PROPAGANDA AND PLRSUASION 132 ,2005,.
16
LA\RLNCL LLSSIG, RLMIX: MAKING AR1 AND COMMLRCL 1lRIVL IN 1lL l\BRID
LCONOM\ ,2008,.
1
\OClAI BLNKLLR, 1lL \LAL1l Ol NL1\ORKS ,200,.

roles Or will they perhaps become o primary importance or many areas o
national lie
1he only honest answer is that it is too early to tell. And yet, at the same time,
the transience o att systems suggests that at least .ove o what we take or
granted right now as intrinsic to our inormation lie and to the nature o the
Internet ritt ade.
1he reasons are many. It might simply be that the underlying ideas just
discussed turn out to hae their limits. Or that they are subject to an almost
natural cycle-excessie decentralization begins to make centralization more
attractie, and ice ersa. More sinisterly, it might be because orces
disadantaged by these ideas seem to undermine their power-whether
concentrated orces, like a powerul state, or more subtle orces, like the human
desire or security, simplicity and ease that has long powered irms rom the
National Broadcasting Corporation to Apple, Inc.
\hateer the reasons, and while I do think the Internet is exceptional ,like the
United States itsel,, I also think it will, come to resemble more normal`
inormation networks-indeed, it has already begun to do so in many ways.
Lxceptionalism, in short, cannot be assumed, but must be deended.
R R R
I began this essay with a comparison between Internet and American
exceptionalism. \et I want to close by suggesting we can learn rom the
comparison in a slightly dierent sense. Iè suggested that there is a natural
tendency or any exceptional system to ade and transition back to obsered
patterns. But een i that`s true, what is natural is not always normatiely good,
not always what we want. lor example, it may ery well be natural` or a
democracy, ater a ew decades or less, to ripen into a dictatorship o some
kind, gien the rustrations and ineiciencies o democratic goernance.
Cromwell and Napoleon are the bearers o that particular tradition, and it has
certainly been the pattern oer much o history.
But the idea o American Lxceptionalism has included a commitment to trying
to aoid that ate, een i it may be natural. Despite a ew close calls, the
United States remains an exception to the old rule that Republics ineitably
collapse back into dictatorship under the sway o a great leader. 1he Internet,
so ar, is an exception to the rule that open networks ineitably close and
become dominated by the State or a small number o mighty monopolists.
1wenty-ie years ater .COM, we might say we still hae a republic o
inormation-i we can keep it.


Bibliography
In lieu o extensie ootnotes, I thought I`d proide here the books and articles
that hae, implicitly or explicitly, taken on the question o Internet
Lxceptionalism. Notice that, or those amiliar in the ield, this may lead to
some unusual groupings-but the undamental question is whether the project
in question tries to argue the Internet is magically dierent or a repeat o age-
old problems.
Exceptionalist Works
PL1LR lUBLR, LA\ AND DISORDLR IN C\BLRSPACL ,199,.
J. l. Saltzer, D. P. Reed & D. D. Clark, va1ova .rgvvevt. iv
,.tev De.igv, 2 ACM 1RANSAC1IONS ON COMPU1LR S\S1LMS ,1OCS,
2-288 ,1984,.
NIClOLAS NLGROPON1L, BLING DIGI1AL ,1996,.
LA\RLNCL LLSSIG, 1lL lU1URL Ol IDLAS ,2002,.
Daid R. Johnson & Daid Post, ar c oraer.: 1be Ri.e of ar iv
C,ber.ace, 48 S1AN. L. RLV. 136 ,1996,.
Mark A. Lemley & Lawrence Lessig, 1be va of vatova: Pre.errivg tbe
.rcbitectvre of tbe vtervet iv tbe roaabava ra, 48 UCLA L. RLV. 925
,2001,.
Susan P. Craword, 1be vtervet ava tbe Pro;ect of Covvvvicatiov. ar, 55
UCLA L. RLV. 359 ,200,.
DAVID POS1, IN SLARCl Ol JLllLRSON`S MOOSL ,2009,.
1IM \U, 1lL MAS1LR S\I1Cl ,2010,.

Anti-Exceptionalist Works
LA\RLNCL LLSSIG, CODL: AND O1lLR LA\S Ol C\BLRSPACL
,2006,.
1IM \U & JACK GOLDSMI1l, \lO CON1ROLS 1lL IN1LRNL1
,2008,.
1im \u, C,ber.ace orereigvt,., 10 lARV. J.L. & 1LCl. 64 ,199,.
Christopher S. \oo, !ovta Mavaativg roaabava êtror/ êvtratit, et
or vrt Covetitiov. . Covvevt ov tbe vatova Debate, 3 J.
1LLLCOMM. & lIGl 1LCl. L. 23 ,2004,.
\OClAI BLNKLLR, 1lL \LAL1l Ol NL1\ORKS ,200,.
COR\ DOC1ORO\, LI11LL BRO1lLR ,2008,.

On the Topic / Mixed
JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\
1O S1OP I1 ,2009,.

Section 230 of the CDA:
Internet Exceptionalism as a
Statutory Construct
By H. Brian Holland
*

Introduction
!"#$% "'( %#)$'*%#' "# +,,-. !%$'"/# 012 /3 '4% 5/**6#"$)'"/#( 7%$%#$8 9$'
4)( :%$/*% ;%<4);( '4% */(' ("=#"3"$)#' (')'6'% "# '4% <%=6>)'"/# /3 /#>"#%
$/#'%#'. )#? /#% /3 '4% */(' "#'%#(%>8 ($<6'"#"@%?A B)#8 %)<>8 $/**%#')'/<(
$<"'"$"@%? :/'4 5/#=<%((. 3/< "'( );;)<%#' "#):">"'8 '/ $<)3' '4% */<% >"*"'%?
(')'6'% "' "#'%#?%?. )#? '4% $/6<'(. 3/< "#'%<;<%'"#= '4% (')'6'% :</)?>8 )#? 3)">"#=
'/ >"*"' "'( <%)$4A C)'%< $/**%#')'/<( 3/$6( */<% $>%)<>8 /# ;/>"$8 $/#$%<#(.
$/#'%#?"#= '4)' '4% 3)">6<% '/ "*;/(% >"):">"'8 /# "#'%<*%?")<"%( 3)">( '/
%33%$'6)'% ;<"#$";>%( /3 %33"$"%#$8 )#? $/(' )D/"?)#$%A B/<% <%$%#'>8.
$/**%#')'/<( 4)D% )<=6%? '4)' !%$'"/# 012 "**6#"'8 (4/6>? :% >"*"'%?
:%$)6(% "' $/#'<":6'%( '/ '4% ;</>"3%<)'"/# /3 )#/#8*/6( 4)'% (;%%$4.
"#'"*"?)'"/#. )#? '4<%)'( /3 D"/>%#$% )=)"#(' '<)?"'"/#)>>8 *)<="#)>"@%? =</6;(A
9$E#/F>%?="#= '4% D)>"?"'8 /3 '4%(% $/#$%<#(. '4"( %(()8 #%D%<'4%>%(( ')E%( '4%
/;;/("#= D"%F. ?%3%#?"#= :</)? !%$'"/# 012 "**6#"'8 )( %((%#'")> '/ '4%
%D/>D"#= ('<6$'6<% /3 G#'%<#%' =/D%<#)#$%A !;%$"3"$)>>8. !%$'"/# 012 ;</D"?%( )
*%)#( /3 F/<E"#= F"'4"# '4% (/D%<%"=# >%=)> (8('%* '/ %33%$'6)'% *)#8 /3 '4%
=/)>(. "?%)>(. )#? <%)>"'"%( /3 '4% G#'%<#%' %H$%;'"/#)>"(' )#? $8:%<I>":%<')<")#
*/D%*%#'(A J8 *"'"=)'"#= '4% "*;/("'"/# /3 $%<')"# %H'%<#)> >%=)> #/<*( "# '4%
/#>"#% %#D"</#*%#'. !%$'"/# 012 4%>;( '/ $<%)'% '4% "#"'")> $/#?"'"/#( #%$%(()<8
3/< '4% ?%D%>/;*%#' /3 ) */?"3"%? 3/<* /3 %H$%;'"/#)>"(*A K"'4 '4% "*;)$' /3
%H'%<#)> #/<*( ?"*"#"(4%?. K%: 0A2 $/**6#"'"%(. (6$4 )( F"E"(
+
)#? (/$")>
#%'F/<E (%<D"$%(.
0
4)D% %*%<=%? '/ 3)$">"')'% ) >"*"'%? *)<E%' "# #/<*( )#?
D)>6%( )#? '/ ;</D"?% "#'%<#)> %#3/<$%*%#' *%$4)#"(*( '4)' )>>/F #%F
$/**6#)> #/<*( '/ %*%<=%A !%$'"/# 012 ;>)8( ) D"')> </>% "# '4"( ;</$%(( /3

L
9((/$")'% M</3%((/<. N%H)( K%(>%8)# !$4//> /3 C)FA 9 */?"3"%? D%<("/# /3 '4"( %(()8
/<"="#)>>8 );;%)<%? "# '4% O#"D%<("'8 /3 P)#()( C)F Q%D"%FA !" $%&%"'% (& )"*+"% !",%-.%/+0-1
!..2"+,13 405+*+,0,+"6 7(..2"+,+%' (& 8(/+&+%/ 9:5%;,+("0*+'.< R- OA P)#A CA Q%DA 1-, S022TU.
!""#$%%#&#'()*))(+*,-.%)-/0%#&#'()*,1.2&3)"(&,"4567898:;0A
+
9 F"E" "( ) F%:("'% ?%("=#%? '/ )>>/F D"("'/<( '/ %)(">8 $<%)'% )#? %?"' )#8 ;)=% /# '4% ("'%A
V/< */<% "#3/<*)'"/#. (%% K"E";%?"). =+>+. !""#$%%'+*<5=5#'65&*-(>%<5=5%?5=5 S>)('
)$$%((%? 7%$A +. 02+2UA
0
!/$")> #%'F/<E (%<D"$%( )<% /#>"#% (%<D"$%( ?%("=#%? 3/< 6(%<( '/ (4)<% *%(()=%(. >"#E(. )#?
*%?") S;4/'/( )#? D"?%/U F"'4 3<"%#?( /< /'4%<( F"'4 ("*">)< "#'%<%('(A !/*% ;/;6>)< (/$")>
#%'F/<E (%<D"$%( )<% V)$%://E. B8!;)$%. )#? NF"''%<A

building heterogeneous communities that encourage collaboratie production
and communication. Lorts to substantially reorm or restrict Section 230
immunity are thereore largely unnecessary and unwise.
1he essay begins with a brie introduction to Section 230. As interpreted and
applied by the judiciary, this statute is now conceied as a broad grant o
immunity rom tort liability-broad not only in terms o those who can claim
its protection but also in terms o predicate acts and causes o action to which
such immunity extends.
\orking rom this oundation, I then seek to position the courts` expansion o
Section 230 immunity within the larger debate oer Internet goernance,
suggesting that proponents o expanded immunity are successully creating what
might be characterized as a modiied, less demanding orm o cyber-libertarian
exceptionalism than what Lric Goldman calls, in his essay in this book, the
lirst \ae o Internet Lxceptionalism` ,one o Internet Utopianism`,, as
articulated in the mid-1990s. 1he dramatic expansion o Section 230 immunity
has in a limited sense eectuated a ision o a community in which norms o
relationship, thought and expression are yet to be ormed. 1he tort liability
rom which Section 230 proides immunity is, together with contract, a primary
means by which society deines ciil wrongs actionable at law. In the near
absence o these external norms o conduct regulating relationships among
indiiduals, the online community is ree to create its own norms, its own rules
o conduct, or none at all. It is a glimpse o an emergent community existing
within, rather than without, the soereign legal system.
linally, I make the case or presering broad Section 230 immunity. As an
initial matter, many o the reorms oered by commentators are both
unnecessary and unwise because the costs o imposing indirect liability on
intermediaries are unreasonable in relationship to the harm deterred or
remedied by doing so. Moreoer, the imposition o liability would undermine
the deelopment o \eb 2.0 communities as a orm o modiied exceptionalism
that encourages the deelopment o communal norms, eicient centers o
collaboratie production, and open orums or communication.
The Expansion of Section 230 Immunity
In May o 1995, a New \ork trial court rocked the emerging online industry
with its decision in trattov Oa/vovt, vc. r. Proaig, errice. Co.,
3
holding the
Prodigy computer network liable or deamatory comments posted on one o its
bulletin boards by a third-party. 1he key actor in this result was Prodigy`s
attempt to create a more amily-riendly enironment through the exercise o
editorial control oer the bulletin boards and moderating or oensie content.

3
No. 31063,94, 1995 \L 32310 ,N.\. Sup. Ct. May 24, 1995,.

Prodigy was thereore treated as a publisher o the inormation, rather than a
mere distributor, and held strictly liable or actionable third-party content.
Representaties o the online industry argued that the Prodigy decision placed
serice proiders in an untenable position by creating a lobson`s choice`
4

between monitoring content and doing nothing, thereby insulating the serice
rom liability. Congress responded to the decision by amending the drat
Communications Decency Act ,CDA, to include a tailored immunity proision
addressing the online industry`s concerns. As one element o what came to be
known as the Good Samaritan proisions o the CDA, Section 230 was
generally intended to proide online serice proiders and bulletin board hosts
with immunity rom tort liability or the deamatory acts o their users. 1his
was accomplished by addressing those speciic elements o common law
deamation at issue in the Prodigy decision-editorial control and the distinct
treatment o publishers and distributors under the law. 1o that end, Section
230 proided that no interactie computer serice should be treated as the
publisher or speaker o third-party content, and that eorts to moderate
content should not create such liability.
In the years ollowing the enactment o Section 230, courts consistently
extended its application. 1his trend began in 199 with the watershed decision
in Zerav r. .verica Ovtive, vc.,
5
in which the lourth Circuit applied Section 230
to claims that America Online ,AOL, should be held liable or the deamatory
content posted by one o its users. 1he plaintis claimed liability arose in part
because AOL had allegedly ailed to remoe third-party deamatory messages
rom its bulletin board system within a reasonable time, reused to post
retractions to deamatory messages, and ailed to screen or similar deamatory
messages thereater. 1he court ound the plainti`s tort claims were preempted
by Section 230, which rendered AOL immune. In reaching this result, the court
rejected a strict reading o Section 230 as being limited to its terms. Although
the statute ailed to make any explicit reerence to distributor liability, which the
Proaig, decision appeared to leae intact, the court read distributor immunity
into the statute, inding distributor liability to be an included subset o the
publisher liability oreclosed by the statute. By collapsing the publisher-
distributor distinction, the lourth Circuit adopted the most expansie reading
possible o both deamation law and Section 230. 1hus, een though AOL
knew the statements were alse, deamatory, and causing great injury, AOL
could simply reuse to take proper remedial and preentatie action without ear
o liability.

4
SAMULL lISlLR, 1lL RUS1ICK`S ALARM 1O 1lL RABBILS ,1660,, as cited in ob.ov`. cboice,
\ikipedia, -../011!(+2*>*/!%*'+#3"12*>*1V#64#(W?A4<,-#*,! ,last accessed Dec. 1,
2010,.
5
129 l.3d 32 ,4th Cir. 199,.

lollowing Zerav, and building on that court`s reading o both the statute and the
policies sought to be eected, courts hae extended the reach o Section 230
immunity along three lines: ,1, by expanding the class who may claim its
protections, ,2, by limiting the class statutorily excluded rom its protections,
and ,3, by expanding the causes o action rom which immunity is proided.
6

As to the irst, courts hae interpreted the proision o immunity to interactie
computer serices to include such entities as \eb hosting serices, email serice
proiders, commercial websites like eBay and Amazon, indiidual and company
websites, Internet dating serices, priately-created chat rooms, and Internet
access points in copy centers and libraries. 1he additional proision o
immunity to users o those serices promises similar results. Already, one
decision has held that a newsgroup user cannot be held liable or re-posting
libelous comments by a third party,
while another court ound a website

message board to be both a proider and a user o an interactie computer
serice.
8

1he second line o extension results rom a narrow reading o the term
inormation content proider,` which deines the class or whom there is no
immunity. Speciically, courts hae held that minor alterations to third-party
content does not constitute the proision o content itsel, so long as the
proider does not induce the unlawul content through the proision o
oending raw materials o authorship and where the basic orm and message o
the original is retained.
9
1he third point o expansion has been to extend
Section 230 immunity beyond causes o action or deamation and related
claims to proide immunity rom such claims as negligent assistance in the
sale,distribution o child pornography,
10
negligent distribution o pornography
o and to adults,
11
negligent posting o incorrect stock inormation,
12
sale o
raudulently autographed sports memorabilia,
13
inasion o priacy,
14
and
misappropriation o the right o publicity.
15

6
vt .ee lair lousing. Council o San lernando Valley . Roommates.com, LLC, 521 l.3d
115 ,9th Cir. 2008, ,declining to extend Section 230 immunity to Roommates.com or
certain categories o content solicited by the site or users in iolation o ederal air housing
laws,.
Barrett . Rosenthal, 146 P.3d 510, 52 ,Cal. 2006,.

8
DiMeo . Max, 433 l. Supp. 2d 523, 531 ,L.D. Pa. 2006,.
9
Batzel . Smith, 333 l.3d 1018, 1031 ,9th Cir. 2003,. ee at.o Donato . Moldow, 865 A.2d
11, 24 ,N.J. Super. Ct. App. Di. 2005, ,quoting atet r. vitb,.
10
Doe . Am. Online, Inc., 83 So. 2d 1010, 101 ,lla. 2001,.
11
Does . lranco Prods., No. 99 C 885, 2000 \L 8169, at 5 ,N.D. Ill. June 22, 2000,, aff à
.vb vov. Doe . G1L Corp., 34 l.3d 655 ,th Cir. 2003,.
12
Ben Lzra, \einstein & Co. . Am. Online, Inc., 206 l.3d 980, 986 ,10th Cir. 2000,.
13
Gentry . eBay, Inc., 121 Cal. Rptr. 2d 03, 15 ,Cal. Ct. App. 2002,.

Section 230, Internet
Governance & Exceptionalism
!"#$%#&' )"#*"+ #*& ,%-.&- '&/%#& 01&- 2+#&-+&# .01&-+%+3&4 #*& 30+3&5# 06
2+#&-+&# &73&5#"0+%,"89 5-&8$9&8 #*%# 3:/&-85%3& 3%++0# /& 30+6"+&' /:
5*:8"3%, /0-'&-8 0- 30+#-0,,&' /: #-%'"#"0+%, 801&-&".+ .01&-+9&+#84 %+' #*$8
#*%# 3:/&-;,"/&-#%-"%+ 3099$+"#"&8 )",, &9&-.& "+ )*"3* +0-98 06 -&,%#"0+8*"54
#*0$.*# %+' &75-&88"0+ %-& :&# #0 /& 60-9&'< =,#*0$.* #*&8& "'&%8 *%1& /&&+
8$/>&3#&' #0 "+#&+8& 3-"#"3"89 %+' 809&)*%# 0/83$-&' /: -&3&+# '&1&,059&+#8
"+ #*& .01&-+%+3& '&/%#&84 #*&: -&9%"+ % #0$3*8#0+& 60- #*& 3:/&-;,"/&-#%-"%+
"'&%,< ?*"8 5%-# 06 #*& &88%: 8&&@8 #0 3,&%- 85%3& "+ #*& .01&-+%+3& '&/%#&8 60-
#*"8 1"8"0+ 06 &73&5#"0+%,"894 %+' %-.$&8 #*%# !&3#"0+ ABC "8 "+ 809& ,"9"#&' )%:
6%3","#%#"+. #*& &9&-.&+3& 06 3:/&-;,"/&-#%-"%+ 3099$+"#"&8 "+ % 90'"6"&'4 ,&88
'&9%+'"+. 60-9<
Foundational Arguments of
Internet Governance
?*& '&/%#& 01&- 2+#&-+&# .01&-+%+3& &10,1&' "+ #)0 8$-5-"8"+.,: '"8#"+3#4 %,/&"#
30+1&-.&+# 8#%.&8< ?*& 6"-8# 8#%.& 06 #*& .01&-+%+3& '&/%#& 603$8&' 0+ ,%) %+'
803"%, +0-984 %+' )*&#*&- #*&8& #-%'"#"0+%, 90'&,8 06 -&.$,%#"+. *$9%+
-&,%#"0+8 30$,' /& 1%,"',: %55,"&' #0 #*& 0+,"+& &+1"-0+9&+#< 2+ #*"8 30+#&7#4
&73&5#"0+%,"89 )%8 30+3&5#$%,"D&' %8 % 8#%#& 06 /&"+. #0 )*"3* #*& 2+#&-+&# *%'
+%#$-%,,: &10,1&'4 %5%-# 6-09 #&--&8#-"%, 85%3&< ?*& 8&30+' 8#%.& 06 #*& '&/%#&
"+#-0'$3&' +&#)0-@ %-3*"#&3#$-& %8 %+ "950-#%+# %+' 50#&+#"%,,: '09"+%+#
9&%+8 06 -&.$,%#"+. #*& 0+,"+& &+1"-0+9&+#< 2+ #*"8 30+#&7#4 &73&5#"0+%,"89
/&3%9& %+ 0/>&3#"1& #0 /& 5$-8$&' %+' 5-0#&3#&' %8 % 9%##&- 06 3*0"3&4 -%#*&-
#*%+ % +%#$-%, 8#%#&< =# % 90-& &7%3#"+. ,&1&,4 #*&8& '&/%#&8 "95,"3%#&'
6$+'%9&+#%, E$&8#"0+8 06 ,&."#"9%3:4 5-&6&-&+3&4 50,"#"384 '&903-%3:4 30,,&3#"1&
'&3"8"0+;9%@"+.4 %+' ,"/&-#%-"%+ "'&%,8<
2+ #*& &%-,: FGGC84 %8 #*& 2+#&-+&# /&.%+ #0 -&%3* #*& 9%88&8 )"#* #*& %'1&+# 06
#*& H0-,' H"'& H&/4 % 5%-#"3$,%- 1"8"0+ 06 #*& 0+,"+& &+1"-0+9&+# &9&-.&' #0
%'103%#& %+' '&6&+' 2+#&-+&# &73&5#"0+%,"89< I&83-"/&' %8 '"."#%,
,"/&-#%-"%+"89 0- 3:/&-;,"/&-#%-"%+"894 #*& 1"8"0+ )%8 0+& 06 6-&&'094 ,"/&-#:4
%+' 8&,6;-&.$,%#"0+< J:/&-;,"/&-#%-"%+8 /&,"&1&' #*& 2+#&-+&# 30$,' %+' )0$,'
'&1&,05 "#8 0)+ &66&3#"1& ,&.%, "+8#"#$#"0+8 #*-0$.* )*"3* -$,&8 )0$,' &9&-.&<
?*&8& +0-98 )0$,' &9&-.& 6-09 30,,&3#"1& '"830$-8& %-0$+' /&*%1"0-4

FK
J%-%6%+0 1< L&#-085,%8*<3094 2+3<4 BBG M<B' FFFG4 FFAK NG#* J"-< ACCBO<
FP
"## $%& %# FFAA4 FFAP N&7#&+'"+. Q ABC "99$+"#: #0 '&6&+'%+# "+ 3,%"9 R%,,&."+. "+1%8"0+ 06
5-"1%3:4 9"8%55-0-"%#"0+ 06 #*& -".*# 06 5$/,"3"#:4 '&6%9%#"0+ %+' +&.,".&+3&SO< "## '()*
T&-6&3# FC4 2+3< 1< JJU",, VVJ4 KWW M<B' FFCA4 FFFWXFG NG#* J"-< ACCYO N6"+'"+. #*%# Q ABC
"99$+"#: &7#&+'8 #0 8#%#&;,%) "+#&,,&3#$%, 5-05&-#: 3,%"984 "+3,$'"+. $+6%"- 3095&#"#"0+4 6%,8&
%'1&-#"8"+.4 %+' -".*# 06 5$/,"3"#:O<

relationship, and content, rather than rom the control and regulation o
network architecture. Control o architecture was seen almost exclusiely as an
instrument by which to enorce emerging social norms, and not as a means o
determining the norms themseles. By the mid-1990s this process o sel-
regulation was well underway.
At the same time, howeer, soereign nations and their constituents increasingly
sought to impose existing oline legal regimes on this emerging, resource-rich
enironment. Many in the online community resisted, perceiing this regulation
as a threat to the exceptional nature o the Internet. Adocates o sel-
regulation enisioned cyberspace as a distinct sphere, apart rom physical space.
1hese cyber-libertarian exceptionalists saw the imposition o existing oline
legal systems grounded in territorially-based soereignty as inappropriate. 1hey
belieed that the online enironment should instead be permitted to deelop its
own discrete system o legal rules and regulatory processes. Sel-regulation was
preerable in its own right because it had proen so eectie in creating the
enironment sought to be presered, and also because the alternatie seemed
deastating. 1he imposition o external, territorially-based legal regimes would
be, the exceptionalists argued, ineasible, ineectie, and undamentally
damaging to the online enironment.
laced with the attempted imposition o oline legal regimes, cyber-libertarians
responded by attacking the alidity o exercising soereign authority and
external control oer cyberspace. According to Proessors Daid Johnson and
Daid Post, two leading proponents o sel-goernance, external regulation o
the online enironment would be inalid because Internet exceptionalism-the
state o being to which the Internet naturally eoled-destroys the link
between territorially-based soereigns and their alidating principles o power,
legitimacy, eect, and notice.
16
Most importantly, the Internet`s decentralized
architecture depries territorially-based soereigns o the power, or ability, to
regulate online actiity. Likewise, extraterritorial application o soereign law
ails to represent the consent o the goerned, or to eectuate exclusiity o
authority based on a relatie comparison o local eects. 1he loss o these
limiting principles results in oerlapping and inconsistent regulation o the same
actiity with signiicant spilloer eect. Depried o these alidating principles,
it would be illegitimate to apply soereign authority and external control in
cyberspace.

16
Daid R. Johnson & Daid Post, ar ava oraer.-1be Ri.e of ar iv C,ber.ace, 48 Stan. L.
Re. 136 ,1996,.

A primary challenge to these cyber-libertarian arguments came rom Proessor
Goldsmith, who engaged both their descriptie and normatie aspects.
1
In
terms o the legitimacy o soereign regulation, Goldsmith criticized Johnson
and Post`s limited iew o soereignty and oer-reliance on the relationship
between physical proximity and territorial eects. Moreoer, he argued that
they had oerstated the impossibility o regulation, mistaking ability or cost,
ailed to recognize the deterrent eect on extraterritorial actors o local
enorcement against end users and network components located within the
territory, and mistakenly equated alid regulation with some measure o near-
perect enorcement. linally, where true conlicts between soereigns existed,
Goldsmith argued that these could be resoled with the same tools used in the
oline world-rules o jurisdiction, conlict o laws, enorcement, etc.
1hroughout, Goldsmith struck at Johnson and Post`s exceptionalist iew o the
Internet, implicitly rejecting the ultimate signiicance o both the technical and
communal aspects o that ideal. 1his critique proed deastating to these early
cyber-libertarian arguments.
1he goernance debate entered its second phase in 1999 with the publication o
Proessor Lessig`s book, Coae ava Otber ar. of C,ber.ace.
18
Prior to Lessig`s
book, the goernance debate had ocused primarily on behaioral and property
norms, with the assumption that either existing soereign law or the law
emerging rom Internet sel-goernance would preail. Network architecture
merely proided the means to enorce these norms, particularly those emerging
rom sel-goernance. Lessig reconceied Internet exceptionalism as a two-part
phenomenon, one regulatory and the other cultural. 1he ormer recognizes that
many o those eatures that make the Internet exceptional ,in the cyber-
libertarian sense, are merely coding choices, and not the innate nature o
cyberspace. \ithin the network, architecture and code are the most basic orms
o regulation. Code can be easily changed. 1hus, Lessig argued, to protect the
cultural aspects o exceptionalism, we must irst recognize the exceptional
regulatory power o architecture and code within cyberspace, and its piotal role
in presering or destroying that culture.
Lessig irst pointed out that law and social norms are but two means o
regulating human behaior. In cyberspace, unlike real space, it is possible or
architecture to dominate regulatory structures. Architecture acts as a regulator
in the oline world as well-in the orm o time, nature, physics, etc.-but our
laws and social norms are generally conceied with these regulators assumed.
Alteration o that architecture is unusually diicult i not practically impossible.
In cyberspace, by comparison, architecture in the orm o code is remarkably

1
Jack L. Goldsmith, .gaiv.t C,beravarcb,, 65 U. Chi. L. Re. 1199 ,1998,, Jack L. Goldsmith,
1be vtervet ava tbe .biaivg igvificavce of 1erritoriat orereigvt,, 5 Ind. J. Global Legal Stud. 45
,1998,.
18

luid. Code eectuates a series o choices, rom data collection, to anonymity,
to access. And code can be changed. Not only is code luid, but within
cyberspace it is a uniquely powerul orm o regulation. Rather than regulating
behaior and relationships through punishment, deterrence and post-iolation
correctie action, code proides the means to exercise perect control and thus
perect regulation-regulation not just o eects, but o the ery unierse o
choices rom which an indiidual actor is able to select.
\ith this shit in ocus, the debate itsel eoled. Lessig cautioned that the
greatest threat to the exceptional culture o cyberspace comes rom the union o
perect control and market orces o commerce. 1he architectural components
that proide the means o perect control are held almost exclusiely by priate
entities with commercial and political interests distinct rom the collectie. 1he
inisible hand, Lessig argued, cannot resist the promise o perect control, and
has little or no motiation to protect the undamental alues promoted by
cyber-libertarian exceptionalism. According to the cyber-libertarian narratie,
barriers that are present in the real world do not exist or are ae vivivv. in the
online enironment. In the context o Internet architecture, exceptionalism can
be ound in original principles o network design that rely on open protocols
and non-discriminatory data transer-a network that is decentralized,
borderless, and with the potential or nearly unlimited data capacity. Indeed,
the digital data lowing through this system is itsel exceptional, because it is
easy to create and manipulate, easy to copy with no degradation in quality, and
easy to access and distribute. In the context o online relationships,
exceptionalism resides ,at the ery least, in the interactiity, immediacy, and
potential scope o interaction, as well as the opportunity or anonymity.
loweer, the ery promise o perect control is to eliminate many o these
choices and the undamental alues they relect as subserient to commercial
goals. In cyberspace, control oer coded architecture supplies the means or
making this election. Building on this assertion, Lessig argued that in order to
protect undamental alues, decisions regarding architecture should emerge
rom the body politic and collectie decision-making, rather than being
concentrated in priate actors.
lor many cyber-libertarians, Lessig`s message presented great problems.
Although many had already abandoned the argument that the exercise o
soereign authority in cyberspace was normatiely inalid, they had not gien
up ,as a matter o preerence, the ision o an emergent, sel-goerned, digital
libertarian space. Soereign legal regimes were still seen as the greatest threat to
that ision. 1erritorial goernments should, the cyber-libertarians argued,
simply leae cyberspace alone to lourish. lrom this perspectie, Lessig`s
arguments about the unique regulatory power o architecture and code in
cyberspace were largely conincing. But his description o the corrupting
inluence o perect control and concentrated priate power, and particularly his

call or goernment regulation to counteract those inluences and presere
undamental alues, were diicult to square with most libertarian iews.
1he debate on net neutrality proides a glimpse o this diision. Many
commentators, including Lessig, are concerned that the priate owners that
control the physical,inrastructure layer o the network will, in pursuit o cross-
layer ertical integration and increased reenues, priilege certain content or
applications. 1hey thereore endorse regulatorily-mandated neutrality as a
means o presering one aspect o Internet exceptionalism. Not surprisingly,
many libertarians reject this approach, endorsing instead market-based solutions
or eectuating indiidual choice.
1he irony o this debate is airly apparent. Many who might otherwise hae
characterized themseles as cyber-libertarian, or at least sympathetic to that
ision, are now conlicted. Net neutrality would necessarily be imposed by
external soereign legal systems and subordinated to the control o commercial
entities, rather than emerging as a common norm. In the extremes, the issue
seems to present a choice between entrenched political power and unregulated
market orces, with neither proiding adequate protection or indiiduals. 1hus,
many o the Internet exceptionalists who sought to segregate the Internet rom
territorial boundaries, who assumed existing soereign goernments and legal
regimes were the greatest threat to the online community, who belieed that the
computer scientist would remain in control o the network ,and thus in control
o enorcement,, ound themseles asking Congress to protect the Internet
rom priate actors and market orces.
Whats Left of Exceptionalism?
\hat then is let o Internet exceptionalism In his reolutionary essay .
Dectaratiov of tbe vaeevaevce of C,ber.ace, John Perry Barlow described
cyberspace as consisting not o computers, wires, or code, but o transactions,
relationships, and thought itsel.`
19
It was this ision, this perception o an
eoling social space, that guided Barlow`s ideal o the culture he sought to
presere-a distinct ision o potential worthy o protection. Indeed, to many
early inhabitants o cyberspace, communal control and regulation o network
architecture appeared a gien, i or no other reason than that perect external
control seemed almost impossible. lreedom o choice in indiidual expression,
human behaior, and relationships were the heart o the online cultural and
social ideal that stirred Barlow and other cyber-libertarians.
As it eoled, the goernance debate ractured this largely uniied ision,
distinguishing alidity rom preerence, law and social norms rom architecture

19
John Perry Barlow, . Dectaratiov of tbe vaeevaevce of C,ber.ace ,leb. 8, 1996,,
-../011-#&!4+!88+#3"1X6'3$#21:!,$'3'.*#(7;*('$+-.&$.

and code, technical exceptionalism rom cultural exceptionalism, goernment
power rom priate commercial power, and een libertarian rom libertarian.
Lessig argued persuasiely that the greatest threat to digital libertarianism arose
rom priate actors, unbounded by undamental alues ,including constitutional
alues, and with the ability to exercise perect control oer choice. Lessig`s
analysis, generally speaking, was ocused on the treatment o data as data, based
primarily on the identity o its owner and the commercial interests represented.
Choice in action was to be controlled by the regulation o owned data,
discriminatory treatment o data to the beneit o certain owners, restriction o
network access, and similar means. 1hese technical controls would then be
bolstered by traditional soereign law alidating those measures.
\hat seems somewhat obscured in Lessig`s architecture-and-code approach
,which clearly remains the central concern o the goernance debate, is Barlow`s
original ision o relational libertarianism, with its ocus on expression o
indiidual choice and the deelopment o new communal social norms within a
system o sel-goernance. 1his is the part o Internet exceptionalism that was,
in a sense, oerwhelmed by the debate oer architecture and code. \et there
are some choices, primarily relational, that remain largely unaected by that
debate. In this sphere, the question is not access to choice, the ability to
choose, or the aailable unierse o choices, but rather what norms apply to the
choices being made outside those controls.
Post argues that undamental normatie alues could best be protected by
allowing the widest possible scope or uncoordinated and uncoerced indiidual
choice among dierent alues and among dierent embodiments o those
alues.`
20
le beliees that the imposition o soereign legal regimes in
cyberspace, rather than promoting undamental alues as Lessig argued, would
instead deny the digital libertarian culture the opportunity to deelop apart rom
the oline world, with its own set o undamental alues. le argues it is better
to sere the priate interest ,een i powerul and commercially motiated, than
the interest o terrestrial soereigns. Indeed, he sees exceptionalism as requiring
sel-goernance, to the exclusion o external legal norms imposed by soereign
powers, as a precondition to the emergence o a new system o norms.
Section 230 as a Form of
Cyber-Libertarian Exceptionalism
Most would say that Barlow and Post lost the battle. loweer, this particular
strain o Internet exceptionalism, enisioned as sel-goernance and emerging
social norms applicable to relationships between indiiduals ,as opposed to data
as data,, has been presered in a modiied, less demanding orm. Ironically, it is
because o soereign law, not in spite o it, that this occurred. 1he dramatic

20
Daid Post, .gaiv.t .gaiv.t C,beravarcb,,` 1 Berkeley 1ech. L.J. 1365 ,2002,.

expansion o Section 230 immunity has eectuated many o the ideals
promoted by Post, Barlow, and others, albeit on a limited scale. 1his expansion
has created an enironment in which many o the norms and regulatory
mechanisms present in the oline world are eectiely inapplicable. 1his is so
not because the ery nature o cyberspace makes such application impossible, or
because soereign law is necessarily ineectie or inalid, but rather because
soereign law has airmatiely created that condition.
1he torts or which Section 230 proides immunity are, together with contract
law, the primary means by which society deines ciil wrongs actionable at law.
1hese norms o conduct regulate relationships among indiiduals: articulating
wrongs against the physical and psychic well-being o the person ,e.g., assault,
battery, emotional distress,, wrongs against property ,e.g., trespass to land,
trespass to chattels, conersion,, wrongs against economic interests ,e.g., raud,
tortious intererence,, and wrongs against reputation and priacy ,e.g.,
deamation, misappropriation o publicity, inasion o priacy,. Section 230 has
been interpreted and applied to proide expansie immunity rom tort liability
or actions taken on or in conjunction with computer networks, including the
Internet. Statutory language deining who may claim the protections o Section
230 immunity, including proiders o interactie computer serices and the
users o such serices, has been broadly extended. In contrast, the primary
limitation on the range o claimants to Section 230 immunity, which is
statutorily unaailable to the allegedly tortious inormation content proider,
has been construed airly narrowly. Moreoer, the immunity proided to this
expansie cross-section o online participants now reaches well beyond
deamation to include a wide range o other tortious conduct and claims. As
such, many o the norms o conduct regulating relationships among indiiduals
in the oline world-those ciil wrongs actionable at ,tort, law-simply do not
apply to many in the online world.
Len where the online entity is alleged to be aware o the illegal acts o their
users, and to be either actiely acilitating those illegal acts or reusing to stop
them, the intermediary retains Section 230 immunity. 1his is true een where
the intermediary has the knowledge, technical ability, and contractual right to
take remedial action. In the oline world, such actie and knowing acilitation
would likely iolate social norms established in tort law. In the online world,
howeer, the deendants are immune rom liability. Lstablished norms, as
expressed through the mechanisms o tort law, are neutralized by Section 230
and its judicial interpretations.
In the near absence o these external legal norms, at least within the range o
choices being made outside the data-as-data architectural controls, the online
community is ree to create its own norms, its own rules o conduct, or none at
all. 1he inhabitants may not hae a blank slate-criminal law, intellectual
property law, and contract law still apply-but much o what Barlow embraced

as central tenets ,mind, identity, expression, remain undeined. Section 230
oers a modiied ersion o cyber-libertarian exceptionalism, less demanding o
the soereign and existing oline social norms, and thereore less satisying.
But it is nonetheless a glimpse o that society, maintained by the soereign legal
regime rather than against it. 1he law now applies to nearly eery tort that can
be committed in cyberspace. It is nibbling at the edges o intellectual property
rights. It protects against the ciil liability components o criminal acts. It
generally extends to all but the irst speaker, who may well get lost in the
network to escape liability een without immunity.
A Case for Preserving
Section 230 Immunity
As interpreted by the courts, the immunity proisions o Section 230 hae been
heaily criticized. Many commentators hae argued that by ailing to impose
indirect liability on intermediaries, signiicant harms will go undeterred or
unremedied, and that Section 230 should be reormed to sere the interests o
eiciency and cost allocation. 1his part o the essay addresses these criticisms
directly, concluding that substantially reorming the statute is both unnecessary
and unwise because the cost o such liability is unreasonable in relation to the
harm deterred or remedied. Indeed, gien Section 230`s role in acilitating the
deelopment o \eb 2.0 communities, reorming the statute to narrow the
grant o immunity would signiicantly damage the online enironment-both as
it exists today and as it could become.
Evaluating Calls for Reform
Larly critics o Section 230 tended to ocus on the issues o congressional intent
and broad interpretation by the courts. More recent commentators hae moed
beyond these issues to engage the larger implications o proiding such
sweeping immunity to online intermediaries, suggesting amendments to Section
230 intended to eectuate policies o eiciency and cost allocation. 1his
critique begins with the premise that in the online enironment, indiidual bad
actors are oten beyond the reach o domestic legal authorities. 1his creates a
situation in which signiicant indiidual harms cannot be legally deterred or
remedied, and the ear that the Internet`s potential as a marketplace will not be
realized. Gien these negatie conditions, where a third party maintains a
certain leel o control, the imposition o indirect liability is desirable. 1he
ailure to do so may create ineiciencies by ailing to detect and deter harmul
behaior where the cost o doing so is reasonable. Commentators hae argued
that, in the online enironment, intermediaries are in the best position to deter
negatie behaior, to track down primary wrongdoers, and to mitigate damages.
1his is particularly true in regard to inormation-based torts, the damages o
which might be mitigated in many circumstances simply by taking down,
prohibiting, or blocking the objectionable content.

At the heart o this attack on Section 230 immunity is the idea that, in the
absence o indirect intermediary liability, signiicant harms will go undeterred or
unremedied. 1hese ears are either misplaced or oerstated. As an initial
matter, it is not clear that a signiicant number o bad actors are beyond the
reach o the law. Adances in technology are making it increasingly possible to
locate and identiy bad actors online, such that online anonymity is diicult to
maintain. Likewise, where the bad actor is identiied but is ound outside the
jurisdiction, soereign goernments hae deeloped methods or resoling
disputes to permit the direct extraterritorial application o domestic law, such as
rules o jurisdiction, conlict o laws, and recognition o judgments. Indeed,
anti-exceptionalists hae strenuously argued that the application o soereign
authority to online actiity originating outside the jurisdiction is legitimate and
alid in large part because o these rules.
Moreoer, although the immunity proided by Section 230 arguably mitigates
the legal incenties or online intermediaries to deter and remedy certain
negatie behaior, it does not eliminate those legal incenties. Section 230
expressly states that it has no eect on criminal law, intellectual property law, or
communications priacy law. 1hese external norms remain applicable to and
enorceable against both content proiders and intermediaries in the online
enironment. Perhaps een more signiicantly, although Section 230 remoes
legal incenties to enorce the norms expressed in tort law, law is certainly not
the only incentie or an intermediary to act. Communal, commercial and other
incenties also play a role. Indeed, Section 230 immunity allows intermediaries
the reedom to interene in a multitude o ways. 1hus, indiidual harms and
marketplace security can be addressed through alternate legal regimes and
internal incenties.
lurthermore, proponents o indirect intermediary liability concede that een
where harms do exist, intermediaries may only rightly be held liable or ailing to
detect and deter harmul behaior where the cost o doing so is reasonable. It
is unclear, howeer, that the costs o intermedial regulation are reasonable. In
terms o remedies and reorms, critics generally suggest some orm o the
detect-deter-mitigate model, imposing a duty upon the intermediary with the
potential or liability in cases o breach. 1he two most common models are
traditional liability ,damages, regimes and notice-and-takedown schemes.
Proponents o traditional liability schemes generally ind theoretical ault with
the exceptionalist iew o the Internet, and analytical ault with broad judicial
interpretations o the statute that collapse distributor-with-knowledge liability
into immunity rom publisher liability. Proponents o a notice-and-takedown
scheme likewise work rom a distributor-with-knowledge model that imposes a
limited duty o care on intermediaries, but generally acknowledge some degree
o exceptionalism that requires a distinct scheme. Most suggest some ariation

utilizing elements o the Digital Millennium Copyright Act ,DMCA,
21
and the
Luropean Union`s L-Commerce Directie,
22
wherein intermediary liability is
triggered by actual notice o the objectionable content or a standard o
reasonable care, and requiring remedial action ,e.g., taking down the content at
issue,.
1he costs o these indirect intermediary liability schemes could be great. Under
traditional liability rules, intermediaries may be orced to adopt a least-common-
denominator approach, resulting in oerly-broad restrictions on expression and
behaior. A modiied distributor-with-knowledge approach, usually in the orm
o a takedown scheme similar to that employed by the DMCA, may produce the
same type o chilling eect. 1his is potentially exacerbated by the use o a
should-hae-known standard that can trigger the need to patrol or harmul
content, raising costs and leading to een greater oerbreadth in application.
Moreoer, indirect liability reduces incenties to deelop sel-help technology,
such as location or identity tracking sotware and end-user ilters, the
deelopment o which was one o Section 230`s primary policy goals. 1hus, i
the scale o undeterred or unremedied harms is minimal, and the negatie
impact o a detect-deter-mitigate model is signiicant, then the cost associated
with the imposition o indirect intermediary liability is not reasonable.
Resisting the Urge Toward Homogeny
1he case or presering Section 230 immunity begins by recasting intermediary
immunity in terms o exceptionalism, sel-goernance and norms, because it is
precisely the gap between the oline social norms expressed in tort law and the
broad immunity proided to online participants that has led to the rather strong
criticism o Section 230. As a conceptual matter, communal enorcement
presents the greatest challenge to eectuating some modiied ersion o the
exceptionalist ideal. \hen external legal norms are excluded, internal
enorcement mechanisms acilitate the emergence o new communal norms to
take their place. Much o the criticism o Section 230 stems rom the lack o
legal enorcement that accompanies immunity, and the resulting inability to
orm new social norms to replace those o the soereign. It is important to
recognize, howeer, that \eb 2.0 communities, such as wikis and social
networks, represent a real and signiicant maniestation o the exceptionalist
ision, because they both acilitate a market in norms and alues, and proide
the internal enorcement mechanisms necessary or internal norms to emerge.
Section 230 plays a ital role in the deelopment o these communities by

21
Digital Millenium Copyright Act, Pub. L. 105-304, 112 Stat. 2860 ,1998,.
22
Directie 2000,31,LC o the Luropean Parliament and o the Council o 8 June 2000 on
certain legal aspects o inormation society serices, in particular electronic commerce, in
the Internal Market ,Directie on electronic commerce`,, -../011!537
$!E+!53#/'+!51Y!EZ3*H!3D1Y!EZ3*H!3D+%#[53*JGIYI\0M?@@@Y@@MB0IF0F]^.

substantially and continually mitigating the primacy o external legal norms
within the conines o the community. 1his permits choice, empowers the
intermediary to create a market in social norms, and allows alternate orms and
gradations o enorcement. 1he architecture o the community gies these
choices orm and substance, backed by an enorcement model, such that
communal norms hae the opportunity to deelop. In this sense, Section 230
and the \eb 2.0 model eectuate the emergence o a modiied orm o
exceptionalism. 1he reorms proposed by most commentators would hae a
negatie impact on these communities, with little beneit beyond those
communal norms that are likely to emerge, and should be rejected.
Exceptionalism, Self-Governance & Social Norms
Lxceptionalism does not argue or the absence o social norms. Instead,
exceptionalism embraces the idea o cyberspace as an enironment in which the
authority o external legal regimes is minimal, and where an open market in
norms and alues works in concert with sel-goernance to permit the online
community to establish its own substantie social norms. Section 230 helps to
eectuate a modiied orm o exceptionalism by moderating the imposition o
external legal norms so as to permit a limited range o choices-bounded, at
least, by criminal law, intellectual property law and contract law-in which the
online community is ree to create its own norms and rules o conduct.
loweer, the deelopment o social norms within this enironment requires
not only the ability to exercise broad indiidual choice among dierent alues
and embodiments o those alues, but also some mechanism o communal
enorcement through which to eectuate some orm o sel-goernance.
Larly proponents o exceptionalism were able to ocus on relational libertarian
ideals, iewing the Internet as a unique social space in which norms goerning
thought, expression, identity, and relationship should be permitted to eole.
1his ocus deeloped precisely because the mechanisms o enorcement
required or sel-goernance and the eoling deinition o emergent social
norms were taken or granted. 1he architecture o enorcement was primarily
controlled by a community inoled in the process as adherents to the
exceptionalist ideal, who could be trusted both to ensure broad indiidual
choice and to utilize the means o enorcement as a tool o sel-goernance as
norms emerged.
As a means o eectuating exceptionalism, the primary weakness o Section 230
is the lack o an enorcement component. Although the modiied
exceptionalism enabled by Section 230 permits a range o choices, it does
nothing to proide enorcement mechanisms to solidiy emerging communal
norms. \here immunity exists, legal enorcement mechanisms are neer
triggered. Likewise, the architecture o enorcement relied upon by early
exceptionalists is no longer communal or likely committed to the ision o a
distinct cyber-libertarian space, but is instead concentrated in priate

commercial entities. As a consequence, Section 230 immunity creates a gap:
Certain external legal norms are excluded, but internal communal norms are
oten unable to coalesce to take their place. It is this gap, resulting rom the
lack o architectural enorcement controls, which uels criticism o the
immunity proision. In application, howeer, an enorcement model has
emerged that mediates the tension between the broad aailability o indiidual
alue choices and the ability to eectiely sel-goern so as to permit the
deelopment o communal norms.
Communities of Modified Exceptionalism
\eb 2.0 communities are structured as a limited commons and are built on an
architecture o participation that operates as a platorm or user-created content
and collaboration. At the core are principles o open communication,
decentralized authority, the reedom to share and re-use, and an idea o the
Internet as a social orum or market or exchanging opinions and ideas in search
o norms to create a culture based on sharing. Section 230 plays a ital role in
the deelopment and maintenance o these architectures by proiding
intermediaries with limited immunity rom liability or the tortious content
proided by users. Indeed, in this sense, Section 230 seems to aor the
deelopment o \eb 2.0 serices and the proision o user-based content oer
the traditional model o proiding irst-party institutional content.
1he parallels between \eb 2.0 and Barlow`s ision o a communal social space
are eident, albeit in modiied orm. Barlow embraced the potential o an
enironment premised upon reedom o choice in indiidual expression, human
behaior and relationships. 1o achiee that potential, he and others belieed
that regulation by existing soereign powers must be rejected in aor o sel-
goernance, so that new communal social norms might hae the opportunity to
emerge. At the heart o this ideal was an airmation that alues participation in
the market o expression, ideas and action without the constraint o
preconceied alue judgments. \eb 2.0 promises a somewhat limited ersion
o this enironment-existing within soereign authority, narrowed by certain
enduring norms, and conined to segmented communities administered by
priate entities-by acilitating the market by which norms are tested.
1wo o the most common models o these \eb 2.0 serices, wikis and social
networks, are indicatie o how Section 230 can eectuate the modiied orm o
cyber-libertarian exceptionalism described aboe. Partly as a result o the
immunity rom liability proided by Section 230, these serices acilitate the
market in social norms by creating enclaes in which users may exercise broad
,although not unbounded, indiidual choice among competing alues. At the
same time, the intermediary retains control oer the architecture and thus the
means o enorcement. As the market deines social good through the
eolution o communal norms, that architecture may be employed as a
mechanism o goernance. In the absence o legal incenties, the enorcement

o communal norms is drien by internal incenties, such as the need or
inancial support rom community donations, a communal desire or
inormation integrity, or the need to build an audience or adertising. In some
communities, participants may be incentiized by credibility and stature in the
orm o temporal seniority, post count, rank within the community`s goerning
body, etc.
1he online encyclopedia \ikipedia is a speciic example o a \eb 2.0
community o collectie action. Lach entry in the \ikipedia database is created
and edited by olunteers who are guided by three primary principles: the
Neutral Point o View policy, the No Original Research policy, and the
Veriiability policy. Registered users can originate new articles, and any user,
whether registered or anonymous, can edit an existing article. In the period
between \ikipedia`s inception in 2001 and 2010, this experiment in oluntary
collaboratie action produced more than ten million articles.
1hese actiities are oerseen by two leels o administrators, administrators and
bureaucrats. Administrators ,historically called sysops, short or system
operators, hae the power to edit pages, delete or undelete articles and article
histories, protect pages, and block or unblock user accounts or IP addresses.
Bureaucrats hae the urther power to create additional sysops with the
approal o the community. In lebruary 2006, in response to a series o
signiicant and persistent acts o andalism, the co-ounder o \ikipedia created
an additional layer o protection: Administrators can protect any article so that
all uture changes must be approed by an administrator.
23
Administrators help
acilitate dispute resolution and enorcement. Low-leel disputes are resoled
in talk pages. lere, moderators guide members to resolution with reerence to
policies and guidelines deeloped oer the lie o the community. 1hus,
principle alues and norms can lead to more speciic rules. 1his approach
works in most cases. More serious iolations, such as malicious editing o an
article ,or andalism,, are addressed through ast-repair mechanisms executed
by community members. \ikipedia administrators are also able to block user
accounts or IP addresses.
As described, the \ikipedia community relects a modiied orm o the
exceptionalist model, initially allowing or indiidual choice among a range o
alues, acilitating a market in social norms, and proiding a means o
enorcement to eectuate norms as they deelop. Indeed, recent studies relect
not only that norms hae emerged rom this market, but that those norms hae
solidiied and expanded. 1hrough this process, the \ikipedia community is

23
ee \ikipedia, !i/ieaia: Protectiov Potic,,
-../011!(+2*>*/!%*'+#3"12*>*1_*>*/!%*'093#.!,.*#(</#$*,= ,last accessed Dec. 1,
2010,.

moing rom an immediate ocus on particular articles to more generalized
concerns or quality o content and community.
Not unexpectedly, open source projects such as \ikipedia are not immune to
abuse. In terms o community health, and to protect against these abuses,
\ikipedia has adopted a code o conduct and principles o etiquette that stress
ciility and discourage personal attacks. As discussed aboe, these norms are
enorced through an architecture that is designed to reinorce those norms with
an eye towards the health o the community. At the most basic leel, this
occurs through routine editing by participants. Oer time, more complex
mechanisms or dispute resolution and enorcement hae deeloped, such that
in the past ew years administratie and coordination actiities hae gained
importance.
1he relationship between architecture and social norms is ascinatingly apparent
both in the \ikipedia`s architectural choice to track and correlate the IP address
o any anonymous user who edits the encyclopedia, as well as the deelopment
o a monitoring system that tracks those changes or analysis. 1his system
seres as a mechanism or enorcing social norms, particularly the norm o
neutrality in more controersial areas. In terms o more ormal enorcement,
some edits that might preiously hae been oerlooked are now being
reexamined in light o the organization rom which they originated. Less
ormally, but perhaps een more eectiely, organizations which are perceied
to hae breached the norms o the community hae aced, and will ace,
recriminations. Moreoer, the entire community is now aware that enorcement
o those norms is now more eectie, presumably creating a deterrence eect.
1he \ikipedia example illuminates a constant process, as choices are narrowed
by communal norms that deelop and are gien lie through enorcement
mechanisms, such that principle norms generate a breadth o more particular
rules. Section 230 immunity plays an important role in this process, permitting
the community to eole and structure itsel in the most eicient manner. 1o a
limited extent, Section 230 immunity permits uncoordinated and uncoerced
indiidual choice among dierent alues and among dierent embodiments o
those alues. It urther allows the intermediary to play an actie role in
acilitating the market in social norms and in creating enorcement mechanisms
as a tool o sel-goernance. 1hose enorcement mechanisms can then
themseles adapt. 1his allows not only or the deelopment o distinct
community alues, but also or a means o tapping into incenties, adapting to
eoling norms and conditions, and reducing costs associated with disputes.
\ithin this ramework, greater ariations in community norms are possible. As
communities grow, niche communities are ormed at low cost. It is not the
global ision o early exceptionalism, but rather a more limited and localized
orm o modiied exceptionalism that unctions as a laboratory or testing social
norms and alues.

Conclusion
Critics o Section 230 hae both oerstated the harms arising rom immunity
and understated the costs o alternate schemes or imposing indirect liability on
online intermediaries. At the same time, they hae ignored the important role
Section 230 plays in the deelopment o online communities. 1he immunity
proided by Section 230 helps to create the initial conditions necessary or the
deelopment o a modiied orm o exceptionalism by mitigating the eect o
external legal norms in the online enironment. \eb 2.0 communities are then
able to acilitate a market in norms and proide the architectural enorcement
mechanisms that gie emerging norms substance. Gien Section 230`s crucial
role in this process, and the growing importance o \eb 2.0 communities in
which collaboratie production is yielding remarkable results, reorming the
statute to substantially narrow the grant o immunity is both unnecessary and
unwise.


Internet Exceptionalism Revisited
By Mark MacCarthy
*

Introduction
In the mid-1990s, commentators began debating the best way or goernments
to react to the deelopment o the Internet as a global communications
medium. Internet exceptionalists argued that the borderless nature o this new
medium meant that the application o local law to online actiities would create
insoluble conlicts o law. 1he exceptionalists belieed that as the Internet
grew, reliance on local goernments to set rules or the new online world would
not scale well. 1heir alternatie was the notion o cyberspace as a separate place
that should be ruled by norms deeloped by sel-goerning communities o
users.
1

Critics o the exceptionalist iew responded with a ision o a bordered Internet
where local goernments could apply local law.
2
In this iew, cyberspace is not
a separate place. It is simply a communications network that links real people in
real communities with other people in dierent jurisdictions. Goernments can
regulate actiity on this new communications network in many dierent ways,
including by relying on the local operations o global intermediaries. Global
intermediaries are the Internet serice proiders ,ISPs,, payment systems, search
engines, auction sites, and other platorm and application proiders that proide
the inrastructure necessary or Internet actiity. Although they are oten global
in character, they also hae local operations subject to local goernment control.
According to critics o the exceptionalist iew, goernments hae the right and
the obligation to use this regulatory power oer intermediaries to protect their
citizens rom harm.
3
Conlicts that might arise rom this regulatory actiity can

Mark MacCarthy is Adjunct Proessor in the Communications Culture and 1echnology
Program at Georgetown Uniersity. lormerly, he was Senior Vice President or Public
Policy at Visa Inc. Substantial portions o this essay were originally published as Mark
MacCarthy, !bat Pa,vevt vterveaiarie. are Doivg .bovt Ovtive iabitit, ava !b, t Matter., 25
BLRKLLL\ 1LCl. L. J. 103 ,2010,, aailable at -../0116.$`+#3"1%'.'1'3.*,$!41?N<?1B@MA7
BB?@W?@P',G'3.-=W?@_Ia+/%8.
1
ee, e.g., Daid R. Johnson & Daid Post, ar ava oraer. - 1be Ri.e of ar iv C,ber.ace, 48
S1AN. L. RLV. 136, 138-92 ,1996,.
2
.g., Jack L. Goldsmith, .gaiv.t C,beravarcb,, 65 U. ClI. L. RLV. 1199 ,1998,.
3
ee ia. at 1238-39.

be resoled through the normal mechanisms goernments use to resole
conlict o law questions.
4

Goernments generally ollowed the adice o the proponents o regulation, not
the regulatory skeptics.
5
And despite some set-backs in lirst Amendment
cases,
6
regulators hae continued a steady march toward controlling the Internet
by regulating intermediaries.
Some legal scholars argue that goernment

reliance on intermediaries to control unlawul behaior on the Internet is
justiied because putting the enorcement burden on intermediaries is the least
expensie way or goernments to eectiely assert jurisdiction.
8
1he key
rationale is that goernments cannot easily ind wrong-doers on the Internet,
but intermediaries can. 1hey are best positioned to monitor their own systems.
As Mann and Belzley put it, they are the least-cost aoider.`
9

1he deenders o local goernment jurisdiction oer the Internet oten rely on
historical analogies to buttress their case that local control is ineitable and
desirable. Debra Spar deeloped the thesis that society`s reaction to new
technologies ollows a predictable sequence o innoation, commercial
exploitation, creatie anarchy, and then goernment rules.
10
In the innoatie
stage a new technology is deeloped, in the second stage it is used in
commercial entures, in the third stage there is a tension between the anarchist
impulse and the need or commercial order and stability, and in the inal stage
society reaches out to regulate the now mature technology to create and

4
a. at 1200-01 ,arguing that regulation o cyberspace is easible and legitimate rom the
perspectie o jurisdiction and choice o law`,.
5
1he U.S. exception is 230 o the 1elecommunications Act o 1996 which immunizes many
Internet actors rom liability in many contexts or the illegal actiity o their users. 4 U.S.C.
230,c, ,2006,.
6
ee, e.g,, Revo r. .C|, 521 U.S. 844, 885 ,199, ,1he interest in encouraging reedom o
expression in a democratic society outweighs any theoretical but unproen beneit o
censorship.`,, Ctr. or Democracy & 1ech. . Pappert, 33 l. Supp. 2d 606, 665 ,L.D. Pa.
2004, ,inding that a statute requiring ISPs to block access to websites displaying child
pornography iolated the lirst Amendment,.
ee geveratt, JACK GOLDSMI1l & 1IM \U, \lO CON1ROLS 1lL IN1LRNL1: ILLUSIONS Ol A
BORDLRLLSS \ORLD ,2006, ,citing many examples o this trend,. 1his Article documents
urther examples in which payment systems were induced by laws, regulations, pressure, and
notions o corporate responsibility to take actions to control the illegal online behaior o
people using their systems.
8
ee, e.g., Ronald J. Mann & Seth R. Belzley, 1be Provi.e of vtervet vterveaiar, iabitit,, 4 \M.
& MAR\ L. RLV. 239, 249-50 ,2005,.
9
a. at 249.
10
DLBORA L. SPAR, RULING 1lL \AVLS: C\CLLS Ol DISCOVLR\, ClAOS, AND \LAL1l lROM
1lL COMPASS 1O 1lL IN1LRNL1 11-22 ,2001,.

maintain the needed stability.
11
1he deelopment o radio is the standard
example o this pattern. Radio`s initial pioneers thought its ability to wirelessly
broadcast inormation rom one point to many made goernment control
diicult and unnecessary.
12
But later commercial enterprises actiely sought out
goernment regulation in order to end the chaos on the airwaes that preented
broadcasters rom reaching their intended audience.
13
Applying Spar`s analysis
here, the Internet is somewhere between stage three and stage our, where we
can expect urther regulation o Internet actiity under the watchul eye o
goernment. 1he historical example demonstrates that although eery new
technology is thought to be outside the jurisdiction o goernment, this belie
usually gies way in time to the realities o goernment control.
In the case o the Internet, the adent o goernment control prompted many
obserers to think the Internet exceptionalists had been routed.
14
loweer,
Internet exceptionalism is still a widely held belie,
15
and the notion that
goernment control o cyberspace is both impossible and illegitimate still
motiates much discussion o Internet policy.
16
Moreoer, the initial legislatie
expression o Internet exceptionalism-Section 230 o the 1996
1elecommunications Act-is still on the books. 1his section proides a sae
harbor rom indirect liability or what might be called pure Internet
intermediaries-those entities proiding Internet access serice or online

11
a., .ee at.o Mann & Belzley, .vra note 9, at 243-44, GOLDSMI1l & \U, .vra note , at 124
,relying on Spar`s work,.
12
ee geveratt, SPAR, .vra note 10, at 124-90 ,describing the history o radio technology
deelopment,.
13
a. at 11-2.
14
ee GOLDSMI1l & \U, .vra note , at 14 ,asserting that notions o a sel-goerning
cyberspace are largely discredited`,.
15
ee geveratt, DAVID G. POS1, IN SLARCl Ol JLllLRSON`S MOOSL ,Daid Kairys ed., 2009,
|hereinater Post, IN SLARCl Ol JLllLRSON`S MOOSL | ,demonstrating an elegant take on
Internet exceptionalism,. 1he heart o the response to Goldsmith is that scale matters and
that while it is physically possible and permissible under current settled` law o cross-
border jurisprudence, it is not workable` to subject all websites to perhaps hundreds o
dierent and possibly conlicting jurisdictions. ee Daid G. Post, .gaiv.t .gaiv.t
C,beravarcb,, 1 BLRKLLL\ 1LCl. L.J. 1365, 1384 ,2002, |hereinater Post, .gaiv.t .gaiv.t
C,beravarcb,|.
16
ee l. Brian lolland, .vra ,adapted rom l. Brian lolland,v Defev.e of Ovtive vterveaiar,
vvvvit,: acititativg Covvvvitie. of Moaifiea cetiovati.v, 56 U. KAN. L. RLV. 369, 39
,200,,. lolland`s ersion o modiied exceptionalism is closely connected with the legal
principle that online intermediaries are not liable or third party conduct. le asserts that the
immunity rom liability created by 230 o the Communications Decency Act helps to
eectuate a modiied orm o exceptionalism by moderating the imposition o external legal
norms so as to permit a limited range o choices-bounded, at least, by criminal law,
intellectual property law and contract law-in which the online community is ree to create
its own norms and rules o conduct.` a. at 39.

serices.
1
Despite a growing call to reisit this immunity,
18
it has been
extended seeral times. 1he Internet gambling law, which creates liability or
traditional intermediaries such as payment systems, contains a limitation on
liability or pure Internet intermediaries.
19
Similarly, the recently passed online
pharmacy law exempts pure Internet intermediaries rom a general duty to aoid
aiding or abetting unauthorized Internet sales o controlled substances.
20
1he
adoption o these proisions in recent laws might be merely 230 on automatic
pilot, but more likely, some ersion o Internet exceptionalism is at work in
these legislatie distinctions.
A recent speech by the Obama Administration`s senior communications
policymaker, Lawrence Strickling, proides urther eidence o the continuing
releance o the Internet exceptionalist perspectie.
21
In deending Section
230`s limitation on liability, Assistant Secretary Strickling argued:
1his limitation on liability has enabled the creation o
innoatie serices such as eBay and \ou1ube, which host
content proided by others, without requiring that those
serices monitor eery single piece o content aailable on their
sites. Absent this protection against liability, it is hard to
imagine that these serices would hae been as successul as
they turned out to be.
22

Internet exceptionalism is the iew that the normal rules that apply to real-
world proiders o goods and serices should not apply to online entities.
Secretary Strickling argues or this iew on policy grounds. \ithout it, he
asserts, the innoatie character o the Internet would come to a halt. 1he next

1
4 U.S.C. 230,c,,1, ,2006, ,No proider or user o an interactie computer serice shall
be treated as the publisher or speaker o any inormation proided by another inormation
content proider.`,. 1he interpretation o this proision is quite broad. ee, e.g., Zeran . Am.
Online, Inc., 129 l.3d 32, 330-31 ,4th Cir. 199, ,inding that plainti `s tort claims o
deamation were preempted by 230,. 1he immunity does not extend to criminal law,
contract law, or intellectual property law. 4 U.S.C. 230,e,,1,-,4, ,2006,.
18
ee, e.g., Doug Lichtman & Lric Posner, otaivg vtervet errice Proriaer. .ccovvtabte, 14 U. ClI.
SUP. C1. LCON. RLV. 221 ,2006,, John Palrey and Urs Gasser, BORN DIGI1AL 106 ,2008,,
and Daniel Soloe, 1lL lU1URL Ol RLPU1A1ION 125-160 ,200,.
19
31 U.S.C. 5365,c, ,2006,.
20
Ryan laight Online Pharmacy Consumer Protection Act o 2008, Pub. L. No. 110-425,
,h,,3,,A,,iii,, 122 Stat. 4829-30.
21
Remarks by Lawrence Strickling, Assistant Secretary o Commerce or Communications and
Inormation, to Internet Society`s INL1 Series: vtervet 2020: 1be êt ittiov |.er. .rit 2,
2010 araitabte at
-../011222+(.*'+%#,+"#D1/3!4!(.'.*#(41?@B@1O(.!3(!.H#,*!.=<@L?K?@B@+-.&$
22
a.

\ou1ube or Google could neer emerge because the legal liabilities associated
with running such a new business would strangle it.
I the Internet exceptionalists rested their case on the literal impossibility o
extending local law to cyberspace then there is not much let to their argument.
A bordered Internet` where intermediaries try to control behaior prohibited
by local law is becoming a reality. Most Internet intermediaries hae explicit
policies that prohibit illegal actiities.
23
1hese general policies are supplemented
with speciic policies and procedures designed to preent the use o these
systems or speciic illegal actiities.
Moreoer, it is not just oluntary eorts by Internet intermediaries that show
how Internet actiity can be controlled. Goernments hae been eectiely
extending their control oer Internet actiity through imposing obligations on
intermediaries. It has been estimated that at least 26 countries impose some
kind o iltering obligations on Internet entities.
24
Recent goernment actions in
lrance and the United Kingdom impose graduated response` obligations on
ISPs, requiring them to cut o Internet access or alleged repeat copyright
iolators.
25
It is possible to challenge these extensions o goernment power

23
Participants in Google`s adertising programs shall not, and shall not authorize any party to
. adertise anything illegal or engage in any illegal or raudulent business practice.` Google
Inc. Adertising Program 1erms 4 ,Aug. 22, 2006,, araitabte at
-../4011'%2#3%4+"##"$!+,#&14!$!,.1.4'(%,48*(%!3. MasterCard has rules or both
merchants and their acquiring banks: A Merchant must not submit or payment into
interchange . and an Acquirer must not accept rom a Merchant or submission into
interchange, any 1ransaction that is illegal.` MAS1LRCARD, MAS1LRCARD RULLS 5.9. ,2008,,
araitabte at -../011222+&!3,-'(.,#5(,*$+#3"1&!3,-'(.7
',,#5(.1%#2($#'%41&'4.!3,'3%1P'4.!3G'3%<b5$!4<N<@T+/%8. MasterCard prohibits
its issuing banks rom engaging in illegal transactions. a. at 3.8.4. Visa has similar rules, or
example: A Merchant Agreement must speciy that a Merchant must not knowingly submit,
and an Acquirer must not knowingly accept rom a Merchant, or submission into the Visa
payment system, any 1ransaction that is illegal or that the Merchant should hae known was
illegal.` VISA, VISA IN1LRNA1IONAL OPLRA1ING RLGULA1IONS 4.1.B.1.c ,2008,, araitabte at
-../01154'+D*4'+,#&1%#2($#'%1&!3,-'(.41D*4'7*(.!3('.*#('$7#/!3'.*("7
3!"5$'.*#(4+/%8. Visa`s regulations also speciy acquirer penalties or merchants engaging in
illegal cross-border transactions. a. 1.6.D.16.
24
RONALD DLIBLR1, JOlN PALlRL\, RAlAL ROlOZINSKI, JONA1lAN ZI11RAIN, ACCLSS
DLNILD: 1lL PRAC1ICL AND POLIC\ Ol GLOBAL IN1LRNL1 lIL1LRING 1
,2008,.
25
Lric Panner, |.K. .rore. Crac/aorv ov vtervet Pirate., NL\ \ORK 1IMLS, April 8, 2010 at
-../011222+(=.*&!4+,#&1?@B@1@L1@K1.!,-(#$#"=1@K/*3',=+-.&$[4,/JBc4dJ%*"*.'$
W?@!,#(#&=W?@6*$$W?@5>c4.J,4!. Lric Panner, ravce .rore. !iae Crac/aorv ov êt
Pirac,, NL\ \ORK 1IMLS, October 22, 2009,
-../011222+(=.*&!4+,#&1?@@K1B@1?M1.!,-(#$#"=1?M(!.+-.&$[<3JB. Sometimes the
ISPs cooperate in a graduated response policy to settle legal claims. lor a reiew o
goernment and priate sector eorts to control online copyright iolations, .ee Christina
Angelopoulos, itterivg tbe vtervet for Co,rigbtea Covtevt iv vroe, IRIS PLUS, March 2009
aailable at -../011222+#64+,#!+*(.1#!'</56$1*3*41*3*4</$541*/$54L<?@@K+/%8+!(

oer Internet actiity as unwise, or as a iolation o a human right to Internet
access or as too costly. But it is no longer plausible to maintain that they are
simply impossible.
1his conclusion is discussed at length in another essay in this collection that
ocuses on the traditional payment intermediaries, payment card companies
such as Visa, MasterCard, and American Lxpress, as an instructie category o
intermediary platorms.
26
Deelopments oer the last seeral years conclusiely
demonstrate that these payment intermediaries can control speciic illegal
actiities on the Internet and goernments can extend their control to these
payment intermediaries.
1hus, the debate oer Internet exceptionalism has shited rom the nature` o
the Internet as something intrinsically beyond the control o goernments to a
problem o choice.
2
Intermediaries can control illegal behaior on the Internet
and goernments can control intermediaries, but .bovta they And i goernment
should exert control oer intermediaries in order to control Internet actiities,
how should the global legal order be restructured to accommodate their role
1his essay explores the extent to which the experience o payment systems in
controlling the illegal online behaior o their users illuminates the debate
among the Internet exceptionalists, deenders o the bordered Internet, and the
internationalists. It concludes that exceptionalism, in either its original or
modiied orms, is not the right ramework or Internet goernance because
intermediaries should not deer to the judgments o sel-goerning communities
o Internet users when the judgments conlict with local law. 1he
exceptionalists are correct that a bordered Internet` will not scale up, but the
experience o traditional payment systems points towards international
harmonization. I goernments are going to use intermediaries to regulate the
Internet, they need to coordinate their own laws to make that role possible.
1he essay addresses each o the three main approaches to Internet goernance:
exceptionalism, the bordered Internet, and internationalism. 1he irst section,
on exceptionalism, begins with a discussion o the original Internet
exceptionalist perspectie, which iewed goernment regulation o the Internet
as ineasible and normatiely less desirable than goernment deerence to the
rules deeloped by sel-goerning Internet communities. 1his is ollowed by a
discussion o Brian lolland`s reised ersion o exceptionalism. Under this
approach, the arious immunities rom intermediary liability established by local
jurisdictions enable the deelopment o autonomous Internet norms. Both
ersions are shown to hae signiicant limitations when iewed in light o

26
ee MacCarthy, Ovtive iabitit, for Pa,vevt ,.tev., ivfra at 230.
2
ee lolland, .vra note 16, at 36- ,In this context, exceptionalism became an objectie to
be pursued and protected as a matter o choice, rather than a natural state.`,.

payment system experiences. 1he next section explores the bordered
Internet,` the idea that in certain cases local goernments may properly and
unilaterally extend their jurisdiction oer Internet actiities through
intermediaries. Payment intermediaries use standard measures to resole
conlicts o law and ollow a practical rule that treats a transaction as illegal i it
is illegal in the jurisdiction o either the merchant or the cardholder. 1his
section then discusses limitations on this method o resoling cross-border
jurisdictional conlicts. 1he inal section concludes with a discussion and
endorsement o the internationalist perspectie, according to which local
goernments should only exercise control oer speciic Internet actiities in a
coordinated ashion.
Internet Exceptionalism:
The Original Version
In lebruary 1996, John Perry Barlow identiied Internet exceptionalism when
he declared cyberspace to be independent o national goernments, roughly on
the grounds that cyberspace does not lie within your borders` and that it is a
world that is both eerywhere and nowhere, but it is not where bodies lie.`
28
Conlicts in cyberspace would be resoled not with the territorially-based legal
concepts o property, expression, identity, moement, and context,` which do
not apply,` to cyberspace because they are all based on matter, and there is no
matter here.`
29
Rather, in cyberspace goernance will arise according to the
conditions o our world, not yours.`
30
Cyberspace is dierent.`
31

Almost concurrently, legal scholars Daid Johnson and Daid Post made a
similar case or Internet exceptionalism.
32
In their iew, the Internet destroys
the link between geographical location` and the orer o local goernments to
assert control oer online behaior, |and| . the tegitivac, o a local soereign`s
eorts to regulate global phenomena ..`
33
1he Internet destroys the power o
local goernments because they cannot control the low o electrons across
their physical boundaries. I they attempted to do so, determined users would
just route around the barriers. Moreoer, i one jurisdiction could assert control
oer Internet transactions, all jurisdictions could, resulting in the impossibility

28
Declaration o John P. Barlow, Cognitie Dissident, Co-lounder, Llec. lrontier lound., .
Dectaratiov of tbe vaeevaevce of C,ber.ace ,leb. 8, 1996,, araitabte at
-../0112?+!88+#3"1G!(4#34-*/1O(.!3(!.<,!(4#34-*/<6*$$416'3$#2<@?KC+%!,$'3'.*#(.
29
a.
30
a.
31
a.
32
ee geveratt, Johnson & Post, .vra note 1.
33
a. at 130 ,emphasis added,.

that all \eb-based actiity, in this iew, must be subject simultaneously to the
laws o all territorial soereigns.`
34
1he Internet destroys the legitimacy o local
jurisdiction because legitimacy depends on the consent o the goerned and
|t|here is no geographically localized set o constituents with a stronger and
more legitimate claim to regulate it than any other local group. 1he strongest
claim to control comes rom the participants themseles, and they could be
anywhere.`
35
Since eents on the Net occur eerywhere but nowhere in
particular . no physical jurisdiction has a more compelling claim than any
other to subject these eents exclusiely to its laws.`
36

Behind these arguments seemed to be an appealing political ision. 1he ideal
enisaged sel-organizing groups o people making the rules that applied to their
conduct. 1hese rules would not be imposed rom the outside, but would be
reely chosen by the actie participation o the community members. 1he key
was deliberation by ree, rational agents in their communities, not imposition o
rules by an arbitrary act o will by a distant soereign. 1his ideal o participatory
democracy was intended, in part, to oset the alienating eects o large-scale
modern democracies, which in practice had long ailed to proide their
members with the sense o community participation that alone seemed to justiy
the imposition o collectie rules.
1he way this ision would be implemented on the Internet would be through
the deelopment o autonomous communities o Internet users. 1hese Internet
communities were largely isolated rom real world` communities. Since it took
special care and eort to reach out to participate in them, only those people
who really wanted to participate would, and the eects o actiities in those
communities would be limited to those who chose to participate. Gien the
structure o the Internet as a communications network, which moed almost all
major decisions on content to the edges o the network, a diersity o law could
arise in cyberspace as each community deeloped its own norms or regulating
the conduct o its members. People would be ree to participate in the
communities they wanted, but could easily aoid those they did not like.
Lnorcement o the community rules would be accomplished through peer
pressure, reputational systems, inormal dispute resolution mechanisms, and
ultimately, banishment. 1he system as a whole would eole through a process
analogous to biological eolution, where dierse and potentially competing rule
sets as embodied in dierent communities would ie or acceptance in a ree
marketplace o rules.

34
a. at 134.
35
a. at 135.
36
a. at 136.

Internet exceptionalism is thus the iew that actiity on the Internet should be
regulated by Internet community norms, not laws o territorial jurisdictions or
globally harmonized laws.
3
It is hard to aoid the sense that the political ision
pre-dated the Internet-that the easibility argument masked the underlying
ision and the arrial o the Internet simply created the possibility o
implementing the ision in a way that the real` world did not. 1o see this,
imagine the reaction o Internet exceptionalists to the idea o a world
goernment that would establish uniorm global laws. 1his would eliminate the
conlict o law problem. But exceptionalists are een more appalled at the idea
o world goernment control oer the Internet than with the idea o nation-
state control oer it. 1his suggests that the issue is not easibility o control, but
the alue o participatie community decision making and diersity.
1his early cyber libertarian ision was immediately attacked by those who
deended the easibility and legitimacy o extending local laws to coer Internet
actiity.
38
As they note, |t|he mistake here is the belie that goernments
regulate only through direct sanctioning o indiiduals.. Goernments can .
impose liability on intermediaries like Internet serice proiders or credit card
companies.`
39
Goernment action against these intermediaries makes it harder
or local users to obtain content rom, or transact with, the law-eading content
proiders abroad. In this way, goernments aect Internet lows within their
borders een though they originate abroad and cannot easily be stopped at the
border.`
40
And these eorts to bring order to the Internet through pressure on
intermediaries are oten legitimate because they proide something inisible
but essential: public goods like criminal law, property rights, and contract
enorcement . that can usually be proided only by goernments.`
41

1he debate took an interesting twist through the work o Larry Lessig. A key
element o the early exceptionalist ramework was the idea that the Internet had

3
Mann and Belzley describe their iew as consciously exceptionalist` because speciic
characteristics o the Internet make intermediary liability relatiely more attractie than it has
been in traditional oline contexts because o the ease o identiying intermediaries, the
relatie ease o intermediary monitoring o end users, and the relatie diiculty o directly
regulating the conduct o end users.` Mann & Belzley, .vra note 9, at 250-51. But this is an
odd way o raming the issue. Internet exceptionalism is not simply the iew that the
Internet should be treated dierently rom the oline world. 1he claim is more speciically
that the Internet should be ree o local jurisdictions. Mann and Belzley`s iew, which implies
that the Internet should be brought under local jurisdictions through the mechanism o
intermediary liability, is thus the ery opposite o exceptionalism. It is one ersion o
Internet non-exceptionalism.
38
ee geveratt, Goldsmith, .vra note 4 ,challenging the regulation skeptics,.
39
Goldsmith, .vra note 4, at 1238.
40
GOLDSMI1l & \U, .vra note , at 68.
41
a. at 140.

a undamental nature, which goernments did not control, could not alter, and
which eectiely preented them rom imposing local rules. In his inluential
book, Coae ava Otber ar. of C,ber.ace,Lessig took aim at this idea.
42
le
pointed out that computer systems, sotware applications, and communications
networks were human creations and that the choices o the architects o these
systems were embodied in the code that made it possible or these systems to
run. lar rom being a natural object, these systems were subject to the
decisions o the parties ,usually non-goernmental entities, that had the right
and the ability to create, maintain and alter them.
1he initial openness and transparency o the Internet was thereore something
that could not be assumed as a act o nature, but something that needed to be
maintained against possible opponents. But unlike the early cyber libertarians,
Lessig did not ocus on the dangers that local goernments might try to control
choices by controlling code. le thought the openness o the Internet had to be
maintained against the interests o non-goernmental parties seeking to adance
their own strategic interests. Lessig`s initial priate sector targets were the
network carriers who were seeking to alter the end-to-end` design o the
network in order to pursue their own strategic interests at the expense o
application proiders, serice proiders and end users who relied on the
neutrality o the Internet to conduct their ordinary actiities. In this way, the
Internet exceptionalist debate merged with the net neutrality debate and the
original deenders o exceptionalism seemed to be aced with the ,to them,
unattractie dilemma o using local goernments to promote Internet alues o
openness or allowing their Internet choices to be dictated by unaccountable
priate entities that controlled the undamental architecture o the Internet.
43

1his attack was so eectie that many beliee that these notions o a sel-
goerning cyberspace are largely discredited.`
44
But modiied ersions accept
the basic premise that the Internet should be ree o local regulation and
goerned instead by its users. One ersion o the reied exceptionalism,
deended by Brian lolland, ocuses on \eb 2.0 communities.
45
1his iew

42
43
ee lolland, .vra note 16 at 108-119 or a summary o this way o connecting the Internet
exceptionalist debate with the net neutrality debate.
44
a. at 14.
45
lolland writes:
By mitigating the imposition o certain external legal norms in the online
enironment, 230 helps to create the initial conditions necessary or the
deelopment o a modiied orm o exceptionalism. \ith the impact o
external norms diminished, \eb 2.0 communities, such as wikis and social
networks, hae emerged to acilitate a limited market in norms and alues
and to proide internal enorcement mechanisms that allow new communal
norms to emerge.

argues that together with the immunity proisions o Section 230 o
Communications Decency Act, these communities hae the potential to allow
internal community norms to take the place o external territorially based laws.
46

Critique of Internet Exceptionalism
1he experience o global payment intermediaries described in a companion
article in this olume conirms the iew that intermediaries cav eectiely
control illegal actiity in cyberspace. 1his still leaes the question o whether
intermediaries .bovta resist goernmental pressure to control the behaior o
their users. As a general matter, they should not deer to the judgments o sel-
goerning communities o Internet users when these judgments conlict with
local law. As corporate citizens, they hae an obligation to obey the laws o the
jurisdictions in which they operate, and they simply hae no basis to excuse
themseles rom that duty in order to let online communities determine their
own ate. But een when local law does not require them to take action against
illegal behaior, their responsibility to keep their systems ree o illegal actiity
means that they oten should take speciic steps to stop these actiities.
1he undamental objection, een to lolland`s modiied exceptionalism, is that
the law` o Internet communities is not really the law o that community. It is
a commercial contract enorceable under the rules o some local jurisdiction,
and the terms o the contract are subject to the same kinds o legal and
regulatory oersight that bind contracts between people in local jurisdictions.
Deerring to these contracts does not usually mean democratic community sel-
goernment. Local regulations are needed to ully protect the members o these
communities.
4
Moreoer, in some cases, the legal discretion granted to
intermediaries to control the conduct o their members may be too broad and
should be limited by replacing intermediary judgment with public authority
decisions. 1he remainder o this section deelops these points.
Len i Internet communities could substantially exclude a signiicant portion
o external legal norms, it still does not ollow that internal norms will
necessarily emerge rom the process o debate and deliberation that lolland
enisages. As lolland notes, external legal norms are excluded, but internal
communal norms are oten unable to coalesce to take their place` because
enorcement is concentrated in priate commercial entities.`
48
1he hope o his
modiied Internet exceptionalism is that the intermediaries who control the new
\eb 2.0 platorms will be drien by internal incenties to accommodate the

lolland, .vra note 16, at 369.
46
a.
4
1his Section ocuses on competition policy, priacy, and consumer protection as examples.
48

wishes o the online communities they create, allowing users to establish norms
or their own communities.
49

But it is not clear that \eb 2.0 platorms are likely to grant this kind o
democratic sel-goernance. lor example, intermediaries can be subject to
pressure. Craig Newmark, the operator o Craigslist, has insisted that he made
his decision to remoe ads or erotic serices as a result o consultation with his
online community.
50
But it is also true that Craigslist was under criminal
inestigation by a number o state attorneys general or iolation o state laws
against prostitution.
51
One could argue immunity in this case, but Craigslist did
not.
52
It complied with a law enorcement request to remoe certain postings
and the decision to remoe these ads will be subject to ongoing oersight by
these law enorcement agencies.
53
loweer, the question remained whether or
not Craigslist would take the legal risk i the community oted to keep these ads
in place.
1hese communities are not typically goerned by democratic oting procedures
that guarantee the consent o the goerned. 1hey are goerned by contractual
terms o serice. Oten prospectie members o these communities hae a
simple take-it-or-leae-it choice when they decide to join.
54

49
1hese internal incenties include the need or inancial support rom community donations,
a communal desire or inormation integrity, or the need to build an audience or
adertising.` a. at 400, .ee at.o Matthew Schruers, ôte: 1be i.tor, ava covovic. of P
iabitit, for 1bira Part, Covtevt, 88 VA. L. RLV. 205, 261 ,ISPs respond to content-based
complaints as a matter o good business practice or the purpose o maintaining customer
goodwill and satisaction.`,.
50
Craig.ti.t ovvaer ee/. arger DC Rote, NA1`L J., June 2, 2009, aailable at
-../011.!,-%'*$=%#4!+('.*#('$`#53('$+,#&1?@@K1@C1,3'*"4$*4.78#5(%!374!!>47
$'3"!+/-/ ,reporting Craig Newmark`s comments to the Computers lreedom and Priacy
Conerence,.
51
ee Brad Stone, Craig.ti.t to Revore rotic` .a., N.\. 1IMLS, May 14, 2009, at B1. Craigslist`s
attorneys asserted immunity under 230, but chose oluntarily to remoe the ads to which
arious state attorneys general had objected. a. State Attorneys General elt conident that
they could bring a case under state criminal law despite the immunity granted by 230. a.
1he case was gien national attention when a medical student was accused o killing a
masseuse whom he met through Craigslist. a.
52
a.
53
a.
54
ee Johnson & Post, .vra note 1, at 1380 ,describing AOL and Compusere terms o serice
as examples o law in cyberspace,. Johnson & Post iew the rules or an Internet community
to be a matter or principled discussion, not an act o will by whoeer has control o the
power switch.` a. But it is hard to see how terms o serice or a typical Internet serice or
application is anything other than an act o will by the person who controls the serice or
application. It might satisy certain legal standards or inormed consent, but it is not the
product o principled discussion. And this might be the way consumers want it. Online

I consumers do not like the terms o serice, then protest can be eectie, as
in the recent case o users objecting to the change in terms o serice
unilaterally oered by lacebook. By threatening the priacy rights o the
community, the platorm stirred up substantial community unrest, and
ultimately the new terms o serice were withdrawn.
55
But this exit right is not
the same as democratic sel-goernance, and it is not always eectie. \hat i
lacebook had not responded to community objections \ould people actually
hae let, and where would they hae gone Lock-in is a real restriction in social
networks.
1he exemption rom liability based on Section 230 does not mean that online
entities are exempt rom local law. Oten, local law is needed to protect
consumers rom the actions o Internet intermediaries. Regulation o online
communities by goernments seems especially timely and urgent in three areas:
competition policy, priacy, and consumer protection.
\ith respect to competition, concentration in particular sectors o the online
world should be examined because it can so signiicantly reduce consumer
choice. 1he Department o Justice has indicated, or example, that it is going to
take a more actie approach in this area.
56
Along with the lederal 1rade

communities might not oer to determine their online laws through a political process
because the members o the community cannot be bothered. People isit many dierent
websites and use many dierent web serices. It is hard to beliee that they want ull
democratic participation rights to set up the rules or each o these serices. And it is
implausible that they would actually spend the time, i they were oered the opportunity.
1he example o priacy policies makes the point. A recent study concluded that i all U.S.
consumers read all the priacy policies or all the web sites they isited just once a year, the
total amount o time spent on just reading the policies would be 53.8 billion hours per year
and the cost to the economy o the time spent doing this would be >81 billion per year.
Aleecia M. McDonald & Lorrie l. Cranor, 1be Co.t of Reaaivg Prirac, Poticie., 4 I,S: J.L. &
POL`\ lOR INlO. SOC`\ 543, 565 ,2008,.
55
N.\. 1imes, aceboo/, vc.,
-../011.#/*,4+(=.*&!4+,#&1.#/1(!241654*(!441,#&/'(*!418',!6##><*(,1*(%!E+-.
&$[Td'c4,/JB74/#.c4dJ8',!6##>c4.J(=. ,last updated May 2, 2009,. In 200, the
company had created a community backlash when it introduced an adertising serice that
allowed a user`s online actiities to be distributed to other community members. Lpic.org:
Llectronic Priacy Inormation Center, Social Networking Priacy,
-../011!/*,+#3"1/3*D',=14#,*'$(!.1%!8'5$.+-.&$ ,last isited leb. 3, 2009,. In the ace
o this protest, it proided a simple way or users to decline to participate. a. In lebruary
2009, it proposed new priacy rules according to which users will own and control their own
inormation, and in April it allowed a ote o its users on these new principles. Oer 5
o those oting endorse them, and on July 1, 2009 it adopted them. a.
56
Press Release, U.S. Dep`t o Justice, Justice Department \ithdraws Report on Antitrust
Monopoly Law: Antitrust Diision to Apply More Rigorous Standard with locus on the
Impact o Lxclusionary Conduct on Consumers ,May 11, 2009,, araitabte at
-../011222+`54.*,!+"#D1'.31/56$*,1/3!44<3!$!'4!41?@@K1?LNAB@+/%8.

Commission ,l1C,, they hae initiated inquiries ocused on the search engine
market.
5

Priacy and security rules need to be deined as well. 1he l1C has taken major
action in this area, and is stepping up its enorcement.
58
1hey are also ocusing
on the deelopment o a new priacy ramework to analyze the basis or the
harms associated with priacy iolations.
59
lurthermore, the l1C has ocused
on deeloping rules or online behaioral adertising.
60
In addition, rules
goerning priacy or online cloud computing serices need to be clariied,
perhaps by additional legislation.
61

Consumer protection rules should be updated to apply more eectiely to new
deelopments in electronic commerce including the growth o mobile
commerce and user-generated content, the greater aailability o digital goods
online, and increased numbers o consumers acting as online sellers, and new
deelopments in accountability and payment protection. A timely deelopment
might be the harmonization o consumer redress and liability rights across
arious payment mechanisms.
62

linally, the discretion gien to Internet intermediaries oer which transactions
to allow must be subject to public scrutiny. 1oday, intermediaries exercise

5
ee, e.g., Miguel lelt, |.. vqvir, . Covfirvea ivto Coogte oo/. Deat, N.\. 1IMLS, July 3, 2009,
at B3, Miguel lelt & Brad Stone, oara 1ie. at .te ava Coogte crvtiviea, N.\. 1IMLS, May
5, 2009, at B1, Peter \horiskey, Coogte .a Deat . |vaer crvtiv,: Yaboo .greevevt vb;ect of
.vtitrv.t Probe, ovrce. a,, \ASl. POS1, July 2, 2008, at D1.
58
ee Press Release, led. 1rade Comm`n, Sears Settles l1C Charges Regarding 1racking
Sotware ,June 4, 2009,, araitabte at -../011222+8.,+"#D1#/'1?@@K1@C14!'34+4-.&
,reporting that in the Sears case the l1C obtained a settlement rom Sears ater charging
that their consent practices in regard to installing an online tracking program on customers`
computers constituted an unair or deceptie practice,.
59
ee Stephanie Cliord, re.b 1ier. at .gevc, Orer.eeivg Ovtive .a., N.\. 1IMLS, Aug. 5, 2009, at
B1 ,stating that Daid Vladeck, the new head o the l1C`s consumer protection diision, is
rethinking priacy,. Vladeck said that |t|he rameworks that weè been using historically or
priacy are no longer suicient.` a. In his iew the l1C will begin to consider not just
whether companies caused monetary harm, but whether they iolated consumers` dignity
because, or example, |t|here`s a huge dignity interest wrapped up in haing somebody
looking at your inancial records when they hae no business doing that.` a.
60
ee Press Release, led. 1rade Comm`n, l1C Sta Reises Online Behaioral Adertising
Principles ,leb. 13, 2009,, araitabte at -../011222+8.,+"#D1#/'1?@@K1@?16!-'D'%+4-.&.
61
ee geveratt, ROBLR1 GLLLMAN, \ORLD PRIVAC\ lORUM, PRIVAC\ IN 1lL CLOUDS: RISKS 1O
PRIVAC\ AND CONlIDLN1IALI1\ lROM CLOUD COMPU1ING ,2009, ,discussing these cloud
computing issues,.
62
Legal payment protections now dier depending on the type o payment product used
,debit or credit, and the nature o the payment proider-traditional payment proiders like
Visa ace legal requirements while new payment proiders such as cell phone companies do
not.

judgment oer which transactions are subject to such legal risk that they cannot
be allowed. 1hese decisions are made in the context o the business interests
and technological capabilities o the intermediaries themseles, but they hae
important eects on the rights and interests o other parties. Some examples,
explained in a companion essay in this olume, include:
Payment systems eectiely decide which Internet gambling
transactions are illegal. By choosing to block all coded gambling
transactions, the system disadantages horseracing, state lottery, and
Indian gaming transactions that are arguably legal.
Payment systems take complaints rom third parties, make an
independent legal assessment o the merits o the case, and withdraw
serice based on these assessments. In eect, they adjudicate these
copyright cases.
1hese decisions are sound and sensible ways to balance complex and competing
interests. loweer, they are priate sector judgments, ineitably subjectie and
inluenced by the particular interests o the parties inoled.
Other intermediaries also hae enorcement abilities that they can use at their
own discretion. lor instance, in June 2009, it was reported that a British ISP
had agreed to disconnect subscribers who were accused o three instances o
inringement by a copyright owner.
63
Allegations o iolations would be made
by a contractor working or the content owner and transmitted to the ISP.
64
At
this point, these decisions are largely up to the payment intermediaries and the
ISPs themseles, although in some jurisdictions they are dictated by goernment
requirements,
65
yet their decisions will hae proound eects on the shape and

63
ee, e.g., Danny O`Brien, ri.b P .gree. to 1bree tri/e. .gaiv.t t. Cv.tover., DLLPLINKS
BLOG, -../011222+!88+#3"1%!!/$*(>41?@@K1@B1*3*4-7*4/7'"3!!47.-3!!74.3*>!47'"'*(4.7
*.4754!34 ,Jan. 28, 2009,.
64
Under the agreement the music labels, instead o going to court to get an order to hae the
ISP shut o a subscriber`s connection, proide eidence o inringement to the ISP directly.
a. As O`Brien noted,
1he dierence is that an ISP is not a court, and its customers will neer hae
a chance to deend themseles against the recording industry`s accusations
and proo.` 1o whom, without judicial oersight, has the ISP obligated
itsel to proide meaningul due process and to ensure that the standard o
proo has been met
a.
65
1he moement toward graduated response would replace this discretion with goernment
processes. Under the recently passed laute Autorit pour la Diusion des Uures et la
Protection des Droits sur Internet` ,ligh Authority o Diusion o the Art \orks and
Protection o the ,Copy,Rights on Internet, ,lADOPI`, law, lrench ISPs would be
required to suspend Internet access or subscribers who hae been subject to three
allegations o copyright iolations. Catherine Saez, revcb .DOP ar, ôr Covtete, Cav
ravai.b t. !eaov., IN1LLL. PROP. \A1Cl, Oct. 23, 2009, -../011222+*/7

direction o electronic commerce. Deerring to the norms o the Internet
community in this context means deerring to these priate judgments by
intermediaries.
1here is a role or Internet community decision-making. 1he best
circumstances or deerence to law constructed or and by particular Internet
communities is when an Internet community`s norms do not undamentally
impinge upon the ital interests o others who neer isit this new space.`
66
1o
the extent that an Internet community is sel-contained or its actiities aect
others only on a oluntary basis, then there is a case or deerring.
6

Payment Systems &
the Bordered Internet
Goldsmith and \u attack Internet exceptionalism, but they also construct a
positie ision o a bordered Internet.`
68
1his world would work pretty much
as the world worked beore the Internet. New regulations would be crated to
deal with the new dangers speciically created by the Internet, but there would
be no undamental need to adjust the basic domestic or international
ramework.
69

2'.,-+#3"12!6$#"1?@@K1B@1?M183!(,-7-'%#/*7$'27(#27,#&/$!.!7,'(763'(%*4-7*.47
2!'/#(41. A court reiew would be required beore suspension. a. A similar graduated
response program was adopted in Britain in April 2010. See Lric Panner, |.K. .rore.
Crac/aorv ov vtervet Pirate., NL\ \ORK 1IMLS, April 8, 2010, araitabte at
-../011222+(=.*&!4+,#&1?@B@1@L1@K1.!,-(#$#"=1@K/*3',=+-.&$. \hether these
graduated response programs are needed is a point o controersy, but they replace ISP
discretion with a system o public accountability.
66
Johnson & Post, .vra note 1, at 1389.
6
ee POS1, IN SLARCl Ol JLllLRSON`S MOOSL, .vra note 15, at 18-86 ,describing massiely
multi-player online games` or MMOGS as good candidates or this eort at online rule
creation,. 1his might be. loweer, Linden Labs, the creator o Second Lie, one o the
most amous MMOGs, ound it necessary to rely on external banking regulators when it
decided to ban the oering o interest or any return on inestment in-world without proo
o an applicable goernment registration statement or inancial institution charter. Kend
Linden, êr Potic, Regaraivg v!orta av/., SLCOND LIlL BLOGS, Jan. 8, 2008 06:43:56
PM, -../40116$#"4+4!,#(%$*8!+,#&1,#&&5(*.=18!'.53!416$#"1?@@T1@B1@T1(!27
/#$*,=73!"'3%*("7*(72#3$%76'(>4. Linden Labs properly concluded that it isn`t, and can`t
start acting as, a banking regulator.` a. New rule-making institutions will emerge only i
people think that they are real. lor this reason, a policy to deer in certain cases should be
public and stable in order to proide the opportunity or the deelopment o alternatie
rules.
68
GOLDSMI1l & \U, .vra note , at iii.
69
a. at 149.

Jurisdictional disputes would be one signiicant problem with the bordered
Internet. 1he initial Internet exceptionalist argument was that Internet actiity
is simultaneously present in multiple oerlapping and inconsistent jurisdictions,
and that no one jurisdiction has a better claim to regulate the actiity than any
other jurisdiction. It would be better to think o the actiity as taking place in a
separate jurisdiction altogether and hae the territorial goernments o the
world deer to the community norms created there. Goldsmith and \u`s
response was that Internet actiity was real world actiity, taking place in
particular jurisdictions, and that local goernments could exert control oer this
actiity by attaching obligations to the local operations o global Internet
intermediaries.
0
1his indirect liability or intermediaries would make it easier to
extend local law to the bad actor.
1
Conlict o laws would be handled by the
normal mechanisms or resoling these disputes, and ultimately enorced by
actions taken against local operations o global intermediaries.
2

Jurisdiction in cyberspace is a complex topic with many dierent approaches to
assigning both the applicable law and the court o jurisdiction.
3
Questions
include determining the location o the transaction, the jurisdiction, and the
interests o the parties.
4
An early attempt to deal with these issues in the
Internet context was the l1C`s approach to consumer protection in the global
marketplace.
5
1he simplest cross-border electronic transaction implicates
transnational concerns. Choice o law debates ineitably ollow. 1he l1C
considered arguments or the country o origin` approach and the country o
destination` approach.
6
Under the country o origin approach, the law o the

0
a. at 68-2.
1
Mann & Belzley, .vra note 9, at 259 ,|On the Internet it is| easier or een solent
maleasors engaged in high-olume conduct to aoid responsibility either through anonymity
or through relocation to a jurisdiction outside the inluence o concerned policymakers.`,.
Mann and Belzley also argue that indirect liability makes sense in cases in which the retailer
is located in a jurisdiction outside the United States that will not cooperate with the releant
state regulators.` a. at 2.
2
GOLDSMI1l & \U, .vra note , at 158-61.
3
ee, e.g., Paul S. Berman, 1orara. a Co.vootitav 1i.iov of Covftict of ar.: Reaefivivg
Corervvevtat vtere.t. iv a Ctobat ra, 153 U. PA. L. RLV. 1819, 1822 ,2005, ,arguing that judges
should adopt a cosmopolitan approach in Internet cases inoling choice o law and oreign
judgment issues, grounded in the idea that goernments hae an interest not only in helping
in-state litigants win the particular litigation at issue, but a more important long-term interest
in being cooperatie members o an international system and sharing in its reciprocal
beneits and burdens`,.
4
ee geveratt, Goldsmith, .vra note 4 ,discussing many o these theories,, .ee at.o Berman, .vra
note 3, at 1839-40 ,discussing arious choice-o-law theories that address these questions,.
5
lLD. 1RADL COMM`N, CONSUMLR PRO1LC1ION IN 1lL GLOBAL LLLC1RONIC
MARKL1PLACL: LOOKING AlLAD ,2000,. 1he l1C`s discussion o applicable law and
jurisdiction is especially releant. a. at 4-11.
6
a.

merchant would apply and the courts o the merchant`s country would
adjudicate any disputes.
Under the country o destination approach, the law

o the consumer would apply and the courts o the consumer`s country would
adjudicate disputes.
8

1he deense o the country o origin approach relied on the diiculty o
applying any other legal ramework to the electronic marketplace.
9
Only this
country o origin ramework seems to allow or the growth o global e-
commerce. 1he ramework considers problems encountered by small
businesses selling in many countries o creating and applying a standard or
some ariety o purposeul` targeting. Creating a deault rule o the country o
origin was deemed to better proide needed uniormity and predictability or
online businesses.
1his approach has deects. lirst, it orces consumers to rely on unamiliar
consumer protections. I merchants cannot be expected to know the laws o
180 countries, neither can consumers. Second, it creates a race to the bottom,`
whereby unscrupulous merchants can simply locate in a country with weak
consumer protections. 1hird, consumers cannot reasonably be expected to
trael to the country o origin to obtain redress. lourth, consumers could not
rely on their own consumer protection agencies or redress either, since these
agencies would also be unable to enorce the consumer`s home jurisdiction
protections.
So neither deault rule seemed to suice. As a practical matter, consumer
education, sel-regulatory eorts, and the deelopment o codes o conduct by
multinational organizations were the means chosen to address the cross-border
consumer protection issue.
80
lor other issues that could not be addressed

a. at 2.
8
a. 1he Luropean Union appeared to take the side o the country o origin in its L-
Commerce Directie. Luropean Commission, L-Commerce Directie,
-../011!,+!53#/'+!51*(.!3('$<&'3>!.1!7,#&&!3,!1%*3!,.*D!<!(+-.& ,last isited leb.
15, 2010,. 1he Directie contains an Internal Market clause which means that inormation
society serices are, in principle, subject to the law o the Member State in which the serice
proider is established.` a.
9
lLD. 1RADL COMM`N, .vra note 5, at 4 ,discussing the two undamental challenges` to a
country-o-destination ramework, including the use o physical borders to determine
rights in a borderless medium` and compliance costs,.
80
In 1999, the OLCD issued its Guidelines or Consumer Protection in the Context o
Llectronic Commerce, which address principles that could be used by electronic commerce
merchants in the absence o global consumer protection rules. ORG. lOR LCON. CO-
OPLRA1ION & DLV., GUIDLLINLS lOR CONSUMLR PRO1LC1ION IN 1lL CON1LX1 Ol
LLLC1RONIC COMMLRCL ,1999, |hereinater OLCD GUIDLLINLS|, araitabte at
-../011222+#!,%+#3"1%#,5&!(.1NB1@eMMLMe!(<?CLK<ML?CA<BT?LLMN<B<B<B<Be@@+-.&$.
1he l1C and the OLCD held a 10th year anniersary o the release o these guidelines in

through these means, the traditional tools o international conlict o law
resolution would hae to suice.
81

Some commentators such as Paul Berman attempted to reach beyond the
traditional dispute resolution mechanisms or resoling conlict o law cases
with principles that take into account the realities o multiple community
ailiations.
82
lis cosmopolitan pluralism` was cosmopolitan` because it
went beyond the laws o any one particular jurisdiction and recognized the
legitimacy o norms created by priate parties and communities.
83
It was plural
because it did not dissole the multiplicity o community ailiations and their
associated norms into a single world-wide standard. Diersity and conlict
would endure and would need to be resoled according to a series o principles
that recognized the need to balance competing national norms.
84

1hese approaches to resoling jurisdictional disputes in cyberspace hae arious
adantages and disadantages. loweer, payment system intermediaries
needed a mechanism to address the jurisdictional question that was easy to
apply, eectie in resoling the dispute, and minimized legal risk to the system
or its members. It could not wait or unpredictable, ater-the-act judgments by
courts. 1he idea they deeloped, discussed in chapter 6 o this book, was that a

December 2009. OLCD, OLCD Conerence on Lmpowering L-Consumers,
-../011222+#!,%+#3"1*,.1!,#(45&!3,#(8!3!(,! ,last isited Sep. 1, 2010,.
81
In an interesting twist, some commentators used the presence o these dispute resolution
mechanisms to argue against indirect liability or intermediaries. \hy deputize intermediaries
to stop illegal actiities on the Internet when goernments can reach the bad actors and
resole any disputes in the normal way Responding to the argument that indirect liability is
needed because the bad actor is unreachable by law enorcement or aggrieed parties,
lolland says:
As an initial matter, it is not clear that a signiicant number o bad actors are
beyond the reach o the law. Adances in technology are making it
increasingly possible to locate and identiy bad actors online, such that online
anonymity is diicult to maintain. Likewise, where the bad actor is identiied
but is ound outside the jurisdiction, soereign goernments hae deeloped
methods or resoling disputes to permit the direct extraterritorial
application o domestic law, such as rules o jurisdiction, conlicts o laws,
and recognition o judgments.
82
Berman, .vra note 3, at 1862.
83
a.
84
a. Berman`s work has ainities with that o political philosophers working in the area o
national soereignty in a global world. ee, e.g., 1homas \. Pogge, \ORLD POVLR1\
AND lUMAN RIGl1S 168-95 ,2002,.

transaction is unacceptable in the payment system i it is illegal in the
jurisdiction o either the buyer or the seller.
85

1he payment card approach proides a simple deault rule or intermediaries to
apply when determining whether to allow transactions in their systems. It
eliminates the heaily act-based balancing assessments needed to determine, on
a case-by-case basis, whose law applies. 1he deault rule also does not simply
adopt a country o origin or country o destination perspectie, each o which is
limited. Nor does it leae the transaction in a legal limbo where no law
applies.
86

1he payment system experience leads to seeral obserations. lirst, direct
conlicts o law are not as requent as some anticipated. 1echnology and
payment system practices eectiely reduce these conlicts to the rare instance

85
Visa`s policy is stated in vtervatiovat Pirac,: 1be Cbattevge. of Protectivg vtettectvat Proert, iv tbe
21.t Cevtvr,: earivg efore tbe vbcovv. ov Covrt., tbe vtervet, ava vtettectvat Proert, of tbe .
Covv. ov tbe ]vaiciar,, 110th Cong. 3-82 ,200, at 1 ,statement o Mark MacCarthy, Senior
Vice President or Global Public Policy, Visa Inc.,. Other payment intermediaries hae
similar procedures, such as eBay`s restriction about selling and shipping illegal goods to the
country where they are illegal. eBay, Oensie Material Policy,
-../011/'"!4+!6'=+,#&1-!$/1/#$*,*!41#88!(4*D!+-.&$ ,last isited leb. 4, 2010,
,|B|ecause eBay is a worldwide community, many o our users lie in countries where the
possession or sale o items associated with hate organizations is a criminal oense. \e can`t
allow the sale or shipping o these items there.`,.
86
1he internal application o this rule inoles system eiciency and the balance o interests
among the stakeholders in the system. I the merchant is in iolation o its own country`s
law, then enorcement is conceptually easy. Merchants discoered in iolation o local law
either hae to stop the transactions or be remoed rom the system. I the merchant is in
iolation o the law in a dierent jurisdiction, things are more complicated. Should the bank
o the merchant or the bank o the customer be burdened with the enorcement
responsibility I the merchant has this responsibility, then he must not introduce the illegal
transaction into the system and the merchant`s bank must not try to process it, then steps
must be taken at the merchant`s end to stop the transaction. 1hese steps could include: a
system decision requiring the merchant to stop these transactions entirely, coding and
programming modiications by the merchant, the merchant`s processor, or the system
operator that would block transactions at the merchant end rom entering the system i the
customer was rom a jurisdiction where the transaction would be illegal, or restricting the
transaction to the merchant`s own jurisdiction. Alternatiely, the enorcement measures
could be put on the cardholder side. Merchants could introduce properly-coded transactions
into the system and rely on action on the cardholder`s side to stop the transaction. 1his
seems to it the case o Internet gambling, where U.S. law makes Internet gambling illegal
or U.S. citizens, and the payment networks responded to the Unlawul Internet Gambling
Lnorcement Act o 2006 ,UIGLA, with a coding and blocking system that allowed
merchants to continue their serices in countries where Internet gambling was illegal, as
discussed earlier in this Article. lor instance, should merchants be responsible or knowing
the laws o all the countries o all the customers they deal with Perhaps not, but i 90 o
their sales are rom an oshore jurisdiction, they should be responsible or knowing that
sales o their product are legal in that jurisdiction. Violations o the policy would largely be
dealt with on a complaint basis.

where the law o one country demands what the law o another country orbids.
Directly contradicting laws are more common in political` areas, where
goernments are seeking inormation rom intermediaries to enorce local laws
against their own citizens.
8

Second, regulating the Internet by ocusing on the local ailiates o global
payment operations does not require the use o either the traditional or the new
cosmopolitan` conlict resolution methods. By relying on global payment
intermediaries, local jurisdictions reach out to the local ailiates that are totally
within their jurisdiction. 1hey do not put burdens on entities in oreign
jurisdictions at all. 1here is literally no conlict and thus nothing to which
normal mechanisms o conlict resolution may attach.
88

Some commentators hae correctly pointed out that when the laws o dierent
jurisdictions apply to a single transaction, the ability o any particular jurisdiction
to unilaterally regulate the Internet is limited.
89
But intermediaries can reduce
these conlicts. Global payment systems can simpliy transactions to eents in
which only a buyer in one jurisdiction and a seller in another are implicated. By
concentrating enorcement on intermediaries instead o indiiduals or
merchants, local jurisdictions can take adantage o the economies that these
institutions make possible.
1he experience o payment intermediaries reeals that, within limits, the
dierences among conlicting jurisdictions can be managed. 1he bordered

8
ee, e.g., Press Release, Priacy Int`l, Lurope`s Priacy Commissioners Rule Against S\Il1
,No. 23, 2006,, araitabte at
-../011222+/3*D',=*(.!3('.*#('$+#3"1'3.*,$!+4-.&$[,&%fMLAgJE7MLA7NLCMCN
,describing the S\Il1 case, where S\Il1 was required to comply with U.S. demands or
access to inancial inormation about Luropean customers in irtue o its operations on US
soil, while such compliance put them in iolation o the Luropean data protection directie,.
In addition, passage o the Global Online lreedom Act ,GOlA, could put Internet
intermediaries in a conlict o law situation with China and other countries. ee Global
Online lreedom Act o 200, l.R. 25, 110th Cong. ,200,. l.R. 25 was introduced by
Representatie Chris Smith on January 5, 200 and would require U.S. intermediaries to resist
certain orders rom countries in which they are doing business. a.
88
Antigua brought a complaint against the U.S. or the enorcement o its gambling laws, but
its success was based only on ,1, the U.S.`s ailure to exclude Internet gambling rom the list
o serices that required open treatment and ,2, the idiosyncrasies o U.S. gambling law
which appear to allow domestic horse racing to engage in Internet gambling while denying
similar opportunities to oshore Internet gambling merchants. But these are technical
obstacles created by the interaction o complex U.S. law and international \1O law and are
not real conlict o law problems. See Appellate Body Report, |vitea tate.-Mea.vre.
.ffectivg tbe Cro..oraer vt, of Cavbtivg ava ettivg errice., 358-64, \1,DS285,AB,R
,Apr. , 2005,. Op cit. .vra note 130.
89
See, e.g., l. Brian lolland, 1be aitvre of tbe Rvte of ar iv C,ber.ace.: Reorievtivg tbe ôrvatire
Debate ov oraer. ava 1erritoriat orereigvt,, 24 J. MARSlALL J. COMPU1LR & INlO. L. 1, 26
,2005,.

Internet works on a small scale. 1he scale is currently small or two reasons:
lirst, he number o cases o goernments reaching across borders to inlict their
laws on Internet merchants in other jurisdictions is still relatiely small.
Moreoer, in contrast to the rhetoric about the Internet creating a global
marketplace, the scope o cross-border commerce itsel is still limited. 1he
reality is that the olume o cross-border transactions is not large enough to
create a truly substantial cross-border jurisdictional crisis. Currently, only our
percent o the sales or electronic commerce merchants in the U.S. come rom
abroad.
90
And data rom Lurope show that cross border online transactions are
not increasing as ast as oerall e-commerce transactions, staying relatiely
stable rom 2006 to 2008 at six to seen percent.
91

As Daid Post has warned, the problem the Internet creates or local
jurisdictions is one o scale.
92
1he bordered Internet simply does not scale up.
Global payment systems cannot accommodate an enorcement burden in which
each jurisdiction uses payment system mechanisms to enorce each o its local
laws on the Internet.
It is not hard to see how we can get into a kind o tragedy o the commons in
this area. Lach indiidual extension o local jurisdiction into cyberspace seems
small and costless, but collectiely the burden becomes unbearable.
Goernments might eel ree to exploit this enorcement mechanism, in the
same way that grazers use the commons-under the impression that it is an
unlimited resource. loweer, one o two outcomes will occur as the cross-
border rules pile up: Lither cross-border transactions will remain small and the
potential or the Internet to be a global channel o commerce will not be
realized, or the political costs o each goernment attempting to regulate the e-

90
1his is based on transaction data rom the Visa system. ee vtervatiovat Pirac, earivg, .vra
note 85, at 5 ,statement o Mark MacCarthy, Senior Vice President or Global Public
Policy, Visa Inc.,.
91
Comm`n o the Luropean Cmtys., Covvi..iov taff !or/ivg Docvvevt: Reort ov Cro..oraer
covverce iv tbe | 3, SLC ,2009, 283 inal ,Mar. 5, 2009,, araitabte at
-../011!,+!53#/'+!51,#(45&!3414.3'.!"=1%#,41,#&<4.'88<2/?@@K<!(+/%8 ,lrom
2006 to 2008, the share o all LU consumers that hae bought at least one item oer the
Internet increased rom 2 to 33 while cross-border e-commerce remained stable ,6 to
,.`,.
92
ee Post, .gaiv.t .gaiv.t C,beravarcb,, .vra note 15, at 13 ,stating that scale matters`,, .ee
at.o lolland, .vra note 89, at 29. lolland states:
1he online actor cannot know, as a practical matter, the many laws applicable
to a particular act, nor when one or more soereigns may decide to attempt
regulatory action. 1his is particularly true in those areas o regulation in
which morality, religion and culture are at their most inluential, such as
speech, race, sex, and een intellectual property. Moreoer, it is not simply
one actor or a ew legal systems. It is an exponential multitude.
a.

commerce actiities o other countries will mount. Lither deelopment reeals
the limitations o the bordered Internet as a long-term ramework or Internet
goernance.
Goldsmith and \u suggest that enorcement o Internet regulations through
intermediaries is necessarily limited in size.
93
1hey suggest that maybe the
system will not be able to scale up, but it won`t hae to.
94
Small countries such
as Antigua cannot enorce Internet rules because global intermediaries can
simply pull up stakes and leae i the rules are too strict.
95
loweer, there are a
suiciently large number o countries that global intermediaries will not eel
capable o abandoning. I all o them use the intermediary enorcement
mechanism, the system will be oerwhelmed.
Internationalism
1he undamentally correct insight o the Internet exceptionalists is that the
unilateral imposition o one nation`s law onto all Internet actiities that cross
borders won`t scale.
96

Internationalism might be the way out. It is the idea that the Internet will
eentually be goerned, at least or some serices, by global institutions and
arrangements, and that this is the right public policy or local goernments to
ollow in their dealings with illegal cross border Internet transactions.
9
1his
policy could be implemented through a uniorm global standard, or any o a
ariety o techniques such as \orld 1rade Organization rules that bring local
laws into harmony. 1he basic justiication or this policy is similar to the
justiication or establishing a single uniorm national policy that preents the
clash o inconsistent rules at the state leel: \hen actiities hae widespread
and signiicant eects on those outside the local jurisdiction, then uniorm
principles or some other coordinating mechanism should be adopted at the
higher leel.
98
1his uniersalism could promise better laws, whereby the

93
GOLDSMI1l & \U, .vra note , at 81-82.
94
a. at 81.
95
ee ia. at 160 ,suggesting that acting as the Internet police is just a normal cost o doing
business or global companies, which they can aoid in a particular case by leaing a country
that tried to impose costs that exceeded the beneits o continued presence in the country
and thus creating another objection to the bordered Internet to eectiely gie larger
countries a greater role in Internet goernance than smaller ones,.
96
ee Johnson & Post, .vra note 1, at 1390 ,One nation`s legal institutions should not
monopolize rule-making or the entire Net.`,.
9
GOLDSMI1l & \U, .vra note , at 26.
98
a. ,I the nations o the world agree to a single global law or questions like libel,
pornography, copyright, consumer protection, and the like, the lies o Internet users

|i|nternational standards could relect a kind o collection o best practices
rom around the world - the opposite o the tyranny o the unreasonable.`
99

Goldsmith and \u make seeral criticisms o internationalism. lirst, a system
o uniersal laws would be unattractie, it would leae the world diided and
discontent because the uniersal law would be unpopular in large segments o
the world population. Second, the system o local national laws would better
relect dierences among people. Diersity is a good thing and cannot be taken
into account by a uniersal code that oerrides local dierences. 1hird, it is not
needed. 1he conlicts o laws, extraterritoriality, and other considerations are
perectly manageable within the current international ramework. lor example,
since most Internet users do not hae assets in other countries, they are
eectiely subject only to the laws o the country where they lie. Only large
multinational companies with assets all oer the world ace the
multijurisdictional problem, and they already hae to lie with that because they
are already global. Compliance with a plurality o international laws is simply a
cost o doing business or global companies. 1here`s nothing new here that
would justiy a moe to a more harmonized global order. 1here are extra costs
to be sure, but nothing so onerous or burdensome that it would require a moe
to global law.
100

1he responses to these criticisms are straightorward. An unpopular global law
is not the goal. Neither is suppression o diersity the goal. 1he idea is to
integrate local laws in some ashion when the regular conlicts among them
proe to be intolerable. \hen diersity does not create this diiculty, there is
no need or integration. I, or example, local goernments alue diersity
enough to rerain rom using intermediaries to enorce local laws against actors
in other jurisdictions, then there is no need or harmonization o these
enorcement eorts. But to the extent that goernments want to take global
enorcement steps, they also need to take steps to integrate the laws they want

become much simpler: no conlicting laws, no worries about complying with 15 dierent
legal systems, no race to the bottom.`,.

99
a. at 2. Reidenberg also argues that as jurisdictions increasingly conlict there will need to
be an oerarching harmonization o international rules:
|O|nline enorcement with electronic blockades and electronic sanctions will
cause serious international political conlicts. 1hese conlicts arise because o
the impact on territorial integrity. Such conlicts are likely to orce
negotiations toward international agreements that establish the legal criteria
or a state to use technological enorcement mechanisms. 1his progression
leads appropriately to political decisions that will deine international legal
rules.
Joel R. Reidenberg, tate. ava vtervet vforcevevt, 1 U. O11A\A L. & 1LCl. J. 213,
230 ,2003-2004,.
100
ee GOLDSMI1l & \U, .vra note , at 152-60.

to enorce. 1he reason or this is that global intermediaries` costs to mediate
the conlicts associated with unilateral attempts at local regulation o the
Internet will be so onerous and burdensome that they will cause an unwarranted
and unnecessary decline in global interaction.
101

Berman also describes how the internationalist hope or global standards aoids
the conlict o law problem: i we constructed one uniersal world
community` with one set o goerning rules, there would neer need to be a
choice o law` in the sense that conlict-o-laws scholars use the term.`
102

loweer, he is critical o this uniersal world community or two reasons.
lirst, he is critical o this community because o its potential to dissole
community ailiations that proide important emotional connections and
opportunities or normatie discussion o those connections. Second, he iews
this uniersal community as undamentally unrealistic gien the dominance o
current notions o nation-state soereignty.
103

1hese objections can be met at the leel o generality at which they are cast. \e
do not need to think o ourseles as primarily world citizens in order to endorse
speciic global approaches. \e can still hae deep attachments to local
communities and can still debate the relatie importance o the oerlapping
communities we participate in. 1he global approach endorses the iew that
sel-goernment requires a politics that plays itsel out in a multiplicity o
settings, rom neighborhoods to nations to the world as a whole` and citizens
who can abide the ambiguity associated with diided soereignty, who can think
and act as multiply situated seles.`
104
But participation in global community
and the wisdom to know when the global perspectie should take precedence
oer more local concerns is essential to this ision o sel-goernment in a
global world.
1he internationalist proposal is to proide global coordination only when
necessary. It is to moe to global standards when, as a practical matter, the
burdens o allowing dierse local rules are too high. 1he model o national
uniorm standards is appropriate: not eerything has to be done at the national
leel, but some things should be done there in order to hae an eicient and air
national system. Similarly, there is no need to moe rom the current system to

101
Interestingly, the earlier Jack Goldsmith seemed more inclined to accept these practical
considerations as a rationale or international harmonization: \hen in particular contexts
the arbitrariness and spilloers become too seere, a uniorm international solution remains
possible.` Goldsmith, .vra note 4, at 1235.
102
Berman, .vra note 3, at 1860.
103
a. at 1860-61.
104
MIClALL J. SANDLL, PUBLIC PlILOSOPl\: LSSA\S ON MORALI1\ IN POLI1ICS 34 ,2005,.

a world goernment. But i there are practical ways to improe Internet
goernance through global harmonization, they should be taken.
I goernments are going to use payment intermediaries as enorcers o local
law, there are a number o steps that could be taken to coordinate their eorts,
including:
In the Internet gambling context, a moe to an internationally-
interoperable licensing system that would require each jurisdiction that
allows Internet gambling to deer to the licensing decisions o other
jurisdictions
In the copyright context, the continued eolution o uniorm copyright
rules.

International agreements are one mechanism to create coordinated action.
Although controersial because o the secrecy inoled in its deelopment, and
the sense that aected parties were excluded rom participation, the Anti-
Countereiting 1rade Agreement ,AC1A, is a reasonable, though lawed, model
or action in this area.
105
1here are many mechanisms or international
coordination. Decisions regarding which mechanisms to use depend on the
issue and the ora aailable or resolution.
Internationalism has its dangers. \hy should each jurisdiction hae the same
regulations on hate speech and the same regulations on alcohol consumption
1he answer is that there will be no harmonization where there are such
undamental dierences. Intermediaries will be called upon to resole the issue
themseles or they will be caught between warring goernments and orced to
choose sides. But eorts should be made to minimize such dierences when
these dierences hae global consequences, especially when they are supericial
dierences that relect no undamental diisions. lor the same reason that we
want uniorm global technical standards or inormation and communications
technologies, i possible, we want similar legal rameworks i goernments are
going to enorce laws on the Internet.
1hese eorts to ease the riction inoled in extending goernment authority to
the Internet through a global ramework are in line with other eorts to create
global rameworks that promote the growth o the Internet. lor example, the
thirty-irst International Conerence o Data Protection and Priacy
Commissioners, held in Madrid in Noember 2009, adopted a set o global

105
ee Media Statement, Participants in AC1A Negotiations, .vtiCovvterfeitivg 1raae .greevevt
;.C1.), June 12, 2009, araitabte at
-../011222+&!%+"#D.+(h1.!&/$'.!419'"!<<<<L@KAL+'4/E. lor a summary o the
AC1A process and the content o the agreement, .ee 1lL AN1I-COUN1LRlLI1ING 1RADL
AGRLLMLN1 - SUMMAR\ Ol KL\ LLLMLN1S UNDLR DISCUSSION ,2009,, araitabte at
-../011222+&!%+"#D.+(h1.!&/$'.!41P5$.*/'"!:#,5&!(.^]G<<<<L@NCM+'4/E.

priacy standards.
106
1here is also likely to be a renewed push or global
consumer protection on the occasion o the tenth anniersary o the
Organisation or Lconomic Co-operation and Deelopment`s Guidelines or
Consumer Protection in the Context o Llectronic Commerce.
10

Both these eorts relate to the growth o the Internet as a ibrant international
marketplace. 1hey do this by building online trust. Global inormation security
standards reassure people that their inormation is sae no matter what the
physical location o the websites they isit. Lstablishing global priacy
standards means that the collection and use o online inormation will be
goerned by common principles regardless o a website`s jurisdiction and will
make it easier or global business to transer inormation rom one jurisdiction
to another in a seamless manner. linally, eectie global consumer protection
rules will mean that people will hae the inormation and redress rights they
need to shop conidently online no matter where the website is located.
Conclusion
1he initial demand rom Internet exceptionalists that the online world be let
alone by goernments has morphed into the idea that goernments should
create a global ramework to protect and spur the growth o the Internet. 1he
interening steps in this deelopment are not hard to trace: Internet
exceptionalists conused their ideal o sel-goerning Internet communities with
the idea that the Internet was ungoernable because it was a global
communications network that crossed borders. 1his idea o an intrinsically
ungoernable Internet was undermined by the recognition that the coding that
underlies Internet applications and serices is a matter o choice, not

106
Artemi R. Lombarte, Dir., Agencia Lspanola de Proteccin de Datos, Slide Presentation:
International Standards on Data Protection & Priacy ,2009,, araitabte at
-../4011222+'"/%+!41/#3.'$2!61,'('$%#,5&!(.',*#(1,#&/'3!,!(,*'41,#&&#(1O
i99<93*D',=<H5&&*.<@K+/%8. le describes one o the main criteria o the global priacy
standards project as 1o elaborate a set o principles and rights aimed to achiee the
vaivvv aegree of ivtervatiovat accetavce, ensuring at once a high leel o protection.` a.
,emphasis in original,. lor the standards adopted, see 1lL MADRID PRIVAC\ DLCLARA1ION
,No. 3, 2009,, -../011.-!/56$*,D#*,!+#3"1^-!P'%3*%93*D',=:!,$'3'.*#(+/%8.
10
OLCD GUIDLLINLS, .vra note 80, .ee at.o Org. or Lcon. Co-operation & De., Conerence
on Lmpowering L-Consumers: Strengthening Consumer Protection in the Internet
Lconomy, Programme ,2009,, araitabte at
-../011222+#!,%+#3"1%'.'#!,%1MM1??1LL@LNMAC+/%8 ,describing the conerence,. 1he
OLCD endorsed steps toward global enorcement o some consumer protection rules in a
2003 report on cross-border raud and a 200 report on consumer dispute resolution and
redress. ee Comm. on Consumer Policy, Org. or Lcon. Co-operation & De., OLCD
Guidelines or Protecting Consumers rom lraudulent and Deceptie Commercial Practices
Across Borders ,2003,, araitabte at -../011222+#!,%+#3"1%'.'#!,%1?L1MM1?KNCLCL+/%8,
Comm. on Consumer Policy, Org. or Lcon. Co-operation & De., OLCD Recommendation
on Consumer Dispute Resolution and Redress ,200,, araitabte at
-../011222+#!,%+#3"1%'.'#!,%1LM1N@1MTKC@B@B+/%8.

unchangeable nature. I something about this system created diiculties or
goernment control, this could be changed. lurther, the idea that goernments
cannot control the Internet was undermined by the need or the local
operations o global intermediaries to proide essential Internet serices and the
practical ability o goernments to control these intermediaries.
Internet intermediaries can control the content o the actiities on their online
communities, and goernment can compel or pressure intermediaries to take
these steps. Intermediaries hae a general obligation to ollow the law, and
except in extreme cases, they hae no right to resist these lawully established
burdens. 1he establishment o these laws needs to ollow all the rules o good
policymaking, including imposing an obligation only when the social beneits
exceed the social costs. loweer, a bordered Internet in which each country
attempts to use global intermediaries to enorce its local laws will not scale.
1his is the undamentally correct insight o the Internet exceptionalists. I
goernments are going to use intermediaries to enorce local laws, they are
going to hae to harmonize the local laws they want intermediaries to enorce.
237

CHAPTER 4
HAS THE INTERNET FUNDAMENTALLY
CHANGED ECONOMICS?
Computer-Mediated Transactions 239
lal R. Varian
Decentralization, Freedom to Operate
& Human Sociality 257
\ochai Benkler
From Dismal Science to Strange Tales 273
Larry Downes
The Regulation of Reputational Information 293
Lric Goldman

238 CHAPTER 4: HAS THE INTERNET FUNDAMENTALLY CHANGED ECONOMICS?


Computer-Mediated Transactions
By Hal R. Varian
*

Lery now and then a set o technologies becomes aailable that sets o a
period o combinatorial innoation.` 1hink o standardized mechanical parts
in the 1800s, the gasoline engine in the early 1900s, electronics in the 1920s,
integrated circuits in the 190s, and the Internet in the last ew decades.
1he component parts o these technologies can be combined and recombined
by innoators to create new deices and applications. Since these innoators
are working in parallel with similar components, it is common to see
simultaneous inention. 1here are many well-known examples such as the
electric light, the airplane, the automobile, and the telephone. Many scholars
hae described such periods o innoation using terms such as recombinant
growth,` general purpose technologies,` cumulatie synthesis` and clusters
o innoation.`
1

1he Internet and the \eb are wonderul examples o combinatorial innoation.
In the last 15 years we hae seen a huge prolieration o \eb applications, all
built rom a basic set o component technologies.
1he Internet itsel was a rather unlikely innoation, I like to describe it as a lab
experiment that got loose.` Since the Internet arose rom the research
community rather than the priate sector, it had no obious business model.
Other public computer networks, such as AOL, CompuSere, and Minitel,
generally used subscription models, but were centrally controlled and oered
little scope or innoation at the user leel. 1he Internet won out oer these
alternaties, precisely because it oered a lexible set o component
technologies that encouraged combinatorial innoation.
1he earlier waes o combinatorial innoation required decades, or more, to
play out. lor example, Daid lounshell argues that the utopian ision o

Uni. o Cal., Berkeley and Google. halischool.berkeley.edu.
1
"##$ #%&%, Martin \eitzman, '#()*+,-.-/ 01)2/3, 113 Q. J. Ol LCON. 331-360 ,1998,, 1imothy
Bresnahan & M. 1rajtenberg, 0#-#1.4 5617)8# 9#(3-)4)&,#8: ;-&,-#8 )< 01)2/3=, 65 J. Ol
LCONOML1RICS 83-108 ,1995,, .>.,4.+4# ./
!""#$%%&'()*+,(#(-+.,/%)%(((%(-.0.1%23456774&6#89:6;8+!"1<, 1imothy Bresnahan,
0#-#1.4 5617)8# 9#(3-)4)&,#8, ,- lANDBOOK Ol 1lL LCONOMICS Ol INNOVA1ION ,Bronwyn
lall & Nathan Rosenberg, eds., 2010,, Nathan Rosenberg, 9#(3-)4)&,(.4 ?3.-&# ,- /3# @.(3,-#
9))4 A-B68/1C, ,- PLRSPLC1IVLS IN 1LClNOLOG\ 9-31 ,196,, ABBO11 PA\SON USlLR, A
lIS1OR\ Ol MLClANICAL INVLN1ION ,reised ed., Doer Publ`ns 1998,, and Joseph A.
Schumpeter, 93# D-.4C8,8 )< ;()-)*,( ?3.-&#, ,- LSSA\S ON LN1RLPRLNLURS, INNOVA1IONS,
BUSINLSS C\CLLS AND 1lL LVOLU1ION Ol CAPI1ALISM 134-149 ,Richard V. Clemence, ed.,
2000, ,originally published in '#>,#2 )< ;()-)*,( "/./,8/,(8, May 1935,.

interchangeable parts took more than a century to be realized.
2
1he \eb was
inented in the early 1990s, but it did not become widely used until the mid-
1990s. Since then, we hae seen a huge number o noel applications-rom
\eb browsers, to search engines, to social networks-to mention a ew
examples. As with the Internet, the \eb initially had no real business model,
but oered a ertile ground or combinatorial innoation.
Innoation was so rapid on the Internet because the component parts were all
bits. 1hey were programming languages, protocols, standards, sotware
libraries, productiity tools and the like. 1here was no time to manuacture, no
inentory management, and no shipping delay. \ou neer run out o l1ML,
just like you neer run out o email. New tools could be sent around the world
in seconds and innoators could combine and recombine these bits to create
new \eb applications.
1his parallel inention has led to a burst o global innoation in \eb
applications. \hile the Internet was an American innoation, the \eb was
inented by an Lnglishman liing in Switzerland. Linux, the most used
operating system on the \eb, came rom linland
3
, as did MySQL, a widely
used database or \eb applications.
4
Skype, which uses the Internet or oice
communication, came rom Lstonia.
5

O course, there were many other technologies with worldwide innoation,
such as automobiles, airplanes, photography, and incandescent lighting.
loweer, applications or the Internet, which is inherently a communications
technology, could be deeloped eerywhere in the world in parallel, leading to
the rapid innoation we hae obsered.
Computer-Mediated Transactions
My interest in this essay is in the economic aspects o these technological
deelopments. I start with a point so mundane and obious, it barely seems
worth mentioning: Nowadays, most economic transactions inole a computer.

2
Daid A. lounshell, lrom the American System to Mass Production, 1800-1932: 1he
Deelopment o Manuacturing 1echnology in the United States ,1985,.
3
"## Linus 1oralds & Daid Diamond, Just or lun: 1he Story o an Accidental Reolution
,2002,.
4
"## Oracle Corporation, E1)* F,8,)-8 /) '#.4,/C: D- A-/#1>,#2 2,/3 G.>,B DH*.1I$ ?)JE)6-B#1 )<
@C"KL DM, July 200, !""#$%%'(2+15*=<+-.1%"(-!:,(*.>,-(*%&0"(,2&(?*%')2&':
)@1),A+!"1<.
5
"## Andreas 1homann, "IC7#: D M.4/,( "6((#88 "/)1C, CRLDI1 SUISSL GROUP, June 9, 2006,
!""#$%%(1)/)B&0(+-,('&":
*>&**(+-.1%)##%),"&-<(%&0'(@+-C1DC>*()-"&.0EF#(0G,"&-<(H).&'E63963IH-.&'EI8
;4H<)0/EJK.

Sometimes this computer takes the orm o a smart cash register, part o a
sophisticated point o sale system, or a website. In each o these cases, the
computer creates a record o the transaction.
1his record-keeping role was the original motiation or haing the computer as
part o the transaction. Creating a record o transactions is the irst step in
building an accounting system, thereby enabling a irm to understand its
inancial status.
Now that computers are in place, they can, howeer, be used or many other
purposes. In this essay, I explore some o the ways that computer mediation
can aect economic transactions. 1hese computer mediated transactions, I
argue, hae enabled signiicant improements in the way transactions are carried
out and will continue to impact the economy in the oreseeable uture.
I classiy the impact o computer mediated transactions into our main
categories according to the innoation they acilitate:
New orms o contract,
Data extraction and analysis,
Controlled experimentation,
Personalization and customization.

Enable New Forms of Contract
Contracts are undamental to commerce. 1he simplest commercial contract
says, I will do X i you do \,` as in I will gie you >1 i you gie me a cup o
coee.` O course, this requires that the actions to be taken are eriiable. Just
asking or coee does not mean that I will get it. As Abraham Lincoln
supposedly remarked, I this is coee, please bring me some tea, but i this is
tea, please bring me some coee.`
6

A computer used in a transaction can obsere and eriy many aspects o that
transaction. 1he record produced by the computer allows the contracting
parties to condition the contract on terms that were preiously unobserable,
thereby allowing or more eicient transactions.
I am not claiming that increased obseration will necessarily lead to more
eicient contracts. 1here are counterexamples to the assertion that more

6
Susan L. Rattiner, lood and Drink: A Book o Quotations ,2002,.

inormation is better` such as the lirshleier example.
I am merely claiming
that additional inormation allows or more eicient contracts.
O course, the study o contracts is a highly deeloped ield in economics. As
such, it is hardly noel to suggest that contractual orm depends on what is
obserable. \hat is interesting, howeer, is the way that progress in
inormation technology enables new contractual orms.
Consider, or example, a rental-car agency that buys insurance based on
accident rates, and that accident rates, in turn, depend on the speed o a ehicle.
All renters would preer to drie within the speed limit i they are compensated
with a lower rental ee. loweer, i there is no way to monitor the speed o a
rental car, such a contractual proision is unenorceable. Putting a computer
transmitter in the trunk o the car that records the ehicle`s speed makes the
contract enorceable and potentially makes eeryone better o..
8

1he transportation sector has capitalized on the aailability o computerized
transmitters to create more eicient contracts in many areas.
Car dealers are selling cars with starter interrupt` deices that inhibit
operations i car payments are missed.
9

Similar interrupt deices attached to breath analyzers are mandated or
drunk driing oenders in many states.
Parents can buy a deice known as MyKey` which allows them to
limit auto speed, cap the olume on the radio, require seat belt use and
encourage other sae-driing habits or teenage driers.
10

In the releant economics literature, lubbard and Baker examine a
ariety o ways that ehicular monitoring systems hae impacted the
trucking industry.
11

Jack lirshleier, 93# 51,>./# .-B ")(,.4 F.46# )< A-<)1*./,)- .-B /3# '#2.1B /) A->#-/,># D(/,>,/C, 61
1lL AM. LCON. RLV. 561-4 ,Sept. 191,, .>.,4.+4# ./
!""#$%%C)-><"5+C>=>)+'>A(+('>%L=-M%NG49M%67I6OM;GJPOM;Q&,*!<(&C(,+#'C.
8
1his is a particularly simple case. I driers hae heterogeneous preerences, those who
preer to speed may be made worse o by the aailability o such a deice.
9
Associated Press, E)1 ")*# N,&3J1,8I D6/) M6C#18$ '#7) @.- ,8 . N,&3J/#(3 0.B&#/, L.A. 1IMLS,
June 13, 2006, !""#$%%),"&-<(*+<)"&1(*+-.1%M;;3%R>0%69%S>*&0(**%C&:<)"(69.
10
Nick Bunkley & Bill Vlasic, ;-861,-& O6-,)1 0)#8 <)1 . @,4B ',B#, N.\. 1IMLS, Oct. 6, 2008,
!""#$%%???+05"&1(*+-.1%M;;8%6;%;I%)>".1.S&<(*%;I)>".+!"1<.
11
1homas N. lubbard, 93# G#*.-B )< @)-,/)1,-& 9#(3-)4)&,#8: 93# ?.8# <)1 916(I,-&, 115 Q. J. Ol
LCON. 533-560 ,2000,,
http:,,???+1&"#,(**R.>,0)<*+.,/%'.&%)S*%6;+663M%;;99449;;44T8T4, George Baker &
1homas N. lubbard. ?)-/1.(/,+,4,/C .-B D88#/ P2-#183,7: P-J+).1B ?)*76/#18 .-B 0)>#1-.-(# ,-
Q" 916(I,-&, 119 Q. J. Ol LCON. 1443-149 ,2004,,
http:,,???+1&"#,(**R.>,0)<*+.,/%'.&%)S*%6;+663M%;;99449;TMTI364M.

1here are many other examples o computer-mediated contracts. 1he work o
Dana & Spier and Mortimer, proides examples that describe the eiciency
gains resulting rom reenue sharing in the ideo tape rental industry.
12

Video tapes were originally purchased by retail stores rom distributors or
about >65 per tape. Since the ideos were so expensie, stores only bought a
ew. As a result, the popular ideos quickly disappeared rom the sheles,
making eeryone unhappy.
In 1998, retailers and distributors adopted a new business model: a reenue
sharing arrangement in which stores paid a small upront ee o >3 to >8, but
split the reenue when the ideo was rented, with 40 to 60 going to the
retailer. Stores no longer had an incentie to economize on purchases, and all
parties to the transaction-retailers, distributors, and customers-were made
better o.
Sharing reenue at point o sale requires that both parties be able to monitor the
transaction. 1he technological innoations o bar code scanning, the
computerized cash register, and computer networks enabled reenue-sharing
arrangements.
O course, when a transaction takes place online, reenue-sharing is much
easier. Online adertising is a case in point where reenue rom an adertiser
or an ad impression or click may be split among publishers, ad exchanges, ad
networks, ailiates and other parties based on contractual arrangements.
Although the beneits rom computers oering *)1# inormation to contracting
parties hae only been discussed thus ar, there are also cases in which
computers can be used to improe contractual perormance by 3,B,-&
inormation using cryptographic methods. A picturesque example is the
cocaine auction protocol` which describes an auction mechanism designed to
hide as much inormation as possible.
13

linally, algorithmic game theory` is an exciting hybrid o computer science
and economic theory that deseres mention. 1his subject brings computational
considerations to game theory ,how a particular solution can be computed, and

12
James D. Dana & Kathryn L. Spier. '#>#-6# "3.1,-& .-B F#1/,(.4 ?)-/1)4 ,- /3# F,B#) '#-/.4
A-B68/1C, XLIX Q. J. Ol LCON. 223-245 ,2001,, .>.,4.+4# ./
!""#$%%???9+&0"(,*-&(0-(+?&<(5+-.1%R.>,0)<%6687IMTT7%)S*",)-", Julie l. Mortimer,
F#1/,(.4 ?)-/1.(/8 ,- /3# F,B#) '#-/.4 A-B68/1C, 5 RLV. Ol LCON. S1UDILS 165-199 ,2008,,
!""#$%%???9+&0"(,*-&(0-(+?&<(5+-.1%R.>,0)<%6679748MM%)S*",)-".
13
lrank Stajano & Ross Anderson, 93# ?)(.,-# D6(/,)- 51)/)()4: P- /3# 5)2#1 )< D-)-C*)68
M1).B(.8/, ,- PROCLLDINGS Ol INlORMA1ION lIDING \ORKSlOP, LLC1URL NO1LS IN
COMPU1LR SCILNCL, 1999, !""#$%%???+-<+-)1+)-+>A% ,R)6T%V)#(,*%-.-)&0(+#'C.

strategic considerations to algorithm design ,whether a particular algorithm is
actually incentie-compatible,.
14

Some History of Monitoring Technologies
1hough I hae emphasized computer mediated transactions, a computer can be
deined quite broadly. 1he earliest example o an accounting technology I
know o that enabled new orms o contract inoles Mediterranean shipping
circa 3300 B.C.
1he challenge was how to write a bill o lading` or long distance trade in
societies that were pre-literate and pre-numerate. 1he brilliant solution was to
introduce small clay tokens, known as bullae,` which were small
representations o the material being transported. As each barrel o olie oil
was loaded onto a ship, a barrel-shaped token was placed in a clay enelope.
Ater the loading was completed, the enelope was baked in a kiln and gien to
the ship`s captain. At the other end o the oyage, the enelope was then
broken open and the tokens were compared to the barrels o oil on the ship as
they were unloaded. I the numbers matched, the contract was eriied. Later,
marks were scratched on the outside o the enelope to indicate the number o
tokens inside. Some authors beliee that this innoation led to the inention o
writing between 3400 and 3300 B.C.
15

A somewhat more recent example is the inention o the cash register in 1883
by James Ritty.
16
Ritty, a saloon owner, discoered that his employees were
stealing money. In response, he deeloped a deice to record each transaction
on paper tape, an inention that he patented under the name o the
incorruptible cashier.`
1
Ritty`s machine became the basis o the National Cash
Register ,NCR, Company ounded in 1884. 1he NCR deice added a cash
drawer and a bell that sounded ka-ching` wheneer the drawer was opened, to
alert the owner o the transaction, thereby discouraging pilering. 1his
improed monitoring technology made retailers willing to hire employees

14
See NOAM NISAN, 1IM ROUGlGARDLN, LVA 1ARDOS, AND VIJA\ V. VAZIRANI, LDS.,
ALGORI1lMIC GAML 1lLOR\ ,200, or a comprehensie collection o articles and lal R.
Varian, ;()-)*,( @#(3.-,8* G#8,&- <)1 ?)*76/#1,R#B D&#-/8, ,- USLNIX \ORKSlOP ON
LLLC1RONIC COMMLRCL 13-21 ,1995,,
!""#$%%???+*&1*+S(,A(<(5+('>%!)<%V)#(,*%1(-!)0&*1:'(*&/0+#'C or an early
contribution to this theory.
15
Jean-Jacques Glassner, Zainab Bahrani, & Marc Van de Miero, 1he Inention o Cuneiorm:
\riting in Sumer ,2005,.
16
MI1 School o Lngineering, A->#-/)1 )< /3# S##I D1(3,>#: O.*#8 ',//C$ ?.83 '#&,8/#1, April 2002,
!""#$%%?(S+1&"+('>%&02(0"%&.?%,&""5+!"1<.
1
Cash Register and Indicator, U.S. Patent 21,363 ,iled leb 15, 1882,, .>.,4.+4# ./
!""#$%%???+/../<(+-.1%#)"(0"*D!<E(0H<,EH2&'EWXVGYMI6939.

outside their immediate amilies, leading to larger and more eicient
establishments.
18

Enabling Online Advertising
Online adertising seres as a poster child or algorithmic mechanism design. A
Pasadena company called Go1o began ranking search results using an auction.
19

Users did not like this particular orm o search, so Go1o switched to using an
auction to rank adertisements. In the original auction, ads were ranked by bid
per click` and adertisers paid the amount they bid. Ater consultation with
auction theorists, Go1o moed to a second-price auction: An adertiser paid a
price per click determined by the bid o the adertiser in the next lower
position.
20

1here is a undamental diergence o incenties in adertising. 1he publisher
,i.e. the content proider, has space on its \eb page or an ad and wants to sell
ad impressions to the highest bidders. 1he adertiser does not care directly
about ad impressions, but does care about isitors to its website, and ultimately,
the sale o its products. lence, the publisher wants to sell impressions, but the
adertiser wants to buy clicks.
1his is similar to an international trade transaction where the buyer wants to pay
in euros and the seller wants to receie dollars. 1he solution in both cases is the
same: an exchange rate. In the context o online adertising, the exchange rate
is the predicted click-through rate, an estimate o how many clicks a particular
ad impression will receie. 1his allows one to conert the adertiser`s oered
bid per click to an equialent bid per impression. 1he publisher can thus sell
each impression to the highest bidder.
1his mechanism aligns the interests o the buyer and the seller, but creates
unintended consequences. I the adertiser only pays or clicks, he has no
direct incentie to economize on impressions. Lxcessie impressions, howeer,

18
"## JoAnne \ates, M68,-#88 Q8# )< A-<)1*./,)- .-B 9#(3-)4)&C <1)* TUUVJTWXV, ,- A NA1ION
1RANSlORMLD B\ INlORMA1ION: lO\ INlORMA1ION lAS SlAPLD 1lL UNI1LD S1A1LS
lROM COLONIAL 1IMLS 1O 1lL PRLSLN1 10-135. ,Alred D. Chandler and James Cortada,
eds., 2000, ,detailing the role o oice machinery in the deelopment o commercial
enterprises,.
19
0)9)%()* 5)8/8 "/1)-& '#4#>.-(C '.-I,-& ,- Y5G "61>#C )< "#.1(3 ;-&,-#8, BUSINLSS \IRL, April
11, 2000, .>.,4.+4# ./ !""#$%%???+!&/!S()1+-.1%'.-%6Z6:36TM9686+!"1<.
20
lor accounts o the deelopment o these auctions, see John Battelle, 1lL SLARCl: lO\
GOOGLL AND I1S RIVALS RL\RO1L 1lL RULLS Ol BUSINLSS AND 1RANSlORMLD OUR
CUL1URL ,2005,, Stee Ley, "#(1#/ )< 0))&4#-)*,(8: G./.J<6#4#B '#(,7# M1#28 51)<,/.+,4,/C, \IRLD,
2009, !""#$%%???+?&,('+-.1%-><">,(%-><">,(,(2&(?*%1)/)B&0(%6I:
;3%0(#[/../<(0.1&-*D->,,(0"V)/(E)<<.

impose an attention cost on users, so urther attention to ad quality is important
to ensure that ad impressions remain releant to users.
Nowadays, the major proiders o search engine adertising all estimate click-
through rates along with other measures o ad quality and use auctions to sell
these ads. Lconomists hae applied game theory and mechanism design to
analyze the properties o these auctions.
21

Enabling Data Extraction & Analysis
1he data rom computer-mediated transactions can be analyzed and used to
improe the perormance o uture transactions.
1he Sabre air passenger reseration system oered by American Airlines is an
example o this. 1he original conception, in 1953, was to automate the creation
o an airline reseration. loweer, by the time the system was released in 1960,
it was discoered that such a system could also be used to study patterns in the
airline reseration process: 1he acronym Sabre stands or Semi-Automatic
Business Research Lnironment.
22

1he existence o airline reseration systems enabled sophisticated dierential
pricing ,also known as yield management`, in the transportation industry.
23

Many irms hae built data warehouses based on transaction-leel data which
can then be used as input or analytic models o customer behaior. A
prominent example is supermarket scanner data which has been widely used in
economic analyses.
24
Scanner data has also been useul in constructing price

21
"##$ #%&%, Susan Athey & Glenn Lllison, 5)8,/,)- D6(/,)-8 2,/3 ?)-86*#1 "#.1(3, 200.
!""#$%%A>B0("*+C)*+!),2),'+('>%L)"!(5%#.*&"&.0+#'C, Benjamin Ldelman, Michael
Ostrosky, & Michael Schwartz, A-/#1-#/ DB>#1/,8,-& .-B /3# 0#-#1.4,R#B "#()-B 51,(# D6(/,)-, 9
AM. LCON. RLV. 242-259 ,March 200,, lal R. Varian, P-4,-# DB D6(/,)-8, 99 AM. LCON.
RLV. 430-434 ,2009,.
22
Sabre, N,8/)1C, .>.,4.+4# ./ !""#$%%?(S+),-!&2(+.,/%?(S%M;;8;MM4636947%
!""#$%%???+*)S,()&,<&0(*.<>"&.0*+-.1%)S.>"%!&*".,5+!"1.
23
Barry C. Smith, John l. Leimkuhler, & Ross M. Darrow, Z,#4B @.-.&#*#-/ ./ D*#1,(.- D,14,-#8,
22 IN1LRlACLS 8-31 ,1992,, .>.,4.+4# ./ !""#$%%???+R*".,+.,/%#**%M4;364I6 ,on the
history o yield management in the airline industry,. Kalyan 1. 1alluri & Garrett J. an
Ryzin, 1lL 1lLOR\ AND PRAC1ICL Ol RLVLNUL MANAGLMLN1 ,Kluwer Academic
Publishers 2004,, !""#$%%S..A*+/../<(+-.1%S..A*D&'E!./.Q4\]15^_ ,a textbook
explanation o yield management,.
24
Ai Neo & Catherin \olram, S3C G) @.-6<.(/61#18 A886# ?)67)-8= D- ;*7,1,(.4 D-.4C8,8 )<
M1#.I<.8/ ?#1#.48, 22 1lL RAND J. Ol LCON. 319-339 ,2002,,
!""#$%%,(*(),-!+-!&-)/.S.."!+('>%1),A("&0/%')")S)*(*%'.1&0&-A*%'.-*%M;;M[`
!5[a.[b)0>C)-">,(,*+#'C, Igal lendel & Ai Neo, @#.861,-& /3# A*74,(./,)-8 )< ".4#8 .-B
?)-86*#1 A->#-/)1C M#3.>,)1, 4 LCONOML1RICA 163-163 ,2006,,
!""#$%%C)-><"5+?-)*+0.,"!?(*"(,0+('>%L&(!I48%1()*>,&0/+#'C.

indexes,
25
since it allows or much more direct and timely access to prices. 1he
act that the data is timely is worth emphasizing, since it allows or real time
analysis and interention or businesses and at the policy leel.
lyunyoung Choi and I hae used real-time publicly-aailable search engine data
to predict the current leel o economic actiity or automobile, real estate, retail
trade, trael, and unemployment indicators.
26
1here are many other sources o
real-time data such as credit card, package deliery, and inancial data. 1his has
been reerred to as nowcasting` to describe the use o real-time data in
estimating the current state o the economy.
2
A ariety o econometric
techniques are used to deal with the problems o ariable selection, gaps, lags,
structural changes and so on. Much o the real-time data is also aailable at
state and city leels, allowing or regional macroeconomic analysis.
In the last 20 years, the ield o machine learning has made tremendous strides
in data mining.` 1his term was once pejoratie, at least among
econometricians, but now enjoys a somewhat better reputation due to the
exciting applications deeloped by computer scientists and statisticians.
28
One
o the main problems with data mining is oer-itting, but arious sorts o
cross-alidation techniques hae been deeloped to mitigate this problem.
Lconometricians hae only begun to utilize these techniques.
29

25
ROBLR1 C. lLLNS1RA & MA11lL\ SlAPIRO, LDS., SCANNLR DA1A AND PRICL INDLXLS
,2003,, larm loundation, E))B ?5A$ 51,(#8$ .-B ;H7#-B,/61#8: D S)1I83)7 )- /3# Q8# )< "(.--#1
G./. ,- 5)4,(C D-.4C8,8, June 2003, !""#$%%???+(,*+>*')+/.2%
S,&(C&0/%_V^c..'G0'J@#(0'&">,(*%X-)00(,_.0C(,(0-(+!"1.
26
lyunyoung Choi & lal R. Varian, 51#B,(/,-& /3# 51#8#-/ 2,/3 0))&4# 91#-B8, GOOGLL RLSLARCl
BLOG, April 2, 2009, !""#$%%/../<(,(*(),-!+S<./*#."+-.1%M;;7%
;T%#,('&-"&0/:#,(*(0":?&"!:/../<(:",(0'*+!"1<, lyunyoung Choi & lal R. Varian,
51#B,(/,-& A-,/,.4 ?4.,*8 <)1 Q-#*74)C*#-/ M#-#<,/8, GOOGLL RLSLARCl BLOG, July 22, 2009,
!""#$%%/../<(,(*(),-!+S<./*#."+-.1%M;;7%;I%#.*"(':S5:!)<:2),&)0:-!&(C:
(-.0.1&*"+!"1<.
2
"## M. P. CLLMLN1S & DAVID l. lLNDR\, GRLA1 BRI1AIN S1A1IS1ICS COMMISSION,
lORLCAS1ING IN 1lL NA1IONAL ACCOUN1S A1 1lL OllICL lOR NA1IONAL S1A1IS1ICS
,2003,, Jennier L. Castle & Daid lendry, Y)2(.8/,-& <1)* G,8.&&1#&./#8 ,- /3# E.(# )< L)(./,)-
"3,</8, June 18, 2009, !""#$%%???+(-.0.1&-*+.@+)-+>A%1(1S(,*%
R(00&C(,+-)*"<(%K.?-)*";7d.c+#'C |hereinater Castle & lendry, Y)2(.8/,-&|.
28
lor a technical oeriew, 8## 1RLVOR lAS1IL, JLROML lRILDMAN, & ROBLR1 1IBSlIRANI,
1lL LLLMLN1S Ol S1A1IS1ICAL LLARNING: DA1A MINING, INlLRLNCL, AND PRLDIC1ION
,2d ed. 2009,.
29
"## Castle & lendry, Y)2(.8/,-&, 8671. note 2.

Enabling Experimentation
As Ronald Coase has said, I you torture the data long enough it will
coness.`
30
It is diicult to establish causality rom retrospectie data analysis.
It is thus noteworthy that computer mediation allows one to measure economic
actiity and also conduct controlled experiments.
In particular, it is relatiely easy to implement experiments on \eb-based
systems. Such experiments can be conducted at the query leel, user leel, or
geographic leel.
In 2008, Google ran 6,000 experiments inoling \eb search which resulted in
450-500 changes in the system.
31
Some o these experiments were with the user
interace and some were basic changes to the algorithm.
32
1he ad team at
Google ran a similar number o experiments, tweaking eerything rom the
background color o the ads, to the spacing between the ads and search results,
to the underlying ranking algorithm.
In the 1980s, Japanese manuacturers touted their kaizen` system that allowed
or continuous improement` o the production process.
33
In a well-designed
\eb-based business, there can be continuous improement o the product
itsel-the website.
Google and other search engines also oer arious experimental platorms to
adertisers and publishers such as Ad Rotation,` which rotates ad creaties
,,%#%$ the wording o the ad, among arious alternaties to choose the one that
perorms best and \ebsite Optimizer,` a system that allows websites to try
dierent designs or layouts and determine which perorms best.
Building a system that allows or experimentation is critical or uture
improement, but it is too oten let out o initial implementation. 1his is
unortunate, since it is the early ersions o a system that are oten most in need
o improement.

30
Gordon 1ullock, D ?)**#-/ )- G.-,#4 [4#,-\8 ]D 54#. /) ;()-)*,8/8 S3) E.>)1 L,+#1/C\, 2
LAS1LRN LCONOMIC JOURNAL 205 ,No. 2, Spring 2001,, .>.,4.+4# ./
!""#$%%-.<<(/(+!.<5-,.**+('>%P(VJ-%((R%G,-!&2(%e.<>1(MI%eMIKMVM;9[M;I+#'C.
31
Rob lo, 0))&4# "#.1(3 0616 ",-&3.4: S# S,44 91C P6/4.-B,83 AB#.8$ MQ"% S;;[, Oct. 2009,
!""#$%%???+S>*&0(**?((A+-.1%"!(["!,()'%
"(-!S()"%),-!&2(*%M;;7%6;%/../<([*(),-![/+!"1<.
32
AB.
33
lor more inormation on the Japanese kaizen philosophy, 8## MASAAKI IMAI, KAIZLN: 1lL
KL\ 1O JAPAN`S COMPL1I1IVL SUCCLSS ,1986,.

Cloud computing, which I will discuss later in the essay, oers a model or
sotware as serice,` which typically means sotware is hosted in a remote data
center and accessed ia a \eb interace. 1here are numerous adantages to this
architecture. It allows or controlled experiments which can, in turn, lead to
continuous improement o the system. Alternaties such as packaged sotware
make experimentation much more diicult.
Ideally, experiments lead to understanding o causal relations that can then be
modeled. In case o \eb applications there are typically two economic
agents`: the users and the applications. 1he applications are already modeled
ia the source code that is used to implement them, so all that is necessary is to
model the user behaior. 1he resulting model will oten take the orm o a
computer simulation that can be used to understand how the system works.
Some examples o this are the Bid Simulator and Bid lorecasting tools oered
by Google and \ahoo!.
34
1hese tools gie an estimate o the cost and clicks
associated with possible bids. 1he cost per click is determined by the rules o
the auction and can be calculated directly, the clicks are part o user behaior
and must be estimated with economic orecasting. Putting them together
creates a model o the auction outcomes.
How Experiments Change Business
Because computer mediation drastically reduces the cost o experimentation,
there hae been changes or the role o management. As Kohai #/ .4% hae
emphasized, decisions should be based on careully controlled experiments
rather than the lighest Paid Person`s Opinion ,liPPO,.`
35

I experiments are costly, utilizing expert opinions by management is a plausible
way to make decisions. \hen experiments are inexpensie, howeer, they are
likely to proide more reliable answers than opinion, een the opinions o
highly paid experts. lurthermore, een when experienced managers hae
better-than-aerage opinions, it is likely that there are more productie uses o
their time than to sit around a table debating which background colors will
appeal to \eb users. 1he right response rom managers to such questions
should be to run an experiment.`

34
lor more inormation on Bid Simulator and Bid lorecasting, 8## Louise Rijk, M,B ",*64./)1
DBB8 @)1# 91.-87.1#-(C /) 0))&4# DBS)1B8 M,BB,-&, IN1LRNL1 MK1G. & BUS. RLV., Aug. 10,
2009, !""#$%%???+)'21('&)#,.'>-"&.0*+-.1%0(?*<(""(,%
K\[/../<(:)'?.,'*:S&':*&1><)".,+!"1<.
35
Ron Kohai, Roger Longbotham, Dan Sommerield, & Randal M. lenne, ?)-/1)44#B
;H7#1,*#-/8 )- /3# S#+: "61>#C .-B 51.(/,(.4 06,B#, 19 DA1A MINING & KNO\LLDGL
DISCOVLR\ 140-181 ,2008, !""#$%%???+*#,&0/(,<&0A+-.1%-.0"(0"%,M81I4AII>6T4664.

Businesses hae always engaged in experimentation in one orm or another.
1he aailability o computer mediated transactions has, howeer, made these
experiments much more inexpensie and lexible than in the past.
Enabling Customization & Personalization
linally, computer mediated transactions allow or customization and
personalization o interactions by basing current transactions on earlier
transactions or other releant inormation.
Instead o a one size its all` model, the \eb oers a market o one.`
Amazon.com, or example, makes indiidual suggestions o items to purchase
based on an indiidual`s preious purchases, or on purchases o consumers like
that indiidual. 1hese suggestions can be based on recommender systems` o
arious sorts.
36

In addition to content, prices may also be personalized, leading to arious orms
o dierential pricing. 1here are certainly welare eects o such personalized
pricing. Acquisiti and Varian examine a model in which irms can condition
prices based on past history.
3
1he ability o irms to extract surplus, they
discoer, is quite limited when consumers are sophisticated. In act, irms hae
to oer enhanced serices` to justiy higher prices.
I hae preiously suggested that there is a third welare theorem` that applies
to ,admittedly extreme, cases with perect price discrimination and ree entry:
Perect price discrimination results in the optimal amount o output sold while
ree entry pushes proits to zero, conerring all beneits to consumers.
38

1he same type o personalization can occur in adertising. Search engine
adertising is inherently customized since ads are shown based on a user`s
query. Google and \ahoo! oer serices that allow users to speciy their areas
o interest and then iew ads related to those interests. It is also relatiely
common or adertisers to use arious orms o re-targeting` that allow them
to show ads based on users` preious responses to related ads.

36
Paul Resnick & lal R. Varian, '#()**#-B#1 "C8/#*8, 3 COMM`CNS Ol 1lL ASSOC. lOR
COMPU1LR MACl. 56-58 ,March 199,, !""#$%%-)-1+)-1+.,/%1)/)B&0(*%677I%9%8T94:
,(-.11(0'(,:*5*"(1*%#'C.
3
Alessandro Acquisiti & lal R. Varian, ?)-B,/,)-,-& 51,(#8 )- 561(3.8# N,8/)1C. 24 @[90% "?A%
36-381 ,2005,, !""#$%%???+*&1*+S(,A(<(5+('>%!)<%V)#(,*%#,&2)-5+#'C.
38
lal R. Varian, ?)*7#/,/,)- .-B @.1I#/ 5)2#1, ,- JOSLPl lARRLLL, CARL SlAPIRO, & lAL R.
VARIAN, LDS., 1lL LCONOMICS Ol INlORMA1ION 1LClNOLOG\: AN IN1RODUC1ION, 1-46
,Cambridge Uni. Press 2005,. lor a theoretical analysis o irst-degree price discrimination,
see Daid Ulph & Nir Vulkan, ;4#(/1)-,( ?)**#1(#$ 51,(# G,8(1,*,-./,)-$ .-B @.88 ?68/)*,8./,)-,
No. 200, !""#$%%2><A)0+?.,-+.@+)-+>A%?#:-.0"(0"%&1)/(*%-.1S&0(':#)#(,+#'C.

Transactions Among Workers
1hus ar, there has been an emphasis on transactions among buyers, sellers and
adertisers. But computers can also mediate transactions among workers. 1he
resulting improements in communication and coordination can lead to
productiity gains, as documented in the literature on the impact o computers
on productiity.
In a series o works, Paul Daid has drawn an extended analogy between the
productiity impact o electricity at the end o the nineteenth century and the
productiity impact o computing at the end o the twentieth century.
39

Originally, actories were powered by waterwheels which droe a shat and all
o the machines in the actory had to connect to this central shat. 1he
manuacturing process inoled moing the piece being assembled rom station
to station during assembly.
1he power source eoled rom waterwheels to steam engines to electric
motors. Lentually electric motors were attached to each machine, which
allowed more lexibility in how the machines were arranged within the actory.
loweer, actories still stuck to the time-honored arrangements, grouping the
same sort o machines in the same location-all the lathes in one place, saws in
another, and drills in yet another.
In the irst decade o the twentieth century, lenry lord inented the assembly
line. 1hen, the lexibility oered by electric motors became well appreciated.
40

As Daid demonstrates, the productiity impact o the assembly line was
signiicant, and oer the last century, manuacturing has become ar more
eicient.
41

39
"## Paul Daid, 93# GC-.*) .-B /3# ?)*76/#1: D- N,8/)1,(.4 5#187#(/,># )- /3# @)B#1- 51)B6(/,>,/C
5.1.B)H, 80 AM. LCON. RLV. 355-61 ,May 1990,
!""#$%%&'()*+,(#(-+.,/%)%)()%)(-,(2%28;5677;&M#944:36+!"1< |hereinater Daid,
51)B6(/,>,/C 5.1.B)H|, Paul Daid, 0#-#1.4 5617)8# ;-&,-#8$ A->#8/*#-/$ .-B 51)B6(/,>,/C 01)2/3:
E1)* /3# GC-.*) '#>)46/,)- /) /3# ?)*76/#1 '#>)46/,)-., ,- L. Deiaco, L. lornel, & G. Vickery,
eds., 1LClNOLOG\ AND INVLS1MLN1: CRUCIAL ISSULS lOR 1lL 90S ,1991,, Paul Daid,
?)*76/#1 .-B /3# GC-.*): 93# @)B#1- 51)B6(/,>,/C 5.1.B)H ,- /3# Y)/J/))JB,8/.-/ @,11)1, ,-
1LClNOLOG\ AND PRODUC1IVI1\: 1lL ClALLLNGL lOR LCONOMIC POLIC\ 315-348
,1991,.
40
lord suggests that the inspiration or the assembly line came rom obsering the
meatpacking plants in Chicago, where animal carcasses were hung on hooks and moed
down a line where workers cared o dierent pieces. I you could use this process to dis-
assemble a cow, lord igured you could use it to assemble a car. See lLNR\ lORD, M\ LIlL
AND \ORK ,Doubleday, Page & Co. 1923,.
41
I do not mean to imply that the only beneit rom electric motors came rom improed
actory layout. Motors were also more eicient than drie belts and the building
construction was simpler. See Daid, 51)B6(/,>,/C 5.1.B)H, 8671. note 39.

I want to extend Daid`s assembly line analogy to examine knowledge worker
productiity.`
42
Prior to the widespread use o the personal computer,
producing oice documents was a laborious process. A memo was dictated to a
stenographer who later typed the document, making carbon copies. 1he typed
manuscript was corrected by the author and circulated or comments. As with
pre-assembly line production, the partially-produced product was carried around
to dierent stations or modiication. \hen the comments all came back, the
document was re-typed, re-produced and re-circulated.
In the latter hal o the twentieth century, there were some productiity
enhancements or this basic process, such as \hite-Out, Post-it Notes, and
photocopy machines. Nonetheless, the basic production process remained the
same or a century.
\hen the personal computer became widespread, editing became much easier,
and the process o collaboratie document production inoled loppy disks.
1he adent o email allowed one to eliminate the loppy disk and simply mail
attachments to indiiduals.
All o these eects contributed to improing the quantity and quality o
collaboratie document production. loweer, they all mimicked the same
physical process: circulating a document to indiiduals or comments. Lditing,
ersion control, tracking changes, circulation o the documents and other tasks
remained diicult.
Nowadays, there is a new model or document production enabled by cloud
computing.`
43
In this model, documents lie in the cloud,` meaning in some
data center on the Internet. 1he documents can be accessed at any time, rom
anywhere, on any deice, and by any authorized user.
Cloud computing dramatically changes the production process or knowledge
work. 1here is now a single master copy that can be iewed and edited by all
releant parties, with ersion control, check points and document restore built
in. All sorts o collaboration, including collaboration across time and space,
hae become ar easier.

42
See Peter l. Drucker, [-)24#B&#J2)1I#1 51)B6(/,>,/C: 93# M,&&#8/ ?3.44#-&#, 41 CAL. MGM1. RLV.
9-94 ,1999,.
43
Michael Armbrust, Armando lox, Rean Griith, Anthony D. Joseph, Randy l. Katz,
Andrew Konwinski, Gunho Lee, Daid A. Patterson, Ariel Rabkin, Ion Stoica, & Matei
Zaharia, D+)># /3# ?4)6B8: D M#1I#4#C F,#2 )< ?4)6B ?)*76/,-&, leb. 10, 2009,
!""#$%%???+((-*+S(,A(<(5+('>%V>S*%Y(-!P#"*%M;;7%JJ_X:M;;7:M8+!"1<
|hereinater D+)># /3# ?4)6B8^. "## .48), \ikipedia, ?4)6B ?)*76/,-&,
!""#$%%(0+?&A&#('&)+.,/%?&A&%_<.>'[-.1#>"&0/.

Instead o passing the document amongst collaborators, a single master copy o
the document can be edited by all interested parties ,simultaneously i desired,.
By allowing worklow to be re-organized, cloud computing changes knowledge
worker productiity the same way that electricity changed the productiity o
physical labor.
Enabling Deployment of Applications
As preiously mentioned, cloud computing oers what is reerred to as
sotware as a serice.` 1his architecture reduces support costs and makes it
easier to update and improe applications.
Cloud computing, howeer, does not only oer sotware as a serice.` It also
oers platorm as a serice,` which means that sotware deelopers can deploy
new applications using the cloud inrastructure.
Nowadays, it is possible or a small company to purchase data storage, hosting
serices, an application deelopment enironment, and Internet connectiity
o the shel` rom endors such as Amazon.com, Google, IBM, Microsot,
and Sun.
1he platorm as a serice` model turns a ixed cost or small \eb applications
into a ariable cost, dramatically reducing entry costs. Computer engineers can
both explore the combinatorial possibilities o generic components to create
new inentions and can actually purchase standardized serices in the market in
order to deploy those innoations.
1his deelopment is analogous to the recent history o the book publishing
industry. At one time, publishers owned acilities or printing and binding
books. 1oday, due to the strong economies o scale inherent in this process,
most publishers hae outsourced the actual production process to a ew
specialized book production acilities.
Similarly, in the uture, it is likely that there will be a number o cloud
computing endors that will oer computing on a utility-based model. 1his
production model dramatically reduces the entry costs o oering online
serices, and will likely lead to a signiicant increase in businesses that proide
such specialized serices.
44

1he hallmarks o modern manuacturing are routinization, modularization,
standardization, continuous production, and miniaturization. 1hese practices
hae had a dramatic impact on manuacturing productiity in the twentieth

44
D+)># /3# ?4)6B8, 8671. note 43.

century. 1he same practices can be applied to knowledge work in the twenty-
irst century.
Computers, or example, can automate routine tasks such as spell-checking and
data retrieal. Communications technology allows tasks to be modularized and
routed to the workers best able to perorm those tasks. Similar to how the
miniaturization o the electric motor allowed physical production to be
rearranged in 1910, the miniaturization o the computer-rom the mainrame,
to the workstation, to the PC, to the laptop, and to the mobile phone-allows
knowledge production to be rearranged on a local and global scale.
Enabling Micro-Multinationals
An interesting implication o computer mediated transactions among
knowledge workers is that interactions are no longer constrained by time or
distance.
Lmail and other tools allow or asynchronous communication oer any
distance, which allows or optimization o tasks on a global basis. Knowledge
work can be subdiided into tasks, much like physical work in Adam Smith`s
hypothetical pin actory.
45
But een more, those tasks can be exported around
the world to where they can most eectiely be perormed.
lor example, consultants at McKinsey routinely send their PowerPoint slides to
Bangalore or beautiication. 1here are many other cognitie tasks o this sort
that can be outsourced, including translation, prooreading, document research,
#/(% Amazon.com`s Mechanical 1urk is an intriguing example o how computers
can aid in matching up workers and tasks.
46
As o March 200, there were
reportedly more than 100,000 workers rom 100 countries who were proiding
serices ia the Mechanical 1urk.
4

1he dramatic drop in communications costs in the last decade has led to the
emergence o what I hae termed micro-multinationals.`
48
Nowadays, a 10- or
12-person company can hae communications capabilities that only the largest

45
ADAM SMI1l, AN INQUIR\ IN1O 1lL NA1URL AND CAUSLS Ol 1lL \LAL1l Ol NA1IONS 18-
21 ,Ldwin Cannan, ed., Methuen & Co., Ltd. 1904, ,16,,
!""#$%%???+(-.0<&S+.,/%<&S,),5%X1&"!%*1`K+!"1<.
46
\ikipedia. D*.R)- @#(3.-,(.4 961I,
!""#$%%(0+?&A&#('&)+.,/%?&A&%G1)B.0[b(-!)0&-)<[Y>,A.
4
Jason Pontin, D1/,<,(,.4 A-/#44,&#-(#$ 2,/3 N#47 E1)* /3# N6*.-8, N.\. 1IMLS, March 25 200,
???+05"&1(*+-.1%M;;I%;9%M4%S>*&0(**%5.>,1.0(5%M4X",()1+!"1<
D(@E699MTI4M;;(0E-'6-(4';S((3TI'4(&E4;88#),"0(,E,**05"(1-E,**.
48
lal Varian, 9#(3-)4)&C L#>#48 /3# M68,-#88 54.C,-& E,#4B, N.\. 1IMLS, Aug. 25 2005,
!""#$%%???+05"&1(*+-.1%M;;4%;8%M4%S>*&0(**%M4*-(0(+!"1<.

multinationals could aord 15 years ago. Using tools like email, websites, wikis,
oice oer IP, and ideo conerencing, tiny companies can coordinate worklow
on a global basis. By sending work rom one time zone to the next, these
companies eectiely work around the clock, giing them a potential
competitie adantage oer irms that are restricted to one time zone.
Many micro-multinationals share a common history: A student comes to the
United States or graduate school. 1hey use the Internet and the collaboratie
tools aailable in scientiic workgroups. Some get bitten by the start-up bug.
1hey draw on their riends and colleagues back home, who hae other contacts
liing abroad. 1he collaboratie technologies preiously mentioned allow such
loose groups to collaborate on producing computer code, which may end up as
a working product.
As Saxenian has pointed out, emigration` means something quite dierent
now than it did 30 years ago.
49
As she puts it, a brain drain` has been replaced
by a brain circulation.` \e now hae a host o collaboratie technologies that
allow an immigrant to maintain ties to his social and proessional networks in
his home country.
Conclusion
I began this essay with a discussion o combinatorial innoation and pointed
out that innoation has been so rapid in the last decade because innoators
around the world can work in parallel, exploring noel combinations o
sotware components. \hen the innoations are suiciently mature to be
deployed, they can be hosted using cloud computing technology and managed
by global teams, and een by tiny companies. Ideally, these new serices can
sere as building blocks or new sorts o combinatorial innoation in business
processes that will oer a huge boost to knowledge worker productiity in the
uture.

49
ANNALLL SAXLNIAN, 1lL NL\ ARGONAU1S: RLGIONAL ADVAN1AGL IN A GLOBAL
LCONOM\ ,2006,.


Decentralization, Freedom to
Operate & Human Sociality
By Yochai Benkler
*

Three Stories of Innovation in the
Networked Information Economy
In 1994, two groups o sotware engineers were working on the next generation
o critical sotware: a \eb serer, the sotware that a website runs to respond to
requests rom users. One group was within Microsot, understanding that the
next generation o critical inrastructure would be the \eb, and trying to extend
its market rom the operating system to the \eb serer. 1he other was a group
o deelopers led by Brian Behlendor, ormerly rom the group o academic
computing engineers rom the Uniersity o Illinois in Urbana-Champagne,
who were patching up the serer deeloped in tandem with the deelopment o
Mosaic, the irst graphical interace to access the \eb, at Urbana-Champagne.
1hey called it a patchy serer, which became the name o the resulting open
source project: Apache serer. Anyone who would hae predicted that the
system, built by a scrappy set o deelopers, who adopted a licensing approach
that asserted no exclusie rights oer their output, and were working in an area
considered strategically critical by the largest company in the ield and that was
deeloping a product in direct competition, would win would hae been
laughed out o the room. And yet, it moes ,as Galileo amously said,
deending his theories o the Larth`s orbit around the Sun,. Oer 15 years,
through two boom and bust cycles, Apache has held 50-60 o the market
share in \eb serers ,in the summer o 2010, it was about 55,, while
Microsot`s \eb serer market share has hoered between 25 and 35
,about 25 in summer o 2010,.
In 1999, two o the most insightul economists looking at the new rules or the
inormation economy opened their book with an analysis o how Microsot`s
moe into the market in encyclopedias embodied the new challenges created by
the digital economy. In lebruary o 2001, the deeloper o one o seeral
ongoing eorts to deelop an online encyclopedia hal gae up and dumped
about 900 stubs onto an open source platorm, under a license that let anyone
edit it and gae no one power to eto. 1his made participation easy, but control
relatiely hard. And no one was paid to write or edit the encyclopedia. It was
probably the ugliest technical system or encyclopedia deelopment being
experimented with at the time. lie years later, this ugly duckling would be

f.-!)& N(0A<(, is the Berkman Proessor o Lntrepreneurial Legal Studies at larard, and
aculty co-director o the Berkman Center or Internet and Society.

identiied by a study done by the sta at Y./61# as haing roughly similar error
rates as Britannica or science articles. By 2009, Microsot`s Lncarta
encyclopedia product was discontinued. \ikipedia has come to embody the
undamental changes we hae to deal with when trying to understand the
networked inormation economy.
In 2001 a Swedish and Danish entrepreneur inested in sotware deeloped by
three Lstonian programmers and released a brilliant new solution or peer-to-
peer ile sharing: Kazaa. 1hanks to the act that the irm was based in the
Netherlands, where Dutch law proided it greater immunity rom suit by record
labels, Kazaa quickly became a major platorm ater the demise o Napster.
1
By
2003, the same group o entrepreneurs and programmers had launched a peer-
to-peer oice telephony application built on the same basic architecture as
Kazaa: Skype. 1heoretically, Skype should not hae worked. lor close to two
decades, the Internet Protocol`s irst-come, irst-sered,` treat-all-packets-on-
a-best-eorts-basis approach was thought to preent serious oice oer Internet
applications rom working well. And yet, here was this small company
proiding better quality, encrypted, end-to-end communications, using the
users` own computers and connections as its basic inrastructure. 1hey did not
need to control the low o packets in the network to proide Quality-o-Serice
assurances. 1hey just proided serice o a quality that was good enough or
the price: ree or calls rom one Skype user to another, soon ollowed by ery
low rates or calls to regular phones. In 2005, eBay bought Skype or oer >2.5
billion.
Radical Decentralization of
Physical, Human ! Social Capital
1he three stories aboe outline the basic transormatie elements o the
networked inormation economy. \e hae seen a radical decentralization o
the most important orms o capital in the most adanced sectors o the
economy: physical, human, and social capital. lor the irst time since the
Industrial Reolution, the most important inputs into the core economic
actiities o the most adanced economies are widely distributed in the
population. 1echnologically, the change begins with physical capital:
Processing, storage, communications, and sensing hardware hae come to be
deeloped in packages o suiciently low cost to be put in serice by indiiduals
or their own personal use. 1hese adances are capable o mixing consumer use
with production actiities. 1he rapid increase in physical capabilities
emphasizes continuous rapid innoation as a core dimension o growth and
welare which, in turn, emphasizes human capital.

1
Napster itsel was a college dorm room experiment, one o many that lourished at that
time, which dramatically and permanently changed the landscape o the music industry.

luman capital too is, by nature, widely distributed in the population, and is
extremely sticky and hard to aggregate or transer eectiely rom one
indiidual to another. \hile we measure education when we try to quantiy
human capital, that is ar rom all o what human capital really entails. Certainly
it does partly entails acquired, codiied knowledge o the kind we get in
education, but that is only one part o it. Creatiity, insight, experience-all
these go into answering the critical question: \ill this indiidual come up with
an idea, and een more importantly, will this interaction and conersation
among a gien set o indiiduals result in an interesting set o ideas emerging
Organizationally, the increased emphasis on interactions among human beings
responding to surprising new opportunities has increased the importance o
loosely-coupled interactions beyond slower-moing group boundaries like
irms. 1hese new organizational rameworks and the cooperatie dynamics they
require depend on lightweight, lexible mechanisms that we all carry or
interacting with other people. 1hat is, they depend on human sociality. 1his
basic set o protocols or non-destructie human interaction is also
undamentally and widely shared in the population. 1hey are not locked up in
the cabinets o smart corporate lawyers` incorporation orms or major deal
documents. 1hey are the core social and psychological eatures o human
beings that hae co-eoled, physically and culturally, to allow us to be the
kinds o social creatures we in act are-warts and all.
1his distributed network o human beings, possessing the physical, human, and
social capital that they do, are now connected in a global network o
communications and exchange that allows much greater low and conersation,
so that many new connections are possible on scales neer beore seen.
1ogether, these mean that conersations and new ideas-but more importantly,
pilots, experiments, and toy implementations o these new ideas-are cheap and
widespread, and innoation happens eerywhere, all the time, at low cost. 1he
ast majority o ideas go nowhere, just as the ast majority o experiments ail.
But the sheer scale o experimentation has meant that the network has reliably
proided the low o innoation that we hae come to expect and depend on,
and it has largely come rom unpredictable corners rather than rom yesterday`s
innoators or the preious decades` large irms.
As a result o these basic dynamics, in the networked inormation economy,
experimentation, continuous learning and improement, low-cost prototyping,
deployment, iteration, and adoption are more important than well-behaed
innoation inestments. Social behaior plays a much larger productie
economic role than it could when physical capital requirements meant that,
howeer good an idea someone had, transitioning it to a platorm that could
actually be adopted by consumers,users was simply too expensie to do except
through a system o contracts and inestment-through a more-or-less ormal
corporate model. In the networked inormation economy, <1##B)* /) )7#1./# ,8

*)1# ,*7)1/.-/ /3.- 7)2#1 /) .771)71,./#, and >)46-/.1,8* .-B 8)(,.4,/C .1# *)1# ,*7)1/.-/
/3.- <)1*.4 ()-/1.(/$ .-B 74.C .- ,*7)1/.-/ 1)4# .4)-&8,B# ()17)1./# )1&.-,R./,)-.
Freedom to Operate is More
Important than Power to Appropriate
1he story o open source sotware is the core story about the importance o
reedom to operate and loosely-coupled association that transcends contract
and corporate structure. \hat makes a sotware deelopment project open
source` is that the output o the deelopment actiity, the code, is released
under a copyright license that allows anyone to look at, modiy, and redistribute
the code and modiications to it. 1his means that anyone, anywhere, can come
to the state o the art in the code, adopt it, adapt it, and release it, building on
the innoatie contributions o others with complete reedom to operate with
and on it. In a networked enironment where human capital resides in many
places, and where it is impossible or anyone irm to hire all the smartest people
,or more to the point, to hire all the people who are likely to hae the most
releant and powerul insights or any new challenge,, a system that depends on
open access to the unierse o aailable resources, projects, and collaborators
on them will outperorm a system that only allows people who hae already
been identiied, recruited, and contracted with based on past projections o
what would be important or working on a new problem.
1he licensing aspect o open source sotware raises another important aspect o
change. listorically, assuring the owner o inancial capital o the soundness o
an entrepreneur was the critical actor. 1o do so, it was necessary to possess
property in core inputs, and a network o contracts or lows o what could not
reliably or eiciently be owned, like supply relations. 1oday, assuring a steady
and reliable low o complementary contributions rom other deelopers is as
important to maintain a high rate o innoation, experimentation, and
adaptation as securing the complementary inancial inputs. At the early stages,
complementary contributions rom other deelopers are *)1# important than
inancial inputs.
\ith the rise o peer production, radically distributed collaboratie production
on the open source model, adoption o licensing terms like those o ree and
open source sotware, or Creatie Commons, becomes an important aenue to
secure those complimentary human inestments in the project. \here it is
impossible to assure that you will always employ the right people, open source
licensing has become an increasingly common strategy or entrepreneurs and
large irms alike to improe the probability that they will be able to attract the
complimentary rapid deelopment contributions they need, in the time rame
they need it, on currently-unpredicted challenges to assure high-elocity
innoation. lormalized reedom to operate, in the orm o open source
licensing, is coupled with a strong pre-commitment by the irms that undertake

the limitations it imposes on their power to appropriate, so as to assure
potential collaborators against deection with regard to the ruits o the
common, irm-boundary-crossing enterprise. Increasingly this is also becoming
a way or irms that compete in some domains, such as sotware serices, to
engage in pre-competitie cooperation on the deelopment o core necessary
tools, like the Linux kernel and operating system, the Apache \eb serer, #/(%
Voluntarism & Sociality Become
More Important than Formal Contract
Another important characteristic o the networked inormation economy is the
critical role o knowledge and creatiity. 1hese require uniquely human inputs,
and are persistently uncontractible. 1hat is, you can neither deine or explicit
codiication nor characterize or monitoring oer time, what it means to be
creatie, or insightul, or usably knowledgeable in context where an innoation
challenge occurs. As a result, tacit knowledge and insight are necessarily and
always imperectly deined or, or monitored through, contract. 1o assure the
right motiations and orientation towards inding new solutions to challenges, it
is necessary or an economy at large, as it is or any gien organization, to
harness the non-contractible motiations o indiiduals to the knowledge and
innoation task at hand. 1his is not new, in the sense that literature on high-
commitment, high-perormance organizations has been around or decades, and
management theory keeps lowing back and orth between periods that
emphasize explicit material rewards and monitoring to control employees
shirking their responsibilities, and periods where the limitations o those
approaches become clearer, and the beneits o models that depend on a more
holistic, human iew o what is required to create a motiated workorce
preail.
In the networked inormation economy, where so much o what needs to be
done is uncontractible and so many o those who need to be engaged are not
een in a position to hae a contractual relationship, the role o sociality and
cooperatie human systems designs that aim to engage, and depend on social,
moral, and emotional motiations as well as, and oten instead o, material
motiations, has become much larger. \ikipedia, in my three stories, stands as
the ultimate example o a system that critically depends on these non-material
motiational ectors, mediated through a technical-social platorm that is
optimized to engage these motiations and allow people to cooperate oer a
system that proides great reedom to operate, no power to appropriate, and
tremendous room or social organization and interaction ,which all hae their
own warts and bumps,.
Rather than the traditional ormal modes o organization-be they a ormal
corporation based on contracts, or ormal, stable associations on the model o
rotary clubs or unions-the new orms o social networks ,not the

lacebook,MySpace-type websites, but the actual social phenomenon, permit
people to hae more loosely-coupled social associations, in which they can
participate or some o the time, and combine their inestments with many
others who are similarly loosely-tied to each other, and may spend their time at
dierent rates, and in dierent enterprises, during the course o their day, week,
or year. 1ogether, these new orms o loose association, based on social signals
rather than price signals or a ormal corporate managerial hierarchy, orm what
I hae called peer production. 1hey are not, by any stretch o the imagination,
going to replace all production actiities built on more ormal, structured
models. Anyone who claims that the argument is one o replacement
misunderstands the claim.
1he new models o production do, howeer, come to play a signiicant
productie role in an enironment that continues and will continue to be
occupied by more traditional orms. 1hey create new sources o competition-
as in the case o \ikipedia displacing Lncarta-and new orms o
complementary sources o innoation and other inputs-as in the case o open
source sotware and the sotware serices industry. 1hey do not herald the
death o traditional market,irm-based production. 1o argue otherwise would
be silly. But it would be equally silly to simply assume away a major new
organizational innoation. 5##1 71)B6(/,)- .-B ())7#1./,># 36*.- 8C8/#*8 .1# . -#2 2.C
/) 3.1-#88 . 4./#-/ +6/ *.88,>#4C 71)B6(/,># <)1(#% 93#C *.I# /3# 4,-# +#/2##- 71)B6(/,)- .-B
()-86*7/,)- <6RRC$ .-B )<<#1 -#2 7./32.C8 /) 3.1-#88 /3# /,*#$ ,-8,&3/$ #H7#1,#-(#$ 2,8B)*
.-B (1#./,>,/C )< 36-B1#B8 )< *,44,)-8 )< 7#)74# .1)6-B /3# 2)14B /) 7#1<)1* /.8I8 /3./$ 6-/,4 .
B#(.B# .&)$ 2# )-4C I-#2 3)2 /) 7#1<)1* /31)6&3 <)1*.4 *)B#48 )< #*74)C*#-/ .-B ()-/1.(/.
1hey are an organizational innoation that anyone ignores at their peril. Just
ask the Departments o Deense or State how they eel about \ikiLeaks.
Innovation Anywhere !
Everywhere Over an Open Network
1he story o Skype rounds out the core changes that the networked inormation
economy presents. In the mid-twentieth-century, the epitome o innoation
was Bell Labs. \ith enough Nobel laureates to make the most ambitious
academic physics departments green with eny and massie inestment rom
monopoly proits, Bell Labs is where we got the transistor on which the entire
inormation economy is built. It is, indeed, where we got inormation theory
itsel. 1he Bell system also epitomizes the organizational model o the mid-
twentieth-century. One System, One Policy, Uniersal Serice` was how the
company`s legendary President, 1heodore Vail, put it 100 years ago.
2
According
to this model o thought, i the Internet was eer to carry that most delay-

2
A1&1, @,4#8/)-#8 ,- D9_9 N,8/)1C, !""#$%%???+-.,#+)""+-.1%!&*".,5%1&<(*".0(*+!"1<
,last accessed Aug. 1, 2010,.

sensitie o all serices, oice, we would hae to change how we manage
packets. Best eort deliery
3
just wouldn`t do it. Someone needed to manage
the network and decide-this packet, which carries oice, is more latency
sensitie, while that packet, which carries email or a \eb page, can wait. But, as
it turns out, this persistent prediction was alse. And the people who proed it
alse were not working or Bell Labs. Indeed, it was probably impossible or
anyone inside one o the current incarnations o the Bell system to hae done
so. It was, instead, let to three Lstonian deelopers and a couple o Dutch and
Danish edgy entrepreneurs to do so. 1hey were not the only ones to try.
Others did too-Vocal1ec in Israel was among the irst, but they were too
early.
1he point is that in a global networked inormation enironment, innoation
can come rom anywhere, insights o arious orms can ind each other, and
experimentation and implementation are cheap to do rom anywhere to
anywhere else. Massie experimentation is ollowed by massie ailures. But
the ailures are generally cheap, at least by societal standards. And the successes
can be readily disseminated, adopted, and generalized on a major global scale in
ery short time rames. Variation, selection, adaptation and surial,replication
through user adoption, rather than planning and high inestment, hae
repeatedly oered the more robust approach in this new complex and chaotic
enironment. Rapid, low cost experimentation and adaptation on a mass scale,
underwritten by the ease o cheap, ast implementation and prototyping, and
cheap widespread ailure punctuated by a steady low o unpredictable successes
hae been more important to innoation and growth in the networked economy
than models o innoation based on higher-cost, more managed innoation
aimed at planning or predictable, well-understood returns.
Implications for Human Systems Design
\e lie our lies through systems: organizational systems, like corporations,
states, or nonproits, technical systems, like the interstate highway system or the
Internet, institutional systems like law, both public and priate, or social
conentions, and cultural, as in our belie systems or how we know things to
be true, such as religion, or science. 1o a great extent, these systems are too
complex or us to construct deterministic, ully understood interentions that
will clearly lead to desired outcomes, along whateer dimension we think is
important: eiciency, reedom, security, or justice. But we nonetheless apply
ourseles to the task. \e try to use management science to design better
organizational strategies, we try to use law to reine and improe our legal

3
Best eort deliery describes a network serice in which the network does not proide any
guarantees that data is deliered or that a user is gien a guaranteed quality o serice leel or
a certain priority.` \ikipedia, M#8/ #<<)1/ B#4,>#1C,
!""#$%%(0+?&A&#('&)+.,/%?&A&%N(*"[(CC.,"['(<&2(,5 ,last accessed Aug. 1, 2010,.

system, we inest enormous amounts in designing better technical systems, and
so orth.
1he characteristics o the networked inormation economy require that in our
eorts at systems design we emphasize openness and reedom to operate oer
control and power to appropriate and that we emphasize human sociality and
dierse motiations or dierse types oer optimizing or material interests and
letting eerything else sort itsel out. At a practical leel, technical open design
has made the largest and most powerul steps. Anchored in the ery decision to
separate 1CP rom IP, and make the core Internet protocol as open as it can be,
and continuing to the central role that open standards hae played in the
deelopment o the \eb, XML, and \ili, to name just a ew, a continuous
emphasis on openness already has substantial support and inertia, although it is
always under pressure rom irms that think they can get an edge by owning a
de-acto standard, or controlling a technical choke point that would allow them
to extract rents. In management science, we are seeing, slowly and in some
senses at the periphery, eorts to learn the lessons o open source sotware and
apply them to collaboration across irm boundaries and strategic management
o the knowledge ecology that a irm occupies.
In law, the most important battleground in the tension between the control-
oriented approach and the reedom-to-operate approach is intellectual property.
Only this year Amazon receied a patent or social networking
4
that reads
more-or-less like a description o lacebook, launched our years beore
Amazon had een iled its patent application. But not eerything is so silly.
1his summer, the Librarian o Congress exempted iPhone jailbreaking rom the
Digital Millennium Copyright Act`s anti-circumention proisions.
5
I there is
any single policy domain in which it is important to apply what we hae learned
about the new networked inormation economy, it is in the area o intellectual
property. It is also the area where there is the largest potential or intellectual
and political programmatic oerlap between libertarians and progressies.

4
Stan Schroeder, D*.R)- 5./#-/8 ")(,.4 Y#/2)1I,-& "C8/#*$ S,-I8 ./ E.(#+))I, MASlABLL,1LCl,
June 1, 2010, !""#$%%1)*!)S<(+-.1%M;6;%;3%6I%)1)B.0:#)"(0"*:*.-&)<:0("?.,A&0/:
*5*"(1%.
5
Copyright Oice, Rulemaking on Lxemptions rom Prohibition on Circumention o
1echnological Measures that Control Access to Copyrighted \orks, July, 28, 2010,
!""#$%%???+-.#5,&/!"+/.2%6M;6%.

A Common Agenda on
Intellectual Property for the
Networked Information Economy
lrom the perspectie o economic analysis, inormation is a public good. Once
someone creates new inormation or knowledge, anyone can use it without
reducing its aailability or anyone else. Its marginal cost is thereore zero, and
that is its eicient price. loweer, or inormation to be aailable at a price o
zero, the person who produced it must ind some other mechanism to extract
alue rom their inestment in creating the inormation. Otherwise, haing
inormation aailable at its marginal cost today ,zero, will lead to less
production tomorrow.
1he oerwhelming majority o inormation, knowledge, and culture is produced
without the need to rely on explicit, intellectual-property-based mechanisms to
appropriate its beneits. lirms continuously innoate in their processes so as to
lower their costs and improe their proits, but they do generally not patent
their innoations and license them or exclude competitors rom using them.
Indiiduals innoate and deelop experience about their workplace to improe
their own perormance, people read news and create commentary or each
other, and appropriate the beneits o what they ind socially. Goernments
inest in R&D and reap the beneits through higher growth, greater military
might, #/(% Nonproits and academic institutions inest in inormation,
knowledge, and cultural production, and so orth. All these approaches hae
their own adantages and disadantages, but economic surey ater surey or
the past ew decades has shown that een in industrial innoation, a minority o
sectors relies on patents, and the majority relies on a range o supply-side and
demand-side improements in appropriability that come rom deeloping the
inormation and either using it without exchanging it or disseminating it and
relying on irst-moer adantages, network eects, marketing and reputational
beneits, #/(%
1he only industries that are still dependent on intellectual property protections
are the pharmaceutical industry or patents and lollywood, the recording
industry, and much o book publishing or copyright. Len newspapers and
magazines are not so dependent on IP. 1hey are, rather, adertising-supported
media. 1hey depend on release o the inormation to capture demand-side
beneits or their paying clients in a two-sided market-the adertisers.
1he reason that it is important to remember this quick recap o innoation and
Patents & Copyrights Lconomics 101 is that it helps us to see that patents and
copyrights represent a goernment decision to prohibit eeryone rom using
ideas or inormation that they can practically use, in order to sere a public
purpose-supporting a subset o business models or the creation o new
inormation, knowledge, and culture. Now, it is perectly acceptable or

goernment to prohibit some actions in order to sere the public good. \e
preent companies rom selling ood unless it is labeled in certain ways to sere
public health, we prohibit iolence to increase public security, and so orth. But
we try to do so only when there is indeed a good reason.
Sometimes, we combine prohibitions with a market in permissions. 1radeable
emissions permits are a classic example where we think it is more eicient to
allow irms to trade in their permissions than to simply hae direct regulation.
Patents and copyrights are exactly like tradeable emissions permits. 1hey are a
market-based approach toward the regulatory problem o how to preent
people who want to use the existing unierse o inormation and knowledge
that they possess in ways that will undermine uture knowledge production and
innoation. \e prohibit eeryone rom using certain classes o inormation and
knowledge, and we create a market in permissions to use that inormation. \e
call these permissions copyrights` or patents.` \hat is important to
remember is that these permissions markets create a drag on reedom to operate
on current innoation and knowledge creation, and they create a drag on
innoation in all industries that, unlike pharmaceutical or blockbuster moie
markets, do not heaily depend on such permissions markets.
lor progressies, the best way to understand patents and copyrights is through
the prism o ree speech: 1hese are goernment regulations on what and how
we can say things, and how we can use what we know, that are implemented in
pursuit o legitimate goernment ends-aiding innoation and creatie
expression by some industries-at the expense o that reedom. As with any
limitation on speech and learning, it has to be supported by ery good reasons.
It is not at all clear whether our contemporary economic understanding o the
unctioning o copyright law in particular, and patent law to a lesser extent,
proide suicient support or such signiicant restrictions on ree speech.
lor libertarians, the best way to understand patents and copyrights is as a
regulatory system that imposes limitations on how indiiduals can act on
knowledge they possess in pursuit o their own goals. It is a regulatory system
that creates and allocates permissions to generate market-based transer
mechanisms, but a regulatory system in pursuit o a goernment program,
which embodies the judgment that certain business models used to sustain
innoation and expression are more eectie than others, and supports those
deemed more eectie ./ /3# #H7#-8# )< /3)8# )/3#1 .771).(3#8.
Both approaches should lead to a signiicant downward reision in the leel o
acceptable intellectual property enorcement that the United States pursues. Let
me oer one example, which proides the basic structure o the problem: low
long should a copyright last I we thought that copyrights were really property,
the answer would be something like oreer. 1he U.S. Constitution, as well as
the laws o practically eery other country, instead limit the term o copyright,

understanding that there is a big dierence between the need o exclusiity in a
thing that, i one person uses, another cannot, and exclusiity in an idea or
expression that anyone can use without making it any less aailable or anyone
else. 1he ormer is a proper object o property. 1he latter is a proper object o
regulation o indiidual reedom or so using the thing, but only to the extent
justiied by property.
So how long should copyright terms be Let`s try this thought experiment:
Imagine that you are someone with an idea or a moie. \ou walk in to a group
o hypothetical inestors and you tell them: lere`s my idea, here`s the audience
or it, and so here is my projection or how much money we will make on it.`
1he inestors ask you: \hat are your assumptions about timing By when will
we see our return` Now, imagine that you answered: \e won`t really break
een in the irst seenty years, but just you wait until years seenty to ninety-
ie: \e`ll be making millions!` \ou would be laughed out o the room. OK,
let`s try it with not making money the irst twenty years, but making a killing in
the years twenty to thirty.` \ou get the point.
I copyright is intended to assure that there is enough appropriability to attract
inestment in creating a new expression, but it is a regulatory orm that restrains
the reedom o others to operate in pursuit o that goal, then its term should be
keyed to the term necessary to attract inestors. Gien today`s discount rates in
the releant industries-that is, how quickly inestors need to turn a proit
beore they will decide to put their money in some other enterprise-that likely
means 18 months, maybe it means three to ie years. It is possible that
dierent industries hae dierent leels o patience. But undamentally, the
oerwhelming majority o the social cost created by the 95-year term o
copyright-let alone the repeated practice o retroactie extension o copyright
or works already created in response to the then-existing incenties-system-is
incurred without any beneit or inestment purposes. No sane inestor today
cares about returns on an inestment in these kinds o ields ,as opposed to,
say, power plants or utilities, that are ten years out. lor sotware, maybe the
correct period is 18 months, or noels, maybe 10 years, although een there,
the releant party is the publisher`s decision to publish, not the author`s
decision to write-because copyright-based monetization runs through the
publisher`s business decision, not the author`s. In patents, maybe the correct
period is 20 years or pharmaceuticals. Maybe more, or maybe less. But the
principle or all these is the same: 1he period o copyright or patent protection
should be backed out o reasonable inestment assumptions and discount rates,
not pulled out o the lobbying process, which is always skewed in aor o the
small number o irms that possess these rights and against the millions o
potential innoators who do not yet know that this or that piece o regulated
access to inormation will get in their way ie years rom now.

1he details o what might go on a major intellectual property reorm that should
be supported by both libertarians and progressies may dier among
commentators. 1he core structure o the reasons or change are the same: ,a,
Strong patents and copyrights beneit some business models oer others, and in
particular place a strong drag on the radically distributed, chaotic, innoation-
eerywhere-by-eeryone model o the networked inormation economy in aor
o twentieth-century models o much more stable and controlled markets like
those o lollywood and the recording industry, ,b, 1here is a big dierence
between the leel o exclusiity needed to attract inestment at the margin, and
the leel o exclusiity that maximizes its owner`s ability to extract rents, the size
o the dierence between the minimal necessary to attract innoation and the
rent-maximizing leel o protection is equal to the amount the incumbents are
willing to spend on lobbying to keep the line at the maximal point, as opposed
to the minimally-necessary point, and ,c, 1he lines in act should be drawn
where the marginal eect is to attract inestment, not where rents can be
maximized. 1he academic community has spent years trying to reine a set o
interentions that could improe access to inormation, knowledge and culture,
while haing minimal impact on incenties to inest. 1he ollowing represent
some o the most promising o these ideas.
_.#5,&/!" "(,1$ Copyright terms should be keyed to actual market
requirements and the discount rate in the business. Copyright that is
any longer than necessary to attract the marginal inestor that makes a
dierence between the project happening or not represents pure rent
extraction and is a drag on innoation and creatiity.
P(0(?)< .C (@&*"&0/ -.#5,&/!"*$ 1here are mountains o existing
materials ,animal shots rom documentaries rom the 1960s, explosions
and action shots rom 190s B moies, #/(%, that could proide the grist
or new models o creatie mashup tools and sites, but instead sit
unused and unusable because the rights are excessiely tied up.
Lxisting copyrights should be required to be renewed periodically,
initially or a nominal price, and later on in the lie o a copyright or
escalating ees, rising to a leel no greater than necessary to make a
copyright owner think: is their .-C real market or this thing, rather than
ocing holders to make ine distinctions about the alue o the work, on
one hand, or simply automatically renewing eerything, whether or not
it has any market, because it`s cheaper to renew than to reiew
continued iability. 1hose works that continue to be o een small
commercial alue will be renewed. 1hose that continue to be o
emotional signiicance will be renewed. All others will become reely
usable upon ailure to re-register.

P(&0*")"( "!( X.05 '.-",&0(
6
S5 <(/&*<)"&2( ,(2(,*)< .C "!( X>#,(1(
_.>,"g* Z,.A*"(, '(-&*&.0
- In the midst o the panic oer peer-to-

peer ilesharing, the Supreme Court moed away rom its long-standing
precedent that an innoator cannot be orced to oresee and preent
the potentially-inringing uses o its new product ,in the ")-C case, the
VCR,. As long as there are substantial noninringing uses, innoators
are immune to suit by copyright owners whose works are being
inringed by users o the innoator`s product. In 01)I8/#1 the Supreme
Court created a more intention-based, act-intensie inquiry that
imposes greater litigation risk on entrepreneurs who innoate on the
Net with anything that can possibly be used to inringe existing
copyrights. 1his is an unnecessary drag on Internet innoation and
entrepreneurship in aor o the moie and recording industries.
J<&1&0)"( S>*&0(** 1("!.'* #)"(0"*$ lew innoations are as
unnecessary as a law intended to gie business people an incentie to
improe their business model. 1he incentie to deelop a new business
model is that it makes more money or its inentor. 1here is no need
or an additional goernment-granted monopoly on doing business in
this way. 1he lederal Circuit, which created this new doctrine 12 years
ago, tried to walk it back in the M,48I, case,
8
but the Supreme Court
recently held
9
that the particular way that the ederal Circuit went about
doing so was indeensible. Nonetheless, it appears that a majority o
the Supreme Court would support some other, better-reasoned
reersal.
J<&1&0)"( *.C"?),( #)"(0"*$ 1here is airly signiicant eidence that
sotware patents are unnecessary, and that sotware deelopment is
heaily based on serice models, time to market, network eects,
customer habits, #/(% On the other hand, patents get in the way o open
source deelopment, and throw a monkey wrench into the model o
rapid innoation by anyone, anywhere, distributable eerywhere. 1hey
create unnecessary barriers to entry that reduce the reedom to operate
and experiment, and thereby harm innoation.

6
")-C ?)17% )< D*#1,(. >% Q-,>#18.4 ?,/C "/6B,)8$ A-(., 464 U.S. 41 ,1984,, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%X.05[_.,#+[.C[G1(,&-)[2+[W0&2(,*)<[_&"5[X">'&.*h[
^0-.
@0@ "/6B,)8$ A-(% >% 01)I8/#1$ L/B., 545 U.S. 913 ,2005,, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%bZb[X">'&.*h[^0-+[2+[Z,.A*"(,h[\"'..
8
A- 1# M,48I,, 545 l.3d 943, 88 U.S.P.Q.2d 1385 ,led. Cir. 2008,, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%^0[,([N&<*A&.
9
M,48I, >% [.77)8, 561 U.S. ___, 130 S. Ct. 3218 ,2010,, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%N&<*A&[2+[i)##.*.

_,()"( ) jS)0'k .C (@(1#" (@#(,&1(0")"&.0h S."! -.11(,-&)<
)0' 0.0:-.11(,-&)<h ?!(,(S5 >*( .C (@&*"&0/ -.#5,&/!"(' .,
#)"(0"(' &0C.,1)"&.0 ., A0.?<('/( '.(* 0." ",&//(, <&)S&<&"5$
1oday, to the extent that there are exemptions they are spare and
niggardly. Using just three notes rom a prior recording and mashing
them up into a completely new song does not, under present copyright
law, count as de minimis.`
10
Academic experimentation on a patented
drug that does not result in any alternatie drug that competes but
merely begins to create the path to one does not come within patent
law`s research exemption.`
11
1hese attitudes-none orced by the
language o the statutes-relect a judicial temperament that seems to
think o copyright and patents in a Blackstonian sole and despotic
dominion`
12
mindrame, an approach that was neer true o real
property under common law, and would, een in terms o pure theory,
be disastrous i applied to knowledge and inormation. 1he idea would
be to deelop a relatiely robust space or experimentation which, i it
led to products and sales, would entitle the owner o the prior, enabling
innoation or creatie expression to claim some share o the proits o
the downstream innoator or creator. 1he critical point o such an
approach would be to allow millions o experiments to run without
liability or its risk, while at the same time assuring that truly enabling
innoations or those experiments that do succeed can share in the
commercial upside o their contributions to downstream innoation.
_.0"&0>( ". (@#)0' "!( (@(1#"&.0* C,.1 "!( a&/&")< b&<<(00&>1
_.#5,&/!" G-"
13
?!(,(2(, "!)" G-"g* #,.2&*&.0* #<)-( ) ',)/ .0
&0"(,.#(,)S&<&"5 )0' &00.2)"&.0 &0 *5*"(1* "!)" '(#(0' .0 )--(**
". (@&*"&0/ #<)"C.,1* )0' *5*"(1*$ lederal courts hae begun to
reject claims under the DMCA that are eorts by copyright owners to
use digital rights management to throw a monkey wrench into the
works o a competitor. lor example, Lexmark tried to make it hard or
competitors who wanted to compete on toner or its printers by
creating a chip and sotware handshake between the printer and the
toner cartridge. \hen a competitor reersed engineered the handshake
so that their microchip-enabled toner cartridge could work with

10
M1,B&#7)1/ @68,($ A-(% >% G,*#-8,)- E,4*8, 410 l.3d 92 ,6th Cir. 2005,, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%N,&'/(#.,"[b>*&-h[^0-+[2+[a&1(0*&.0[c&<1*.
11
@.B#C >% G6I# Q-,>#18,/C, 30 l.3d 1351 ,2002,.
12
SIR \ILLIAM BLACKS1ONL, COMMLN1ARILS ON 1lL LA\S Ol LNGLAND, Book II Ch. II ,
Clarendon Press ,Oxord, 165-169, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%_.11(0"),&(*[.0["!([\)?*[.C[J0/<)0'.
13
1 U.S.C. 1201-1205, aailable at
!""#$%%(0+?&A&#('&)+.,/%?&A&%a&/&")<[b&<<(00&>1[_.#5,&/!"[G-"lG0"&:
-&,->12(0"&.0[(@(1#"&.0*.

Lexmark printers, Lexmark argued that in order to build their
competing cartridge, the competitor had to make a copy o the
handshake sotware, which in turn required them to get around the
encryption protecting that piece o copyrighted sotware. In other
words, the competitor had iolated the DMCA by circumenting the
digital rights management encryption that protected their copyrighted
handshake sotware. 1he court rejected the argument, emphasizing
that a program copy whose core unction was interoperability did not
iolate the DMCA.
14

In this short a piece, I neither aim or an exhaustie list nor oer a detailed
analysis o each o the proposals identiied. Instead, I oer these as an initial
drat o a range o policies that would increase reedom to operate in the
networked inormation economy, and reduce the drag o the current system o
copyrights and patents on both commercial and social enterprises that hae
played a critical role in the explosie innoation we hae experienced on the
Internet in the past decade and a hal.
Conclusion
1he networked inormation enironment has introduced a period o radically
decentralized capitalization o some o the core economic sectors in the most
adanced economies. As a result, growth is coming to depend increasingly on
innoation rom indiiduals and companies at the edges, operating as the ew
successul experiments out o thousands o similar experiments that go
nowhere. Many o these experiments are commercial. Many are non-
commercial. Many combine the two. As a system, this open, chaotic, complex
innoation system requires reedom to operate. It needs to take adantage o its
technical, economic, and social structure more than it needs power to control
uses as in prior models o well-behaed appropriation. Moreoer, the diersity
o models o experimentation, and the increasingly uzzy line between
production and consumption, between the social and the economic, suggest
that or purposes o economic production and growth, ormal contract and
corporate structure are playing a less important role than they did in the prior
century relatie to the increasingly important role played by loosely-structured
oluntarism and human sociality.

14
Lexmark International, Inc. . Static Control, 38 l.3d 522 ,6th Cir. 2004,.


From Dismal Science
to Strange Tales
By Larry Downes
*

Heroes
It was a ight oer nothing.
In 2008, 12,000 members o the \riters Guild o America staged a withering
strike against the major lollywood studios. It lasted three months, interrupted
dozens o 1V series, and delayed seeral big-budget ilms. 1he two sides
reportedly lost more than >2 billion. \et the sole issue in the dispute was when
and how reenues rom the Internet and other digital distribution o
entertainment would be allocated.
1

So ar, no such reenues exist.
Online distribution o moies and especially 1V is a recent phenomenon,
powered by eer-aster data transmission speeds, the continued spread o
broadband technologies into the home, and improed protocols or ile
compression. It seems certain that proitable models or deliering lollywood
content to computers, personal digital assistants ,PDAs,, cell phones, and other
non-1V deices will emerge. But in these early days, as with music beore it, it
isn`t clear what those models will be. \ill they be supported by adertising
\ill content be pay-per-iew or based on all-you-can-eat subscriptions \ill
consumers preer to own or rent
As industry ponders these unanswerable questions, consumers are doing much
o the innoating themseles as they did with earlier, less bandwidth-intensie
content such as text and music. Users o \ou1ube, Bit1orrent, and all
ariations o ideo streaming or ile-sharing applications, in the interest o speed

\),,5 a.?0(* is an Internet analyst and consultant, helping clients deelop business
strategies in an age o constant disruption caused by inormation technology. le is the
author o WK\JGXQ^KZ YQJ i^\\JP GVV$ a^Z^YG\ XYPGYJZ^JX cFP bGPiJY
aFb^KGK_J ,larard Business School Press 1998, and, most recently, o YQJ \G`X Fc
a^XPWVY^FK$ QGPKJXX^KZ YQJ KJ` cFP_JX YQGY ZFeJPK \^J GKa NWX^KJXX ^K YQJ
a^Z^YG\ GZJ ,Basic Books 2009,. 1his essay is adapted rom 1lL LA\S Ol DISRUP1ION.
1
Michael \hite & Andy lixmer, N)44C2))B S)1I#18 '#/61- /) S)1I D</#1 ;-B,-& "/1,I#,
BLOOMBLRG, leb. 13, 2008, !""#$%%???+S<..1S(,/+-.1%)##*%0(?*D
#&'E0(?*),-!&2(H*&'E)i'?P7._4T`b.

and experimentation, do not bother with the niceties o obeying the law
,Viacom`s >1 billion lawsuit against \ou1ube and Google is currently on
appeal,.
2

\hy did the two sides risk so much ighting oer reenue that doesn`t yet exist
rom channels that haen`t been inented 1he writers say they took a stand in
large part because they did not do so in the early days o ideocassette sales and
rentals. \hen the proitable models inally arried, the writers belieed they got
a much worse deal. Because media sales and rentals now represent the largest
share o entertainment income, missing the boat has been painul or writers.
1he studios argued that until it is clear how and when money is to be made
rom digital distribution, pre-assigning residual royalties to writers would limit
the studios` ability to experiment with dierent distribution and partnership
models. 1hey made the same argument with ideotapes.
Ultimately, new rates or residual royalties were agreed upon or categories
including downloaded rentals and sales, ad-supported streaming media, short
clips, and promotional uses. \hether these proe to be aorable rates, or een
the right categories, remains to be seen. Lither way, media will continue to
migrate to the Internet at the expense o other orms o distribution.
The Strange Behavior
of Non-Rivalrous Goods
It is hard to say i anyone made the right decisions in the writers` strike, in part
because the tools or aluing inormation products and serices, een or
present uses, are terrible. \ou will look in ain at the balance sheets o
companies whose sole assets are inormation-including much o the
entertainment industry, as well as proessional serices such as doctors, lawyers,
and consultants-to ind any useul measure o the current or uture alue o
the company`s real assets. \hile management gurus sing the praises o
deeloping a company`s intellectual capital, inancial reporting systems ignore it.
Accountants reer to all the aluable inormation in a business-its inormation
assets-as intangibles. As the name suggests, these are assets that neer take a
physical orm as do actories and inentory. Unlike physical assets, inormation
assets are generally not counted in calculating the total worth o an enterprise.
lor the most part, a company`s human resources, brands, and good
relationships with customers and suppliers-let alone its copyrights, patents,
trade secrets, and trademarks-are let o its balance sheet. 1he alue o the

2
"## Adam Ostrow, F,.()* L)8#8 `T M,44,)- D&.,- Z)696+#, MASlABLL, June 23, 2010,
!""#$%%1)*!)S<(+-.1%M;6;%;3%M9%5.>">S(:?&0*:2&)-.1:<)?*>&"%.

company`s inormation, at least as ar as accounting is concerned, is basically
nothing.
\hy Accountants hae argued or years that inormation and other intangibles
are so dierent economically rom material goods that traditional methods o
aluation just don`t apply. As an asset, the explanation goes, inormation
behaes precisely the opposite o its tangible counterparts: Capital assets lose
alue as they are used, equipment becomes obsolete, and raw materials are
depleted. Brands and reputations, by contrast, become more aluable the more
they are exercised, in theory generating reenue oreer. \ou cannot determine
the price o a logo or a customer relationship with the same tools you use to
depreciate a tractor.
lair enough. But that doesn`t explain why accountants hae done so little to
deelop aluation techniques that apply to inormation assets. A dangerous
result o that ailure is that ew managers understand inormation or how it
generates alue. Len CLOs o large companies regularly get it wrong when
they talk casually about trademarking an idea` or copyrighting a word.` ,\ou
cannot do either.,
As inormation becomes more central to economic perormance, the ailure to
account or its alue has become dangerous. Lxecuties, especially in public
businesses, are compensated based on the health o their companies` balance
sheets. 1o the extent that inormation alue doesn`t appear there, it`s
understandable that many companies don`t put much, i any, eort into
deeloping or managing those assets.
1hat`s unortunate because the strategic cultiation o inormation assets is
beneicial in many ways. Consider larrah`s Lntertainment, which operates
casinos worldwide in places where gambling is legal.
\hen ormer business school proessor Gary Loeman joined the company as
chie operating oicer in 1998, he decided to look or underutilized assets on
the company`s balance sheet. le ound them in larrah`s data warehouse. Like
most casino chains, larrah`s had implemented a rewards program that gae
customers special beneits or using their membership cards while playing slot
machines. larrah`s was collecting ast amounts o inormation on its actory
loor,` but had done ery little to put that data to use.
A detailed reiew o the collected inormation upended seeral long-standing
myths about where larrah`s made the most money. Most o the company`s
proits came rom a quarter o its customers. 1hose customers were not,
howeer, the cu-linked, limousine-riding high rollers |larrah`s| and |their|

competitors had awned oer or many years.`
3
Instead, larrah`s discoered
that the high-proit customers were regular isitors, many o them recent
retirees. 1hey made requent trips to the casino and spent steadily, i modestly,
at its gaming tables, restaurants, and hotels.
larrah`s quickly reconigured its customer-acing actiities, including check-in,
complimentary meals, and special promotions, orienting them toward the actual,
as opposed to the presumed, best customers. 1he result was a changed
enterprise, one that consistently outperorms its competition.
Len though the balance sheet neer reported the alue o the diamonds
Loeman ound when he looked in his data mine, his inormation assets were
by no means worthless. In 2006, larrah`s was sold to a priate equity
partnership at a price that alued the inormation at more than >1 billion,
representing a 30 premium in the total purchase price.
4
1oday, Gary
Loeman remains CLO o the company, a position he has held since 2003.
1he writers` strike and the larrah`s story teach an important lesson about the
economics o inormation. Just because inormation alue is indeterminate
doesn`t mean it`s worthless. Not by a long shot. 1he lollywood writers and
producers clearly did not think so, nor did the buyers o larrah`s.
Consider another example: Search giant Google has >20 billion in assets, mostly
cash, on its balance sheet. 1he company, howeer, een on the lowest day o
the stock market in ten years, was worth nearly >100 billion-more than ie
times its book alue. Somebody has igured out, at least in part, how to alue
the company`s inormation assets.
Digital lie is made up o inormation. It comes in a wide range o types,
including priate data, speech, news and entertainment, business practices, and
inormation products and serices such as ilms, music, inentions, and
sotware. But all inormation operates under a common set o economic
principles. So to thrie in the next digital decade, you must understand the
basic elements o inormation economics.
In modern economic terminology, goods are categorized as either priate` or
public` goods. Most goods in our industrial economy are priate goods.
Purely priate goods are those that can be possessed by only one person at a

3
Gary Loeman, Diamonds in the Data Mine, lARV. BUS. RLV., May 2003% "## .48) Julie
Schlosser, 9#.(3#1\8 M#/$ lOR1UNL, March 8, 2004, !""#$%%1.0(5+-00+-.1%1)/)B&0(*%
C.,">0(%C.,">0([),-!&2(%M;;T%;9%;8%939388%&0'(@+!"1.
4
Ryan Nakashima, N.11.3\8 ;-/#1/.,-*#-/ D((#7/8 M6C)6/ M,B <1)* 51,>./# ;a6,/C 01)67, USA
1ODA\, Dec. 19, 2006, !""#$%%???+>*)".')5+-.1%1.0(5%&0'>*",&(*%M;;3:6M:67:
!),,)!S>5.>"[@+!"1.

time ,rialrous`, and whose use can be limited to that person or with
whomeer she might share it ,excludable`,. I I own a barrel o oil, then you
don`t own it, unless I sell it to you, in which case I no longer hae it. Once it`s
used, it`s gone oreer-no one has it anymore.
Public goods, by contrast, can be used by more than one person at the same
time ,non-rialrous`,, and limiting access to them is diicult, i not impossible
,non-excludable`,. 1he classic example in economics textbooks is national
deense. Lither eeryone has this good or nobody does. 1he military protects
eeryone, including those who do not pay taxes. Deensie missiles cannot be
programmed to leae a single house unguarded.
Inormation is an archetypically non-rialrous good. As 1homas Jeerson
amously wrote, le who receies an idea rom me, receies instruction himsel
without lessening mine, as he who lights his taper at mine, receies light without
darkening me.`
5
Once a composer completes a song, there`s no physical limit
to how many people can perorm it simultaneously. 1here is a cost associated
with its creation, but the composer incurs no additional cost no matter how
many times the work is played. Regardless o how oten it is perormed, the
composition still exists. In act, it becomes more aluable the more reely it`s
shared-it becomes more popular, maybe een a hit.`
So inormation is non-rialrous, but is it also excludable Until recently, the
answer in practice was oten no. 1hat`s because many inormation products
that sprang rom the creatiity o the human mind could not easily be
distributed without irst being copied to physical media such as books,
newspapers, or, in the case o music, CDs and records. In that transormation
,demassiication,` in Alin 1oler`s terminology
6
,, inormation lost its
excludable property, looking more like the barrel o oil than like national
deense. It`s easy to limit access to the barrel o oil-there`s only one, ater all.
1he song, once recorded and duplicated, is harder to control, but it`s still
possible to exclude those who didn`t pay or a copy or pay or the right, as in
radio, to broadcast it.
Inormation, until recently, was a public good in theory but in practice behaed
more like a priate good. 1he need to reduce it to physical media masked its
true nature, and gae rise to seemingly incongruous terminology that includes
stealing an idea,` pirating content,` and, most signiicantly, intellectual
property.` Ater more than ie hundred years o Gutenberg`s moeable type,
we`re so conditioned to experiencing inormation through mass-produced
media that we equate the cost o the media with the alue o the content. As

5
Letter rom 1homas Jeerson to Isaac McPherson ,Aug. 13, 1813,, .>.,4.+4# ./
!""#$%%#,(**:#>S*+>-!&-)/.+('>%C.>0'(,*%'.->1(0"*%)6[8[8*6M+!"1<
6
"## &#-#1.44C ALVIN 1OllLLR, 1lL 1lIRD \AVL ,1980,.

John Perry Barlow poetically put it, until the adent o electronic distribution
through the Internet, the bottle was protected, not the wine.`

lor inormation products, that protection-the ability to exclude-is almost
entirely a unction o law: 1he law o copyright makes it a crime to copy`
inormation without permission. Copyright gies the composer the exclusie
right to make or authorize perormances o a song, or example, or to record it
and produce copies. At the same time, copyright outlaws the production o
copies by anyone else, including someone who purchased a legal copy.
By limiting both the perormance and production o a song, copyright
transorms non-rialrous inormation into a rialrous physical good. But the
alchemy o copyright is starting to ail as digital technology makes it easier to
distribute songs electronically. Gien the Internet, it`s now much harder to
limit who gets to hear a song and when. A copy no longer requires expensie
recording and pressing equipment, or access to a costly and ery isible retail
distribution network.
Although the composer can legally exclude those who do not buy authorized
copies o his work, his ability to police that right is increasingly expensie, oten
costing more than it`s worth. \ou can`t realistically stop people rom humming
your tune, een i they do it out loud. And now you can`t really stop them rom
sharing copies o a digital recording, either.
Other than the composer hersel, howeer, consumers are also potentially
harmed by the transormation o inormation goods back to their non-rialrous
state. 1here were and remain important reasons or legal systems that treat
inormation as i it were a rialrous good.
Copyright, or example, is designed to maximize the alue o the up-ront
inestment that inormation producers must make. I copyright didn`t exist,
you could simply buy a recording o the song, reproduce it, and sell your own
copies. Because your total inestment would be only the cost o a single copy,
your ersion would likely be cheaper than the one marketed by the composer
himsel. In theory, the composer would ind it diicult to recoer his creatie
inestment, making him less likely to undertake his important work in the irst
place. Ultimately, eeryone would be worse o.
But copyright`s alue comes at a high price. By imposing costs on the exchange
o inormation that otherwise would not exist, the law neutralizes many o the
aluable eatures o non-rialrous goods.

John Perry Barlow, 93# ;()-)*C )< AB#.8$ \IRLD 2.03, 1994,
!""#$%%???+?&,('+-.1%?&,('%),-!&2(%M+;9%(-.0.15+&'()*[#,+!"1<.

lortunately, this special power is limited. Len with copyright, some orms o
sharing are perectly legal. Libraries can loan out the same copy o a recording
to as many people as want to hear or play it, one at a time. lans who purchased
their own copy are likewise ree to loan it to their colleagues, or een to resell
their copy to a used record store or through online serices such as Amazon
Marketplace or eBay.
Copyrights also hae an expiration date. In the United States, or example,
copyrights lasts 0 years beyond the lie o the author or 120 years or certain
works.
8
Ater this period, the work is no one`s property-the public can use it
howeer they want to. Anyone can perorm the work, make copies o it, adapt
it, or incorporate it into new works. It becomes oreer ater a purely non-
rialrous good.
Copyright protection is also limited to the producer`s particular expression and
not the underlying ideas. 1he ideas in a song ,loe conquers all, loe stinks, are
non-rialrous rom the moment the song is written.
Consider a 1996 court case inoling sports statistics. In the days beore the
\eb and wireless data deices, sports ans who were not attending a game
could get up-to-the-minute inormation rom a dedicated paging deice rom
Motorola called Sportstrax. Sportstrax employees watched sporting eents on
1V and entered key inormation ,#%&%$ who had the ball or who had scored, into
a computer system. A ew minutes later, Sportstrax customers would be paged
with short updates.
Motorola was sued by the National Basketball Association, which claimed the
transmission o inormation by pager iolated its copyright in the broadcast o
games.
9
1he court disagreed. Sporting eents are not authored,` the judges
noted, and are thereore not protected by copyright in the irst place. Game
data, including interim scores, are acts, not a particular expression o an
inormation producer. lacts are non-rialrous, outside the protection o
copyright.
1oday, pagers hae gien way to cell phones that can take photographs and
ideos and share them ia the Internet. Popular teleision programs such as
American Idol` hae armies o ans who watch the show and write blogs
about each perormance een as they`re watching them. So long as the actual
perormances aren`t being copied, howeer, the commentary is perectly legal.

8
1 U.S.C 302- 03.
9
Nat`l Basketball Assoc. . Motorola, Inc., 105 l.3d 841 ,2d Cir. 199,,
!""#$%%???+<)?+-.,0(<<+('>%-.#5,&/!"%-)*(*%6;4[c9'[8T6+!"1.

As these examples suggest, the challenge or copyright and other laws
controlling inormation has always been to strike the right balance between
incenties or creators and the alue that the public deries rom unettered use.
It`s a balance that is constantly being unsettled by new technologies, a problem
that has accelerated with the adent o the digital age.
On one hand, inormation technology has greatly lowered the cost o creating
and distributing inormation including books, moies, and recorded music. But
the same technologies hae also made it easier to make and distribute
unauthorized copies which are, in many cases, perect replicas o the original.
Should inormation laws be tightened or relaxed in the next digital decade Do
producers need more protection rom copyright laws, or do consumers desere
greater reedom Are new inormation uses made possible by sotware
applications such as \ou1ube, lacebook and llickr, creating more alue than
they destroy, and or whom
The Five Principles
of Information Economics
Unortunately, many o those debating these questions-and there are many,
including lawmakers, industry leaders, and consumer groups-don`t understand
the economic properties o inormation any better than do the accountants who
reuse to measure it. So it`s worth summarizing the ie most important
principles o inormation economics. It`s een better to memorize them:
Renewability
Inormation cannot be used up. It can be enhanced or challenged, it can
become more or less aluable oer time, but once it has been created, it can be
used oer and oer again. In the end it exists as it began. Most new
inormation, moreoer, is created rom other inormation, making it a
renewable energy source. In electronic orm, neither its production nor its use
generates waste products that damage the enironment. In that sense,
inormation is the ultimate green` energy.
1he online encyclopedia \ikipedia, or example, isn`t written by hired experts.
It`s written by olunteers who post articles on subjects they either know or
think they know something about. \ithin certain limits, anyone else can edit,
correct, or change entries. Oer time, the articles eole into a useul and
reliable orm. No money changes hands in either the creation o \ikipedia or
its use.


Universality
Leryone has the ability to use the same inormation simultaneously. Blog
entries, news articles, and \ou1ube ideos can be enjoyed simultaneously by an
unlimited number o people. 1he only distribution costs are the photons on a
screen. Lach consumer, moreoer, may hae a completely dierent reason or
consuming the same inormation, and perhaps her own response to it. She may
be inspired to respond with inormation o her own.
lacebook, or example, is comprised almost entirely o user-generated content.
Users are constantly commenting on status updates, photographs, and other
inormation posted by their riends, and initing each other to join interest
groups or play inormation games.
Magnetism
Priate goods operate under the law o supply and demand: 1he greater the
supply, the lower the price you can charge or it, and ice ersa. 1he alue o
inormation, on the other hand, increases as a unction o use. Inormation
alue grows exponentially as new users absorb it. 1he more places my brand or
logo appears, the higher the alue customers attach to all my goods. Use makes
the brand more, not less, aluable.
1his increase in alue accelerates as the inormation spreads, creating a kind o
magnetic pull that generates network eects. Since no one owns the Internet`s
protocols, or example, these standards hae spread easily, resulting in the
explosie growth that began in the 1990s. 1he standards are now more aluable
than when only a ew people used them.
Friction-Free
1he more easily inormation lows, the more quickly its alue increases. In
electronic orm, inormation can moe in any direction at the speed o light. It
experiences no decay along the way, arriing at its destinations in the same orm
as when it departed. lor many kinds o inormation, including languages,
religious doctrines, and adertising, the ease o transer helps to improe
society-or at least the proits o those who disseminate it. 1he cheaper it is to
spread the word, the more likely and quickly it will be spread.
1here is, howeer, an inherent paradox: 1he rictionless spread o inormation
can undermine the incenties or its production. Content producers, including
authors, musicians, news organizations, and moie studios, inest heaily in the
production o new inormation. 1o recoer their inestment, inormation
producers must charge or its use. Lconomically, howeer, een the simplest
payment schemes ,subscriptions, or example, slow the natural tendency o
inormation to moe reely. Since inormation lows along the path o least

resistance, markets look or ways to aoid ees. In that sense, new technologies
oten subert old business paradigms, een when the inentors o those
technologies didn`t intend or them to do so.
Vulnerability
Inormation`s alue does share one property with tangible goods: It`s not
indestructible. Value can be destroyed through misuse. I you license your
company`s name ,and thus its reputation, to an inerior product or a product
that does not hae a clear connection to your brand, you risk conusing
consumers about what your brand stands or. Value can also be destroyed by a
third party, perhaps a competitor oering a knocko product that looks like
yours but is o lesser quality. Or an identity thie can appropriate your name
and credit history to borrow money rom banks or credit card companies.
\hen the thie disappears, not only is the money gone, so is your reputation.
Inormation can also be a ictim o its own success. It is now so easy to
produce, distribute, and consume inormation that users are experiencing
oerload. 1oday, websites, e-mails, blogs, text messages, and een tweets`-
brie messages that relect the thoughts o a user on 1witter-all compete or
users` limited time. As the sources o inormation and the olume produced
expand rapidly, consumers ind it increasingly diicult to limit their exposure to
inormation o real alue to them.
m m m
It`s easy to see these ie principles in action in digital lie. Consider Google.
One might wonder how a company can be worth anything, let alone >100
billion, when it charges absolutely nothing or its products and serices. \ou
can search Google`s databases and use its e-mail serice day and night without
spending a penny, you can store photos on its Picasa photo serice, create
documents with its online word processing sotware, iew Google Maps, and
share ideos on its \ou1ube serice or ree. Indeed, the company is
determined to hae as many people as possible take complete adantage o it.
Len though databases and other serices are what consumers want and
thereore represent the source o the company`s alue, Google does not hoard
those assets as i they were barrels o oil to be used only when necessary. It
treats them instead as non-rialrous goods that increase in alue as more people
use them.
1he company isn`t being generous. Google makes nearly all its money by
renting out adertising space to companies whose products and serices
complement the things consumers do when they are using Google. I
inormation wants to be ree,` then let it be as ree as possible, and make all
the proits rom the collateral eects o the network. 1hat`s the company`s

simple business strategy, one that has created remarkable new alue een as it
disrupts the assumptions o eery industry the company touches.
And or social networking sites, just giing away content isn`t good enough.
1hese companies also hae to ind ways to get users to help them deelop their
sites in the irst place. Companies like lacebook, MySpace, and their
proessional equialent, LinkedIn, are constantly adding ree tools and gadgets
to make their products more compelling. Once a serice reaches the tipping
point ,lacebook now has 500 million users!,, the search or ways to make
money begins in earnest, including premium serices and targeted
adertisements. But thanks to the weird economics o inormation, there
remain powerul reasons not to charge the users or the core product-eer.
The Problem of Transaction Costs
1here`s one additional aspect o inormation economics that is essential to
understand. 1he rictionless transer o inormation and the problem o
inormation oerload suggest that the economy o digital lie is a kind o
machine. Like the best engines, it can operate with remarkable eiciency-
proided its parts are kept lubricated and ree o oreign matter. Already,
technologies perected during the last digital decade hae ruthlessly eliminated
waste in our increasingly eicient online lies. Still, the inormation economy is
not perect. It suers, like its physical counterpart, rom a kind o ineiciency,
what economist Ronald Coase irst called transaction costs.`
Coase came to the United States rom Lngland as an economics graduate
student in 1931. Only twenty years old, Coase had a reolutionary agenda.
Struggling to reconcile the socialism o his youth with the ree-market sensibility
o his proessors, Coase saw big companies as proo that centralizing actiities
could work on a grand scale. I he could learn how big companies did it, Coase
imagined, then perhaps the lessons could be applied to big goernments as well.
Oddly enough, no one had eer asked why companies existed, and certainly no
one had eer thought to ask the people who were running them.
\hat Coase learned made him swear o socialism oreer, and led to the
publication o an article that changed economic thinking oreer-an article
cited as reolutionary sixty years later, when Coase receied the Nobel Prize in
Lconomics.
In 1he Nature o the lirm,` Coase argued that there is a price not only to
what companies buy and sell, but also to the process o buying and selling it.
10
Buyers and sellers hae to ind each other, negotiate deals, and then

10
Ronald l. Coase, 93# Y./61# )< /3# E,1*, 4 LCONOMICA 368-405 ,No., 193,,
!""#$%%)("'*+!0>-+('>+-0%>#<.)'C&<(%M;;8;963M66769TTT+#'C.

consummate them. 1his actiity was neither especially easy nor without costs.
Coase thereore argued that companies were becoming bigger because markets
were, relatiely speaking, too expensie.
Coase called the price o doing a deal its transaction cost.` 1he existence o
transaction costs, he belieed, explained why companies were internalizing more
and more actiities, especially repeated unctions like buying raw materials and
marketing. lor these actiities, maintaining an inside unction such as a
purchasing department was cheaper than relying or each indiidual purchase on
whoeer might happen to be in the market.
1o understand why, let`s take a simple example. Say you work or an aerage-
size company and youè run out o paper clips. Almost assuredly, you will get
your paper clips not by leaing your oice to drie to the oice supply store but
by going down the hall to the supply cabinet, where your company`s purchasing
department maintains an inentory o basic supplies. \our company will, in
act, keep such basic supplies on hand as a matter o course, without giing
much thought to the cost o carrying this inentory. 1his holds true een i
buying and distributing oice supplies hae nothing to do with what your
business does. \our company is likely to keep paper clips on hand een i there
is no discount or buying in bulk.
\hy Len i you could get paper clips on your own or the same price, you
still hae to go out and get them. 1his means inding the stores that carry them
and learning how much they charge. 1hen you hae to choose between the
closest store and the one with the best price. At the checkout stand, you need
to make sure you are really charged what the store adertises. I the clips are
somehow deectie, you hae to take them back and demand replacements or
some other remedy.
And that`s just or a simple transaction. Imagine instead that you`re buying raw
materials needed to manuacture a jet airplane. 1here is the additional eort o
negotiating a price, writing a contract, inspecting the goods, and, potentially,
inoking the legal system to enorce the terms and conditions. It`s better, you
say, to own the supplier or at least to buy in bulk and aoid all that trouble.
1hat trouble` is transaction costs.
\orking rom Coase`s basic idea, economists hae identiied six main types o
transaction costs:
X(),-! -.*"*: Buyers and sellers must ind each other in increasingly
dierse and distributed markets.
^0C.,1)"&.0 -.*"*: lor buyers, learning about the products and
serices o sellers and the basis or their cost, proit margins, and

quality, or sellers, learning about the legitimacy, inancial condition,
and needs o the buyer, which may lead to a higher or lower price.
N),/)&0&0/ -.*"*: Buyers and sellers setting the terms o a sale, or
contract or serices, which might include meetings, phone calls, letters,
axes, e-mails, exchanges o technical data, brochures, meals and
entertainment, and the legal costs o contract negotiations.
a(-&*&.0 -.*"*: lor buyers, comparing the terms o one seller to
other sellers, and processes such as purchasing approal designed to
ensure that purchases meet the policies o the organization, or sellers,
ealuating whether to sell to one buyer instead o another buyer or not
at all.
V.<&-&0/ -.*"*: Buyers and sellers taking steps to ensure that the good
or serice and the terms under which the sale was made, which may
hae been ambiguous or een unstated, are translated into the behaior
expected by each party. 1his might include inspecting the goods and
any negotiations haing to do with late or inadequate deliery or
payment.
J0C.,-(1(0" -.*"*: Buyers and sellers agreeing on remedies or
incomplete perormance. 1hese include eerything rom mutual
agreements or a discount or other penalties to expensie litigation.

As this list suggests, transaction costs range rom the triial ,turning oer a box
o paper clips to see the price, to amounts greatly in excess o the transaction
itsel ,imagine i you were seriously injured by a deectie paper clip lying o
the shel and sticking you in the eye,. In act, economists Douglass North and
John \allis hae estimated that up to 45o total economic actiity consists o
transaction costs.
11
Lliminating them entirely would translate to a staggering
>4.5 trillion in annual saings in the United States alone, eliminating much o
the work done by accountants, lawyers, adertisers, and goernment agencies.
One needn`t go that ar to improe economic perormance, howeer. lirms are
created, Coase concluded, because the additional cost o organizing and
maintaining them is cheaper than the transaction costs inoled when
indiiduals conduct business with each other using the market. lirms, while
suering ineiciencies o their own, are more eicient at certain types o
actiities than the market. 1echnologies-in 193, Coase had in mind
telephones, in particular-improed the perormance o one or both, constantly
resetting the balance between what was best to internalize and what was best
let to the market.

11
John Joseph \allis & Douglass C. North, @#.861,-& /3# 91.-8.(/,)- "#(/)1 ,- /3# D*#1,(.-
;()-)*C, ,- LONG-1LRM lAC1ORS IN AMLRICAN GRO\1l 95-162 ,Stanley L. Lngerman &
Robert L. Gallman, eds. 1986,, !""#$%%???+0S(,+.,/%-!)#"(,*%-73I7.

So which unctions should a irm perorm internally 1he deceptiely-simple
answer is only those actiities that cannot be perormed more cheaply in the
market or by another irm. In act, as Coase says, a irm will tend to expand
precisely to the point where the costs o organizing an extra transaction within
the irm become equal to the costs o carrying out the same transaction by
means o an exchange on the open market.`
12

lor some actiities, say plumbing, the open market works relatiely well, and
the need or plumbers to orm large irms to aoid transaction costs has neer
arisen. lor the large-scale operations o integrated manuacturers, such as
Boeing and General Motors, which require coordination, heay capital
inestment, and complex distribution systems, the irm is the only economically
iable solution.
Coase belieed economists should turn their attention to the practical problem
o uncoering transaction costs whereer they occur and eliminating those that
are unnecessary. Doing so, he hoped, would, among other things, help reduce
the need or goernment interention. A great deal o regulation and liability
laws, Coase argued, were unconscious eorts to oercome transaction costs or
certain types o actiities, such as accidents and pollution. But the regulations
themseles generate so many transaction costs that in many cases doing nothing
at all would hae produced a better result. 1o ind out how much law and
regulation are optimal requires a better understanding, once again, o the costs
inoled.
Coase had hoped his elegant proo would get economists working on the real
problem at hand. Ironically, all he did was make economics more esoteric.
Instead o lowering themseles to the kind o empirical research that was
common in other social sciences, economists simply dispose o Coase in an
opening ootnote. 1hey assume a rictionless economy` and then proceed to
deelop elaborate mathematical models o behaior in a purely theoretical
unierse. Rather than join his quest, most economists retreated to more
abstract models o economic behaior, which Coase dismisses as little more
than a ast mopping-up exercise` o loose ends let by Adam Smith`s seminal
18
th
century work, 93# S#.4/3 )< Y./,)-8.
Increasingly rustrated with his economist colleagues, Coase instead took up
residence at the Uniersity o Chicago`s law school. Lconomists, he came to
see, aoided inormation, and misused the ew sources, such as goernment
data, that were readily aailable. Lconomics had become a shell game. I you
torture the data enough,` he wrote, dismissing much o modern economic

12
Coase, 93# Y./61# )< /3# E,1*, 8671. note 10.

analysis, nature will always coness.`
13
I that was all economists could do,
Coase decided he was no economist. Awarded the Nobel Prize in 1991, Coase
began his acceptance speech on a note o despair. In my long lie I hae
known some great economists,` he told the committee, but I hae neer
counted mysel among their number nor walked in their company.`
14

Look at the perormance o the economy oer the past twenty years and it`s
easy to sympathize with Coase`s rustration. 1he rational` stock market still
booms and busts. Cyclical industries continue to oerexpand and then
oercontract. Lorts at creating an open global economy without trade barriers
are met with rioting mobs. National banking regulators read eery tea lea they
can ind and still go to bed wondering i they hae cut rates too soon or too
late, too much or too little, or een i their cuts hae made an iota o dierence.
\hile most economists iddle with ormulas, the economy is burning. \ithout
a better understanding o the nature o transaction costs, we`ll neer be able to
predict-let alone improe-what seem to be the most basic elements o
economic behaior.
1hat, in any case, is the real world. In the digital world, the problem is not only
less seere, but also solable. 1he ree low o inormation made possible by
digital technology is decreasing the riction o transaction costs in a ariety o
interactions. lrom global price comparisons to searches o much o the world`s
knowledge to auctions or anything, the cost o deal-making is plummeting.
1he Internet is driing down all six types o transaction costs. 1hat`s what`s
made the Internet so disruptie in the last decade, and what will continue to
drie dramatic consumer, business, and regulatory changes in the next digital
decade.
Consider a ew examples:
1. X(),-! -.*"*: 1echnology connects people across geographical, time, and
national borders. Automatic notiications or obscure collectibles on eBay,
inding old riends through the People \ou May Know` eature on
lacebook, or letting your 1iVo pick programs or you that it thinks you
might like to watch-each o these reduces search costs, sometimes
dramatically. Restaurant and other business reiews aailable directly on
cell phones make it easier to ind just the right place no matter where you
are. 1here`s een an iPhone application uses GPS technology to help you
ind your car in a crowded parking lot!

13
Ronald l. Coase, N)2 "3)64B ;()-)*,8/8 ?3))8#=, G. \arren Nutter Lecture in Political
Lconomy, American Lnterprise Institute ,1982,.
14
Ronald l. Coase, Y)+#4 51,R# L#(/61#, Dec. 9, 1991,
!""#$%%0.S(<#,&B(+.,/%0.S(<[#,&B(*%(-.0.1&-*%<)>,()"(*%6776%-.)*(:<(-">,(+!"1<.

2. ^0C.,1)"&.0 -.*"*: 1echnology creates standard data structures that can be
searched and consolidated oer a growing network o computers. 1he
asymmetry o sellers concealing data has eroded, radically changing the way
people buy cars, real estate, and inestment securities. lree or subscription
serices including Carlax, Zillow, and \ahoo! linance gie buyers an
abundance o aluable inormation that was preiously inaccessible at any
price. Online dating serices such as Chemistry.com increasingly use
sophisticated proiling technology to suggest compatible matches.
3. N),/)&0&0/ -.*"*: 1he exchange o inormation can now take place
digitally and is captured in databases or easy reuse in subsequent
transactions. Instant publication o classiied ads on Craigslist means many
local transactions are completed within minutes. Business-to-business
transactions increasingly rely on libraries o standard terms. 1he nonproit
Association or Cooperatie Operations Research and Deelopment
,ACORD,, or example, uses the XML data standard to create standard
orms used by insurance and reinsurance agents and brokers oering lie,
property, and other lines o products.
4. a(-&*&.0 -.*"*: Visibility to expanded online markets gies both buyers
and sellers a better picture o minute-to-minute market conditions. Seeral
insurance websites, including Progressie.com, proide instant quotes and
comparisons to the prices o their competitors. Cell phone users can
compare prices rom online merchants while shopping at retail stores,
putting added pressure on merchants to match or beat those prices or oer
other incenties, including deliery or ater-sales support. Online gamers
can check the reputation o potential participants to decide whether to
allow them to join their teams.
5. V.<&-&0/ -.*"*: 1ransactions conducted with system-to-system data
transers create a more complete record o the actual perormance o the
participants, which can then be captured and queried. lor goods purchased
online, most merchants now proide direct access to detailed shipping and
tracking inormation rom expediters such as UPS or ledLx or een
standard deliery rom the postal serice. Some merchants, including Dell
Computers, proide inormation about the manuacturing process, allowing
customers to track their products beore they are een shipped. Most
sotware products now collect bug and other ailure inormation in real
time, automatically installing updates and repairs. Players o the online
\orld o \arcrat game can speak` directly to in-game employees or
robots wheneer they hae a problem.
6. J0C.,-(1(0" -.*"*: Llectronic records can simpliy the process o
resoling disputes oer what was agreed upon or what did or did not occur.
Online payment serices such as PayPal oer elaborate dispute resolution

unctions that include mediation and arbitration when buyers and sellers
cannot resole their dierences, along with insurance and guaranteed
satisaction. 1hese are all supported by the collection o end-to-end
transaction data documenting the actual perormance o buyers and sellers.
Bloggers can quickly whip up electronic mobs to put pressure on
companies, politicians, or celebrities whose behaior they eel does not
comply with agreed-upon standards.
Conclusion: Conflicts at the Border
Digital technology, as I argued in my 1998 book, Unleashing the Killer App,`
15

has created a corollary to Coase`s obseration about business organizations. As
transaction costs in the open market approach zero, so does the size o the
irm-i transaction costs are nonexistent, then there is no reason to hae large
companies. lor products constructed entirely or largely out o inormation, we
now stand on the erge o what Don 1apscott and Anthony \illiams call peer
production,` where just the right group o people come together to apply the
right set o skills to sole complex problems, whether in business or
otherwise.
16
I called this phenomenon 1he Law o Diminishing lirms.`
1echnology is changing the dynamics o irms, making them smaller but more
numerous. 1his, howeer, is good or the oerall economy. Liciency
translates to saings o time, money, and decreased waste. Productiity,
customer satisaction, and the aailability o customized products and serices
hae improed dramatically. Keeping in touch across time zones and long
distances gets easier, as does organizing dierse groups o people or social,
political, or business reasons. 1he aerage consumer can now edit an online
encyclopedia, post news and photos as a citizen journalist, or operate a home-
based business that can produce and distribute just about anything.
Now or the bad news. Our current legal system, orged in the actories o the
Industrial Reolution, was designed to maximize the alue o rialrous goods.
It cannot be easily modiied to deal with the unique economic properties o
inormation. \orse, the crushing oerhead o regulations and lawsuits, which
may no longer be cost-eectie een in the physical world, adds een less alue
when applied to the lower-transaction-cost-enironment o digital lie.
Increasingly, the old rules do little more than hold back innoation or the
beneit o those who cannot or do not know how to adapt to the economics o
digital lie. In many cases, ineicient laws are propped up by ailing businesses

15
LARR\ DO\NLS & ClUNKA MUI, UNLLASlING 1lL KILLLR APP: DIGI1AL S1RA1LGILS lOR
MARKL1 DOMINANCL ,1998,.
16
DON 1APSCO11 & AN1lON\ \ILLIAMS, \IKINOMICS: lO\ MASS COLLABORA1ION
ClANGLS LVLR\1lING ,2003,.

that are not eager to see their adantages erased. Sometimes those ighting
transormation are powerul business interests, including large media
companies, real estate agents, and een some o the technology companies
whose products uel the digital reolution.
Resistance may also come rom the users themseles. In digital lie, priate
inormation can be inaluable in deciding who to interact with, either or
business or or interpersonal transactions. As adances in technology bring
more priate inormation online, powerul emotions hae been actiated.
Citizens in much o the world beliee their rights to priacy are being iolated,
not only by businesses but by their classmates and neighbors.
Perhaps most worrisome, goernments are taking adantage o lower
transaction costs to improe the technology o sureillance, raising ears o the
dystopic world described by George Orwell in his noel 1984.`
Lower transaction costs hae also proen useul to criminals and terrorists, who
operate reely and anonymously in digital realms. Sometimes their crimes
exploit the ulnerability o inormation, as in the case o identity thet and other
orms o Internet raud. More ominously, irtual gangs are able to attack the
inrastructure o the Internet itsel, releasing iruses and other harmul sotware
that incapacitate serers, destroy data, or, in the case o spam, simply waste
people`s most precious resource: time.
Perhaps the most diicult problems o inormation economics, howeer,
inole the plasticity o inormation in electronic orm. 1echnology has made it
possible to realize the remarkable potential o inormation to be shared and
een enhanced by as many people as are interested. Ineitably, eery new
innoation that supports this creatie urge runs headlong into laws protecting
inormation as property-laws that treat public goods as i they were priate
goods. Although such laws may be necessary, they hae proen unduly rigid in
their current orm, sparking some o the most itriolic ights on the digital
rontier.
1he explosion o digital technology at home, at work, and in goernment,
coupled with the economics o inormation, has created a perect storm. Our
industrial-age legal system will not surie this social transormation. Ater the
lood, as in preious technological reolutions, a new legal paradigm will emerge
to guide the construction o laws better suited to digital lie.
Implementing these new laws will require a great deal o coordination and
collaboration. Most o all, it will require considerable courage on the part o
those who lie in both the physical and digital worlds. 1he next digital decade,
like the last one, will proceed in its and starts, with surprising changes o cast
and characters, allies becoming enemies and enemies inding common ground.

Some winners and losers will proe, in retrospect, to hae been easily predicted.
Others will come rom nowhere.
1he only thing certain is the author o the script: the poorly-understood but
increasingly critical economic properties o inormation.


The Regulation of
Reputational Information
By Eric Goldman*
Introduction
1his essay considers the role o reputational inormation in our marketplace. It
explains how well-unctioning marketplaces depend on the ibrant low o
accurate reputational inormation, and how misdirected regulation o
reputational inormation could harm marketplace mechanisms. It then explores
some challenges created by the existing regulation o reputational inormation
and identiies some regulatory options or the uture.
Reputational Information Defined
1ypical deinitions o reputation` ocus on third-party cognitie perceptions o
a person.
1
lor example, M4.(I\8 L.2 G,(/,)-.1C deines reputation as the esteem
in which a person is held by others.`
2
Bryan Garner`s D G,(/,)-.1C )< @)B#1-
L#&.4 Q8.&# deines reputation as what one is thought by others to be.`
3
1he
lederal Rules o Lidence also relect this perception-centric iew o
reputation.`
4

Associate Proessor and Director, ligh 1ech Law Institute, Santa Clara Uniersity School o
Law. Lmail: (/.<'1)0n/1)&<+-.1. \ebsite: !""#$%%???+(,&-/.<'1)0+.,/. In
addition to a stint as General Counsel o Lpinions.com, a consumer reiew website now
part o the eBay enterprise, I hae proided legal or consulting adice to some o the other
companies mentioned in this essay. I prepared this essay in connection with a talk at the
1hird Annual Conerence on the Law and Lconomics o Innoation at George Mason
Uniersity, May 2009.
1
As one commentator explained:
1hrough one`s actions, one relates to others and makes impressions on them.
1hese impressions, taken as a whole, constitute an indiidual`s reputation-
that is, what other people think o you, to the extent that their thoughts arise
rom what they know about you, or think they know about you.
Llizabeth D. De Armond, E1)/3C ?3.)8: @)B#1- G./. S.1#3)68,-& .-B P4BJE.83,)-#B G#<.*./,)-,
41 VAL. U.L. RLV. 1061, 1065 ,200,.
2
BLACK`S LA\ DIC1IONAR\ ,8th ed. 2004,.
3
BR\AN A. GARNLR, A DIC1IONAR\ Ol MODLRN LLGAL USAGL ,1990,.
4
"##, #%&%, lLD. R. LVID. 803,19,, 803,21,.

Although this deinition is useul so ar as it goes, I am more interested in how
inormation aects prospectie decision-making.
5
Accordingly, I deine
reputational inormation` as ollows:
inormation about an actor`s past perormance that helps
predict the actor`s uture ability to perorm or to satisy the
decision-maker`s preerences.
1his deinition contemplates that actors create a pool o data ,both subjectie
and objectie, through their conduct. 1his pool o data-the reputational
inormation-can proide insights into the actor`s likely uture behaior.
Reputation Systems
Reputation systems` aggregate and disseminate reputational inormation to
consumers o that inormation. Reputation systems can be mediated or
unmediated.
In unmediated reputation systems, the producers and consumers o reputational
inormation communicate directly. Lxamples o unmediated reputation systems
include word o mouth, letters o recommendation and job reerences.
In mediated reputation systems, a third-party publisher gathers, organizes and
publishes reputational inormation. Lxamples o mediated reputation systems
include the Better Business Bureau`s ratings, credit reports,scores, inestment
ratings ,such as Morningstar mutual und ratings and Moody bond ratings,, and
consumer reiew sites.
1he Internet has led to a prolieration o mediated reputation systems, and in
particular consumer reiew sites.
6
Consumers can reiew just about anything
online, examples include:
eBay`s eedback orum,
which allows eBay`s buyers and sellers to rate

each other.
Amazon`s product reiews, which allows consumers to rate and reiew
millions o marketplace products.
\elp.com, which allows consumers to reiew local businesses.

5
Luis M.B. Cabral, 93# ;()-)*,(8 )< 9168/ .-B '#76/./,)-: D 51,*#1 ,June 2005 drat,,
!""#$%%#)/(*+*"(,0+05>+('>%L<-)S,)<%,(#>")"&.0%P(#>")"&.0[d>0(;4+#'C ,treating
inormation about reputation as inputs into Bayesian calculations,.
6
Indeed, this has spurred the ormation o an industry association, the Rating and Reiew
Proessional Association. !""#$%%???+,),#)+.,/.
!""#$%%#)/(*+(S)5+-.1%*(,2&-(*%C.,>1%C(('S)-A+!"1<.

1ripAdisor.com, which allows consumers to reiew hotels and other
trael attractions.
RealSel.com, which allows consumers to reiew cosmetic surgery
procedures.
Ao.com, which allows consumers to rate and reiew attorneys.
Glassdoor.com, which allows employees to share salary inormation
and critique the working conditions at their employers.
lonestly.com,
8
which allows co-workers to reiew each other.
RateMyProessors.com, which allows students to publicly rate and
reiew their proessors.
DontDatelimGirl.com, which allows people to create and ind
proiles o men who are alleged cheaters.`
9

1heLroticReiew.com, which allows johns to rank prostitutes.
10

Why Reputational Information Matters
In theory, the marketplace works through an inisible hand`: consumers and
producers make indiidual and autonomous decisions that, without any
centralized coordination, collectiely determine the price and quantity o goods
and serices. \hen it works properly, the inisible hand maximizes social
welare by allocating goods and serices to those consumers who alue them
the most.
A properly unctioning inisible hand also should reward good producers and
punish poor ones. Consumers allocating their scarce dollars in a competitie
market will transact with producers who proide the best cost or quality
options. Oer time, uncompetitie producers should be drummed out o the
industry by the aggregate but uncoordinated choices o rational and inormed
consumers.
loweer, gien the transaction costs inherent in the real world, the inisible
hand can be subject to distortions. In particular, to the extent inormation

8
lonestly.com was preiously called Unarnished. "## Lelyn Rusli, Q->.1-,83#B: D ?4#.-$
S#44JL,&3/#B 54.(# E)1 G#<.*./,)-, 1LClCRUNCl, Mar. 30, 2010,
!""#$%%"(-!-,>0-!+-.1%M;6;%;9%9;%>02),0&*!(':):-<()0:?(<<:<&/!"(':#<)-(:C.,:
'(C)1)"&.0%.
9
PlayerBlock is a similar serice, tracking undesirable dating prospects by their cellphone
number. "## Leslie Katz, A8 Z)61 G./# . 54.C#1= "#-B . 9#H/ .-B E,-B P6/, CNL1 News.com,
Oct. 22, 200, !""#$%%0(?*+-0("+-.1%89;6:6;I8T[9:78;M;M4:I+!"1<.
10
"## Matt Richtel, "#H 91.B# @)-,/)18 . [#C E,&61#\8 S)#8, N.\. 1IMLS, June 1, 2008. PunterNet
is another website in this category, proiding reiews o British sex workers. John Omizek,
56-/#1Y#/ 93.-I8 N.11,#/ <)1 @.88,># Q782,-&, 1lL RLGIS1LR, Oct. 5, 2009,
!""#$%%???+"!(,(/&*"(,+-.+>A%M;;7%6;%;4%#>0"(,0("[!),1)0%.

about producers is costly to obtain or use, consumers may lack crucial
inormation to make accurate decisions. 1o that extent, consumers may not be
able to easily compare producers or their price,quality oerings, in which case
good producers may not be rewarded and bad producers may not be punished.
\hen inormation is costly, reputational inormation can improe the operation
o the inisible hand by helping consumers make better decisions about
endors. In this sense, reputational inormation acts like an inisible hand
guiding the inisible hand ,an eect I call the secondary inisible hand`,,
because reputational inormation can guide consumers to make marketplace
choices that, in aggregate, eectuate the inisible hand. 1hus, in an inormation
economy with transaction costs, reputational inormation can play an essential
role in rewarding good producers and punishing poor ones.
Gien this crucial role in marketplace mechanisms, any distortions in
reputational inormation may eectiely distort the marketplace itsel. In eect,
it may cause the secondary inisible hand to push the inisible hand in the
wrong direction, allowing bad producers to escape punishment and ailing to
reward good producers. 1o aoid this unwanted consequence, any regulation
o reputational inormation needs to be careully considered to ensure it is
improing, not harming, marketplace mechanisms.
Note that the secondary inisible hand is, itsel, subject to transaction costs. It
is costly or consumers to ind and assess the credibility o reputational
inormation. 1hereore, reputation systems themseles typically seek to
establish their own reputation. I describe the reputation o reputation systems
as a tertiary` inisible hand-it is the inisible hand that guides reputational
inormation ,the secondary inisible hand, to guide the inisible hand o
indiidual uncoordinated decisions by marketplace actors ,the primary inisible
hand,. 1hus, the tertiary inisible hand allows the reputation system to earn
consumer trust as a credible source ,such as the \all Street Journal, the New
\ork 1imes or Consumer Reports, or to be drummed out o the market or
lack o credibility ,such as the now-deunct anonymous gossip website
JuicyCampus,.
11

Thinking About Reputation Regulation
1his part explores some ways that the regulatory system interacts with
reputation systems and some issues caused by those interactions.

11
Matt Iester, D O6,(C "36/B)2-, JUIC\CAMPUS BLOG, leb. 4, 2009,
!""#$%%R>&-5-)1#>*+S<./*#."+-.1%M;;7%;M%R>&-5:*!>"'.?0+!"1<.

Regulatory Heterogeneity
Regulators hae taken diergent approaches to reputation systems. lor
example, consider the three dierent regulatory schemes goerning job
reerences, credit reporting databases and consumer reiew websites:
Job reerences are subject to a mix o statutory ,primarily state law, and
common law tort regulation.
Credit reporting databases are statutorily micromanaged through the
oluminous and detailed lair Credit Reporting Act.
12

Consumer reiew websites are irtually unregulated, and many potential
regulations o consumer reiew websites ,such as deamation, are
statutorily preempted.

1hese dierent regulatory structures raise some related questions. Are there
meaningul distinctions between reputation systems that support heterogeneous
regulation Are there best practices` we can obsere rom these
heterogeneous regulatory approaches that can be used to improe other
regulatory systems 1hese questions are important because regulatory schemes
can signiicantly aect the eicacy o reputation systems. As an example,
consider the dierences between the job reerence and online consumer reiew
markets.
A ormer employer giing a job reerence can ace signiicant liability whether
the reerence is positie or negatie.
13
Giing unaorable reerences o ormer
employees can lead to deamation or related claims,
14
and there may be liability
or a ormer employee giing an incomplete positie reerence.
15

Lmployers may be statutorily required to proide certain objectie inormation
about ormer employees.
16
Otherwise, gien the potentially no-win liability
regime or communicating job reerences, most knowledgeable employers

12
15 U.S.C. 1681-81x.
13
"## 1resa Baldas, D '.83 )< 51)+4#*8 )>#1 O)+ '#<#1#-(#8, NA1`L L.J., Mar. 10, 2008 ,Lmployers
are inding that they are being sued no matter what course they take, whether they gie a bad
reerence, a good reerence or stay entirely silent.`,.
14
1-2 LMPLO\MLN1 SCRLLNING 2.05 ,Matthew Bender & Co. 2008, ,hereinater
LMPLO\MLN1 SCRLLNING`,.
15
Randi \. . Muroc Joint Uniied Sch. Dist., 14 Cal. 4th 1066 ,199,.
16
1hese laws are called serice letter statutes.` "## LMPLO\MLN1 SCRLLNING, 8671. note 14.
Germany has a mandatory reerence law requiring employers to urnish job reerences, but
in response German employers hae deeloped an elaborate system or coding the
reerences. Matthew \. linkin & Kenneth G. Dau-Schmidt, ")4>,-& /3# ;*74)C## '#<#1#-(#
51)+4#*, 5 AM. J. COMP. L. 38 ,2009,.

reuse to proide any subjectie recommendations o ormer employees,
positie or negatie.
1

1o curb employers` tendency towards silence, many states enacted statutory
immunities to protect employers rom lawsuits oer job reerences.
18
loweer,
the immunities hae not changed employer reticence, which has led to a irtual
collapse o the job reerence market.
19
As a result, due to mis-calibrated
regulation, the job reerence market ails to proide reliable reputational
inormation.
In contrast, the online consumer reiew system is one o the most robust
reputation systems eer. Millions o consumers reely share their subjectie
opinions about marketplace goods and serices, and consumer reiew websites
keep prolierating.
1here are seeral possible reasons why consumer reiew websites might succeed
where oline reputation systems might ail. My hypothesis, discussed in a
companion essay in this collection, is that the dierence is partially explained by
4 U.S.C. 230, passed in 1996-at the height o Internet exceptionalism-to
protect online publishers rom liability or third party content. Section 230 lets
websites collect and organize indiidual consumer reiews without worrying
about crippling legal liability or those reiews. As a result, consumer reiew
websites can motiate consumers to share their opinions and then publish those
opinions widely-as determined by marketplace mechanisms ,,%#%$ the tertiary
inisible hand,, not concerns about legal liability.
1he success o consumer reiew websites is especially noteworthy gien that
indiidual reiewers ace the same legal risks that ormer employers ace when
proiding job reerences, such as the risk o personal liability or publishing
negatie reputational inormation. Indeed, numerous indiiduals hae been
sued or posting negatie online reiews.
20
As a result, rational actors should
ind it imprudent to submit negatie reiews, yet, millions o such reiews are
published online. A number o theories might explain this discrepancy, but one
theory is especially intriguing: Mediating websites, priileged by their own
liability immunity, ind innoatie ways to get consumers oer their ears o
legal liability.

1
"## Baldas, 8671. note 13.
18
1he immunizations protect employer statements made in good aith. LMPLO\MLN1
SCRLLNING, 8671. note 14.
19
"## linkin & Dau-Schmidt, 8671. note 16.
20
"##, #%&%, \endy Dais, Z#47 '#>,#28 "7.2- D/ L#.8/ E,># L.286,/8, MLDIAPOS1 ONLINL MLDIA
DAIL\, Jan. 21, 2009,
!""#$%%???+1('&)#.*"+-.1%#>S<&-)"&.0*%DC)EG,"&-<(*+#,&0"c,&(0'<5H),"[)&'E78II
8, Agard . lill, 2010 U.S. Dist. LLXIS 35014 ,L.D. Cal. 2010,.

\hat lessons can we draw rom this comparison One possible lesson is that
reputation systems are too important to be let to the market. In other words,
the tertiary inisible hand may not ensure accurate and useul inormation, or
the costs o inaccurate inormation ,such as denying a job to a qualiied
candidate, may be too excessie. I so, extensie regulatory interention o
reputation systems may improe the marketplace.
An alternatie conclusion-and a more conincing one to me-is that the
tertiary inisible hand, aided by a powerul statutory immunity like Section 230,
works better than regulatory interention. I so, we may get better results by
deregulating reputation systems.
System Configurations
Gien the regulatory heterogeneity, I wonder i there is an ideal` regulatory
coniguration or reputation systems, especially gien the tertiary inisible hand
and its salutary eect on publisher behaior. 1wo brie examples illustrate the
choices aailable to regulators, including the option o letting the marketplace
operate unimpeded:
G0"&:Z)1&0/. A endor may hae inancial incenties to distort the low o
reputational inormation about it. 1his reputational gaming can take many
orms, including disseminating alse positie reports about the endor,
21

disseminating alse negatie reports about the endor`s competitors, or
manipulating an intermediary`s sorting or weighting algorithm to get more credit
or positie reports or reduce credit or negatie reports. Another sort o
gaming can occur when users intentionally lood a reputation system with
inaccurate negatie reports as a orm o protest.
22

Do regulators need to curb this gaming behaior, or will other orces be
adequate 1here are seeral marketplace pressures that curb gaming, including
competitors policing each other,
23
just as they do in alse adertising cases.
24
In

21
Liestyle Lit lolding, Inc. . RealSel Inc., 2:08-c-10089-PJD-RS\ ,answer,counterclaims
iled March 3, 2008,, !""#$%%???+,()<*(<C+-.1%C&<(*%G0*?(,+#'C ,alleging that Liestyle
Lit posted ake positie reiews about its own business to an online reiew website,.
22
lor example, consumers protesting the digital rights management ,DRM, in LA`s Spore
game looded Amazon`s reiew site with one-star reiews, een though many o them
actually enjoyed the game. "## Austin Modine, D*.R)- E4.83 @)+ @.648 "7)1# G'@, 1lL
RLGIS1LR, Sept. 10, 2008,
!""#$%%???+"!(,(/&*"(,+-.+>A%M;;8%;7%6;%*#.,([',1[)1)B.0[(CC(-"%. A similar
protest hit Intuit`s 1urbo1ax 2008 oer its increased prices. "## Steen Musil, D*.R)-
'#>,#2#18 "4.* 961+)9.H E## ?3.-&#8, CNL1 NL\S.COM, Dec. , 2008,
!""#$%%0(?*+-0("+-.1%89;6:6;;6[9:6;66I9M9:7M+!"1<.
23
"## Cornelius . DeLuca, 2010 \L 109928 ,D. Idaho Apr. 26, 2010, ,a marketplace endor
sued oer alleged shill online reiews posted by competitors,.

addition, the tertiary inisible hand may encourage reputation systems to
proide adequate policing` against gaming. loweer, when the tertiary
inisible hand is weak, such as with ake blog posts where search engines are the
only mediators,
25
goernment interention might be worth considering.
P&/!" .C P(#<5. A endor may wish to publicly respond to reputational
inormation published about it in an immediately adjacent ashion. Many
consumer reiew websites allow endors to comment or otherwise reply to
user-supplied reiews, but not all do. lor example, \elp initially drew
signiicant criticism rom business owners who could not eectiely reply to
negatie \elp reiews because o \elp`s architecture,
26
but \elp eentually
relented and oluntarily changed its policy.
2
As another example, Google
permitted quoted sources to reply to news articles appearing in Google News as
a way to correct the record.`
28

Regulators could require consumer reiew websites and other reputation
systems to permit an adjacent response rom the endor.
29
But such
interention may not be necessary, the tertiary inisible hand can prompt
reputation systems to oluntarily proide a reply option ,as \elp and Google
did, when they think the additional inormation helps consumers.
Undersupply of Reputational Information
1here are three primary categories o reasons why reputational inormation may
be undersupplied.

24
"##, #%&%, Lillian R. BeVier, D 56RR4# ,- /3# L.2 )< G#(#7/,)-, 8 VA. L. RLV. 1 ,1992,.
25
"## Press Release, New \ork Oice o the Attorney General, Attorney General Cuomo
Secures Settlement \ith Plastic Surgery lranchise 1hat llooded Internet \ith lalse Positie
Reiews, July 14, 2009,
!""#$%%???+)/+05+/.2%1('&)[-(0"(,%M;;7%R><5%R><56TS[;7+!"1<.
26
See Claire Cain Miller, 93# '#>,#2 ",/# Z#47 G1.28 ")*# P6/(1,#8 )< A/8 P2-, N.\. 1IMLS, Mar. 3,
2009.
2
"## Claire Cain Miller, Z#47 S,44 L#/ M68,-#88#8 '#87)-B /) S#+ '#>,#28, N.\. 1IMLS, Apr. 10,
2009.
28
"## Dan Meredith & Andy Golding, 5#187#(/,>#8 D+)6/ /3# Y#28 <1)* 5#)74# ,- /3# Y#28,
GOOGLL NL\S BLOG, Aug. , 200,
!""#$%%/../<(0(?*S<./+S<./*#."+-.1%M;;I%;8%#(,*#(-"&2(*:)S.>":0(?*:C,.1:
#(.#<(:&0+!"1<.
29
"## lrank A. Pasquale, '.-I,-&8$ '#B6(/,)-,8*$ .-B '#87)-8,+,4,/C, 54 CLLV. S1. L. RLV. 115
,2006,, lrank A. Pasquale, D8/#1,8I '#>,8,/#B: G#+./,-& . ',&3/ )< '#74C )- "#.1(3 '#864/8, 3 J. BUS.
& 1LCl. L. 61 ,2008,.

Inadequate Production Incentives
Much reputational inormation starts out as non-public ,,%#%$ priate`,
inormation in the orm o a customer`s subjectie mental impressions about
his,her interactions with the endor. 1o the extent this inormation remains
non-public, it does not help other consumers make marketplace decisions.
1hese collectie mental impressions represent a ital but potentially
underutilized social resource.
1he act that non-public inormation remains locked in consumers` heads could
represent a marketplace ailure. I the social beneit rom public reputational
inormation exceeds the priate beneit rom making it public, then
presumptiely there will be an undersupply o public reputational inormation.
I so, the goernment may need to correct this ailure by encouraging the
disclosure o reputational inormation-such as by creating a tort immunity or
sites that host that disclosure, as Section 230 does, or perhaps by going urther.
But there already may be market solutions to this problem, as eidenced by the
prolieration o online reiew websites eliciting lots o ormerly non-public
reputational inormation.
lurther, relatiely small amounts o publicly disclosed reputational inormation
might be enough to properly steer the inisible hand. lor example, the irst
consumer reiew o a product in a reputation system creates a lot o alue or
subsequent consumers, but the 1,000
th
consumer reiew o the same product
may add ery little incrementally. So een i most consumer impressions remain
non-public, perhaps mass-market products and endors still hae enough
inormation produced to keep them honest. At the same time, endors and
products in the long tail`
30
may hae inadequate non-public impressions put
into the public discourse, creating a aluable opportunity or comprehensie
reputation systems to ix the omission. loweer, reputation systems will tackle
these obscure marketplace options only when they can keep their costs low
,gien that consumer interest and traic will, by deinition, be low,, and
reputation system deregulation helps reduce both the costs o litigation as well
as responding to takedown demands.

30
Chris Anderson, 93# L)-& 9.,4, \IRLD, Oct. 2004,
!""#$%%???+?&,('+-.1%?&,('%),-!&2(%6M+6;%")&<+!"1<.

Vendor Suppression of Reputational Information
Vendors are not shy about trying to suppress unwanted consumer reiews ex
post,
31
but endors might try to suppress such reiews ex ante. lor example,
one ca owner grew so tired o negatie \elp reiews that he put a No
\elpers` sign in his ca`s windows.
32

1hat sign probably had no legal eect, but Medical Justice oers an ex ante
system to help doctors use preemptie contracts to suppress reiews by their
patients. Medical Justice proides doctors with a orm agreement that has
patients waie their rights to post online reiews o the doctor.
33
lurther, to
bypass 4 U.S.C. 230`s protectie immunity or online reputation systems that
might republish such patient reiews, the Medical Justice orm prospectiely
takes copyright ownership o any patient-authored reiews.
34
,Section 230 does
not immunize against copyright inringement,. 1his approach eectiely allows
doctors-or Medical Justice as their designee-to get reputation systems to
remoe any unwanted patient reiews simply by sending a DMCA takedown
notice.
35

Lx ante customer gag orders may be illegal. In the early 2000s, the New \ork
Attorney General challenged sotware manuacturer Network Associates` end
user license agreement, which said the customer will not publish reiews o
this product without prior consent rom Network Associates, Inc.` In
response, the New \ork Supreme Court enjoined Network Associates rom
restricting user reiews in its end user license agreement.
36
Medical Justice`s
scheme may be equally legally problematic.
lrom a policy standpoint, ex ante customer gag orders pose serious threats to
the inisible hand. I they work as intended, they stare reputation systems o
the public inormation necessary to acilitate the marketplace. 1hereore,

31
"## Lric Goldman, P-4,-# S)1B )< @)6/3 .-B A/8 A*74,(./,)-8 <)1 91.B#*.1I L.2, ,- 1RADLMARK
LA\ AND 1lLOR\: A lANDBOOK Ol CON1LMPORAR\ RLSLARCl 404 ,Graeme B.
Dinwoodie and Mark D. Janis eds., ,2008, ,discussing lopsided databases where all negatie
reiews are remoed, leaing only positie reiews,.
32
Steanie Olsen, Y) G)&8$ Z#47#18 D44)2#B, CNL1 NL\S.COM, Aug. 14, 200,
!""#$%%0(?*+-0("+-.1%89;6:6;I8T[9:7I47799:I+!"1<.
33
Lindsey 1anner, G)(/)18 "##I 0.& P1B#18 /) "/)7 5./,#-/8\ P-4,-# '#>,#28, ASSOCIA1LD PRLSS,
Mar. 3, 2009, !""#$%%???+>*)".')5+-.1%0(?*%!()<"!%M;;7:;9:;4:'.-".,:
,(2&(?*[K+!"1.
34
Michael L. Carbine, 53C8,(,.-8 Q8# ?)7C1,&3/ A-<1,-&#*#-/ 931#./ /) M4)(I 5./,#-/ './,-&8 )- /3# S#+,
AIS`S lLAL1l BUSINLSS DAIL\, Mar. 30, 2009,
!""#$%%???+)&*!()<"!+-.1%N0.?%!S';99;;7+!"1<.
35
1 U.S.C. 512,c,,3,.
36
People . Network Associates, Inc., 58 N.\.S.2d 466 ,N.\. Sup. Ct. 2003,.

regulatory eorts might be required to preent ex ante customer gag orders
rom wreaking haoc on marketplace mechanisms.
Distorted Decision-Making
from Reputational Information
Reputational inormation generally improes decision-making, but not always.
Most obiously, reputational inormation relies on the accuracy o past
inormation in predicting uture behaior, but this predictie power is not
perect.
lirst, marketplace actors are constantly changing and eoling, so past behaior
may not predict uture perormance. lor example, a person with historically
bad credit may obtain a well-paying job that puts him or her on good inancial
ooting. Or, in the corporate world, a business may be sold to a new owner
with dierent management practices. In these situations, the predictie
accuracy o past inormation is reduced.
3

Second, some past behaior may be so distracting that inormation consumers
might oerlook other inormation that has more accurate predictie power. lor
example, a past crime or bankruptcy can oerwhelm the predictie inormation
in an otherwise-unblemished track record o good perormance.
Ultimately, a consumer o inormation must make smart choices about what
inormation to consult and how much predictie weight to assign to that
inormation. Perhaps regulation can improe the marketplace`s operation by
shaping the inormation that consumers consider. lor example, i some
inormation is so highly prejudicial that it is likely to distort consumer decision-
making, the marketplace might work better i we suppress that inormation
rom the decision-maker.
38

At the same time, taking useul inormation out o the marketplace could create
its own aderse distortions o the inisible hand. 1hereore, we should tread
cautiously in suppressing certain categories o inormation.

3
?<. Note, M.B2,44, 116 lARV. L. RLV. 1845 ,2003, ,describing how companies can mask a
track record o bad perormance through corporate renaming,.
38
?<. lLD. R. LVID. 403 ,Although releant, eidence may be excluded i its probatie alue is
substantially outweighed by the danger o unair prejudice, conusion o the issues, or
misleading the jury.`,. 1his ear underlies a lrench proposal to enact a right to orget`
statute. "## Daid Reid, E1.-(# 5)-B#18 ',&3/J/)JE)1&#/ L.2, BBC CLICK, Jan. 8, 2010,
!""#$%%0(?*+SS-+-.+>A%M%!&%#,./,)11(*%-<&-A[.0<&0(%8TTIITM+*"1.

Conclusion
Although reputation` has been extensiely studied in a ariety o social science
disciplines, there has been comparatiely little attention paid to how regulation
aects the low o reputational inormation in our economy. Understanding
these dynamics would be especially aluable in light o the prolieration o
Internet-mediated reputation systems and the irresistible temptation to regulate
noel and innoatie reputation systems based on emotion, not necessarily
sound policy considerations.
305

CHAPTER 5
WHO WILL GOVERN THE NET IN 2020?
Imagining the Future of Global Internet Governance 307
Milton Mueller
Democracy in Cyberspace: Self-Governing Netizens &
a New, Global Form of Civic Virtue, Online 315
Daid R. Johnson
Whos Who in Internet Politics: A Taxonomy of
Information Technology Policy & Politics 327
Robert D. Atkinson

306 CHAPTER 5: WHO WILL GOVERN THE NET IN 2020?


Imagining the Future of
Global Internet Governance
By Milton Mueller
*

\hen discussing who ,or what, will goern the Internet in 2020, people tend to
want predictions. 1hey want authoritatie statements rom experts. 1hey want
you to tell them what ritt happen. But an honest scholar o Internet goernance
would neer attempt to meet that demand. 1he problem is not just that the
uture o Internet goernance is uncertain, subject to the inluence o many
complex ariables. 1he problem is that its uture is, literally, indeterminate.
\hile it is correct that there is an ongoing struggle oer the goernance o the
Internet, we cannot know how it will come out.
lorget about predictions and orecasts. It`s better to hae a clear conception o
how we ravt the Internet to be goerned. 1his means that we need to be able to
ivagive easible utures and to create strategies to realize them.
Let`s step back. \hy is Internet goernance an interesting problem in the irst
place \hy does contemplating the Internet`s uture require imagination and
creatiity Because there is a tension, een a contradiction, between the existing
institutions or regulating communications and inormation, and the technical
capabilities and processes o open internetworking. Lxisting institutions are
organized around territorial, hierarchical nation-states, the process o
internetworking, on the other hand, proides globalized and distributed
interoperation amongst all the elements o an increasingly powerul and
ubiquitous system o digital deices and networks.
1his technical capability puts pressure on the nation-state in ie distinct ways.
1. It globalizes the .coe o communication. Its distance-insensitie cost
structure and non-territorial addressing and routing architecture make
borderless communication the deault, any attempt to impose a
jurisdictional oerlay on its use requires additional ,costly,
interentions.
2. It acilitates a quantum jump in the .cate o communication. It massiely
enlarges our capacity or message generation, duplication, and storage.
As a programmable enironment, it industrializes inormation serices,
inormation collection, and inormation retrieal. 1he sheer olume o

!"#$%& !()##)* teaches and does research on the political economy o communication and
inormation. lis new book +,-./012 3+4 2-3-,25 -6, 78/938 :/8;-;<2 /=
;+-,0+,- 7/>,0+3+<, ,MI1 Press, 2010, proides a comprehensie oeriew o the
political and economic driers o a new global politics.

transactions and content on the Internet oten oerwhelms the capacity
o traditional goernmental processes to respond-but that same
scalability can transorm goernmental processes as well.
3. It ai.tribvte. covtrot. Combined with liberalization o the
telecommunications sector, the Internet protocols hae decentralized
and distributed participation in and authority oer networking and
ensured that the decision-making units oer network operations are not
necessarily closely aligned with political units, as they were in the days
o post, telephone and telegraph monopolies.
4. It gror. ver iv.titvtiov.. Decision-making authority oer standards and
critical Internet resources rests in the hands o a transnational network
o actors that emerged organically alongside the Internet, outside o the
nation-state system. 1hese relatiely young but maturing institutions,
such as the Internet Lngineering 1ask lorce ,IL1l,, the Internet
Corporation or Assigned Names and Numbers ,ICANN,, and
Regional Internet Address Registries ,RIRs, proide a new locus o
authority or key decisions about standards and critical resources.
5. It cbavge. tbe otit,. By conerging dierent media orms and acilitating
ully interactie communication, the Internet dramatically alters the cost
and capabilities o group action. As a result, radically new orms o
collaboration, discourse, and organization are emerging. 1his makes it
possible to mobilize new transnational policy networks and enables
new orms o goernance as a solution to some o the problems o
Internet goernance itsel.
1ransnational scope, boundless scale, distributed control, new institutions and
radical changes in collectie action capabilities-these actors are transorming
national control and soereignty oer communication and inormation policy,
setting in motion new institutional orms and new kinds o geopolitical
competition. 1he goernance o global Internetworking is thus a relatiely new
problem created by socio-technical change. 1he uture o Internet goernance
will be drien by the clash between its raw technical potential and the desire o
arious incumbent interests-most notably nation-states-to assert control
oer that potential.
\hile the Internet poses noel goernance problems, how we sole them
cannot be predicted. It depends itally on our ability to accurately diagnose the
economic, technical and political orces at work and on our ability to imagine
strategies, mechanisms and techniques that can harness those orces to do what
we want to do. 1hus, to repeat, it is better to inest our mental resources in
conceptualizing and enacting easible isions o how we ravt the Internet to be
goerned than it is to inest in making deterministic predictions.

The Pace of Change
2020 is not ery ar away. 1en years is the blink o an eye when it comes to
institutional deelopment at the global leel. Proo o this can be ound by
glancing back ten years rom our current antage point o 2010. Despite the
Internet`s reputation or rapid change, the basic issues and problems o Internet
goernance hae not changed much since 2000. \es, there has been turbulence
in the market economy, with speciic irms rising and alling. But the cast o
institutional characters that regulate or goern Internetworking was already well
in place by 2000. 1he organically deeloped Internet institutions such as the
IL1l, the Internet Society, and the Internet address registries
1
already existed.
On the other hand, the U.S. goernment and its rial nation-states were
entering the scene. ICANN, with its unilateral control by the U.S. goernment,
had already emerged as the uncomortable compromise between the a-national
Internet community` and the community o states. 1he seeds o the tensions
among the U.S., the L.U. and the BRIC
2
nations caused by U.S. pre-eminence
in that regime were already sown. 1here was already theoretical talk o cyberwar
,though this has picked up dramatically in the last ew years,. 1here were already
eorts to block and ilter Internet content, though these hae become
increasingly reined. Peer-to-peer ile sharing was already beginning to drie
copyright holders mad ,Napster was started in 2000,. \hateer change has
taken place since 2000 has been eolutionary rather than reolutionary.
Disruption or Continuity?
It is possible to identiy some aspects o the current Internet goernance regime
that could disrupt the existing eolutionary trajectory. I diide them into two
distinct categories: the geopolitical and the techno-economic.
Geopolitical Factors
-?) 0%%$: One o the most important geopolitical actors is still U.S. control o
the root o the name and numbering hierarchies. 1his control is bound up with
the issue o the singularity o those roots and the uniersal interoperability o
the Internet. lere the U.S. is pre-eminent, and along with that pre-eminence
come orms o responsibility and danger. A policy misstep or mistake can
disrupt the .tatv. qvo. \ill the U.S. inally ully priatize ICANN, or will it
internationalize` it \ill the Internet Assigned Numbers Authority ,IANA,
contract
3
be competitiely bid, or routinely reassigned to ICANN, keeping it

1
Users are assigned IP addresses by Internet serice proiders ,ISPs,, who usually obtain
larger blocks o IP addresses rom a Regional Internet Registry ,RIR,.
2
Brazil, Russia, India, and China
3
1he IANA contract is a contract between the U.S. Commerce Department that authorizes
ICANN to perorm what is commonly reerred to as the IANA unction, a bundle o
technical operations that includes the registration and allocation o IP addresses,

subordinate to the United States goernment 1he goernance issues related to
the root and control,coordination hierarchies hae intensiied as the technical
community ,usually unded by U.S. goernment contracts, has moed to
secure` the Internet by making access to and use o critical identiier resources
reliant upon cryptographic key hierarchies. 1he harder the U.S. goernment
tries to rigidiy its existing orms o institutional control oer Internet resources,
the more likely it becomes that the Internet will ragment.
<@A)*B.C*DC*): Military conlict is always a potent source o institutional
disruption. Geopolitically, the U.S. is pursuing a dangerously contradictory
agenda. On the one hand it insists on retaining pre-eminent control o Internet
standards, protocols and irtual resources and maximizing the dependence o
the rest o the world on them. On the other hand it wants to treat cyberspace as
a national asset` and deelop an oerwhelming cyber-warare and cyber-
weapons capability based on those ery same standards, protocols and
resources. But insoar as cyberspace is militarized, its status as a globalized
platorm or inormation and communication among the business and ciil
society is undermined. Contradictions abound here, and as they play out, the
chances o a structural change increase.
=*)) -*CE) "& ;&D%*FC$"%& 2)*G"H)I: 1he U.S. approach to Internet reedom
is drien as much by economics as by ideology and ethics. Due to its liberal
policies, the U.S. leads the world in the supply o Internet-based inormation
serices. O course the rest o the world will gradually catch up, but the natural
state o Internet-based inormation serices is to be transnational, accessible
anywhere in the world, and so suppliers who would challenge the Googles and
lacebooks must be transnational as well. 1he contradiction between the open
Internet and arious orms o trade protectionism in the content industries-
including the cultural protectionism that is oten disguised as support or
diersity`-could be a key drier o Internet goernance. Adocates o ciil
liberties and communication rights need to orge common ground with
adocates o ree trade and market liberalism or anything important to happen
here.
Techno-Economic Factors
J&#"H)&I)E ."*)#)II 9*%CEAC&E: A great deal o the consolidation o control
oer the Internet is contingent upon the access bottleneck. 1he ewer market
players in the Internet serice proider space, the easier it becomes or states
and state-aored monopolies to blunt and channel the potential o inormation
and communication technology. 1hus, new access technologies like unlicensed
wireless broadband become critical actors shaping the uture. I they can take

management o the Internet root serer system, and maintenance o the authoritatie root
zone ile or the domain name system.

root and disrupt current market structures around the supply o Internet access,
the arrangements or goernance and control will need to be reconsidered. \ith
greater choice o access arrangements, the less easible it becomes or
goernments to impose onerous regulatory arrangement upon consumers
through intermediaries.
4:; C&E +)$ +)($*C#"$@: A key characteristic o the Internet so ar has been
the end-to-end` principle, which put the processing intelligence or
applications and serices at the end points and made the network a relatiely
simple packet-orwarding system. Deep packet inspection ,DPI, is a new
technological capability that could lead to a wholesale departure rom that
principle. Deeloped in response to legitimate concerns about eicient
bandwidth management and the detection and interception o malware, it
increases the awareness and control o the network intermediary oer the traic
coursing through its system. 1his is a ateul shit o control. Needless to say,
there are demands to extend its capabilities to less technical orms o
interention, such as censorship, copyright protection or national security-
oriented sureillance o communications. At the same time, concerns about
priacy, network neutrality, and competition policy hae put legal and regulatory
checks upon the usage o DPI. 1his is an arena that goes to the heart o
Internet goernance in the uture.
Two Visions
1wo isions o possible utures should help to illustrate how these themes vigbt
play out, but more importantly, how I think they ovgbt and ought vot play out.
The Dark Vision
Picture a world in a long-term global recession, one that lasts the better part o
the decade we are discussing. 1here is growing conseratism-by which I mean
increasingly nationalist and ethnocentric attitudes, a growing impatience with,
and rejection o, the rigors o market liberalism, and a greater willingness to
trade reedom and innoation or security and stability. In such a scenario o
recession-drien reaction, trade barriers rise. lostility to immigration and
oshoring` grows. Internet-based communications become increasingly
conined to national spaces. 1here is blocking and iltering o content at the
national leel, the ull linkage o online identity to national identity, the licensing
o content, application and serice proiders at the national leel, the
subordination o inormation lows to the sureillance needs o national
goernments. Inrastructure proiders stop expanding and rely on national
broadband subsidy plans. As this happens, the major U.S. Internet,media
corporations succeed in minimizing competition and maximize rent-seeking in
an increasingly mature, stable market. \ith the number o players winnowed
down, these corporations will make disastrous concessions to goernments
seeking to extend their authority oer cyberspace in areas such as online identity

and identiication, online sureillance, security practices, protectionist standards
and content regulation. Some ariant o a Google-Verizon merger spawns the
A1&1 o the 21
st
century, a dominant priate sector entity with its own
commercial interests, but one whose markets and ortunes ollow the lag o
U.S. policy worldwide. Reacting to its quasi-sponsorship by the State
Department, other countries erect barriers. In this context, with national
security and a cyber- ersion o the military-industrial complex becoming the
main drier o international policy, the U.S. goernment eentually participates
in the strangling o its own progeny.
As the U.S. deelops an oerwhelming cyber-warare and cyber-weapons
capability, the rest o the world reolts. 1he U.S. prookes a cyber-cold war, or
perhaps een a short hot war` with Russia and China, and uses it to rationalize
and extend many o the controls. 1he Luropean Commission-but not
Luropean ciil society-will side with the U.S., eectiely paralyzing and
subordinating Lurope`s ability to contribute anything constructie, much less
innoatie, to the Internet goernance debates. 1he Internet world ragments
on linguistic grounds, with the Lnglish-speaking or Lnglish-dominant world
driting away rom the Chinese, Korean, Russian and Japanese societies.
The Bright Vision
It`s easy enough to describe that scenario because it seems to be the road we are
already on. It is much harder to imagine a better uture, one that is both easible
and consistent with the interests and capabilities o current actors. But let`s gie
it a try. In another work, Iè tried to describe the basic nature o what I call a
aevatiovatiea tiberati.v as the guide to the uture o Internet goernance.
4

At its core, a denationalized liberalism aors a uniersal right to receie and
impart inormation regardless o rontiers, and sees reedom to communicate
and exchange inormation as undamental, primary elements o human choice
and political and social actiity. Political institutions should seek to build upon,
not undermine or reerse, the limitless possibilities or orming new social
aggregations around digital communications. In line with its commitment to
reedom, this ideology holds a presumption in aor o networked, associatie
relations oer hierarchical relations as a mode o transnational goernance.
Goernance should emerge primarily as a byproduct o many unilateral and
bilateral decisions by its members to exchange or negotiate with other members
,or to reuse to do so,. 1his networked liberalism thus moes decisiely away
rom the dangerous, conlict-prone tendency to build political institutions
around linguistic, religious, and ethnic communities. Instead o rigid, bounded
communities that conceal domination with the pretense o homogeneity and a

4
MIL1ON MULLLLR, NL1\ORKS AND S1A1LS: 1lL GLOBAL POLI1ICS Ol IN1LRNL1
GOVLRNANCL ,MI1 Press: 2010,.

collectie will,` this liberalism oers goernance o communication and
inormation through more lexible and shiting social aggregations.
Although committed to globalism in the communicatie sector, networked
liberalism recognizes that, or the time being, people are deeply situated within
national laws and institutions regarding such basic matters as contracts,
property, crime, education, and welare. It is characterized not by absolute
hostility to national and subnational goernments as such, but rather by an
attempt to covtaiv them to those domains o law and policy suited to localized or
territorialized authority. It seeks to detach the transnational operations o
Internet inrastructure and the goernance o serices and content rom those
limited jurisdictions as much as possible, and to preent states rom ensnaring
global communications in interstate rialries and politico-military games. 1his
requires a complete detachment o Internet goernance institutions rom
nation-state institutions, and the creation o new, direct accountability
relationships or Internet institutions.
Such an ideology needs to answer tough questions about when hierarchical
exercises o power are justiied and through which instruments they are
exercised. A realistic denationalized liberalism recognizes that emergent orms
o control will arise rom globally networked communities. It recognizes that
authoritatie interentions will be needed to secure basic rights against coercie
attacks, and that network externalities or bottlenecks oer essential acilities may
create a concentrated power with coercie eect. It should also recognize the
exceptional cases where the goernance o shared resources requires binding
collectie action. Insoar as collectie goernance is necessary and unaoidable,
a denationalized liberalism stries to make Internet users and suppliers an
autonomous, global polity, with what might be called veoaevocratic rights to
representation and participation in these new global goernance institutions.
1he concept o democracy is qualiied by the realization that the speciic orm
o democratic goernance associated with the territorial nation-state cannot and
should not be directly translated into the global leel. loweer, it aoe. maintain
the basic objecties o traditional democracy: to gie all indiiduals the same
ormal rights and representational status within the institutions that goern
them so that they can presere and protect their rights as indiiduals. Such a
liberalism is not interested, howeer, in using global goernance institutions to
redistribute wealth. 1hat would require an oerarching hierarchical power that
would be almost impossible to control democratically, its mere existence would
trigger organized political competition or its leers, which would, in the current
historical context, deole into competition among preexisting political and
ethnic collectiities-the ery opposite o networked liberalism.
In short, we need to ind ways to translate classical liberal rights and reedoms
into a goernance ramework suitable or the global Internet. 1here can be no

cyber-liberty without a political moement to deine, deend, and institutionalize
indiidual rights and reedoms on a transnational scale.

Democracy in Cyberspace: Self-
Governing Netizens & a New, Global
Form of Civic Virtue, Online
By David R. Johnson
*

1he Internet can be iewed as a set o wires, wireless pipes` and serers, a set
o protocols, or as a ast array o content and applications to which these lower
layers o the stack proide access. None o those tangible and intangible things
can be goerned.` 1hey may or may not be owned or manipulated. But it is
the actions o the people inoled in creating and using the Internet that are the
proper subject o a question regarding goernance.` Viewed with respect to
the social and legal relationships among the people who are creating and using
it, and who would be goerned,` the Internet is a complex system-so making
accurate predictions about its uture state is impossible.
But it is possible to answer the question: \ho covta and who .bovta goern the
Internet in 2020` My answer is, in a word: netizens-the global polity o those
who collaborate online, seek to use the new aordances o the Internet to
improe the world, and care about protecting an Internet architecture that
acilitates new orms o ciic irtue.
The Internet Governance Debate
1he debate about Internet Goernance` has continued or more than iteen
years and settled into an unsatisying rut. 1he established trope is that early
isionaries ,e.g., John Perry Barlow
1
, claimed that cyberspace was a new realm o
reedom, poised to escape rom regulation by local goernments. 1hen, later
realists` ,e.g., Jack Goldsmith and 1im \u
2
, discoered that soereign
goernments indeed had ways to regulate online speech and een use the
Internet or sureillance and tyranny. Larly idealists enisioned the Internet
Corporation or Assigned Names and Numbers ,charged with setting policy or

Daid Johnson joined New \ork Law School`s aculty in spring 2004 as a isiting proessor
o law. le is a aculty member o the Institute or Inormation Law and Policy.
1
ee John Perry Barlow, . Dectaratiov of tbe vaeevaevce of C,ber.ace ,leb. 8, 1996,,
?$$KI5LLK*%M)H$IN)DDN%*OLPAC*#%QL4)H#C*C$"%&B="&C#N?$F# |hereinater . Dectaratiov of
tbe vaeevaevce of C,ber.ace|, John Perry Barlow, Dectarivg vaeevaevce, 4.06 !RD 121-22
,June 1996,, araitabte at
?$$K5LLQQQNQ"*)ENH%FLQ"*)ELC*H?"G)LRNSTL"&E)K)&E)&H)N?$F#.
2
ee JACK GOLDSMI1l & 1IM \U, \lO CON1ROLS 1lL IN1LRNL1: ILLUSIONS Ol A
BORDLRLLSS \ORLD ,2006, |hereinater \lO CON1ROLS 1lL IN1LRNL1|.

the domain name system and allocating blocks o IP numbers,
3
as a new
democratic global institution constrained by consensus reached in a global
community. Later, we obsere an expensie bureaucracy imposing complex
regulations-as one wag has put it: recapitulating the lCC and doing it
badly.`
4
Internet Goernance as an empowering and liberating democratic ideal
is a total ailure-or so it would seem.
1his debate has missed a undamental point by asking the wrong question. 1he
key question is not \ho will goern the Internet` Instead, it is \ill the
global Internet aect the way in which we ,the global polity, goern ourseles`
One way or another, we act through goernments, NGOs, priate corporations
and many other types o groups to collectiely set the rules by which we lie our
lies. \e all strie or a world in which our own choices determine how we are
goerned. So the more salient way to put the real question at issue is: Can the
Internet make society more democratic`
The Vision of the Internets Founders
1he ounders o the Internet ,technologists, adocates, policy makers, and
isionaries, saw its democratic potential. 1hey were not ,mostly, seeking to
create a lawless rontier.` 1hey were instead seizing a moment o lexibility
during which new modes o association or community improement might
lourish. 1hey opposed rigid regulation and unaccountable power, o course.
But they also aored collaboratie decision-making to establish rules, group
eort to write empowering and constraining code, and respectul deliberation to
orge new norms. 1hey aored ciility and ciic irtue ,good netizenship, as
much as indiidual liberty.
\e`re not just talking about the ounding technologists. Many indiiduals,
nonproit organizations and companies came together, in the late 1980s and
throughout the 1990s, to create a technology and policy ramework to enable a
democratic Internet: open access, limitations on intermediaries` powers and
liabilities, priacy protections or communications, limitations on centralized
leers o power ,like the control oer the domain name system, and, at least in
the United States, establishment o strong lirst Amendment rights ,in contrast
to regulation o the Internet as a orm o mass media,. 1o a considerable

3
ICANN is generally accepted as the authority or decisions about what top leel domains
will be placed in the root zone ile.` By establishing contractual conditions in connection
with such additions, it can establish rules that low down onto registries, registrars and
registrants. lor example, this capability has been used to require registrants in generic 1op
Leel Domains to submit to a Uniorm Dispute Resolution Policy` that decides and takes
action on disputes about cybersquatting.`
4
larold leld, quoted in Jonathan \einberg, C.^^, vtervet tabitit,, ava êr 1o eret
Dovaiv. 1 n. 1, ?$$K5LLDCH(#$@N#CQNQC@&)N)E(L.)"&A)*OL"HC&&)$HNKED.

degree, rom that perspectie, the Internet`s ounders hae, up to this point,
succeeded.
I want to re-emphasize that, while the Internet relies on wires and protocols, it
most undamentally consists o connections among people. Local goernments
may control who has the right to proide access. Standard setting bodies may
hae some say in what protocols become widely adopted. Law plays a role-and
the law o local goernments does constrain the actions o people oer whom
they can assert jurisdiction. But the goernance` o the Internet is
undamentally a question about how we all constrain the manner in which we
do whateer it is we do in groups online, including establishing new structures
o society, new orms o social organization, and new roles and rules that
incentiize our eorts and ocus our minds. No goernment could een hope
to make the rule set or a global web o relationships inoling billions o
people interacting in complex and eer eoling ways. 1he goernance o the
social layer o the Internet will be, perorce, decentralized.
The Democratic Nature of the Internet
1hus, contrary to Larry Lessig`s suggestion in Coae,
5
I submit that the Internet
o today has a nature: It is inherently democratic. Not ineitably so, in the sense
that any global communications network would necessarily be democratic. And
not necessarily so in the uture. But historically so and by design-in the sense
that tbi. Internet, the one we hae and the one that scaled globally in no time,
was successul precisely because it was open, decentralized, tolerant o
innoation and disagreement, oluntary, and empowering o anyone who cared
to use it to join with others to improe the world. Lery time we address an
email, or establish a blog, or agree` to some terms o serice,` we are creating
the rules or our online society.
lrom \ikipedia to PatientsLikeMe,
6
we are continuously learning how to use
the Internet to come together to share knowledge, improe education, sole
health care problems, and proide charitable assistance to those in need around
the world. \e use the Internet to participate in local politics and explore new
ways to make global society energy eicient. In these and countless other ways,
the Internet is an engine o democratic ciic irtue.

5
ee LA\RLNCL LLSSIG, CODL: AND O1lLR LA\S Ol C\BLRSPACL, VLRSION 2.0 ,Basic Books
2006,.
6
lor example, PatientsLikeMe allows users to create proiles and then share, interact, and
learn rom the experience o other users on health and wellness issues.

Self-Governance Online
As noted, and sometimes in deiance o repressie regulations, we choose many
o the rules under which these collaborations occur simply by logging in to one
website or platorm rather than another. \e can rise up together in protest
when a site like lacebook changes its terms o serice` and priacy policy` in
ways we don`t like. And we increasingly accept obligations to spend our
attention and eort in support o groups we ind online with whom we share a
persistent purpose or goal.
Len an email listser inoles social duties ,not always honored, o course, to
ellow participants. Open source eorts hae eoled complex, hierarchical, yet
open and democratic ,or meritocratic, sel-goernance structures. \e are
beginning to understand that, at least or those things that can happen online,
the choices we make about which groups to join hae as much impact on
setting the terms o our relationships with others than any goernmental laws or
regulations.
Goldsmith and \u and other anti-exceptionalists` hae cited the \ahoo!
lrance case, Australia`s imposition o its libel laws on a New Jersey-based
publisher, and Italy`s coniction o Google executies or the proposition that
the Net cannot escape rom the real world` goernance o soereign states.

1heir examples, rather than making a case or a bordered Internet, in act proe
the opposite: A world in which eery local soereign seeks to control the
actiities o netizens beyond its borders iolates the true meaning o sel-
goernance and democratic soereignty.
Attention Governance
& Global Civic Virtue
Len though goernments still hae the guns and hae not yet uniormly agreed
to deer to sel-goerning online groups, \e the Netizens` are still mostly in
control o what happens online. 1he eeryday actions o millions o bloggers
and tweeters ,and re-tweeters, and senders o emails and instant messages draw
the attention o the entire world to new and interesting ,though not always
important, deelopments eery day. 1he distributed collaboration arising rom
ratings, rankings and reiews disclose and shame, or shine a lattering light on,
eery action o eery author, seller, politician, organization and anyone else who
wields any orm o power. \e are learning to use the Internet to engage in a
decentralized orm o democracy that might be called attention goernance.`
Democracy is about decentralization and equalization o power-particularly
the power to inluence the rules under which we lie our lies together. 1his

\lO CON1ROLS 1lL IN1LRNL1, .vra note 2.


requires the absence o centralized, unaccountable power-tyranny-whether
exercised by goernment or corporations. It also requires that indiiduals
participate in collectie action and adopt a rame o mind that asks how to
improe society rather than only how best to achiee their own priate goals.
1hat rame o mind is called ciic irtue.` It creates a eedback loop-showing
us all ourseles in a mirror, thereby enhancing our ethical standards or both
indiidual actions and the actions we take together in organizations and groups
,including ia the global corporations in which we inest, sere as employees or
participate as customers,. Perhaps the single most powerul contribution to the
soereignty o the people made by the Internet is its ability to direct our
collectie attention. At one time, mass media held that power. Now we all do, in
potentially equal measure. \hile online anonymity may ,sometimes
unortunately, gie us the power to act without disclosing our identity, the
Internet simultaneously makes it irtually impossible or groups o people to act
together without being conronted with the consequences o their actions and
the iews o others regarding the moral and social alue ,or lack thereo, o
those consequences.
Democracy is not just about how we organize political campaigns or
goernmental institutions. It is about how we sel-organize all aspects o
society-and what we can do in collaboration with others to improe the world.
It is not just about reedom rom arbitrary control by goernment ,or corporate
tyrants,-it is, rather, about the myriad ways in which we come together to
construct society. 1he Internet has had a proound impact on political
campaigns by making it much easier or indiiduals to contribute small amounts
and get inoled in local actiities. And netizens could and should use their
newound collectie oice to instruct their local goernments to protect the
Internet and its new reedoms, rather than using it to restrict our reedom. But
the Internet has done een more to decentralize decision-making-about how
we spend our attention and eort and how we organize our collaboratie
eorts. 1oday, anyone can orm a purposeul group online on lacebook,
\ahoo! Groups, Google Groups, 1witter, etc. and attract adherents to a cause
or een start a new organization. \hen we can more easily act together, we
become more powerul and more ree-we enjoy what 1ocqueille would hae
called a new equality o condition.`
8

1he interesting thing about attention is that, as long as you are awake, you hae
to spend it! It is a non-renewable resource. \ou can ritter it away on the
entertainments o teleision or tweeting. \ou can gie it away to an employer
whose goals and ethics don`t match your own. Or, ater the basic necessities o
lie hae been arranged or, you can apply a higher standard. 1hrough the

8
ALLXIS DL 1OCQULVILLL, DLMOCRAC\ IN AMLRICA ,1835,.

Internet, you can join with others in groups that try to make the world a better
place, and talk together about what that means.
9

Good Netizenship
Goerning the Internet well undamentally entails goerning ourseles-making
sure that more o our time, attention and eort is spent in roles that are deined
in relation to social organizations and purposie groups that make society more
productie, congruent, ethical, and, yes, interesting, complex and empowering
or eeryone. It inoles deending the new ciic religion o the Internet,
including preseration o indiidual choice and deerence to the sel-goernance
o online groups. It is no light duty to be a good netizen.
Goldsmith and \u hae shown that open communications ia the Internet can
indeed be shut down by goernments.
10
But they don`t say anything about how
to presere and enhance democracy in the context o a global Internet. 1hey
postulate the extension o local and state power onto people who hae no
opportunity to participate in making these policies. But they ail to take account
o the possibility that once \e the Netizens understand the threat, we could
reuse to allow that to happen.
low could we netizens preent the tyranny o local goernments or o
corporate intermediaries with a new kind o power generated by network
eects It may be a struggle at times. But we use our minds online-we direct
our attention and support to groups we alue. It takes a ery seriously repressie
goernmental regime to regulate minds rather than behaior. And not een
goernments, much less corporations, can stand oreer in opposition to what
large segments o the people they regulate tbiv/-especially when they are
thinking, and talking, together. As Victor lugo amously remarked, One
resists the inasion o armies, one simply cannot resist the inasion o ideas.`
11

Len those who purport to want to presere ciil ciic dialogue and sel-
goernance by the people sometimes suggest that the Internet has broken our
existing ,U.S., democratic institutions-polarizing political actions, eliminating
the possibility o political compromise, ostering hate speech and inciting the

9
As 1ocqueille obsered about America in the 1830s: v tbeir otiticat a..ociatiov. tbe .vericav.,
of att covaitiov., viva., ava age., aait, acqvire a geverat ta.te for a..ociatiov ava gror accv.tovea to tbe v.e
of it. 1here they meet together in large numbers, they conerse, they listen to one another,
and they are mutually stimulated to all sorts o undertakings. 1hey aterwards trav.fer to cirit
tife tbe votiov. tbe, bare tbv. acqvirea ava va/e tbev .vb.errievt to a tbov.ava vro.e.. 1hus it i. b, tbe
ev;o,vevt of a aavgerov. freeaov tbat tbe .vericav. tearv tbe art of revaerivg tbe aavger. of freeaov te..
forviaabte.` a. ,emphasis added,.
10
\lO CON1ROLS 1lL IN1LRNL1, .vra note 2.
11
VIC1OR lUGO, 1lL lIS1OR\ Ol A CRIML ,18,.

mob.
12
Len those who aor ciic irtue suggest that the Internet has led to a
retreat into indiidualism, mindless narcissism, pornography, intellectual
distraction and worse.
13

1he democratic potential o the Internet is under threat-as democracy always
is. Some local soereigns attempt to limit its reedoms. People lacking the
requisite ciic irtue may lapse into sel-indulgent indiidualism. \e might all
decide to lie under the beneolent dictatorship o search engines and online
app stores that make our lies a little more conenient and secure. \e might all
decide it is too much trouble to help our ellow netizens in oreign countries
who are ighting repressie local goernments. Large corporations with
monopolies born o network eects might gain enough power to become a new
aristocracy, purporting to beneit the people, but ruling as they please and giing
priority to proit. Bad actors might turn this new communications medium into
a social nightmare. So any theory o the democratic potential ,and actual
achieement, o the Internet must include a iew regarding how we can all use
the Internet itsel to presere and enhance gains achieed up till now in popular
soereignty and ciic collaboration.
The Trajectory of Freedom
I hae such a theory-one deried rom relecting on 1ocqueille`s iews
regarding the new democracy that he discoered in the America o 1830: 1be
vtervet e.tabti.be. a ver eqvatit, of covaitiov ava evabte. v. to eerci.e tibert, to forv
a..ociatiov. to vr.ve ver ciric, .ociat, ava cvttvrat goat.. Such a world can produce
wonders simply because we become more powerul when we act together in
groups. Moreoer, to paraphrase Olier \endell lolmes, Man`s mind, once
stretched to a new democratic practice, neer regains its original dimensions.`
14

1he theory, then, is that haing discoered and exercised new ways to improe
the world, whateer they mean by improe,` netizens will collaborate in
myriad ways to protect their newound powers.
1he actual state o society may, o course, periodically regress. Some groups will
adopt deinitions o improement` that are so intrusie upon and
unacceptable to other groups that goernmental and corporate powers will be
rightly inited in to constrain such non-congruent actions. ,lor example, almost
eeryone agrees that the Internet should not proide a sae haen or child

12
ee, e.g., CASS SUNS1LIN, RLPUBLIC.COM 2.0 ,200,.
13
ee, e.g., NIClOLAS CARR, 1lL SlALLO\S: \lA1 1lL IN1LRNL1 IS DOING 1O OUR BRAINS
,2010,.
14
1he original quotation is as ollows: Man`s mind, once stretched by a new idea, neer
regains its original dimensions.` Olier \endell lolmes, qvotea iv l. JACKSON BRO\N, JR., A
lA1lLR`S BOOK Ol \ISDOM ,1989,.

pornography or terrorism and goernmental powers will need to be used to
address these problems.,
But the trajectory o reedom and een ciic irtue has been, in broad terms,
oer time, constantly upward-because eeryone who gets a chance to
experience an increased leel o democratic sel-goernment-a new equality
o condition`-a new kind o power that comes rom the ability to direct and
control one`s own attention and combine one`s eorts with those o others-
comes to share a desire to hae a oice in shaping the world or the better ,een
when we don`t all agree on what better` means,. And eeryone has now tasted
an empowering opportunity to join with others, online, to do so.
Acting together, the ounding netizens created a global network, and thus,
ineitably, a global economy, society and politics. 1he isionary ounders o the
Internet did not seek to liberate selish indiidualism, rontier justice based on
orce, or mere wilderness escape. 1hey were ciically irtuous themseles and
oresaw the creation o great schools and libraries, social serices, cultural
enues, and eerything else a prosperous and democratic global township might
want, online.
15
Perhaps they were nae, a bit too optimistic that eeryone else
shared their ciility. Perhaps they assumed that most online groups would make
rules and take actions designed to beneit those who were aected by those
rules and actions. Conronted with criminals or tyrants, these Internet optimists
would be ,and are, as quick as anyone to call or a rule o law.` But they can
now enision a law` consisting in part o globally applicable rule sets and
globally accessible sel-goerning organizations that exist only because netizens
hae deoted their time, attention and eort to support or shun new online
institutions.
Copyright law won`t disappear, but we now also hae Creatie Commons. Laws
against spam will surie, but we also hae sotware ilters. Local content
regulation will persist, but we now hae proxy serers. Banks will be regulated,
but online currencies can also lourish. Goernments will still regulate and tax
the shipment o physical goods, but most long ago gae up trying to establish
custom houses at their irtual border. Lery netizen is still a citizen, subject to
local regulation. But, increasingly, we can trael online to irtual places that hae
rules no local legislature would adopt. Land use in Second Lie will not become
a subject o any real world goernment`s zoning laws. 1opic moderation in an
online discussion group is not likely to become a matter o local regulation.

15
ee, e.g., lO\ARD RlLINGOLD, 1lL VIR1UAL COMMUNI1\: lOMLS1LADING ON 1lL
LLLC1RONIC lRON1ILR ,1993,, lRLD 1URNLR, lROM COUN1LR CUL1URL 1O
C\BLRCUL1URL: S1L\ARD BRAND, 1lL \lOLL LAR1l NL1\ORK, AND 1lL RISL Ol
DIGI1AL U1OPIANISM ,2006,, KA1IL lAlNLR & MA11lL\ L\ON, \lLRL \IZARDS S1A\ UP
LA1L: 1lL ORIGINS Ol 1lL IN1LRNL1 ,1996,.

Online schools will establish their own rules or participation in the classes they
oer.
\here we congregate online, most o the releant rules will originate with our
shared support o the manner in which the proprietor ,the owner o the serer,
the writer o constraining code, a moderator, goerns` that online space.
Goernments will, in general, deer to online spaces that are mostly minding
their own business, rather than inlicting harms on outsiders. 1hey hae enough
to do in securing our physical saety. So congruent rule sets oluntarily
adopted` by willing users` will become most o the applicable law` o online
lie. 1hat new law will be undamentally democratic in character, not because
we elected representaties to a global legislature, but because we all hae new
powers to decide where to go online and to persuade others to join us.
Because global scale and interconnection make it easier to ind each other,
aluable online groups can exist een i there are ery ew who share the
group`s goals and interests. By the same token, online sites that no one isits
lose social salience. An online source o destructie code or spam may still
intrude upon our inite attention, but almost eeryone agrees that we can and
should get better at constraining the actions o those who use the Internet to
inlict that kind o harm on others. ISPs proide centralized ilters but these
depend in part on user actions to lag spam and malicious code. 1his same
dependence on our collectie attention happens to apply to goernments and
corporations, the ery real legal ictions` that we call into existence by means
o a shared act o imagination. 1hey thrie only insoar as we allocate our own
time, eort and attention to acilitate their purposes. I we reuse to play along,
goernments and corporations are doomed to lose ,legitimate, power and
cannot ultimately impose their will on an unwilling global polity.
In short, there was neer any possibility, or dream, that global online society
would be a society without some amount o order. Len the most idealistic
Internet Lxceptionalists, like John Perry Barlow, neer denied that there might
be problems in Cyberspace that needed to be soled, only that \e in
Cyberspace` would sole them by orming our own Social Contract.`
16
1he
questions hae always been: low much order \here would it come rom
And, i it came rom the online community, how closely would its mechanisms
approximate the democratic ideal o giing eeryone at least a potentially equal
say in what particular kind o order ,rules, norms, incenties, roles, were to be
established \ho gets to tell who else what they can and cannot do \ill the
Internet presere, or een enhance, the soereignty o the people 1hose were
the key questions rom the Internet`s ery beginning.

16
See . Dectaratiov of tbe vaeevaevce of C,ber.ace, .vra note 1.

\u, Goldsmith and others triumphantly declare the death o Internet
Lxceptionalism based on the reality that goernments seek to regulate the
Internet-and succeed in doing so to a much greater extent than some Internet
Lxceptionalists might hae imagined possible. But this does not mean the
Internet does not present prooundly new opportunities or sel-goernment-a
new potential orm o democracy. It is up to us to seize that opportunity. I we
do so, the Internet will hae proed itsel as exceptional as its ounders hoped.
Cosmopolitan Pluralism
Because people hae many dierent alues ,ideas about the social good,, the
global online society will need to be pluralistic, cosmopolitan, and tolerant o
dierse coexisting groups. As the Internet`s ounding netizens urged, online
society could and should be based on the moral norm that all groups should be
conseratie in what they send, liberal in what they accept.`
1
Aboe all, they
imagined ,correctly, in my iew, that such a global online society would become
eer more complex-proiding increasingly dierse roles or people to play
while, simultaneously, presering connections ,causal and communicatie,
among all its parts and the whole-thereby creating ast new wealth ,o all
types, or all to share. All wealth ultimately comes rom trade. And trade
requires two people who alue whateer they hae to exchange dierently-so
that the bargain makes them both better o.
18
\e veea our dierences.
Democratic Internet goernance can presere them, by allowing us to tolerate
dierse groups, rather than seeking to impose a single rule set on eeryone.
I you think democracy is about oting, then the apogee o democratization is
when eeryone has an equal ote. 1hat just doesn`t scale globally. \e will
ote` online with our clicks, not with ballots. I you think democracy is about
deliberation and discussion, then the ideal would seem to be a continuous global
town meeting, with eeryone getting an equal turn at the microphone. Ptea.e,
.are v.!
I, howeer, you think that democracy is about equalization o ,potential, power
to hae an inluence on how society is structured, on how we will improe the
world, then you hae to ask: \hat is it that we all hae in equal measure, the
deployment o which cav shape our world and its rules. 1he answer is attention
and eort. 1hat is why attention goernance is inherently democratic. 1hat is

1
1he Internet Society, RlC 46, 1be 1ao of 1: . ôrice`. Cviae to tbe vtervet vgiveerivg
1a./ orce 6 ,Sept. 2006, ,quoting Jon Postel,, araitabte at
?$$K5LL$%%#IN")$DN%*OLKEDL*DHRTUUNKED.
18
ee geveratt,, LRIC D. BLINlOCKLR, 1lL ORIGIN Ol \LAL1l: LVOLU1ION, COMPLLXI1\, AND
1lL RADICAL RLMAKING Ol LCONOMICS ,2006,, DAVID \ARSl, KNO\LLDGL AND 1lL
\LAL1l Ol NA1IONS: A S1OR\ Ol LCONOMIC DISCOVLR\ ,2006,.

how the Internet can make sel-goernance real in the context o a new global
society o netizens.
Lery social organization with power ,including goernments and corporations,
depends critically on a collectie act o imagination. \e create our social
institutions by going along together with the idea that they exist, adopting roles
that constrain our time and attention in the serice o their goals and inesting
our attention and eort in ways that urther empower them. 1he Internet is,
centrally, a way or us to deploy our attention and eort together, in a context
in which we can see the resulting eects and correct or constrain social
organizations that don`t share our alues. It makes us all citizens ,and, indeed,
global netizens, in a new way. It increases our power and, by doing so, our
responsibilities.
A New Sovereignty of the People
Internet goernance will not be about oting, or complex goernmental
regulations, or een treaties among states. It will be more mundane, more
perasie, and more prooundly important than that. As long as some states
create haens or reedom ,whether the lirst Amendment and Section 230 o
the Communications Decency Act
19
in the United States or the Icelandic
protection o \ikiLeaks
20
,, netizens will ind ways to exercise those reedoms.
Insoar as some states create haens or wrongdoers ,think Nigerian phishers,
Russian botnets,, goernments, corporations and indiiduals will all respond, in
their own ways, to aoid or suppress such eils. It would be olly to suggest that
no organizations ,and, thereore, people in organizational roles, will be more
powerul than others, or that no one will use power or eil so widely
condemned that it cannot and should not be tolerated.` But no collectie
action can persist online oer the long term i it is not tolerated` by those who
decide where to direct their attention, what products to buy, what 1erms o
Serice to accept, what jobs to take, what companies to inest in, and what local
politicians to send packing.
America was ounded on the idea o soereignty o the people. 1he global
Internet gies the people more tools to exercise that soereignty and greater
isibility on when and how and where to do so. Good netizenship isn`t
eortless. Ciic irtue requires, well, irtue. It is about responsibilities, not
rights. \hether the Internet will realize its democratic potential will ultimately
depend, o course, on the character o the people-the new global polity.

19
4 U.S.C. 230 ,proiding liability immunity or proiders and users o an interactie
computer serice` who publish third party inormation,.
20
ee Robert Mackey, 1ictor, for !i/iea/. iv cetava`. Partiavevt, 1be eae tog, N.\. 1IMLS, June
1, 2010, araitabte at ?$$K5LL$?)#)E)NA#%OIN&@$"F)INH%FLVSWSLSTLWULG"H$%*@BD%*B
Q"X"#)CXIB"&B"H)#C&EIBKC*#"CF)&$L.

I`m optimistic about the prospects or enhanced sel-goernance by a global
polity, or one simple reason: Most o us, whateer our nationality, ravt to be
empowered. 1o communicate. 1o associate to make the world better or
ourseles, our children and eeryone else. \eè had a taste o the new
democracy o the Internet. \e`ll neer willingly, or or long, go back. \e
certainly .bovtav`t turn away rom this new opportunity, the Internet ounders`
shared dream o a more empowering, democratic, global society. I we
remember the democratic isions o the ounders, and commit to continue to
act as good netizens, then we could and should att goern the Internet,`
together, in 2020-and beyond.

Whos Who in Internet Politics:
A Taxonomy of Information
Technology Policy & Politics
By Robert D. Atkinson
*

\here`s the Internet in the United States going to be in a decade Gien the
important role o public policy in shaping a host o Internet issues, one way to
answer this question is to understand the political constellation that now shapes
U.S. and-to some extent-international Internet policy.
Debates hae erupted oer myriad inormation technology ,I1, issues such as
copyright protection, priacy, open source sotware procurement, cybersecurity,
Internet taxation, media ownership, Internet goernance, electronic oting,
broadband deployment and adoption, anti-trust, spectrum reorm, net
neutrality, Internet censorship, and equality o access. 1hese issues raise amiliar
legal and political questions in some unamiliar contexts, and hae gien rise to
a liely, increasingly shrill, and important digital politics. 1oday, interest groups
o all kinds, including a host o single-issue adocacy organizations, routinely
weigh in on a range o Internet and digital economy issues. Vexing policy
conundrums arise constantly, with each new business model and Internet
innoation creating a new wrinkle in the abric o the debate.
low we resole these issues will hae important implications or what the
Internet o 2020 looks like. 1he debate oer I1 policy issues does not take
place in a acuum or only in the corridors o Congress. lrom think tanks to
trade associations to single-issue adocacy groups, a prolieration o
organizations ights to shape digital policy debates. 1his essay is a ield guide to
help the reader understand the politics o I1.
1
It describes the major groups o
players in the I1 policy debate and discusses how they dier along two key

Robert D. Atkinson is the ounder and president o the ;&D%*FC$"%& -)H?&%#%O@ C&E
;&&%GC$"%& =%(&EC$"%&, a \ashington, DC-based technology policy think tank. le is also
author o the State New Lconomy Index series and the book, 1lL PAS1 AND lU1URL Ol
AMLRICA`S LCONOM\: LONG \AVLS Ol INNOVA1ION 1lA1 PO\LR C\CLLS Ol GRO\1l
,Ldward Llgar, 2005,.
1
lor other useul attempts at creating Internet policy typologies, .ee Adam 1hierer & Berin
Szoka, C,beribertariavi.v: 1be Ca.e for Reat vtervet reeaov, 1lL 1LClNOLOG\ LIBLRA1ION
lRON1, Aug. 12, 2009, ?$$K5LL$)H?#"A)*C$"%&NH%FLVSSYLSZLWVLH@A)*B#"A)*$C*"C&"IFB
$?)BHCI)BD%*B*)C#B"&$)*&)$BD*))E%FL, and Adam 1hierer, .re Yov av vtervet Otivi.t or
Pe..ivi.t. 1be Creat Debate orer 1ecbvotog,`. vact ov ociet,, 1lL 1LClNOLOG\ LIBLRA1ION
lRON1, Jan. 31, 2010, ?$$K5LL$)H?#"A)*C$"%&NH%FLVSWSLSWL[WLC*)B@%(BC&B"&$)*&)$B
%K$"F"I$B%*BK)II"F"I$B$?)BO*)C$BE)AC$)B%G)*B$)H?&%#%O@\,V\ZS\YYIB"FKCH$B%&B
I%H")$@.

dimensions shaping policy: indiidual empowerment s. societal beneit, and
laissez-aire s. goernment regulation. It then uses our timely and important
policy cases ,priacy, taxation, copyright protection, and net neutrality, to
illuminate how these politics play out today in the United States. \hile primarily
ocused on American digital politics, this ramework is not entirely unique to
the United States.
The Major Players
1he primary players in the I1 policy debate all into eight basic groups:
1. ;&$)*&)$ ,]H)K$"%&C#"I$I: 1hese Netizens` beliee that they launched the
Internet reolution. 1ypiied by groups such as the lree Sotware
loundation and the Llectronic lrontier loundation, and dedicated readers
o \ired magazine, they beliee inormation wants to be ree`
2
and that all
sotware should be open-source. 1hey think technology itsel can sole
many problems that it might create ,i users are only smart enough to
program sotware to protect themseles,, and that cyberspace should be
goerned by the inormally enorced social mores ,i.e., netiquette`, that
eoled among early users. Like John Perry Barlow in his 1996 4)H#C*C$"%&
%D ;&E)K)&E)&H) %D <@A)*IKCH),
3
they deplore both goernment
inolement in the Internet and its widespread commercialization. In their
iew, anyone who suggests that society, through its legitimately elected
goernment leaders, might hae a role to play in shaping the Internet,
including deending copyright, just doesn`t get it.` Internet exceptionalists
beliee the Internet should be goerned by its users. Araid your priacy is
being iolated 1echnologically-empowered users are the best solution, as
they set their \eb browser to reject cookies, use anonymizer tools and
encrypt their web traic. \orried about the recording industry losing
money rom Internet piracy Lncourage artists to ind a new business
model, like selling 1-shirts and putting on more concerts. \orried oer
lackluster I1 industry competitieness in the U.S. Don`t make waes,
Goernment interention generally makes things worse. Ater all, Silicon
Valley didn`t need \ashington to get where it is.
2. 2%H"C# ,&O"&))*I: 1hese liberals beliee the Internet is empowering but
they worry that its growth is haing unintended and sometimes dire
consequences or society-whether they inoke the so-called Digital
Diide ,between the wired` and the unwired`, the purported loss o
priacy, net neutrality, or concern that corporations are controlling the use

2
Stewart Brand, speaking at the irst lacker`s Conerence, 1984. Roger Clarke, vforvatiov
!avt. to be ree, ?$$K5LLQQQN*%O)*H#C*X)NH%FL;;L;.$A=N?$F#.
3
John Perry Barlow, A Declaration o the Independence o Cyberspace, leb. 8, 1996,
?$$KI5LLK*%M)H$IN)DDN%*OLPAC*#%QL4)H#C*C$"%&B="&C#N?$F#.

o digital content. 1hey mistrust both goernment and corporations, the
latter especially-particularly large telecommunications companies and
Internet companies making money rom the use o consumer data ,to,
ironically, proide ree content and serices,. A large array o groups and
indiiduals carry this mantle, including the Benton loundation, Center or
Democracy and 1echnology ,on some issues,, Center or Digital
Democracy, Ciil Rights lorum on Communication Policy, Consumer
Project on 1echnology, Llectronic Priacy Inormation Center, lree Press,
Media Access Project, and Public Knowledge, and scholars such as
Columbia`s 1im \u and most o those hanging their hats at larard`s
Berkman Center ,among them, Larry Lessig and \ochai Benkler,. Social
engineers tend to beliee the Internet should sere mainly as an educational
and communications tool. 1hey ear that its empowering capabilities will be
taken away by powerul multinational corporations and statist goernments
that will reshape it to sere their own narrow purposes ,either to steal our
priacy, limit our reedom on the Internet, spy on us, or all three,. As such,
they minimize the role o I1 as an economic engine, and ocus more on the
impact o I1 on social issues, such as priacy, community, access to
inormation and content, and ciil liberties.
3. =*)) !C*X)$)*I: 1his group iews the digital reolution as the great third
wae o economic innoation in human history ,ater the agricultural and
industrial reolutions,. I1 reduces transaction costs and acilitates the
application o markets to many more areas o human actiity. lree
marketers enision a dramatically reduced role or goernment as the
Internet empowers people, liberates entrepreneurs, and enables markets.
Inluenced by groups such as the Cato Institute, the Mercatus Center, the
Paciic Research Institute, the Phoenix Center, 1he Progress & lreedom
loundation, and the 1echnology Policy Institute, they consider the
emergence o the Internet as a ehicle or commerce ,e.g., exchanging
goods, serices, and inormation in the marketplace, and a liberating and
progressie orce. 1hey are skeptical o the need or goernment
inolement, een goernment partnering with industry to more rapidly
digitize the economy.
4. !%E)*C$)I: 1his group is staunchly and unabashedly pro-I1, seeing it as
this era`s driing orce or both economic growth and social progress.
\hile they iew the Internet as a unique deelopment to which old rules
and laws may not apply, they beliee appropriate guidelines must be
deeloped i it is to reach its ull potential. Likewise, they argue that while
rules and regulations should not aor bricks-and-mortar companies ,see
48 below, oer Internet ones, neither should they aor Internet companies
oer bricks-and-mortars. Moreoer, they argue that while goernment
should do no harm` to limit I1 innoations, it should also actiely do
good` by adopting policies to promote digital transormation in areas such

as broadband, the smart electric grid, health I1, intelligent transportation
systems, mobile payments, digital signatures, and others. Lxamples o
moderates include the Center or Adanced Studies in Science and
1echnology Policy, the Center or Strategic and International Studies, the
Inormation 1echnology and Innoation loundation ,I1Il,, and the
Stilwell Center.
5. !%*C# <%&I)*GC$"G)I: 1his group sees the Internet as a dangerous place, a
irtual den o iniquity, populated by pornographers, gamblers, child
molesters, terrorists, and other degenerates. Unlike the ree marketers, the
moral conseraties hae no qualms about enlisting goernments to
regulate the Internet. 1hey hae been the driing orce behind the
Communications Decency Act`s censorship restrictions and Child Online
Protection Act ,both deemed unconstitutional,, Internet iltering in
libraries, and worked to push legislation to ban online gambling. 1hey hae
also joined orces with the liberal social engineers ,Group 42, in pushing
or strong net neutrality` regulations, earing that Internet Serice
Proiders ,ISPs, will somehow discriminate against Christians online. 1his
group argues that, because the Internet is a public space, some rules and
laws are necessary to goern anti-social behaior. 1hey do not beliee that
technology can sole all social problems-on the contrary, they beliee that
the Internet is generally urthering the decline o culture. \et, in some
instances they embrace the Internet as a tool, as eidenced by ormer
Secretary o Lducation \illiam Bennett`s K-12 Internet-based home
schooling project. In general, moral conseraties don`t want indiiduals
empowered to engage in antisocial behaior, nor do they want corporations
to acilitate such behaior. Lxamples are groups like the Christian Coalition
and locus on the lamily, and around the world with countries like
Indonesia, 1hailand, Saudi Arabia and other religiously conseratie nations
that seek to limit actiity on the Internet.
6. /#E ,H%&%F@ 0)O(#C$%*I: 1his group beliees that there is nothing
inherently unique about the Internet and that it should be regulated in the
same way that goernment regulates eerything else, including past
technologies. 1here is a certain sense o urgency among certain elected
oicials, goernment bureaucrats, and public interest` adocates who
beliee that cyberspace is in a state o near-anarchy-a haen or criminals,
con artists, and rapacious corporations. Lxemplars o this group include,
law enorcement oicials seeking to limit use o encryption and other
innoatie technologies, eterans o the telecom regulatory wars that
preceded the breakup o Ma Bell, legal analysts working or social
engineering think tanks, as well as goernment oicials seeking to impose
restrictie regulatory rameworks on the broadband Internet. As ar as old
economy regulators are concerned, the 1934 Communications Act ,or
perhaps its 1996 update, answered all the questions that will eer arise

regarding the Internet. Moreoer, Luropean, Chinese and other old
economy regulators oerseas ear that, absent more regulation, their nations
will be bypassed by the American Internet leiathan.
. -)H? <%FKC&")I ^ -*CE) 3II%H"C$"%&I: 1his group encompasses a
range o organizations rom the politically say hardware, sotware and
communications giants to Internet start-ups. 1hese businesses, rom old
stalwarts like IBM, A1&1, and lewlett Packard to teenagers` like Cisco
Systems and Microsot, and youngsters` like Google and lacebook,
understand that trade, tax, regulatory, and other public policy issues
increasingly aect their bottom line and competitie position. \hile the
players in this group ,and in Bricks and Mortars, don`t hae the same leel
o ideological cohesion as the aboe groups, they share a certain set o
interests which justiies their grouping. 1hey realize that getting one`s way
in politics takes more than being right: It requires playing the game and
making one`s case persuasiely. lrom time to time, some tech businesses
may take the Internet exceptionalist position that the Internet should be let
ree rom goernment interention. Generally, they do so only to aoid
regulation that might put them at a competitie disadantage. On the
whole, tech companies tend to beliee that regulation can be both
adantageous and detrimental, they do not ight against all regulations and
are in aor o the right ones or them ,and increasingly support the
wrong` ones or their competitors,.
4
1o some extent, they also adocate
policies that are good or the technology industry or the economy as a
whole. \hile communication companies, being in a traditionally regulated
industry, hae long recognized the importance o goernment, most I1
companies hae ignored goernment and policy issues, being too busy
creating the technologies that drie the digital world. But as these
companies hae matured and become aware, oten through painul
experience, o how issues in \ashington can aect their bottom line, many
hae eoled into political sophisticates. And while indiidual tech
companies can hae dierent iews on dierent issues, these dierences
are largely rooted in business model interests, rather than ideological iews
about the market or goernment.
8. 9*"HXIBC&EB!%*$C*I: 1his group includes the companies, proessional
groups, and unions that gain their lielihood rom old-economy, ace-to-
ace business transactions. 1hese include both producers ,such as
automobile manuacturers, record companies, and airlines, and distributors
and middlemen ,such as retailers, car dealers, wine wholesalers, pharmacies,

4
lor a discussion o how technology companies iew public policy see AC1`s |vaer.tavaivg tbe
1 obb,: .v v.iaer`. Cviae, 2008,
?$$K5LLCH$%&#"&)N%*OLK(A#"HC$"%&ILVSSZLSZLS_L(&E)*I$C&E"&OB$?)B"$B#%AA@BC&B
"&I"E)*IBO("E)L.

optometrists, real estate agents, or unions representing workers in these
industries,. Many o them ear, oten correctly, that the Internet is making
them obsolete, while others hae worked to transorm their business
models to take adantage o e-commerce. In recent years, there has been a
widening rit between the bricks-and-mortar producers and the distributors
and middlemen ,and the unions that represent their workers,. Producers
hae begun to realize that they can use the Internet to go directly to their
consumers, bypassing ,or at least minimizing, the role o bricks-and-mortar
middlemen. 1he middlemen and unions, working actiely to keep this rom
happening or at least to orestall the day o reckoning, are not shy about
enlisting the aid o goernment to leel the playing ield.` Certainly, the
long running battle oer taxing Internet sales represented a ight between
bricks-and-mortars and tech companies. Likewise, the grocery store
workers` union in Caliornia has recently worked to pass legislation making
it more diicult or stores to use sel-serice checkout systems.
5

The Dividing Lines
1he aboe groups` attitudes about Internet policy can be placed along two axes:
Individual Empowerment vs. Societal Benefit
1his line separates groups on the basis o belies about the Internet`s oerriding
purpose. In some ways this is a ariant on the classic tension between liberty
and equality. loweer, it goes beyond this to represent the tension between
indiidualism and communitarianism, with the ormer being a ocus on
indiidual rights, and the latter inoking community beneits like economic
growth, security, and improed quality o lie.
1hose in the indiidual empowerment category beliee that I1`s chie unction
is to liberate indiiduals rom control by, or dependence on, big organizations.
lor them the Internet is a ast, open global communications medium designed
principally to enable indiiduals to reely communicate and access inormation.
\hen debating any issue, they examine it principally through the lens o how it
aects indiiduals, not society as a whole. 1hus, the issue o net neutrality is
seen in terms o its eect on indiidual reedom to act in any way desired on
broadband networks. Such groups want to put the little guy on the same playing
ield as the big boys, whether this means supporting small ISPs, small media
outlets, or indiidual open source coders.
1hose belonging in the societal beneit camp beliee I1 and the Internet`s main
job is to increase economic productiity, promote goernment responsieness

5
Robert D. Atkinson, vvoratiov ava t. .rv, of Oovevt., BUSINLSS\LLK, Sept. 23, 2010,
!""#$%%&'()*!+,-&./'&&0''1+*23%4'()*!5&'()*!6')37.//28(".2/9(/:9."&9()3;92
<92##2/'/"&=)'&->"&?')?(@'7AB.

and eiciency, and enable the deelopment new and better serices or
consumers as a societal whole. 1hey tend to examine indiidual I1 policy issues
through the lens o how they aect the communitarian interest and are willing
to accept tradeos to indiidual liberty or reedom i they boost oerall
economic or societal well-being. lor example, they see the actions o ISPs to
manage their broadband networks as being necessary to help the majority o the
users, een i it means that a ew bandwidth hogs` hae to wait a minute or
two longer to download their pirated copy o ora of tbe Rivg.. 1hey also beliee
that both goernment and corporations can sere as proxies or community
interests, and that what`s good or, say, Cisco, A1&1, Microsot or Google or
the ederal goernment can be good or America as whole. Some groups all in
between the two extremes and argue that tradeos between particular
indiidual`s beneit ,or harm, and community interests are ineitable.
Internet exceptionalists and social engineers generally beliee the Internet is all
about indiidual empowerment. 1he ormer resent its commercialization and
iew empowerment as ineitable. 1he latter, as stated earlier, beliee the
Internet should mainly be an educational and social networking tool and ear its
empowering capabilities will be taken away by powerul multinational
corporations and statist goernments that will reshape the Internet to sere
their own narrow purposes ,proit in the ormer, control in the latter,. Both see
hackers and pirates as lone champions standing tall against greedy corporate and
inept goernment leiathans.
Bricks-and-mortars and old economy regulators see I1 in instrumental terms as
designed or commerce and, by extension, or the community beneit. 1hey just
don`t like how the Internet has eoled, whether it`s competition rom Dot-
Coms or the spread o strong encryption that rustrates goernment
sureillance, censorship, and other control. 1ech companies also see I1 in more
instrumental terms, arguing that its rules should acilitate robust commerce.
Moral conseraties don`t want indiiduals empowered, since this will just
enable een more antisocial behaior, and they also don`t want corporations to
acilitate such behaior.
Moderates and ree marketers occupy the middle ground. 1hey beliee that the
digitization o the economy holds great promise or boosting productiity and
improing society. At the same time, they see the Internet as creating
communities, boosting education, and giing people more control oer their
lies. lree marketers don`t beliee that indiidual interests should necessarily
trump corporate interests-they see corporations as persons under the law.


Laissez-Faire vs. Government Regulation
1he groups diide along this line oer the degree to which the goernment
should impose ormal rules on I1 and the Internet.
Internet exceptionalists, and to a lesser degree ree marketers, beliee the
Internet should be goerned by its users. 1hese groups lie on the tai..efaire side
o the diiding line. 1hey consider the Internet unique and capable o creating
spontaneous order, a model or how the rest o society should be organized.
lree marketers beliee the Internet is what allows Coase`s ision o a society
with low transaction costs and ubiquitous markets to become a reality.
6

At the other extreme are groups on the goernment regulation side o the line,
who see the Internet as a new \ild \est` calling or a man with a badge to
protect ulnerable citizens against intrusie goernments and proit-hungry
corporations. Moral conseraties, social engineers, and old economy regulators
tend to hold this iew, arguing or an array o goernment actions to limit what
companies can do. So do bricks-and-mortars, although less as a matter o
principle than as a way o clinging to their eer-weakening economic position.
Moderates and tech companies occupy the middle ground. 1hey beliee the
Internet is unique and generally requires a light regulatory touch i I1
innoation is to thrie. But in some key areas such as cybersecurity and
copyright protection, they beliee that the Internet needs stronger rules,
especially to enable law enorcement to go ater bad actors. In still other areas,
such as the priacy o non-sensitie data and net neutrality, they beliee that
sel-regulating goernment,business partnerships are the best way to protect
consumers while giing companies needed lexibility.
I1Il was ormed to adance a set o pragmatic solutions to the growing
number o technology-related policy problems. \e beliee the growth o the
digital economy and society depends on a synthesis o these iews: the correct
position will tend to lie at the intersection o the two axes. 1he dichotomy
between indiidual empowerment and institutional eiciency is not a zero-sum
game. Indiiduals beneit both socially and economically when goernments
and corporations work more eiciently and eectiely, and institutions beneit
when indiiduals are inormed and able to make choices. A light touch on
regulation is important to maintain the lexibility required to operate in this
high-speed economy, but goernment action is also necessary to gie businesses
and consumers conidence that the Internet is not a den o thiees or a market
tilted against air competition, and to help speed digital transormation ,e.g., the
ubiquitous use o I1 throughout the economy and society,.

6
Lconomist Ronald Coase postulated that high transaction costs engendered large
organizations. ee, e.g., RONALD COASL, 1lL lIRM, 1lL MARKL1 AND 1lL LA\ ,1988,.

Ongoing Policy Debates
O course, the aboe typology is imperect-with many indiiduals and
organizations alling into more than one group or no group at all. But as one
looks at the central political ights about the uture o inormation technology,
the inluence o these competing actions is clear. As case studies, we consider
the recent debates oer our key issues: priacy, taxation, copyright protection,
and net neutrality.
Privacy
\hile the recent laps oer lacebook and Google Street View are the most
isible examples, the collection and use o personal inormation about Internet
users by corporations and goernment is the source o many heated and
emotional debates. Old economy regulators and social engineers want to impose
sweeping regulations that would gie indiiduals control oer their` personal
data. And while they tolerate, grudgingly, adertising as the one true business
model or Internet content and serices ,they oppose ISPs allowing content or
application companies to oluntarily pay or prioritized serice, they want to
limit the eectieness o online adertising, and the reenue it can raise,
because o priacy ears.
Many tech companies want complete reedom to collect personal data, proided
they comply with priacy policies they write themseles. And while some tech
companies hae supported moderate notice and choice` legislation, most
companies remain wary o any ederal regulation o priacy, een as they
recognize the need or ederal laws to preempt increasingly antsy state
legislators rom passing a patchwork o dierent Internet priacy bills.
Internet exceptionalists expect technology to sole the problem. As ar as
they`re concerned, users should take responsibility or their own priacy and
apply the tools aailable to protect their personal data.
lree marketers reject the need or priacy legislation, asserting that the harms
rom regulation would ar outweigh the beneits, and that goernment
regulation is likely to be an imposition on indiidual liberty and choice,
including basic rights o ree speech. \hile moderates worry that oerly-strict
priacy laws would stile innoation and increase costs or consumers, they also
beliee that, absent any rules, users will not deelop the trust needed or the
digital economy and society to lourish.
1he recent uror oer lacebook is a perect example o how these issues play
out. 1his social network company announced two new eatures in 2010: instant
personalization, which allows users to share data rom their lacebook proile
with partner websites, and social plug-ins or third party websites, which allow

users to more easily share web pages they like with their social network outside
o lacebook.

Social engineers howled in protest, demanding restrictie goernment
regulations to bar such practices. Some, like Danah Boyd, a ellow at larard`s
Berkman Center or Internet and Society, went so ar as to claim that lacebook
unctioned as a public utility and should be regulated like one.
8

lacebook was slow to react, initially ocusing more on highlighting the beneits
o its innoatie new tools. loweer, it quickly responded more appropriately,
rolling out a much more user-riendly and transparent system o user priacy
controls.
I1Il and other moderates as well as ree marketers argue that goernment
control oer the priacy policies o social networks is not necessary to protect
consumers and moreoer, would be harmul to uture innoation. In the heated
political enironment o the priacy debate, goernment interention would
probably become regulatory oerkill. At the same time, moderates argue that
legitimate priacy concerns about personally identiiable data and sensitie data
,inancial or medical inormation, or example, need to be addressed through
comprehensie industry-wide codes o sel-regulation, enorceable by
goernment action ,e.g., l1C action against companies that do not lie up to
their own priacy policies or unair and deceptie trade practices,.
\hen it comes to the collection and use o data by goernment, the coalitions
reconigure. lere the Internet exceptionalists, social engineers, and ree
marketers make common cause in their crusade against Big Brother.` It largely
does not matter whether the goal is to crack down on deadbeat dads, catch red
light runners, or preent terrorist attacks: I it inoles the goernment
collecting more inormation or using existing inormation or new purposes,
these groups will generally oppose it. In protesting against the growing practice
o cities installing red light cameras, ormer Republican louse majority leader
Dick Armey railed: 1his is a ull-scale sureillance system. Do we really want a
society where one cannot walk down the street without Big Brother tracking our
eery moe`
9

lor more see: Daniel Castro, Inormation 1echnology and Innoation loundation, 1be Rigbt
to Prirac, i. ôt a Rigbt to aceboo/, April 2010,, !""#$%%.".<+2)@%#-,>.*(".2/&%<(*',221C
/2"C).@!", and Daniel Castro, Inormation 1echnology and Innoation loundation, aceboo/
i. ôt tbe vev,, 2010, !""#$%%.".<+2)@%#-,>.*(".2/&%<(*',221C/2"C'/'3;.
8
Danah Boyd, aceboo/ i. a vtitit,; vtititie. get regvtatea, APOPlLNIA, May 15, 2010,
!""#$%%000+D'#!2).(+2)@%"!2-@!"&%()*!.8'&%ABEB%BF%EF%<(*',221C.&C(C-".>.";C
-".>.".'&C@'"C)'@->("':+!"3>.
9
1homas C. Greene, Co. |.ivg igb1ecb vrreittavce iv toriaa, 1lL RLGIS1LR, July 2, 2001,
!""#$%%000+"!')'@.&"')+*2+-1%ABBE%BG%BA%*2#&H-&./@H!.@!"'*!H&-)8'.>>(/*'%.

ligh-tech companies hae engaged in the debate oer goernment use o and
access to data based in large part on their business interests. 1echnology
companies with direct business interests in proiding goernment technologies
to collect inormation ,e.g., smart card and biometrics companies, hae been
strong supporters o particular initiaties. Other technology companies,
worrying that goernment access to data can restrict commerce or reduce
consumer trust in the Internet ,e.g., in cloud computing applications where
consumer data is remotely stored, hae called or limitations or procedural
saeguards on goernment access to data.
\hether a middle position in the debate can be ound remains an ongoing
question. Moderates support the adoption o new technologies by goernment,
i it is clearly demonstrated that they ulill an important public mission and i
potential priacy problems are eectiely addressed, especially by designing
priacy protections into systems. At the same time, they support putting into
place adequate rules and protections goerning the access to that data by
goernment.
Internet Sales Taxes
1ax policy is controersial in any setting, but perhaps particularly so with regard
to the Internet. 1he collection o state and local sales taxes or Internet
transactions is so controersial that 15 years ater it was irst raised, the issue
continues to be debated. Old economy regulators want sales taxes to be
collected on Internet purchases and want high taxes on telecommunications
serices to maintain their reenue. 1he state o Colorado has gone so ar as to
require Internet retailers to share the names and purchase inormation o
Colorado residents with the state goernment ,so the state can collect a use`
tax rom Internet shoppers,. Bricks and mortar companies want sales taxes
imposed to maintain their competitie position against pure-play Internet
retailers. Some social engineers aor not only sales tax collection, but also
special taxes on broadband use to subsidize access or low-income and rural
households.
By contrast, the tech companies inoled in selling oer the Internet do not
want the burden o collecting taxes oer thousands o jurisdictions, and they do
not want to lose their price adantage. Likewise, they do not want broadband or
telephone serice unairly taxed at higher rates. Others-like many ree
marketers and Internet exceptionalists-oppose Internet sales taxes on
principle. 1hey beliee the ewer taxes the better,` especially when it comes to
promoting the new digital economy.
Internet exceptionalists, tech companies, and ree marketers will likely continue
to oppose giing states the right to tax Internet sales to their residents rom
companies outside their borders. State goernments will press hard or the right,
citing their large budget shortalls. And pragmatists will likely aor state sales

taxes, particularly i they are tied to a qvia ro qvo deal orcing states to rescind
laws and regulations that discriminate against e-commerce sellers, and i taxation
is administered in ways that minimize administratie burden. lor now, howeer,
the debate continues, with states legally unable to collect sales taxes and most
states imposing high, discriminatory taxes on telecommunications serices.
Copyright Protection
As irtually all media hae become digital, protecting copyrights has become a
nightmare. 1he controersy oer the ile sharing system Napster almost a
decade ago was just the beginning. 1he ubiquity o ile-sharing technologies,
coupled with computers that can rip digital iles rom CDs or DVDs, and high-
speed broadband networks that can quickly transer large iles, has meant that
digital piracy` has grown like wildire. Internet exceptionalists argue that the
Internet Age marks the end o intellectual property rights because enorcing
copyright protections on digital media is too diicult ,hence their mantra
inormation wants to be ree.`, 1hese adocates claim that non-commercial
ile sharing` o copyrighted media is a orm o air use, which they assert is
legal under copyright law. lor example, the Llectronic lreedom lorum`s Let
the Music Play` campaign protests the music and ilm industries` prosecution o
ile copiers. In their ideal world, some rich dot-com entrepreneur would
establish a separate country on a desert island, linked to the rest o the world by
high speed iber-optic cable and hosting a massie computer with a cornucopia
o pirated digital content, all beyond the reach o national copyright laws.
Many social engineers side with the Internet exceptionalists, though or ery
dierent reasons. 1hey ear that technology will let copyright holders exact such
strict control on content that traditional notions o air use will become
obsolete. And they ear that digital rights management ,DRM, technologies will
become so stringent that actiities consumers hae long enjoyed ,like the ability
to play music iles on more than one deice, will be prohibited. Both argue
strongly against any eorts to better control digital copyright thet that may
impinge on indiidual liberty or indiidual rights like ree speech ,e.g., permitting
ISPs to ilter or illegal content or crating international treaties like AC1A to
strengthen and harmonize anti-piracy eorts,. And both would loe to see the
Digital Millennium Copyright Act ,DMCA, enter the dust bin o I1 policy
history, particularly the academics and engineers who eel the DMCA restricts
their ability to hack DRM technology in the name o research.
10

Because o their emphasis on property rights, most ree marketers tend to
strongly support eorts to limit digital copyright thet. But with their ocus on
reedom, a ew come all the way around to the let, arguing that because liberty

10
Michele Boldrin, Against Intellectual Monopoly, No. 10, 2008,
!""#$%%000+*("2+2)@%'8'/"+#!#5'8'/".:7FIJA.

trumps property, the grant o intellectual property rights by goernment
amounts to the proision o a state-sanctioned monopoly.
11
In their iew,
indiiduals should be ree to use digital content in ways they want and content
owners-not others such as digital intermediaries-should be responsible or
policing the use o their content.
Moderates also support eorts to limit digital copyright thet, belieing that
such thet is wrong, and that a robust digital ecosystem requires economic
incenties to produce oten expensie digital content. At the same time,
howeer, they are not absolutists, and in particular seek to balance the costs and
beneits o copyright deense, especially through air use.
1he bricks and mortar companies-including the Recording Industry
Association o America-initially worked to block the deelopment o new
technologies that acilitate playing downloaded and possibly pirated music. But
more than a decade later the content industries are not so much ighting against
such technologies as they are working to deelop and use technologies that can
counter copyright thet, and going ater organizations that enable widespread
digital content thet ,e.g., the Swedish Pirate Bay,.
12
And een as they hae
struggled to cope with music and moie piracy, content producers hae largely
come to terms with the realities o the digital era: 1hey hae begun proiding
legal, aordable, and consumer-riendly means or consumers to buy or iew
copyright-protected digital content, with Apple`s i1unes music store and lulu
being the most prominent examples.
Although generally sympathetic to the content proiders` copyright concerns,
many high-tech companies ,e.g., ISPs, search engines, social networks, ear that
the ederal goernment will require them to adjust their businesses to become
copyright enorcers, either by haing to take action against their customers or
by building in expensie content protection technologies. Once again, the
question is whether a compromise can be ound, ensuring that content holders
hae the legal protections and economic incenties they need to continue
producing copyrighted materials without imposing oerly-large burdens on
technology companies, and by extension their customers.

11
ee, e.g., Cato Institute, .gaiv.t vtettectvat Movoot,, 2008,
!""#$%%000+*("2+2)@%'8'/"+#!#5'8'/".:7FIJA.
12
Lric Panner, Mv.ic vav.tr, Covvt. tbe Co.t of Pirac,, 1lL NL\ \ORK 1IMLS, Jan. 21, 2010,
!""#$%%000+/;".3'&+*23%ABEB%BE%AA%,-&./'&&%@>2,(>%AA3-&.*+!"3>5)'<7)'*2):./
@H./:-&");H(&&2*.(".2/H2<H(3').*(, Lamonn lorde, rov Peer to tervit,: !itt ]vvivg tbe
egat Diriae otre .v,tbivg., 1lL MUSIC NL1\ORK, Dec. 6, 2010,
!""#$%%000+"!'3-&.*/'"02)1+*23%3-&.*C<'("-)'&%./:-&");%ABEB%EA%BJ%<)23C
#'')C"2C'"')/.";C0.>>CK-3#./@C"!'C>'@(>C:.8.:'C&2>8'C(/;"!./@%.

Net Neutrality
\hat has become a highly contentious issue, net neutrality, reers to the idea
that the indiidual networks collectiely orming the Internet be controlled by
users rather than by their owners and operators. \hile network operators are in
a unique position to manage their resources, proponents o net neutrality
beliee they cannot be trusted to utilize their knowledge or the good o the
Internet user community.
Social engineers are the most passionate about net neutrality, but they make
common cause with the eterans o the old economy regulator group and
Internet exceptionalists. Indeed, social engineer 1im \u coined the still-
mystiying term net neutrality.`
13
1hese groups ear that the Internet`s unique
nature is under threat by the orces o incumbent telecommunications and cable
companies proiding broadband serice. I Big Broadband` gets its way,
neutralists ear the Internet will go the way o cable 1V, the ast wasteland`
14

where elitist programming such as 1be !ire competes with adertising-
supported, populist programming such as .vericav aot.
lree marketers see net neutrality as one more attack by big goernment
regulators on the Internet, the last bastion o reedom rom regulation. 1hey
argue that market orces and consumer choice will always discipline any anti-
consumer iolations o net neutrality, while antitrust or tort law will sere as a
handy tool to remedy any anti-business iolations.
1ech companies are split on the issue, largely around which side o the network
they are on. 1hose tech companies proiding network serices ,e.g., ISPs and
major equipment makers, are generally against strong regulations in support o
network neutrality ,at least with regard to the network itsel, while companies
whose business model depends on using the network to gain access to
customers ,e.g., content & serice proiders like Google, are either neutral or in
aor o a stronger regulatory regime ,at least with regard to the inrastructure
layers, as opposed to other parts o the Internet stack,` such as applications.,

13
1im \u, êtror/ êvtratit,, roaabava Di.crivivatiov, 2 JOURNAL Ol 1LLLCOMMUNICA1IONS
AND lIGl 1LClNOLOG\ LA\ 141 ,2003,,
!""#$%%#(#')&+&&)/+*23%&2>I%#(#')&+*<35(,&")(*"H.:7ILLLJI.
14
On May 9, 1961, in a speech to the National Association o Broadcasters, newly-appointed
lCC chairman Newton N. Minow reerred to teleision as a ast wasteland.` Newton N.
Minow, 1eleision and the Public Interest,` address to the National Association o
Broadcasters, \ashington, D.C., May 9, 1961.

loweer, these dierences hae begun to blur somewhat, as eidenced by the
October 2009 joint statement on net neutrality issued by Google and Verizon.
15

Moderates generally see the Internet as a work-in-progress. Moderates beliee it
is good that network equipment producers are improing the Internet and
recognize operators as possessing the highly specialized knowledge needed to
proide equitable access to the Internet`s pool o resources. But moderates
realize that competition doesn`t operate as eiciently in some network markets
as it does in the markets or general-purpose consumer goods and serices. In
other words, some network markets are under-competitie ,because network
eects create market power,, so markets alone aren`t suicient to guarantee an
open Internet or eeryone.
16
1he role o goernment in Internet regulation is
to ensure that all consumers enjoy the ruits o inestment and innoation, but
only in ways that don`t limit continued inestment and innoation.
As these and other issues continue to be ought in legislatures and communities
around the country, goernment oicials should seek solutions that balance the
needs o indiiduals with those o society, and that oer the security o codiied
laws when necessary and the lexibility o inormal rules when appropriate. As
the technology policy debates go on and the arious actions push or the
solutions that it their ideologies and interests, the policies that promote the
growth and itality o the digital economy will not be ound at the extremes, but
instead in the ital center.
The Future of Digital Politics
Some might argue that these issues are transitory and will recede in importance
as the digital economy matures. But there is good reason to beliee otherwise:
1he debates that pit online consumers against resistant middlemen are likely to
continue as new orms o online distribution eole. 1he emergence o much
aster and ubiquitous wired and wireless broadband networks will mean more
Americans using these networks and more business models deeloping to take
adantage o them. Data generated by emerging new technologies such as
wireless location systems, digital signature systems, intelligent transportation
systems, the smart electric grid, health I1, and radio requency identiication
,RlID, deices-some used by goernment, others by the priate sector-will
drie new priacy concerns among social engineers and their ellow traelers. In
some ways, the digital reolution has been so successul that many preiously

15
Lowell McAdam, CLO Verizon \ireless & Lric Schmidt, CLO Google, linding Common
Ground on an Open Internet, Verizon PolicyBlog, Oct. 21, 2009, !""#$%%#2>.*;,>2@+
8').D2/+*23%N>2@?2&"%JGF%O./:./@P2332/Q)2-/:2/(/R#'/S/"')/'"+(&#T.
16
Richard Bennett, Inormation 1echnology and Innoation loundation, 1 Covvevt. ov
CC roaabava Recta..if,ivg, August 10, 2010, !""#$%%000+.".<+2)@%#-,>.*(".2/&%.".<C
*233'/"&C<**C,)2(:,(/:C)'*>(&&.<;./@.

analog political issues hae become digital issues, on the other hand, the
political issues o the uture remain unormed, precisely because the
technologies are changing so quickly.
1he public policy issues surrounding the I1 reolution are no longer sideshows
or mere theoretical discussions or a handul o technologically say people,
nor are they the royal road to a utopia o untold wealth and perect reedom.
1he battle lines hae been drawn, and the issues are both serious and complex.
Digital politics, i not the great issue o our age, will be central to the lie o our
nation in the decade ahead-and well beyond.

343

PART II
ISSUES & APPLICATIONS

345

CHAPTER 6
SHOULD ONLINE INTERMEDIARIES
BE REQUIRED TO POLICE MORE?
Trusting (and Verifying) Online Intermediaries Policing 347
lrank Pasquale
Online Liability for Payment Systems 365
Mark MacCarthy
Fuzzy Boundaries: The Potential Impact
of Vague Secondary Liability Doctrines
on Technology Innovation 393
Paul Szynol

346 CHAPTER 6: SHOULD ONLINE INTERMEDIARIES BE REQUIRED TO POLICE MORE?


Trusting (and Verifying)
Online Intermediaries Policing
By Frank Pasquale
*

Introduction
Internet Serice Proiders ,ISPs, and search engines hae mapped the \eb,
accelerated e-commerce, and empowered new communities. 1hey can also
enable intellectual property inringement, harassment, stealth marketing, and
rightening leels o sureillance. As a result, indiiduals are rapidly losing the
ability to control their own image on the web, or een to know what data others
are presented with regarding them. \hen \eb users attempt to ind
inormation or entertainment, they hae little assurance that a carrier or search
engine is not subtly biasing the presentation o results in accordance with its
own commercial interests.
1

None o these problems is readily susceptible to swit legal interention.
Instead, intermediaries themseles hae begun policing their own irtual
premises. eBay makes it easy or intellectual property owners to report
inringing merchandise. A carrier like Comcast has the technical power to slow
or block traic to and rom a site like Bit1orrent, which is oten accused o
inringement.
2
Google`s StopBadware program tries to alert searchers about
malware-ridden websites,
3
and \ou1ube employs an indeterminate number o
people to police copyright inringement, illegal obscenity, and een many
grotesque or humiliating ideos.
4
Reputable social networks do the same or
their own content.

Proessor o Law, Seton lall Law School, Visiting lellow, Princeton Center or Inormation
1echnology Policy.
1
Benjamin Ldelman, araCoaivg ia. iv Coogte .tgoritbvic earcb Re.vtt., No. 15, 2010,
aailable at !""#$%%&&&'()*)+),-.*'/01%!.0+2/+3*1% ,I present categories o searches
or which aailable eidence indicates Google has hard-coded` its own links to appear at
the top o algorithmic search results, and I oer a -)"!/+/,/14 or detecting certain kinds
o tampering by comparing Google results or similar searches. I compare Google`s hard-
coded results with Google`s public 5".")-)*"5 and promises, including a dozen denials but
at least one admission.`,.
2
ee, e.g., Comcast Corp. . lCC, No. 08-1291, 2010 U.S. App. LLXIS 039 ,D.C. Cir. April
6, 2010,.
3
lor more inormation, isit !""#$%%5"/#(.+&.0)'/01%.
4
\ou1ube, Yov1vbe Covvvvit, Cviaetive.,
!""#$%%&&&'4/6"6()'2/-%"%2/--6*3"47163+),3*)5 ,\ou1ube sta reiew lagged

\et all is not well in the land o online sel-regulation. loweer competently
they police their sites, nagging questions will remain about their airness and
objectiity in doing so. Is Comcast blocking Bit1orrent to stop inringement,
or to decrease access to content that competes with its own or iewers low
much digital due process does Google need to gie a site it accuses o harboring
malware I lacebook eliminates a ideo o war carnage, is that a token o
respect or the wounded or one more relexie eort o a major company to
ingratiate itsel with a \ashington establishment currently committed to
indeinite military engagement in the Middle Last
Questions like these will persist, and erode the legitimacy o intermediary sel-
policing, as long as key operations o leading companies are shrouded in
secrecy. Administrators must deelop an institutional competence or
continually monitoring rapidly-changing business practices. A trusted adisory
council charged with assisting the lederal 1rade Commission ,l1C, and
lederal Communications Commission ,lCC, could help courts and agencies
adjudicate controersies concerning intermediary practices. An Internet
Intermediary Regulatory Council ,IIRC, would spur the deelopment o what
Christopher Kelty calls a recursie public`-one that is itally concerned with
the material and practical maintenance and modiication o the technical, legal,
practical, and conceptual means o its own existence as a public.`
5
Questioning
the power o a dominant intermediary is not just a preoccupation o the
anxious. Rather, monitoring is a prerequisite or assuring a leel playing ield
online.
Understanding Intermediaries Power
Internet intermediaries goern online lie.
6
ISPs and search engines are
particularly central to the web`s ecology. Users rely on search serices to map
the web or them and use ISPs to connect to one another. Lconomic
sociologist Daid Stark has obsered that search is the watchword o the
inormation age.`
ISPs are oten called carriers` to relect the parallel

ideos 24 hours a day, seen days a week to determine whether they iolate our Community
Guidelines. \hen they do, we remoe them.`,
5
ClRIS1OPlLR M. KLL1\, 1\O BI1S: 1lL CUL1URAL SIGNIlICANCL Ol lRLL SOl1\ARL 3
,Duke Uni. Press 200,.
6
lor a deinition o intermediary, .ee 1homas l. Cotter, ove Ob.erratiov. ov tbe ar ava
covovic. of vterveaiarie., 2006 MICl. S1. L. RLV. 6, 68-1 ,|A|n intermediary` can be any
entity that enables the communication o inormation rom one party to another. On the
basis o this deinition, any proider o communications serices ,including telephone
companies, cable companies, and Internet serice proiders, qualiy as intermediaries.`,.
DAVID S1ARK, 1lL SLNSL Ol DISSONANCL: ACCOUN1S Ol \OR1l IN LCONOMIC LIlL 1

,Princeton Uni. Press 2009, ,Among the many new inormation technologies that are
reshaping work and daily lie, perhaps none are more empowering than the new technologies
o search.`,.

between their own serices in the new economy and transportation
inrastructure. Online intermediaries organize and control access to an
extraordinary ariety o digitized content. Content proiders aim to be at the
top o Google Search or Google News results.
8
Serices like i1unes, lulu, and
\ou1ube oer audio and ideo content. Social networks are extending their
reach into each o these areas. Cable-based ISPs like Comcast hae their own
relationships with content proiders.
9

\hen an Internet connection is dropped, or a search engine ails to produce a
result the searcher knows exists somewhere on the web, such ailures are
obious. loweer, most web experiences do not unold in such a binary, pass-
ail manner. An ISP or search engine can slow down the speed or reduce the
ranking o a website in ways that are ery hard or users to detect. Moreoer,
there are many points o control, or layers, o the \eb.
10
Len when users`
experience with one layer causes suspicion, it can blame others or the problem.
1he new power o intermediaries oer reputation and isibility implicates
seeral traditional concerns o the American legal system.
11
Unortunately,
Internet intermediaries are presently bound only by weak and inadequate
enorcement o consumer protection and alse adertising statutes, which were
designed or ery dierent digital inrastructures.

8
ee Deborah lallows & Lee Rainie, Pew Internet & Am. Lie Project, Data Mevo: 1be
Povtarit, ava vortavce of earcb vgive. 2 ,Aug. 2004,,
!""#$%%&&&'#)&3*")0*)"'/01%#+85%9:97;.".7<)-/7=).02!)*13*)5'#+8 ,1he
aerage isitor scrolled through 1.8 result pages during a typical search.`,, Leslie Marable,
at.e Oracte.: Cov.vver Reactiov to earvivg tbe 1rvtb .bovt or earcb vgive. !or/: Re.vtt. of av
tbvograbic tva,, CONSUMLR \LB\A1Cl, June 30, 2003, at 5, araitabte at
!""#$%%&&&'2/*56-)0&)(&."2!'/01%#+85%8.,5)>/0.2,)5'#+8 ,1he majority o
participants neer clicked beyond the irst page o search results. 1hey trusted search
engines to present only the best or most accurate, unbiased results on the irst page.`,.
9
ROBLR1 \. MCClLSNL\, RICl MLDIA, POOR DLMOCRAC\: COMMUNICA1ION POLI1ICS IN
DUBIOUS 1IMLS 123 ,2000, ,describing how conergence o digital technology eliminates
the traditional distinctions between media and communications sectors`,.
10
JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1-AND lO\ 1O S1OP I1 6 ,2008,
,describing a physical layer, the actual wires or airwaes oer which data will low,` an
application layer, representing the tasks people might want to perorm on the network,` a
content layer, containing actual inormation exchanged among the network`s users,` and a
social layer, where new behaiors and interactions among people are enabled by the
technologies underneath`,.
11
\ochai Benkler, Covvvvicatiov. vfra.trvctvre Regvtatiov ava tbe Di.tribvtiov of Covtrot orer Covtevt,
22 1LLLCOMM. POL`\ 183, 185-86 ,1998, ,describing the power o intermediaries oer
inormation low: technology, institutional ramework, and organizational adaptation .
determine . who can produce inormation, and who may or must consume, what type o
inormation, under what conditions, and to what eect`,, Cotter, .vra note 6, at 69-1
,discussing some o the unctions o technological intermediaries, including their control o
inormation low rom suppliers to consumers,.

In the space o a brie essay, I cannot surey the entire range o intermediary
policing practices. But it is worthwhile to drill down a bit into the tough
questions raised by one intermediary-the dominant search engine, Google-as
it decides what is and is not an acceptable practice or search engine optimizers
who want their clients` sites to appear higher in the rankings or gien queries.
Search engineers tend to diide the search engine optimization ,SLO, business
into good guys` and bad guys,` oten calling the ormer white hat SLO` and
the latter black hat SLO.`
12
Some degree o transparency regarding the search
engine`s algorithm is required to permit white hat SLO. 1hese rules are
generally agreed upon as practices that make the web better,` i.e., hae resh
content, don`t sell links, don`t stu metatags` with extraneous inormation just
to get attention. loweer, i there were complete transparency, black hat`
SLOs could unairly eleate the isibility o their clients` sites-and een i this
were only done temporarily, the resulting churn and chaos could seerely reduce
the utility o search results. Moreoer, a search engine`s competitors could use
the trade secrets to enhance its own serices.
1his secrecy has led to a growing gray zone o Internet practices with uncertain
eect on sites` rankings. Consider some o the distinctions below, based on
search engine optimization literature:
?!3") A." ,acceptable,
13
B0.4 C0). ,unclear how these
are treated,
14

D,.2E A." ,unacceptable, can
lead to down-ranking in Google
results or een the Google
Death Penalty` o De-Indexing,
Asking blogs you like to
link to you, or engaging in
reciprocal linking between
your site and other sites in a
legitimate dialogue.
15

Paying a blogger or site to link
to your blog in order to boost
search results and not just to
increase traic.
Creating a link arm` o spam
blogs ,splogs, to link to you, or
linking between multiple sites
you created ,known as link
arms, to boost search results.
16

12
Llizabeth an Couering, . Reteravce Reteravt.,
!""#$%%F2-2'3*+3.*.')+6%G/,HI%3556)J%G.*2/6G)03*1'!"-, ,search engineers`
animosity towards the . guerilla ighters o spamming and hacking, is more direct` than
their hostility toward direct business competitors,, Aaron \all, Coogte 1biv/. YO| .re a
tac/ at O. bovta Yov 1rv.t 1bev., SLOBOOK, Apr. 1, 2008,
!""#$%%&&&'5)/(//E'2/-%"/>1//1,)>4/6>.0)>.>5#.--)0 ,claiming that Google
discriminates against sel-identiied SLOs,.
13
Phil Craen, Lthical` Search Lngine Optimization Lxposed!, \eb\orkshop,
!""#$%%&&&'&)(&/0E5!/#'*)"%)"!32.,>5).02!>)*13*)>/#"3-3K."3/*'!"-, ,last isited
Jun. 8, 2009,.
14
Grey lat SLO, !""#$%%10)4!."5)/'2/-% ,last isited Jun. 5, 2006, ,claiming a Grey lat
SLO is someone who uses black hat techniques in an ethical way.,
15
iv/ cbeve., GOOGLL \LBMAS1LR CLN1RAL,
!""#$%%&&&'1//1,)'2/-%56##/0"%&)(-.5")05%(3*%.*5&)0'#4L.*5&)0MNNJON ,1he

13
are treated,
14

Running human-conducted
tests o search inquiries
with permission rom the
search engine.
Doing a ew queries to do
elementary reerse
engineering. ,1his may not be
permitted under the 1erms o
Serice,.
Using computer programs to
send automated search queries
to gauge the page rank
generated rom arious search
terms ,1erms o Serice
prohibit this,
1

Creating non-intentional
duplicate content ,through
printer-riendly ersions,
pages aimed at mobile
deices, etc.,
18

Intentionally creating
permitted duplicate content to
boost search results
Intentionally creating
unnecessary duplicate content
on many pages and domains to
boost results
Generating a coherent site
with original and
inormatie material aimed
at the user
Creating content or additional
pages that walk the line
between useul inormation
and doorway pages`
Creating doorway pages` that
are geared towards popular
keywords but that redirect to a
largely unrelated main site.
19

best way to get other sites to create releant links to yours is to create unique, releant
content that can quickly gain popularity in the Internet community. 1he more useul content
you hae, the greater the chances someone else will ind that content aluable to their
readers and link to it.`,.
16
Duncan Riley, Coogte Dectare. ]ibaa Ov tog iv/ arv., 1LClCRUNCl, Oct. 24, 200,
!""#$%%&&&'")2!206*2!'2/-%IPPQ%HP%IR%1//1,)>+)2,.0)5>F3!.+>/*>(,/1>,3*E>
8.0-5%.
1
.vtovatea Qverie., GOOGLL \LBMAS1LR CLN1RAL,
!""#$%%&&&'1//1,)'2/-%56##/0"%&)(-.5")05%(3*%.*5&)0'#4L.*5&)0MNNJOQ
,Google`s 1erms o Serice do not allow the sending o automated queries o any sort to
our system without express permission in adance rom Google.`,, Google 1erms o
Serice: Use o the Serices by you, !""#$%%&&&'1//1,)'2/-%.22/6*"5%ST= ,last isited
Jun. 4, 2009, ,\ou agree not to access ,or attempt to access, any o the Serices by any
means other than through the interace that is proided by Google, unless you hae been
speciically allowed to do so in a separate agreement with Google.`,.
18
Dvticate Covtevt, GOOGLL \LBMAS1LR CLN1RAL,
!""#$%%&&&'1//1,)'2/-%56##/0"%&)(-.5")05%(3*%.*5&)0'#4L.*5&)0MNNJOU
,Lxamples o non-malicious duplicate content could include: Discussion orums that can
generate both regular and stripped-down pages targeted at mobile deices, Store items
shown or linked ia multiple distinct URLs, Printer-only ersions o web pages`,.
19
Google Blogoscoped, Cervav M! avvea rov Coogte, leb. 4, 2006,
!""#$%%(,/1/52/#)+'2/-%.02!3G)%IPPN>PI>PR>*NP'!"-,, Matt Cutts, Ravivg v ov
vtervatiovat !eb.av, MA11 CU11S: GADGL1S, GOOGLL, AND SLO, leb. 4, 2006,
!""#$%%&&&'-.""26""5'2/-%(,/1%0.-#3*1>6#>/*>3*")0*."3/*.,>&)(5#.-% ,Google
employee conirming BM\`s ban,.

13
are treated,
14

1argeting an appreciatie
audience
20

Putting random reerences to
salacious or celebrity topics on
a blog primarily deoted to
discussing current aairs
21

Distracting an inoluntary
audience with completely
misleading indexed content
,akin to initial interest
conusion` in Internet
trademark law,
22

Inluencing search engine
by making pages easier to
scan by automated bots
23

Creating hidden pages` when
there may be a logical reason
to show one page to search
engine bots and another page
to users who type in the page`s
URL
Using hidden pages` to show a
misleading page to search
engine bots, and another page
to users who type in the page`s
URL

As these practices show, search engines are reerees in the millions o contests
or attention that take place on the web each day. 1here are hundreds o
entities that want to be the top result in response to a query like sneakers,`
restaurant in New \ork City,` or best employer to work or.` Any academic
who writes on an obscure subject wants to be the go-to` authority when it is
Googled-and or consultants, a top or tenth-ranked result could be the

20
!ebva.ter Cviaetive.: De.igv ava covtevt gviaetive., GOOGLL \LBMAS1LR CLN1RAL,
!""#$%%&&&'1//1,)'2/-%56##/0"%&)(-.5")05%(3*%.*5&)0'#4L.*5&)0MJOQNU ,last
isited Jun. 4, 2009, ,1hink about the words users would type to ind your pages, and make
sure that your site actually includes those words within it.`,.
21
Daniel Soloe, 1bav/., ]evvifer .vi.tov ;or tbe Mavifota !a,. to Do tbe ave earcb),
CONCURRING OPINIONS,
!""#$%%&&&'2/*26003*1/#3*3/*5'2/-%.02!3G)5%IPPN%PH%"!.*E57F)**38)0'!"-, ,One
o my more popular posts is one entitled V)**38)0 C*35"/* W6+) 9!/"/5 .*+ "!) C*"3>
9.#.0.KK3 C2". It seems to be getting a lot o readers interested in learning about the
workings o the Anti-Paparazzi Act and the law o inormation priacy. It sure is surprising
that so many readers are eager to understand this rather technical statute. Anyway, or the
small part that Jennier Aniston plays in this, we thank her or the traic.`,, Dan liller, Coffee
Or ^vae Cetebrit, Pboto.: . 1ate Of 1ro rergreev Po.t., 1lL lACUL1\ LOUNGL,
!""#$%%&&&'"!)8.26,"4,/6*1)'/01%IPPX%PR%2/88))>/0>*6+)'!"-, ,signiicant amounts
o traic arried in the orm o web surers seeking out pictures o Jennier Aniston`,.
22
Jason Preston, Coogte vvi.be. qviaoo for barivg too vvcb av, BLOG BUSINLSS SUMMI1, Jul. 11,
200, !""#$%%(,/1(653*)5556--3"'2/-%IPPQ%PQ%1//1,)>#6*35!)5>5Y63+//>8/0>
!.G3*1>"//>-62!>5#.-'!"-.
23
!ebva.ter Cviaetive.: De.igv ava Covtevt Cviaetive., GOOGLL \LBMAS1LR CLN1RAL,
!""#$%%&&&'1//1,)'2/-%56##/0"%&)(-.5")05%(3*%.*5&)0'#4L.*5&)0MJOQNU ,last
isited Jun. 4, 2009, ,Create a useul, inormation-rich site, and write pages that clearly and
accurately describe your content.`,, a. ,1ry to use text instead o images to display
important names, content, or links. 1he Google crawler doesn`t recognize text contained in
images.`,.

dierence between lucratie gigs and obscurity. 1he top and right hand sides o
many search engine pages are open or paid placement, but een there the
highest bidder may not get a prime spot because a good search engine stries to
keep een these sections ery releant to searchers.
24
1he organic results are
determined by search engines` proprietary algorithms, and preliminary eidence
indicates that searchers ,and particularly educated searchers, concentrate
attention there. Businesses can grow reliant on good Google rankings as a way
o attracting and keeping customers.
lor example, John Battelle tells the story o the owner o 2bigeet.com ,a seller
o large-sized men`s shoes,, whose site was knocked o the irst page o
Google`s results or terms like big shoes` by a sudden algorithm shit in
Noember 2003, right beore the Christmas shopping season. 1he owner
attempted to contact Google seeral times, but said he neer got a response.`
Google claimed the owner may hae hired a search engine optimizer who ran
aoul o its rules-but it would not say precisely what those rules were.
25
Like
the IRS`s unwillingness to disclose all o its audit lags,` the company did not

24
Steen Ley, ecret of Coogtevovic.: Datavetea Recie rer. Profitabitit,, \IRLD, May 2, 2009,
!""#$%%&&&'&30)+'2/-%26,"60)%26,"60)0)G3)&5%-.1.K3*)%HQ>
PN%*)#71//1,)*/-325 ,in Google`s Ad\ords program,`1he bids themseles are only a
part o what ultimately determines the auction winners. 1he other major determinant is
something called "!) Y6.,3"4 52/0). 1his metric stries to ensure that the ads Google shows
on its results page are true, high-caliber matches or what users are querying. I they aren`t,
the whole system suers and Google makes less money.`,, .ee at.o Google, !bat i. tbe Qvatit,
core ava or i. it Catcvtatea,
!""#$%%.+&/0+5'1//1,)'2/-%56##/0"%.&%(3*%.*5&)0'#4L!,M)*Z.*5&)0MHPIHO ,last
isited Sept. 1, 2009, ,1he Ad\ords system works best or eerybody-adertisers, users,
publishers, and Google too-when the ads we display match our users` needs as closely as
possible.`,.
25
JOlN BA11LLLL, 1lL SLARCl: lO\ GOOGLL AND I1S RIVALS RL\RO1L 1lL RULLS Ol
BUSINLSS AND 1RANSlORMLD OUR CUL1URL ,Portolio 1rade 2005,. ee at.o Joe Nocera,
tvc/ iv Coogte`. Dogbov.e, N.\. 1IMLS, Sept. 13, 2008,
!""#$%%&&&'*4"3-)5'2/-%IPPX%PU%HJ%")2!*/,/14%HJ*/2)0.'!"-, ,In the summer o
2006 . Google pulled the rug out rom under |web business owner Dan Saage, who had
come to rely on its reerrals to his page, Sourcetool|. . \hen Mr. Saage asked Google
executies what the problem was, he was told that Sourcetool`s landing page quality` was
low. Google had recently changed the algorithm or choosing adertisements or prominent
positions on Google search pages, and Mr. Saage`s site had been identiied as one that
didn`t meet the algorithm`s new standards. . Although the company neer told Mr. Saage
what, precisely, was wrong with his landing page quality, it oered some suggestions or
improement, including running ewer AdSense ads and manually typing in the addresses
and phone numbers o the 600,000 companies in his directory, een though their \eb sites
were just a click away. At a cost o seeral hundred thousand dollars, he made some o the
changes Google suggested. No improement.`,. Saage iled suit against Google on an
antitrust theory, which was dismissed in March 2010. ee 1radeComet, LLC . Google, Inc.,
2010 U.S. Dist. LLXIS 20154 ,S.D. N.\. March 5, 2010,,
!""#$%%&&&'2/60"!/65)*)&5'2/-%IPHP%PJ%PX%B//1,)[IP/#3*3/*'#+8.

!"#$ $& '()*+$ *"#+',-"$&). $& /"+# $&& /)("$ "# ,#0().$"#0+#/ &1 2&! +$
0($(3$(0 $2(+) $"3$+3.4
5& 1")6 3-"+*. -+7( 89+/1(($43&*:. 2";( #&$ 9((# 1,--< (="*+#(0 +# $2( >,0+3+"-
.<.$(*6 -")/(-< 9(3",.( ?&&/-( 2". .,33(..1,--< 0(1-(3$(0 $2(* 9< 3-"+*+#/ $2"$
+$. .(")32 )(.,-$. (*9&0< &'+#+&#. ')&$(3$(0 9< $2( @+).$ A*(#0*(#$4 5(;()"-
")$+3-(. 2";( B,(.$+&#(0 !2($2() 9-"#7($ @+).$ A*(#0*(#$ ')&$(3$+&# 3&;(). "--
.(")32 (#/+#( "3$+&#.6 "#0 $2"$ 3&#3-,.+&# 2". #&$ <($ 9((# (*9)"3(0 &# $2(
"''(--"$( -(;(- +# $2( C#+$(0 5$"$(.4
8D
E2( @EF:. /,+0"#3( $& .(")32 (#/+#(.6
')&*&$+#/ $2( 3-(") .('")"$+&# &1 &)/"#+3 "#0 '"+0 )(.,-$.6 .,//(.$. $2"$ .(")32
(#/+#(.: @+).$ A*(#0*(#$ .2+(-0 +. #&$ +#.,)*&,#$"9-(4
8G
H2+-( " 3)("$+;( &)
&''&)$,#+.$+3 -+$+/"#$ 3&,-0 3&#3(+;"9-< "0;"#3( " @+).$ A*(#0*(#$ )+/2$ $&
')&*&$( ')&0,3$. &) '&.+$+&#. !+$2&,$ +#0+3"$+#/ $2"$ $2( ')&*&$+&# 2". 9((#
'"+0 1&)6 .,32 " 32"--(#/( 2". #&$ <($ (-+*+#"$(0 1"-.( "0;()$+.+#/ -"!6 "#0 (;(#
'&-+$+3"- .'("7(). 2";( 9((# )(B,+)(0 $& )(;("- $2(+) 1,#0+#/ .&,)3(.4
8I

Qualified Transparency for Carrier
& Search Engine Practices
J&$2 .(")32 (#/+#(.: )"#7+#/ ')"3$+3(. "#0 3"))+().: #($!&)7 *"#"/(*(#$
.2&,-0 9( $)"#.'")(#$ $& .&*( (#$+$< 3"'"9-( &1 0($(3$+#/ 9+".(0 '&-+3+#/ 9<
$2(.( +#$()*(0+")+(.4
8K
E2()( ")( .&*( +#.$+$,$+&#"- ')(3(0(#$. 1&) $2( 7+#0 &1
*&#+$&)+#/ $2"$ !&,-0 9( #(3(..")< $& "33&*'-+.2 $2(.( /&"-.4 @&) (="*'-(6
$2( @)(#32 F&**+..+&# L"$+&#"-( M( N:O#1&)*"$+B,( ($ 0(. N+9()$(. PFLONQ
2". .(;()"- ')()&/"$+;(. 0(.+/#(0 $& ')&$(3$ $2( ')+;"3< "#0 )(',$"$+&# &1

8D
@)"#7 R".B,"-(6 Rav/ivg., Reavctiovi.v, ava Re.ov.ibitit,6 FNSTSNALM 5E4 N4 UST4 P8VVDQW
@)"#7 R".B,"-( X Y)(# J)"32"6 eaerat earcb Covvi..iov6 KZ FYULSNN N4 UST4 [[\K P8VVIQW
](##+1() A4 F2"#0-()6 . Rigbt to Reacb av .vaievce: .v .roacb to vterveaiar, ia. ov tbe
vtervet6 Z^ _Y@5EUA N4 UST4 [VK^6 [[VK P8VVGQ4
8G
ee J)"32" X R".B,"-(6 eaerat earcb Covvi..iov6 .vra #&$( 8D P0+.3,..+#/ $2( +*'-+3"$+&#. &1
S--(# ?&&0*"#:. !&)7 &# `.$("-$2 *")7($+#/a 1&) .(")32 (#/+#(.6 "#0 2&! $2( _+''.-(<
N($$() &1 8VV8 +#"0(B,"$(-< "00)(..(0 .,32 3&#3()#. +# $2( +#0,.$)<4Q4
8I
O# (")-< 3".(. "--(/+#/ "# "))"< &1 ,#1"+) 3&*'($+$+&# "#0 9,.+#(.. $&)$. 3-"+*. "/"+#.$ .(")32
(#/+#(.6 $2( @+).$ A*(#0*(#$ 2". ')&;(# " 1&)*+0"9-( .2+(-0 "/"+#.$ -+"9+-+$<4 5(")32
(#/+#(. 32")"3$()+b( $2(+) )(.,-$. ". &'+#+&#6 "#0 -&!() 3&,)$. 2";( 9((# )(-,3$"#$ $& '(#"-+b(
$2(* 1&) $2(.( 1&)*. &1 (=')(..+&#4 O# &$2() !&)76 O 2";( 0(.3)+9(0 !2< $2+. @+).$
A*(#0*(#$ 9"))+() $& "33&,#$"9+-+$< .2&,-0 #&$ 9( +#.,)*&,#$"9-(4 5(")32 (#/+#(. $"7(
"0;"#$"/( &1 " !(9 &1 /&;()#*(#$"- +**,#+$+(. $2"$ $2(< !&,-0 9( -&"$2 $& .,))(#0()4
.R r. Rvv.feta6 ^\G C454 \G P8VVDQ "#0 3&/#"$( 3".(. .$"#0 1&) $2( ')&'&.+$+&# $2"$ .,32
+**,#+$+(. 3"# 9( 3&#0+$+&#(0 &# "/)((*(#$ $& 3()$"+# 3&#0+$+&#. &# "# (#$+$<:. .'((324
H2"$(;() $2( 1(0()"- /&;()#*(#$:. !+--6 +$ +. !+$2+# +$. '&!() $& )(/,-"$( )"#7+#/ "#0 )"$+#/
(#$+$+(. +# .&*( !"< !2(# $2(< ")( .& 0(('-< 0('(#0(#$ &# /&;()#*(#$"- "3$+&#4 @)"#7
R".B,"-(6 ..teri./ Reri.itea6 Z ]4 JC54 X ESF_4 NAH D[ P8VVIQ4
8K
O *("# '")$+"- +# $!& .(#.(. &1 $2( !&)0c,#0,-< .(-1d+#$()(.$(06 &) &#-< '")$-< .&-;+#/
')&9-(*. $2(< 3-"+* $& 9( .&-;+#/4

lrench citizens, and to enorce standards o air data practices.
30
CNIL
ensure|s| that citizens are in a position to exercise their rights through
inormation` by requiring data controllers to ensure data security and
conidentiality,` to accept on-site inspections by the CNIL,` and to reply to
any request or inormation.`
31
CNIL also grants indiidual persons the right to
obtain inormation about the digital dossiers kept on them and the use o this
inormation. lor example, CNIL explains that lrench law proides that:
Lery person may, on simple request addressed to the
organisation in question, hae ree access to all the inormation
concerning him in clear language.
Lery person may directly require rom an organisation
holding inormation about him that the data be corrected ,i
they are wrong,, completed or clariied ,i they are incomplete
or equiocal,, or erased ,i this inormation could not legally be
collected,.

30
Law No. 8-1 o January 6, 198, J.C.P. 198, III, No. 44692. Lnglish translation o law as
amended by law o August 6, 2004, and by Law o May 12, 2009,
!""#$%%&&&'2*3,'80%83,).+-3*%+/26-)*"5%)*%C2"QX>HQ\C'#+8, lrench language text
modiied through Law No. 2009-526 o May 12, 2009, J.O., May 13, 2009,
!""#$%%&&&'2*3,'80%,.>2*3,%Y63>5/--)5>*/65%, lrench language consolidated ersion as
o May 14, 2009,
!""#$%%&&&',)1380.*2)'1/6G'80%.8832!S)]")'+/L23+S)]")MVT^_SàSPPPPPPXXNRNP
Z8.5"9/5MHZ8.5"^)Y:+MXINJNXIJRZ2.")1/03)b3)*M23+Z/,+C2"3/*M0)2!S)]").
Commission Nationale de lÌnormatique et des Liberts ,CNIL,, ounded by Law No. 8-1
o January 6, 198, .vra, is an independent administratie lrench authority protecting
priacy and personal data held by goernment agencies and priate entities. Speciically,
CNIL`s general mission consists o ensuring that the deelopment o inormation
technology remains at the serice o citizens and does not breach human identity, human
rights, priacy, or personal or public liberties.
31
CNIL, Rigbt. ava Obtigatiov., !""#$%%&&&'2*3,'80%)*1,35!%"!)>2*3,%031!"5>.*+>
/(,31."3/*5% ,last isited Mar. 12, 2010,. Speciically, Chapter 6, Article 44, o the CNIL-
creating Act proides:
1he members o the Commission nationale de lìnormatique et des
liberts` as well as those oicers o the Commission`s operational serices
accredited in accordance with the conditions deined by the last paragraph o
Article 19 ,accreditation by the commission,, hae access, rom 6 a.m to 9
p.m, or the exercise o their unctions, to the places, premises, surroundings,
equipment or buildings used or the processing o personal data or
proessional purposes, with the exception o the parts o the places,
premises, surroundings, equipment or buildings used or priate purposes.
Law No. 8-1 o January 6, 198, J.C.P. 198, III, No. 44692, ch. 6, art. 44, at 30,
!""#$%%&&&'2*3,'80%83,).+-3*%+/26-)*"5%)*%C2"QX>HQ\C'#+8.

Lery person may oppose that inormation about him is used
or adertising purposes or or commercial purposes.
32

\hile the United States does not hae the same tradition o protecting priacy
prealent in Lurope,
33
CNIL`s aims and commitments could proe worthwhile
models or U.S. agencies.
U.S. policymakers may also continue to experiment with public-priate
partnerships to monitor problematic behaior at search engines and carriers.
lor instance, the National Adertising Diision ,NAD, o the Council o Better
Business Bureaus is a oluntary, sel-regulating body` that ields complaints
about allegedly untruthul adertising.
34
1he ast majority o companies
inestigated by NAD comply with its recommendations, but can also resist its
authority and resole the dispute beore the l1C.
35
Rather than oerwhelming
the agency with adjudications, the NAD process proides an initial orum or
adertisers and their critics to contest the alidity o statements.
36
NAD is part
o a larger association called the National Adertising Reiew Council ,NARC,,
which promulgates procedures or NAD, the Children`s Adertising Reiew
Unit ,CARU,, and the National Adertising Reiew Board ,NARB,.
3

32
CNIL, Rights and Obligations, .vra note 31.
33
James \hitman, 1be 1ro !e.terv Cvttvre. of Prirac,: Digvit, 1er.v. ibert,, 113 \ALL L.J. 1151,
1155 ,2004, ,comparing U.S. and Luropean priacy law,.
34
Seth Steenson, or êr . êr. or vrorea . vrorea. 1be Peote !bo Kee .arerti.er.
ove.t, SLA1L, July 13, 2009, !""#$%%&&&'5,.")'2/-%3+%IIIHUNX.
35
a. ,\hen an ad is brought to their attention, the NAD`s lawyers reiew the speciic claims
at issue. 1he rule is that the adertiser must hae substantiated any claims beore the ad was
put on the air, so the NAD will irst ask or any substantiating materials the adertiser can
proide. I the NAD lawyers determine that the claims aren`t alid, they`ll recommend that
the ad be altered. 1he compliance rate on this is more than 95 percent. But i the adertiser
reuses to modiy the ad ,this is a oluntary, sel-regulating body, not a court o law,, the
NAD will reer the matter to the lederal 1rade Commission. One such l1C reerral
resulted in an >83 million judgment against a weight-loss company.`,.
36
a.
3
NA1IONAL ADVLR1ISING RLVIL\ COUNCIL, 1lL ADVLR1ISING INDUS1R\`S PROCLSS Ol
VOLUN1AR\ SLLl-RLGULA1ION: POLICILS AND PROCLDURLS 2.1,a, ,July 2, 2009, ,1he
National Adertising Diision o the Council o Better Business Bureaus ,hereinater NAD,,
and the Children`s Adertising Reiew Unit ,CARU,, shall be responsible or receiing or
initiating, ealuating, inestigating, analyzing ,in conjunction with outside experts, i
warranted, and upon notice to the parties,, and holding negotiations with an adertiser, and
resoling complaints or questions rom any source inoling the truth or accuracy o
national adertising.`,. 1hough billed as sel-regulation,` it is diicult to see how the policy
would hae teeth were it not sel-regulation in the shadow o an l1C empowered by the
Lanham Act to aggressiely police alse adertising. 1he l1C has seeral mechanisms by
which to regulate unair business practices in commerce. ee, e.g., 15 U.S.C. 45,b, ,2006,

Instead o an Innoation Lnironment Protection Agency ,iLPA,` ,the agency
Lawrence Lessig proposed to supplant the lCC,, I would recommend the
ormation o an Internet Intermediary Regulatory Council ,IIRC,, which would
assist both the lCC and l1C in carrying out their present missions.
38
Like the
NARC, the IIRC would ollow up on complaints made by competitors, the
public, or when it determines that a practice deseres inestigation. I the sel-
regulatory council ailed to reconcile conlicting claims, it could reer complaints
to the l1C ,in the case o search engines, which implicate the l1C`s extant
expertise in both priacy and adertising, or the lCC ,in the case o carriers,.
In either context, an IIRC would need not only lawyers, but also engineers and
programmers who could ully understand the technology aecting data, ranking,
and traic management practices.
An IIRC would research and issue reports on suspect practices by Internet
intermediaries, while respecting the intellectual property o the companies it
inestigated. An IIRC could generate oicial and een public understanding o
intermediary practices, while keeping crucial proprietary inormation under the
control o the companies it monitors. An IIRC could deelop a detailed
description o saeguards or trade secrets, which would preent anyone outside
its oices rom accessing the inormation.
39
Another option would be to allow
IIRC agents to inspect such inormation without actually obtaining it. An IIRC
could create reading rooms` or use by its experts, just as some courts allow
restrictie protectie orders to goern discoery in disputes inoling trade
secrets. 1he experts would reiew the inormation in a group setting ,possibly
during a period o days, to determine whether a gien intermediary had engaged
in practices that could constitute a iolation o priacy or consumer protection
laws. Such reiew would not require any outside access to sensitie
inormation.
I preer not to speciy at this time whether an IIRC would be a priate or public
entity. Lither approach would hae distinct costs and beneits explored ,in
part, by a well-deeloped literature on the role o priate entities in Internet

,giing the commission the authority to register an oicial complaint against an entity
engaged in unair business methods,.
38
It could include a search engine diision, an ISP diision ocusing on carriers, and eentually
diisions related to social networks or auction sites i their practices begin to raise
commensurate concerns.
39
1his is the way that the NAD proceeds. It proides speciic procedures under which the
participants can request that certain sensitie inormation be protected. ee NA1`L
ADVLR1ISING RLVIL\ COUNCIL, 1lL ADVLR1ISING INDUS1R\`S PROCLSS Ol VOLUN1AR\
SLLl-RLGULA1ION 2.4,d,-,e,, at 4-5 ,2009,,
!""#$%%&&&'*.+0)G3)&'/01%PQ790/2)+60)5'#+8 ,discussing procedure or conidential
submission o trade secrets,.

goernance.
40
Regardless o whether monitoring is done by a goernmental
entity ,like CNIL, or an NGO ,like NARC,, we must begin deeloping the
institutional capacity to permit a more rapid understanding o intermediary
actions than traditional litigation permits.
41

It is not merely markets and antitrust enorcement that are insuicient to
constrain problematic intermediary behaior-the common law is also likely to
all short. It is hard to imagine any but the wealthiest and most sophisticated
plaintis` attorneys attempting to understand the tweaks to the Google
algorithm that might hae unairly diminished their clients` sites` salience. 1rade
secrets hae been deployed in the context o other litigation to rustrate
inestigations o black box algorithms.
42
Lxamination o Google`s algorithms
subject to ery restrictie protectie orders would amount to a similar barrier to
accountability. Gien its recent string o litigation ictories, it is hard to imagine
rational litigants continuing to take on that risk. Moreoer, it makes little sense
or a court to start rom scratch in understanding the complex practices o
intermediaries when an entity like the IIRC could deelop lasting expertise in
interpreting their actions.
A status quo o unmonitored intermediary operations is a eritable ring o
Gyges,`
43
tempting them to push the enelope with policing practices which

40
ee, e.g., Philip J. \eiser, vtervet Corervavce, tavaara ettivg, ava etfRegvtatiov, 28 N. K\. L.
RLV. 822, 822 ,2001, ,examining in particular the nature and limits o a key priate
regulator o the Internet: standard-setting organizations and their institution o open,
interoperable standards`,.
41
Google has already recognized the need or some kind o due process in response to
complaints about its labeling o certain websites as harmul` ,due to the presence o iruses
or other security threats at the sites, ia the StopBadware program. ee ZI11RAIN, lU1URL
Ol 1lL IN1LRNL1, .vra note 10, at 11 ,Requests or reiew-which included pleas or
help in understanding the problem to begin with-inundated StopBadware researchers, who
ound themseles oerwhelmed in a matter o days by appeals rom thousands o \eb sites
listed. Until StopBadware could check each site and eriy it had been cleaned o bad code,
the warning page stayed up.`,. Google`s cooperation with the larard Berkman Center or
Internet Research to run the StopBadware program could preigure uture intermediary
cooperation with NGOs to proide rough justice` to those disadantaged by certain
intermediary practices.
42
ee Jessica Ring Amunson & Sam lirsch, 1be Ca.e of tbe Di.aearivg 1ote.: e..ov. frov tbe
]evvivg. r. vcbavav Covgre..iovat tectiov Covte.t, 1 \M. & MAR\ BILL R1S. J. 39, 39-98
,2008, ,|1|he litigation ultimately was utterly inconclusie as to the reason or the 18,000
electronic underotes because discoery targeting the deectie oting system was thwarted
when the oting machines` manuacturer successully inoked the trade-secret priilege to
block any inestigation o the machines or their sotware by the litigants.`,.
43
1he Ring o Gyges is a mythical magical artiact mentioned by the philosopher Plato in
book 2 o his Republic ,2.359a-2.360d,. It granted its owner the power to become inisible
at will. 1hrough the story o the ring, Republic discusses whether a typical person would be
moral i he did not hae to ear the consequences o his actions.` \ikipedia, Rivg of C,ge.,
!""#$%%)*'&3E3#)+3.'/01%&3E3%^3*17/87B41)5 ,last accessed Dec. 1, 2010,.

cannot be scrutinized or challenged. Distortions o the public sphere are also
likely. \hile a commercially-inluenced ast-tracking` or up-ranking` o
some content past others might raise suspicions among its direct ,but dispersed,
ictims, the real issues it raises are ar broader. I an online ecology o
inormation that purports to be based on one mode o ordering is actually based
on another, it sets an unair playing ield whose biases are largely undetectable
by lay obserers. Stealth marketing generates serious negatie externalities that
menace personal autonomy and cultural authenticity. Moreoer, the degree o
expertise necessary to recognize these externalities in the new online
enironment is likely to be possessed by only the most committed obserers.
1his potent combination o expertise and externalities is a classic rationale or
regulation. As Danny \eitzner`s proposal or extreme actinding` ,in the
context o the Google-DoubleClick merger reiew, recognized, only a
dedicated group o engineers, social scientists, attorneys, and computer
scientists are likely to be adept enough at understanding search engine decisions
as a whole to understand particular complaints about them.
44
Someone needs
to be able to examine the iner details o the publicly undisclosed operation o
culturally signiicant automated ranking systems-that is, to watch those who
watch and inluence us.
45

44
ee geveratt,, Danny \eitzner, !bat to Do .bovt Coogte ava Dovbtectic/. ota Coogte to t`. !ora
!itb ove treve actfivaivg .bovt Prirac, Practice., GOOGLL OPLN IN1LRNL1 POLIC\ BLOG,
Oct. 8, 200, !""#$%%+31'25.3,'-3"')+6%(0).+206-(5%*/+)%IPJ:
In the 1990s, the l1C under Christine Varney`s leadership pushed operators
o commercial websites to post policies stating how they handle personal
inormation. 1hat was an innoatie idea at the time, but the power o
personal inormation processing has swamped the ability o a static statement
to capture the priacy impact o sophisticated serices, and the leel o
generality at which these policies tend to be written oten obscure the real
priacy impact o the practices described. It`s time or regulators to take the
next step and assure that both indiiduals and policy makers hae
inormation they need.
\eitzner proposes that |r|egulators should appoint an independent panel o technical, legal
and business experts to help them reiew, on an ongoing basis the priacy practices o
Google.` a. 1he panel would be made up o those with technical, legal and business
expertise rom around the world.` a. It would hold public hearings at which Google
technical experts are aailable to answer questions about operational details o personal data
handling.` a. 1here would be sta support or the panel rom participating regulatory
agencies,` real-time publication o questions and answers,` and |a|n annual report
summarizing what the panel has learned.` a.
45
In the meantime, Google has been deeloping a tool that would help consumers detect i
their Internet serice proider was running aoul o Net neutrality principles.` Stephanie
Condon, Coogteac/ea 1oot Detect. êt itterivg, toc/ivg, CNL1 NL\S, Jan. 28, 2009,
!""#$%%*)&5'2*)"'2/-%XJPH>HJOQX7J>HPHOIHHQ>JX'!"-, ,|1he tool, M-Lab,| is running
three diagnostic tools or consumers: one to determine whether Bit1orrent is being blocked
or throttled, one to diagnose problems that aect last-mile broadband networks, and one to

Why Dominant Search Engines &
Carriers Deserve More Scrutiny than
Dominant Auction Sites & Social Networks
1hose skeptical o the administratie state may ind this proposal to watch the
watchers` problematic. 1hey think o intermediaries as primarily market actors,
to be disciplined by market constraints. loweer, the deelopment o
dominant \eb 2.0 intermediaries was itsel a product o particular legal choices
about the extent o intellectual property rights and the responsibilities o
intermediaries made in legislatie and judicial decisions in the 1990s. As
intermediaries gained power, arious entities tried to bring them to heel-
including content proiders, search engine optimizers, trademark owners, and
consumer adocates. In traditional inormation law, claims under trademark,
deamation, and copyright law might hae posed serious worries or
intermediaries. loweer, reisions o communications and intellectual property
law in the late 1990s proided sae harbors that can trump legal claims sounding
in each o these other areas.
46
Some basic reporting responsibilities are a small
price to pay or continuing enjoyment o such immunities.
An argument or treating internet intermediaries more like regulated entities
owes much to the trail-blazing work o legal realists. Among these, Robert
lale`s work on utilities remains especially inspirational.
4
lale deeloped many
o the theoretical oundations o the New Deal, ocusing on the ways in which
the common law became inadequate as large business entities began ordering

diagnose problems limiting speeds.`,. It remains to be seen whether Google itsel would
submit to a similar inspection to determine whether it was engaging in stealth marketing or
other problematic practices.
46
1 U.S.C. 512,d, ,2000, ,Digital Millennium Copyright Act o 1998 sae harbor,, 4 U.S.C.
230,c,,1, ,2000, ,Communications Decency Act o 199 sae harbor or intermediaries,.
lor critical commentary on the latter, .ee Michael L. Rustad & 1homas l. Koenig, Rebootivg
C,bertort ar, 80 \ASl. L. RLV. 335, 31 ,2005, ,An actiist judiciary, howeer, has radically
expanded 230 by conerring immunity on distributors. Section 230,c,,1, has been
interpreted to preclude all tort lawsuits against ISPs, websites, and search engines. Courts
hae . haphazardly lump|ed| together web hosts, websites, search engines, and content
creators into this amorphous category.`,.
4
Ilana \axman, Note, ate`. egac,: !b, Prirate Proert, i. ôt a ,vov,v for ibert,, 5
lAS1INGS L.J. 1009, 1019 ,lale`s most undamental insight was that the coercie power
exerted by priate property owners is itsel a creature o state power. . By protecting the
owner`s property right . the goernment`s unction o protecting property seres to
delegate power to the owners` oer non-owners, so that when the owners are in a position
to require non-owners to accept conditions as the price o obtaining permission to use the
property in question, it is the state that is enorcing compliance, by threatening to orbid the
use o the property unless the owner`s terms are met.` . .|A|ll property essentially
constitutes a delegation o state power to the property owner.,. lor a powerul application
o these ideas to Internet law, .ee Julie Cohen, ocbver iv C,ber.ace: 1be êr covovic
Ortboao, of Rigbt. Mavagevevt,` 9 MICl. L. RLV. 462 ,1998,.

increasing proportions o the national economy.
48
lale`s crucial insight was
that many o the leading businesses o his day were not extraordinary
innoators that desered` all the proits they made, rather, their success was
dependent on a network o laws and regulation that could easily shit aor
rom one corporate player to another.
49
lale ocused his theoretical work on
the utilities o his time, expounding an economic and philosophical justiication
or imposing public serice obligations on them. Regulatory bodies like state
utility commissions and the lCC all learned rom his work, which showed the
inadequacy o priate law or handling disputes oer inrastructural utilities.
Market adocates may worry that monitoring o search engines and carriers will
lead to more extensie sureillance o the aairs o other intermediaries, like
social networks and auction platorms. 1hey may eel that competition is
working in each o those areas, and should be the oundation o all intermediary
policy. loweer, competition is only one o many tools we can use to
encourage responsible and useul intermediaries. \e should rely on
competition-promotion ia markets and antitrust only to the extent that ,a, the
intermediary in question is an economic ,as opposed to cultural or political,
orce, ,b, the oice` o the intermediary`s user community is strong,
50
and ,c,
competition is likely to be genuine and not contried. 1hese criteria help us
map older debates about platorms onto newer entities.
lor search engines and carriers, each o these actors strongly militates in aor
o regulatory interention. Broadband competition has ailed to materialize
beyond duopoly serice or most Americans. 1here are seeral reasons to
suspect that Google`s dominance o the general purpose search market will
continue to grow.
51
Just as past policymakers recognized the need or common

48
Duncan Kennedy, S!) =".E)5 /8 b.&c /0c A.,) .*+ _/62.6,", 15 LLGAL S1UDILS lORUM
,4, ,1991,.
49
BARBARA lRILD, 1lL PROGRLSSIVL ASSAUL1 ON LAISSLZ lAIRL: ROBLR1 lALL AND 1lL
lIRS1 LA\ AND LCONOMICS MOVLMLN1 ,1998,, aailable at
!""#$%%&&&'!6#'!.0G.0+')+6%2.".,/1%_^:9^C'!"-,.
50
Competition is designed to proide users an exit` option, regulation is designed to gie
them more o a oice` in its goernance. lirschman ALBLR1 O. lIRSClMAN, it ava
1oice: .v avaivg bere of vftvevce, iv RIVAL VIL\S Ol MARKL1 SOCIL1\ AND O1lLR
RLCLN1 LSSA\S 8-80 ,1986, ,describing exit` and oice` as two classic options o
reorm or protest,. 1o the extent exit is unaailable, oice ,inluence, within the releant
intermediary becomes less necessary, to the extent oice is aailable, exit becomes less
necessary.
51
Bracha & Pasquale, eaerat earcb Covvi..iov, .vra note 26, at 119. Section III o the article,
\hy Can`t Non-Regulatory Alternaties Sole the Problem,` addresses the many actors
impeding competition in the search market. Present dominance entrenches uture
dominance as the leading search engine`s expertise on user habits grows to the extent that no
competitor can match its understanding o how to target ads well. a. Since that article was
published, larard Business School Proessor Ben Ldelman has inestigated another sel-
reinorcing aspect o Google`s market power: the non-portability o AdSense data, which

carrier obligations or concentrated communications industries, present ones
will need to recognize carriers` and search engines` status as increasingly
essential acilities or researchers, adertisers, and media outlets.
52

1he parallel is apt because, to use the three dimensions discussed aboe, carriers
and dominant general-purpose search engines a, are just as important to culture
and politics as they are to economic lie, b, conceal key aspects o their
operations, and are essentially credence goods, itiating user community
inluence, and c, do not presently ace many strong competitors, and are
unlikely to do so in the immediate uture. 1he irst point-regarding cultural
power-should lead scholars away rom merely considering economies o scale
and scope and network eects in ealuating search engines. \e need to
consider all dimensions o network power-the ull range o cultural, political,
and social obstacles to competition that a dominant standard can generate.
53

Moreoer, policymakers must acknowledge that competition itsel can drie
practices with many negatie externalities. 1he bottom line here is that someone
needs to be able to look under the hood` o culturally signiicant automated
ranking systems.
\hat about auction platorms, another important online intermediary
54
lere,
a purely economic, antitrust-drien approach to possible problems is more
appropriate. 1o use the criteria mentioned aboe: ,a, a site like eBay is a ery
important online marketplace, but has little cultural or political impact and ,b,
the user community at eBay understands its reputation rankings ery well, and
has shown remarkable capacities or cohesion and sel-organization to protest

makes it diicult or Google customers to apply what they hae learned about their Internet
customers to ad campaigns designed or other search engines. Ben Ldelman, PPC Ptatforv
Covetitiov ava Coogte`. Ma, ôt Co,` Re.trictiov, June 2, 2008,
!""#$%%&&&'()*)+),-.*'/01%*)&5%PNIQPX>H'!"-,. As Ldelman shows, Google has
tried to make the data it gathers or companies sticky,` inextricable rom its own
proprietary data structures.
52
1IM \U, 1lL MAS1LR S\I1Cl ,Knop, 2010, ,promoting separations principle` in the
digital landscape.,.
53
DAVID GRL\AL, NL1\ORK PO\LR: 1lL SOCIAL D\NAMICS Ol GLOBALIZA1ION 45 ,\ale
Uni. Press 2008, ,|1|he network power o Lnglish isn`t the result o any intrinsic eatures
o Lnglish ,or example, it`s easy to learn`,: it`s purely a result o the number o other people
and other networks you can use it to reach. . 1he idea o network power . explains how
the conergence on a set o common global standards is drien by the accretion o
indiidual choices that are ree and orced at the same time.`,.
54
Daid S. Lans, .vtitrv.t ..ve. Rai.ea b, tbe vergivg Ctobat vtervet covov,, 102 N\. U. L.
RLV. COLLOQU\ 285, 291 ,2008, ,Luropean Community law and decisional practice .
impose special obligations and signiicant scrutiny on irms that hae market shares as low as
40 percent.`,. Lans compiles data demonstrating that some leading auction platorms
,such as eBay, are well aboe this market share in Lurope and the U.S. a. ,citing comScore,
M,Metri qearcb 2.0 Ke, Mea.vre. Reort, Dec. 200,
!""#$%%&&&'2/-52/0)'2/-%-)"!/+%-)"!/+'.5#,.

,and occasionally oerturn, policies it dislikes. 1hese actors oerwhelm the
possibility that ,c, competition in the general auction market ,as opposed to
niche auctions, may be unlikely to deelop. I real competitors ail to
materialize due to illicit monopolization, antitrust judgments against Microsot
,and parallel requirements o some orms o operating system neutrality`, can
guide uture litigants seeking online auction platorm neutrality. \hile eBay`s
user community successully pressured Disney to end its 2000 special-
preerence deal with eBay, in the uture antitrust judgments or settlements
might require the ull disclosure o ,and perhaps put conditions on, such deals.
55

In social networks, another area where tipping can quickly lead to one or a ew
players` dominance,
56
the situation is more mixed. \hile Rebecca Mackinnon
and danah boyd hae compared lacebook to a utility, the amously market-
oriented Lconomist magazine has compared it to a country, possibly in need o
a constitution and ormal input rom users. Social networks are closer to search
engines than auction sites with respect to actor a: they are becoming crucial
hubs o social interactions, cultural distribution and promotion, and political
organizing.
5

On the other hand, social networks proide a some leerage to their members
to police bad behaior, opening up oice` options, with respect to actor b, ar
more potent than those aailable to the scattered searchers o Google. A group
named lacebook: Stop Inading My Priacy` became ery popular within
lacebook itsel, catalyzing opposition to some proposed eatures o its Beacon
program in 2008.
58
lacebook`s priacy snaus in early 2009 led the company to
organize ormal user community input on uture alterations to the company`s
terms o serice. On the inal actor, competitie dynamics, it appears that
competition is more likely to deelop in the social network space than in the
broadband, search engine, or auction platorm industries. 1here is a more

55
In 2000, eBay granted special perks to Disney on a platorm within its auction site. Ater
protest rom the eBay community,` the perks ceased. eBay CLO Meg \hitman said o the
special Disney deal: \eè concluded that eBay has to be a leel playing ield. 1hat is a
core part o our DNA, and it has to be going orward.` ADAM COlLN, 1lL PLRlLC1
S1ORL: INSIDL LBA\ 292 ,Back Bay Books 2006,.
56
In early 2008, 98 o Brazilian social networkers used Google`s Orkut, 9 o South
Korean social networkers used Cy\orld, and 83 o American social networkers used
MySpace or lacebook. Lans, .vra note :1 at 292.
5
James Grimmelmann, arivg aceboo/, 94 IO\A L. RLV. 113, 52-59 ,2009,,
!""#$%%&/0E5'()#0)55'2/-%F.-)57103--),-.**%IP%.
58
\illiam McGeeran, Di.cto.vre, vaor.evevt, ava aevtit, iv ociat Mar/etivg, 2009 ILL. L. RLV.
1105, 1120 ,2009,,
!""#$%%&&&',.&'6362')+6%,0)G%#6(,32."3/*5%IPPP5%IPPU%IPPU7R%<2B)G)0.*'#+8.

dierse playing ield here than in the carrier or search space, with more than
4,000 social networks in the United States.
59

Any policy analysis o dominant intermediaries should recognize the sensitie
cultural and political issues raised by them. 1he cultural, communal, and
competitie dynamics surrounding dominant search engines and carriers dey
easy or stereotyped responses. Qualiied transparency will assist policymakers
and courts that seek to address the cultural, reputational, and political impact o
dominant intermediaries.
Conclusion
As Daid Brin predicted in 1he 1ransparent Society, urther disclosure rom
corporate entities needs to accompany the scrutiny we all increasingly suer as
indiiduals.
60
\hile the l1C and the lCC hae articulated principles or
protecting priacy, they hae not engaged in the monitoring necessary to
enorce these guidelines. 1his essay promotes institutions designed to deelop
better agency understanding o priacy-eroding practices. \hether public or
priate, such institutions would respect legitimate needs or business
conidentiality while promoting indiiduals` capacity to understand how their
reputations are shaped by dominant intermediaries.

59
Lans, .vra note 54, at 290.
60
DAVID BRIN, 1lL 1RANSPARLN1 SOCIL1\: \ILL 1LClNOLOG\ lORCL US 1O ClOOSL
BL1\LLN PRIVAC\ AND lRLLDOM ,Basic Books 1999,.

Online Liability for
Payment Systems
By Mark MacCarthy

Introduction
U.S. policy toward the liability o Internet intermediaries or online harms was
set in the late 1990s. It consisted o two parts. 1he irst part was Section 230 o
the 1996 1elecommunications Act, proiding a sae harbor rom indirect
liability or online serice proiders.
1
1his sae harbor is an exception rom a
range o normal liabilities that would apply to traditional proiders o media
content such as broadcasters and newspapers. It does not apply to all
intermediaries or platorm proiders, but to what might be called pure`
Internet intermediaries. 1hat is, it coers intermediaries to the extent they are
proiding serices that are somehow intrinsic to the Internet. Under its terms,
except or requirements o contract law, criminal law, and intellectual property
law, online entities are not responsible or the content o the material that is
ound on their systems as long as it has been proided by another inormation
content proider.
1he second part o the U.S. policy toward Internet intermediary liability was set
out in 1998 with the Digital Millennium Copyright Act.
2
DMCA allows a
complete exemption rom copyright liability or entities inoled in pure
transmission actiities. It also creates a notice-and-takedown regime or web
hosts and other online serice proiders. It also allows recipients o these
notices to challenge them. Upon receipt o a response, the online serice
proiders are required to reinstate the allegedly inringing material unless the
rights holder has iled a legal inringement action. Online serice proiders are
exempt rom liability or good aith remoal o material ollowing a notice. It
also proides or penalties i a rights holder iles a notiication that knowingly

Mark MacCarthy is Adjunct Proessor in the Communications Culture and 1echnology
Program at Georgetown Uniersity. lormerly, he was Senior Vice President or Public
Policy at Visa Inc. Substantial portions o this essay were originally published as Mark
MacCarthy, !bat Pa,vevt vterveaiarie. .re Doivg .bovt Ovtive iabitit, ava !b, t Matter., 25
BLRKLLL\ 1LClNOLOG\ LA\ JOURNAL 1039 ,2010,.
1
4 U.S.C. 230,c,,1, ,2006, ,No proider or user o an interactie computer serice shall
be treated as the publisher or speaker o any inormation proided by another inormation
content proider.`,. 1he interpretation o this proision is quite broad. ee, e.g., Zeran . Am.
Online, Inc., 129 l.3d 32, 330-31 ,4th Cir. 199, ,inding that plainti`s tort claims o
deamation were preempted by 230,. 1he immunity does not extend to criminal law,
contract law, or intellectual property law. 4 U.S.C. 230,e,,1,-,4, ,2006,.
2
1 U.S.C. 512 aaiable at !""#$%%&&&R',.&'2/0*),,')+6%652/+)%HQ%OHI'!"-,.

misrepresents that the material is inringing. DMCA also requires online serice
proiders to hae in place some procedures to respond to repeat inringers,`
including termination o accounts in appropriate circumstances.
3

Many commentators think that DMCA represents a balanced compromise.
4

loweer, controersy persists. Content proiders hae successully lobbied or
laws imposing more robust responsibilities or stopping copyright inringement
on Internet serice proiders ,ISPs,. lrance and the United Kingdom, or
example, hae adopted graduated response` mechanisms.
5
1hese liability
regimes require ISPs to orward copyright inringement notices to alleged
inringers, and to disconnect alleged repeat inringers.
On the other hand, deenders o ciil liberties and the lirst Amendment think
DMCA notice-and-takedown requirements are too strong, arguing that a large
proportion o the complaints iled under the law are improper,
6
and that they
contains an inherent imbalance toward takedown, een when lirst Amendment
alues are implicated.

In another essay in this collection, Brian lolland strongly deends Section 230
as a modiied ersion o Internet exceptionalism,
8
and as proiding the basis or
the deelopment o innoation on the Internet.
9
loweer, it, too, has been

3
1 U.S.C. 512,i, conditions the eligibility o the sae harbor. It applies only i the serice
proider has adopted and reasonably implemented, and inorms subscribers and account
holders o the serice proider`s system or network o, a policy that proides or the
termination in appropriate circumstances o subscribers and account holders o the serice
proider`s system or network who are repeat inringers.` Intermediaries such as Google,
\ou1ube and A1&1 appear to hae established termination policies or copyright
inringement.
4
See, or example, United States and Canada Oeriew,` in RONALD DLIBLR1, JOlN
PALlRL\, RAlAL ROlOZINSKI, AND JONA1lAN ZI11RAIN, ACCLSS CON1ROLLLD 38 ,
2010, and JONA1lAN ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1 AND lO\ 1O S1OP I1 119
,2008,.
5
Lric Panner, |.K. .rore. Crac/aorv ov vtervet Pirate., NL\ \ORK 1IMLS, April 8, 2010 at
!""#$%%&&&'*4"3-)5'2/-%IPHP%PR%PU%")2!*/,/14%PU#30.24'!"-,L52#MHZ5YM+313".,
[IP)2/*/-4[IP(3,,[IP6EZ5"M25). Lric Panner, ravce .rore. !iae Crac/aorv ov êt
Pirac,, NL\ \ORK 1IMLS, Oct. 22, 2009,
!""#$%%&&&'*4"3-)5'2/-%IPPU%HP%IJ%")2!*/,/14%IJ*)"'!"-,L70MH.
6
Jennier M. Urban & Laura Quilter, fficievt Proce.. or Cbittivg ffect.`. 1a/eaorv ôtice. |vaer
ectiov :12 of tbe Digitat. Mittevvivv Co,rigbt .ct, 22 SAN1A CLARA lIGl 1LCl LJ 621 ,2006,.
\endy Seltzer, ree eecb |vvoorea iv Co,rigbt`. afe arbor: Cbittivg ffect. of tbe DMC. ov
tbe ir.t .vevavevt, BLRKMAN CLN1LR RLSLARCl PUBLICA1ION NO. 2010-3, p. 16, March
2010, aailable at !""#$%%#.#)05'550*'2/-%5/,J%#.#)05'28-L.(5"0.2"73+MHOQQQXO.
8
ee chapter 3, .ee at.o l. Brian lolland, v Defev.e of Ovtive vterveaiar, vvvvit,: acititativg
Covvvvitie. of Moaifiea cetiovati.v, 56 U. KAN. L. RLV. 369, 39 ,200,.
9
Remarks by Lawrence Strickling, Assistant Secretary o Commerce or Communications and
Inormation, to Internet Society`s INL1 Series: vtervet 2020: 1be êt ittiov |.er., .rit 2,

controersial. Some argue that it allows ISPs to aoid making socially-desirable
inestments necessary to proide security on their networks.
10
Others think it
allows hosting sites to escape their responsibility or deamation and other
harms caused by people who use their sites to spread alse and damaging
inormation.
11

Commentary on the controersies inoled in these two pillars o the U.S.
policy toward online liability is growing.
12
\ork on whether to reise the
consensus position on intermediary liability is underway.
13

1his essay attempts to contribute to this debate by looking at what payment
systems hae been doing about online liability. 1his will proide an illuminating
perspectie on the debate or a ery straightorward reason: Payment systems
hae operated outside this ramework or online liability. 1hey are not coered
by Section 230 and they are not subject to the notice-and-takedown proisions
o the DMCA. low hae they handled issues relating to the use o payment
systems or illegal actiity online 1his essay explores this question through an
examination o two cases in which they hae been called upon to take steps to
control illegal actiity inoling their payment systems: Internet gambling and
copyright inringement.
Some hae argued that payment systems should hae legal responsibility or
keeping their systems ree o illegal online actiity.
14
Payment systems can keep
track o those who use their system online - both merchants and cardholders
hae contracts with inancial. 1he online transactions using payment systems
can be tracked electronically by type. Goernments and aggrieed parties might
not be able to ind wrong-doers who use payment systems or illegal online
actiity, but the payment system proiders can. 1hey are the least-cost`

2010, araitabte at
!""#$%%&&&'*"3.'+/2'1/G%#0)5)*"."3/*5%IPHP%:*")0*)"=/23)"47PRIUIPHP'!"-,.
10
Doug Lichtman & Lric Posner, otaivg vtervet errice Proriaer. .ccovvtabte, 14 U. ClI. SUP.
C1. LCON. RLV. 221 ,2006,.
11
JOlN PALlRL\ AND URS GASSLR, BORN DIGI1AL 106 ,2008,, and DANILL SOLOVL, 1lL
lU1URL Ol RLPU1A1ION 125-160 ,200,.
12
See, Adam 1hierer, Diatogve: 1be vtvre of Ovtive Ob.cevit, ava ociat êtror/., ARS 1LClNICA,
March 5, 2009, !""#$%%.05")2!*32.'2/-%")2!>#/,324%*)&5%IPPU%PJ%.>803)*+,4>
)]2!.*1)>.(/6">"!)>86"60)>/8>/*,3*)>,3.(3,3"4'.05.
13
See, or instance, Organization or Lconomic Cooperation and Deelopment, 1be covovic
ava ociat Rote of vtervet vterveaiarie., April 2010,
!""#$%%&&&'/)2+'/01%+."./)2+%RU%R%RRURUPIJ'#+8.
14
Ronald J. Mann & Seth R. Belzley, 1be Provi.e of vtervet vterveaiar, iabitit,, 4 \M. & MAR\
L. RLV. 239, 249-50 ,2005,, aailable at
!""#$%%52!/,.05!3#',.&'&-')+6%213%G3)&2/*")*"'213L.0"32,)MHIIOZ2/*")]"M&-,0.

aoider o the damage done by this illegal online actiity and so should bear the
burden o controlling it.
1his perspectie seems wrong to me. Still, payment system practices toward
illegal online actiity on their systems suggest seeral lessons. lirst, regardless o
the precise legal liabilities, intermediaries hae a general responsibility to keep
their systems ree o illegal transactions and they are taking steps to satisy that
obligation. Second, the decision to impose legal responsibilities on
intermediaries should not be based on the least cost aoider principle.
Assessments o intermediary liability must take into account market ailures, as
well as an analysis o costs, beneits and equities. 1hird, i intermediaries are
shouldered with responsibilities to control illegal online actiity, these
responsibilities needed to be clearly spelled out. lourth, i goernments are
going to use intermediaries to enorce local laws, they must harmonize these
local laws.
Part II o this essay outlines a ramework or the analysis o intermediary
liability. 1his ramework calls or a thorough analysis, including an assessment
o market ailure and an analysis o the costs, beneits, and equities inoled in
imposing intermediary liability. Part III applies this ramework to the policies
and practices o payment intermediaries in the areas o Internet gambling and
online copyright inringement. Part IV draws some conclusions rom these
experiences.
Indirect Intermediary
Liability Regimes
Most legal regimes hold parties liable or their own misconduct. In contrast, an
indirect liability regime holds a person responsible or the wrongs committed by
another. 1here are usually seeral parties inoled in an indirect liability regime:
the bad actor, the wronged party and a third party. 1he bad actor is the person
directly inoled in causing the harm to the wronged party. A third party,
neither the bad actor nor the wronged party, is assigned responsibility in an
attempt to preent the harmul conduct o the bad actor or to compensate the
wronged party or the harm. In the case o copyright inringement, or example,
the bad actor would be the inringer, the wronged party would be the record
company that owned the music copyrights, and the third party would be an ISP
or a payment system that acilitates the inringement. Indirect liability can be
imposed through a ariety o legal mechanisms.
15

15
ee, e.g., Douglas Lichtman, otaivg vtervet errice Proriaer. .ccovvtabte, 2 RLG. 54, 59 ,2004,
,proposing that ISP liability or cyber security issues could be established in a regime o
negligence or strict liability, whether it is best implemented by statute or ia gradual
common law deelopment`,, Mann & Belzley, .vra note 14, at 269-2 ,suggesting three
possible regimes: traditional tort regime, a takedown requirement, and a hot list,.

A Framework for Analysis
Indirect liability holds a party responsible or wrongs committed by another
person. \hy should there be any such rule \hy not simply hold the bad actor
responsible 1he economic analysis o indirect liability attempts to answer this
question using some standard economic tools and concepts.
16
A standard
economic ramework considers issues o market ailure, costs and beneits, and
equity to assess the need or an indirect liability regime in speciic cases.
1

Market Failure Analysis
Beore imposing an indirect liability regime, economic analysis asks whether
there is really any market ailure. I there is no market ailure, there is no need
or an indirect liability rule. In particular, there need not be an indirect liability
rule when the law or the wronged party can eectiely reach the bad actor
directly
18
and transaction costs are not signiicant.
Len i the wronged party cannot easily reach a bad actor that a third party can
reach, it is still not necessary to impose liability on the third party. \hen the
wronged party and the intermediary can easily negotiate an arrangement,
eiciency will guide the third party to undertake enorcement eorts on behal
o the wronged party. 1his is a key aspect o a market ailure analysis. Unless
transaction costs interere with contracting, aected parties can allocate liability
eiciently through contractual design.
19

16
ee geveratt, Lichtman & Posner, .vra note 10 ,summarizing this perspectie,, Douglas
Lichtman & \illiam Landes, vairect iabitit, for Co,rigbt vfrivgevevt: .v covovic Per.ectire,
16 lARV. J.L. & 1LCl. 395, 396-99 ,2003,.
1
ee Lichtman & Posner, .vra note 10, at 228-33.
18
1he eectie reach condition is ealuated prior to an assessment o the ability o a third
party to eectiely control the bad actiity. ee ia. at 230-31. I the law or the wronged party
can easily reach the bad actor, then why een consider whether to impose a duty on a third
party O course, the bad actors are neer totally out o reach o the law or wronged parties.
\ith some inite expenditure o resources, perhaps ery large, the direct bad actors could be
brought to justice or harms could be preented. 1he real economic question is whether
those costs are larger than the costs o assigning that enorcement role to a third party. And
this means that the eectie reach condition collapses into the control actor discussed,
inra. Landes and Lichtman put the comparatie point accurately, applied to the speciic case
o contributory copyright liability: lolding all else equal, contributory liability is more
attractie . the greater the extent to which indirect liability reduces the costs o copyright
enorcement as-compared to a system that allows only direct liability.` Lichtman &
Landes, .vra note 16, at 398.
19
Lichtman & Posner, .vra note 10, at 235. Lichtman and Posner also ocus on what the
parties might do: 1he right thought experiment is to imagine that all the releant entities
and all the ictims and all the bad actors can eiciently contract one to another and then to
ask how the parties would in that situation allocate responsibility or detecting and deterring
bad acts.` a. at 25. But there is no need to conduct this thought experiment in the abstract.
lree, equal, and rational parties can bargain to allocate responsibility and so we can answer

Cost Benefit Analysis
Some arguments or indirect liability ollow a least cost analysis. A least cost`
perspectie puts the burden o enorcing the law on the party that can stop the
illegal transactions at the lowest cost. locusing on costs is desirable in order to
create an eicient enorcement regime. In a least cost` ramework, the cost to
the intermediary itsel and to the direct customers o the intermediary must be
taken into account. I ISPs or payment systems hae to incur costs to monitor
their system or illegal content, those costs will be passed down to their direct
customers. \ith the price increase, some customers stop using the serice or
reduce their usage o it. I the serice proided is a network serice, then the
external network eects on other users o the serice rom an oerall reduction
in use also hae to be counted.
20
According to the least cost idea, when these
costs are less than the cost o enorcement actiity by the wronged party or by
enorcement oicials, then liability rests with the intermediary.
1his least-cost analysis is limited. It ignores the size o the harms that can be
aoided by intermediary action. 1he mistake is to think that i eorts by third
parties proide more enorcement than eorts by the wronged parties then it
must be worthwhile or the third parties to take these enorcement steps.
Similarly, it is sometimes thought that i third parties can more easily reach bad
actors than the wronged parties, then they should be required to do so. But this
is wrong. It is almost always possible to spend more on enorcement and obtain
some return. lrom an economic point o iew, the question is whether that
extra spending proides commensurate reductions in damages. 1hereore, the
least cost rule is not the right decision rule, een in a strictly economic analysis.
Instead, a ull cost-beneit analysis is more appropriate.
21

the question o what the parties would do in this thought experiment by looking at what they
actually do. 1he releant inquiry is whether the bargaining situation is ree o signiicant
transaction costs or other obstacles to reaching an agreement.
20
I there are ewer Internet subscribers then the serice is less aluable to e-commerce
merchants as well since there are ewer potential customers.

ee Matthew Schruers, 1be
i.tor, ava covovic. of P iabitit, for 1bira Part, Covtevt, 88 VA. L. RLV. 205, 250-52 ,2002,,
.ee at.o Lichtman & Posner, .vra note 10, at 241-43 ,seeming to minimize the importance o
these external, network eects in assessing liability regimes: Immunizing ISPs rom liability
is not the correct mechanism or encouraging them to proide positie externalities.` a. at
243,. loweer, the loss o the ISP-generated external beneits is a potential cost o assigning
liability that has to be taken into account when assessing whether to assign liability. Mann
and Belzley`s article gets the oerall point right, noting: 1o the extent the regulation aects
conduct with positie social alue, as is likely in at least some o the contexts this essay
discusses, the direct and indirect eects on that conduct must be counted as costs o any
regulatory initiatie.` Mann & Belzley, .vra note 14, at 24.
21
1he least-cost analysis seems to unction like a cost eectieness analysis, where a gien
leel o enorcement is assumed and the question is how that goal can be reached at the
lowest cost. ee Mann & Belzley, .vra note 14, at 250 ,adopting that perspectie as a
mature scheme o regulation that limits the social costs o illegal Internet conduct in the

1here is a dierence between the costs and beneits to priate parties inoled
and the costs and beneits to society. 1he costs and beneits o third party
enorcement eorts all on dierent parties. A wronged party beneits rom
third party enorcement eorts and the third party pays the costs. 1he wronged
party has a natural incentie to hae the third party do as much as possible in
the way o enorcement-een past the point where there is a corresponding
reduction in damages-because the wronged party appropriates the damage
reduction but pays no costs. lrom an economic eiciency point o iew,
enorcement eorts that do not yield a commensurate reduction in damages are
wasted. Priate beneits may not be worth it rom a social point o iew when
balanced against the costs to other parties.
Equity Analysis
1he cost beneit ramework just described lacks a normatie dimension. It does
not take into account questions o airness, rights, and justice. And it does not
consider who deseres the beneit o protection rom harm or who is at ault or
blameworthy or ailing to take preentie measures.
1he iew that an economic eiciency standard, by itsel, is suicient to create
indirect liability is too strong. 1he ocus on parties who had no part in creating
the problem and who are not responsible or the illegal actiity puts a burden
on people who are innocent o any wrong-doing. Burdening innocent people
seems unair, and arguments that justiy this approach on grounds that it is
good or society as a whole iolate widely accepted moral principles and are
unlikely to withstand public scrutiny.
22

\e should require a person to right the wrongs committed by others only i we
think that person is somehow responsible or those wrongs. Determining who
is responsible or righting wrongs committed by others is controersial in both
moral and political philosophy.
23
Libertarians generally maintain that people
need to ix only the problems that they themseles directly created.
24
\ithout

most cost-eectie manner`,. But a ull cost-beneit analysis gies up the assumption o a
ixed beneit goal and takes the alue o beneits into account as well.
22
ee, e.g., JONA1lAN \OLll, AN IN1RODUC1ION 1O POLI1ICAL PlILOSOPl\ 5 ,1996,
,stating that utilitarianism will permit enormous injustice in the pursuit o the general
happiness`,. A more sophisticated indirect or rule utilitarian approach can attempt to meet
this diiculty, but that approach is subject to diiculties o its own. ee geveratt, JOlN RA\LS,
A 1lLOR\ Ol JUS1ICL ,191, ,critiquing utilitarianism,. 1he underlying intuition behind this
alternatie account o social justice is that |e|ach person possesses an iniolability ounded
on justice that een the welare o society as a whole cannot oerride.` a. at 3
23
ee ivfra notes 24-2 and accompanying text.
24
ee Jim larper, .gaiv.t P iabitit,, 28 RLG. 30, 30-31 ,2005, ,arguing that ISPs should be
liable or harms to third parties only i they hae a duty to these parties and that eiciency`
considerations do not oerride the lack o such a duty ounded on justice,. Libertarians

this limitation, it is hard not to slide into a doctrine that requires all actors to
stop misconduct wheneer they can.
25
Others think that one has a duty to
correct injustices to the extent that one participates in an institutional
ramework which produces injustice.
26
Still others beliee in general positie
duties to eliminate harms een when one has no direct role in causing them.
2

Ultimately, the analysis o indirect liability cannot aoid considerations o
airness, rights, and justice. 1he key actors in this assessment will be those that
hae been used traditionally: directness o the inolement by third parties in
actiities that lead to harm to another person, an assessment o the degree o
harm inoled, the knowledge that third parties hae or should hae about the
speciic harm inoled, what their intentions are, whether they are consciously
acting in urtherance o a crime or other illegal act, and other similar
considerations.
28
1hese complicated normatie and empirical questions cannot
be aoided by a single principle that purports to look at costs and beneits
alone.
29

generally reject the idea that we hae positie duties to ameliorate harms we did not cause.
.g., ia.
25
Mann & Belzley, .vra note 14, at 22 ,noting that the principle that liability should be
assigned regardless o blameworthiness easily could shade into judicial doctrines that would
obligate all actors to stop all misconduct wheneer possible` and thinking that this
unbounded principle` is unduly disruptie`,. But it is hard to see how their proposal to
implement indirect liability through regulation wheneer it would be less expensie than
leaing liability with the wronged party would be less disruptie.
26
ee, e.g., 1lOMAS \. POGGL, \ORLD POVLR1\ AND lUMAN RIGl1S 12 ,2002, ,arguing
that those inoled in an institutional order that authorizes and upholds slaery hae a duty
to protect slaes or to promote institutional reorm, een i they do not own slaes
themseles,.
2
ee, e.g., Daid Luban, ]v.t !ar ava vvav Rigbt., iv IN1LRNA1IONAL L1lICS 195, 209
,Charles R. Beitz et al. eds., 1985, ,stating that all humans in a position to eect` a human
right hae an obligation to do so,.
28
Mann and Belzley criticize the myopic ocus on the idea that the inherent passiity o
Internet intermediaries makes it normatiely inappropriate to impose responsibility on them
or the conduct o primary maleasors.` Mann & Belzley, .vra note 14, at 261-62. But
passiity is releant to the knowledge and control actors needed to assess liability rom an
equity point o iew. Lichtman and Landes seem to criticize the ocus o current law on
knowledge, control, the extent o any non-inringing uses, and other actors` because they
are not particularly clear as to why those issues are central.` Lichtman & Landes, .vra note
16, at 405. But these actors are crucial because they relate to the way in which the equity
issues can be resoled.
29
1hese equity considerations can interact with the cost analysis. Consider the ollowing:
suppose transaction costs make it impossible or the wronged parties to negotiate
enorcement deals with a third party-they are too numerous or lack the resources to
compensate the third party. Suppose urther it is possible that the cost saings inoled in
assigning liability to a third party are substantial. And inally stipulate that the third party`s
inolement in the harm is so remote that assigning blame is a mistake. \e might in that

An economic ramework, broadly construed and supplemented with suitable
considerations o equity, can be a useul way to assess the need or indirect
liability or intermediaries in speciic cases. 1he elements o the ramework are
as ollows:
<.0E)" _.3,60) C*.,4535: Are there substantial transaction costs Can
enorcement be achieed without an indirect liability rule Can priate
parties work out enorcement arrangements among themseles Can
third parties eectiely work with law enorcement without an indirect
liability mandate
d/5">D)*)83" C*.,4535: Does the burden on the wronged party or on
law enorcement to take enorcement steps exceed the burden on the
third parties Are the costs o enorcement eorts reasonable in light
o the reduction in harm Are there longer-term or dynamic
considerations to take into account
`Y63"4 C*.,4535: Do third parties exercise such close control oer the
harm that they should be held responsible or its mitigation or
elimination Are they blameworthy or not taking steps against it Is
the harm particularly egregious

Applying the Framework
to Payment Intermediaries
Payment intermediaries hae deeloped and reined policies and practices to
deal with illegal Internet transactions in their payment networks. 1wo general
conclusions can be drawn rom an analysis o these policies and practices.
1he irst is that payment intermediary action has been eectie. As the
ollowing discussions demonstrate, Internet gambling websites hae been
denied access to the U.S. market, and their current and projected reenues are
in decline. As a result o the payment system action in the .ttofv.cov
copyright inringement case, Allomp3.com was conined to a domestic market
and experienced a dramatic reduction in the olume o actiity at its website.
1he second conclusion is that the widespread assumption that payment system
action in this area is simple and almost cost-ree deseres more careul
consideration.
30
1he discussion o payment intermediaries` actiities to control

circumstance neertheless assign liability to the third party. 1he gains to the rest o us are
just too great. loweer, should we not compensate the third party or taking the
enorcement steps he is required to take Assigning indirect liability when there is not this
leel o control or ault to justiy blameworthiness might be so eicient under a cost analysis
that it is worth considering, but in that case the use o compensation mechanisms should
also be considered.
30
ee, e.g., Perfect 10, 494 l.3d at 824 ,Kozinski, J., dissenting,.

illegal actiity on their systems reeals substantial costs that should gie policy
makers pause beore moing ahead with the imposition o an indirect liability
scheme or payment proiders. 1hese include:
1he cost to maintain and enorce an Internet gambling coding and
blocking scheme that is entirely manual and cannot be automated,
1he cost rom oer-blocking legal transactions,
1he cost to screen and check the business actiity o merchants
participating in the payment systems,
1he cost to monitor the use o payment systems or speciic illegal
actiity, where the payment systems are in no better position than
anyone else to conduct this monitoring actiity,
1he cost to assess complaints o illegality, where the intermediary has
no special expertise and is oten less amiliar with the legal and actual
issues than the wronged party and the allegedly bad actor,
1he cost to deend against legal challenges to enorcement actions,
where the challenge typically comes in an o-shore jurisdiction, and
Longer-term costs to the United States rom taking unilateral action in
this area, including the encouragement o copycat regimes in other
areas o law and in other jurisdictions.

1he reasonableness o these costs in light o the beneits achieed has not yet
been seriously studied. Instead, it seems to be assumed that small compliance
costs are justiied by large enorcement beneits. Although precision in the
estimates o costs and beneits is unlikely in this area, a more disciplined
qualitatie analysis is required.
Internet Gambling Legislation
1he deelopment o the Internet as a commercial medium presented a
challenge to local gambling laws. \ith access to the Internet, indiiduals could
reach gambling serices rom their homes, without the need to trael to a
gambling merchant`s physical operation. 1he Internet proided a way or
gambling merchants who were legal in their own jurisdictions to proide serice
to customers in dierent jurisdictions where gambling was not allowed.
1he United States Congress began its consideration o how to react to illegal
Internet gambling in the late 1990s.
31
One early proposal was to put an

31
ee General Accounting Oice, vtervet Cavbtivg, .v Orerrier of tbe ..ve., Dec. 2002,
!""#$%%&&&'1./'1/G%*)&'3")-5%+PJXU'#+8. Many state laws made Internet gambling
illegal and lederal law also appeared to outlaw at least some orms o it in interstate
commerce. But the legal situation was ambiguous with respect to some orms o Internet
gambling. 1he Interstate \ire Act o 1961 applied to Internet gambling and appeared to
prohibit the use o the Internet or the placing o bets or wages on any sporting eent or
contest.` ee 1he Interstate \ire Act ,18 U.S.C. 1084, at

enorcement burden on ISPs. It would hae required ISPs to terminate
domestic Internet gambling merchants and to block oreign Internet gambling
merchants upon request o law enorcement.
32
1his initial eort ailed to pass,
in part because o concerns about the eectieness and appropriateness o
putting an enorcement burden on ISPs.
33

In 2006, Congress passed the Unlawul Internet Gambling Lnorcement Act
,UIGLA,, which imposed a system o indirect liability on inancial institutions
or the purpose o preenting illegal Internet gambling transactions.
34
Prior to
the passage o UIGLA, payment card networks deised a coding and blocking
system in order to manage the risks o Internet gambling.
35
Lach merchant in
the payment system is normally required to identiy its major line o business
and to include a our digit merchant category code` in each authorization
message.
36
lor gambling, this merchant category code was 995.
3
In addition,
merchants were required to use an electronic commerce indicator when an
Internet transaction was inoled.
38
1ogether, these two pieces o inormation

!""#$%%&&&',.&'2/0*),,')+6%652/+)%HX%65275)27HX7PPPPHPXR>>>>PPP>'!"-,. 1he U.S.
lith Circuit Court o Appeals ruled in 2002 that the \ire Act applied only to sports betting
and not to other types o online gambling. See In re MasterCard, 313 l.3d 25 ,5th Cir.
2002,. 1he status o horseracing was similarly unclear. 1he Interstate lorse Racing Act
appeared to allow the electronic transmission o interstate bets. It was amended in
December 2000 to explicitly include wagers through the telephone or other electronic media.
See the Interstate lorse Racing Act ,15 U.S.C. 3001-300, at
!""#$%%&&&',.&'2/0*),,')+6%652/+)%HO%652756#7PH7HO7HP7OQ'!"-,. 1hese statutes
appeared to allow the Internet to be used or both non-sports gambling and or gambling on
horse races. 1he U.S. Department o Justice, howeer, thought, and still thinks, that existing
statutes bar all orms o Internet gambling. ee Letter rom \illiam L. Moschella, Assistant
Attorney General to Rep. John Conyers Jr., July 14, 2003 at
!""#$%%&&&'31.-3*1*)&5'2/-%.0"32,)5%83,)5%;TV7,)"")0>PJHQHR'#+8 ,1he
Department o Justice beliees that current ederal law, including 18 U.S.C. 1084, 1952,
and 1955, prohibits all types o gambling oer the Internet.`,.
32
l.R. 3125 at !""#$%%"!/-.5',/2'1/G%213>(3*%Y6)04%;L2HPN$I$'%")-#%e2HPN-E"Y-&
33
ee the loor debate on l.R. 3125, CR l605-6068, July 1, 2000,
!""#$%%"!/-.5',/2'1/G%213>(3*%Y6)04%^L0HPN$_b;PPH$AONPOX.
34
Unlawul Internet Gambling Lnorcement Act o 2006, Pub. L. No. 109-34, 120 Stat. 1884
,codiied at 31 U.S.C. 5361-536 ,2006,,.
35
linancial Aspects o Internet Gaming: Good Gamble or Bad Bet: learing Beore the
Subcomm. on Oersight and Inestigations o the l. Comm. on linancial Sers., 10th
Cong. 25-2, 34-35 ,2001, |bereivafter linancial Aspects o Internet Gaming learing|
,statement and testimony o Mark MacCarthy, Senior Vice President, Public Policy, Visa,
U.S.A., Inc., ,describing this system o coding and blocking Internet gambling transactions,,
U.S. GLN. ACCOUN1ING OllICL, supra note 31, at 20-25.
36
U.S. GLN. ACCOUN1ING OllICL, .vra note 31, at 22.
3
VISA MLRClAN1 CA1LGOR\ CLASSIlICA1ION ,MCC, CODLS DIRLC1OR\, araitabte at
!""#$%%&&&'+.'65+.'1/G%#0/260)-)*"%2.0+%2.0+7]%-22'#+8.
38

in the authorization message allowed payment networks or issuing banks to
identiy transactions inoling Internet gambling merchants.
39

Gien this system, it was entirely easible or the issuing bank or the payment
network to block Internet gambling transactions. 1he system could
accommodate conlicting laws in dierent jurisdictions in the ollowing way: I
it was illegal in one country, such as the United States, or cardholders to engage
in Internet gambling, then the issuing banks based in that country could decline
authorization requests or all properly coded Internet gambling transactions.
1his would eectiely block these transactions. loweer, the banks in other
countries who permit Internet gambling, such as the United Kingdom, could
allow the use o their cards or Internet gambling by not declining properly
coded Internet gambling transactions.
1he system was limited in detecting nuances in illegal ersus legal Internet
gambling. I a jurisdiction recognized some Internet gambling transactions as
legal and others as illegal, the system would not detect it.
40
1he merchant
category code described a type o business, not the legal status o the
transaction inoled.
41
I a particular jurisdiction allowed casino gambling, but
not sports betting, both transactions would neertheless be labeled 995. And i
the system was set up to block these coded transactions, then both transactions,
legal and illegal, would be blocked.
42

Another weakness in the system was enorcement. I an Internet gambling
merchant realized that his transactions would be blocked in a large jurisdiction
such as the United States, then he would hae eery incentie to hide.
43
Instead
o describing itsel as a gambling merchant, it would just code itsel as a 1-shirt
sales site or some other legal merchant. \ithout the proper merchant category
code, the system was blind and could not eectiely block the merchant`s
transactions.
44

1he payment networks addressed this enorcement issue with a special program
to eriy that Internet gambling merchants coded their transactions correctly.
45

39
a.
40
41
ee VISA MLRClAN1 CA1LGOR\ CLASSIlICA1ION ,MCC, CODLS DIRLC1OR\, .vra note 3
,listing all the MCC codes by merchant type`,.
42
43
a. at 26.
44
a.
45
a. at 31-32. 1he ines or incorrectly identiying authorization requests or online gambling
transactions are set out at page 55 o the Visa International Operating Regulations. VISA,
VISA IN1LRNA1IONAL OPLRA1ING RLGULA1IONS ,April 2010,,

Payment network personnel would test transactions at popular Internet
gambling sites. 1hey would enter a transaction at the web site and track the
transaction through the payment system. 1hey would be able to tell whether the
transaction was coded properly or not ater they identiied the transaction in the
system. I the transaction was not properly coded, the network would contact
the bank that worked with the merchant and tell the bank that its merchant was
out o compliance with the coding rule. 1he payment network would ask the
bank to take steps to bring the merchant into compliance. linally, the network
would retest the site or proper coding.
46

1he UIGLA required payment systems to hae policies and procedures
reasonably designed to stop illegal Internet gambling transactions.
4
1he statute
creates a sae harbor or payment systems that adopt a coding and blocking
scheme.
48
1he lederal Resere Board and the Department o the 1reasury
implemented this sae harbor with a non-exclusie description o one way in
which a payment system can demonstrate that its policies and practices are
reasonably designed to stop illegal Internet gambling transactions.
49
1his non-
exclusie description tracked the existing industry practices.

!""#$%%65.'G35.'2/-%+/&*,/.+%-)02!.*"5%G35.>3*")0*."3/*.,>/#)0."3*1>0)16,."3/*5>
-.3*'#+8fOOQ. In addition, Visa requires online gambling merchants to post certain notices:
a \ebsite or an Online Gambling Merchant must contain . |t|he statement Internet
Gambling may be illegal in the jurisdiction in which you are located, i so, you are not
authorized to use your payment card to complete this transaction.`` a. at 594.
46
U.S GLN. ACCOUN1ING OllICL, .vra note 31, at 32.
4
Unlawul Internet Gambling Lnorcement Act o 2006, Pub. L. No. 109-34, 120 Stat. 1884
,codiied at 31 U.S.C. 5361-536 ,2006,,.
48
12 C.l.R. 233.6,d,,1,,ii, ,2009,.
49
1he code`s releant section reads:
,ii, Implementation o a code system, such as transaction codes and
merchant,business category codes, that are required to accompany the
authorization request or a transaction, including-
,A, 1he operational unctionality to enable the card system operator or the
card issuer to reasonably identiy and deny authorization or a transaction
that the coding procedure indicates may be a restricted transaction, and
,B, Procedures or ongoing monitoring or testing by the card system
operator to detect potential restricted transactions, including-
,1, Conducting testing to ascertain whether transaction authorization
requests are coded correctly, and
,2, Monitoring and analyzing payment patterns to detect suspicious payment
olumes rom a merchant customer . .
a.

Implementation Challenges with
the Internet Gambling Act
UIGLA deines illegal Internet gambling as whateer is illegal under current
U.S. state and lederal law. It thereore continues the uncertainty regarding the
illegality o some Internet gambling actiities.
50
linancial intermediaries hae
the discretion to block or not block these transactions based upon their own
judgment and the strength o the legal arguments presented to them. UIGLA
also proides them with protection rom liability i they oer-block Internet
gambling sites that turn out to be legal. 1he current law thereby allows
substantial oer-blocking and puts substantial discretion in the hands o the
payment companies.
Impact of UIGEA
A large percentage o non-U.S. companies that deried extensie reenues rom
their operations in the United States let the market ater the passage o
UIGLA. All Luropean companies that had been actie in the U.S. market let it
ater the passage o UIGLA.
51
By December 2008, all the publicly-trade online
gambling irms had let the U.S. market, een though most o the priate irms
remained.
52

1hree major Luropean online gambling merchants lost >3 billion in 2006 rom
this withdrawal rom the U.S market.
53
Measured traic at particular sites
declined as well. In September 2006, Party Poker, or example, which deried
much o its traic rom the United States, had an aerage o about 12,000 actie
players. By Noember 2006, that number had dropped to about 4,000.
54

50
1hese uncertainties aect seeral types o gambling, including horse racing, state lotteries,
Indian gaming, and games o skill.
51
Luropean Commission Directorate-General or 1rade, avivatiov Proceavre Covcervivg av
Ob.tacte to 1raae, !itbiv tbe Meavivg of Covvcit Regvtatiov ;C) ô 2,1, Cov.i.tivg of Mea.vre.
.aotea b, tbe |vitea tate. of .verica .ffectivg 1raae iv Revote Cavbtivg errice. Covtaivt, Reort
to tbe 1raae arrier. Regvtatiov Covvittee ;Covvi..iov taff !or/ivg Paer) 59, June 10, 2009,
aailable at !""#$%%"0.+)')2')60/#.')6%+/2,3(%+/25%IPPU%F6*)%"0.+/27HRJRPO'#+8
|hereinater LC Gambling Report|.
52
Casino City, Ovtive Cavbtivg iv tbe |vitea tate. ]vri.aictiov, 2009,
!""#$%%/*,3*)'2.53*/23"4'2/-%F6035+32"3/*5%6*3")+>5".")5%.
53
ee C Cavbtivg Reort, supra note 51, at 9 ,the direct losses in reenue due to the loss o
the US market or just these three companies were aboe >3 billion in 2006.`,.
54
ee \hichPoker.com, |C. ffect.,
!""#$%%&&&'&!32!#/E)0'2/-%5"."5%g:B`C`88)2"5 ,last accessed Oct. 18, 2010,.
\hichPoker attributes the departure o the biggest publicly-traded online poker sites rom
the US market to stock market rules.


Shortly ater UIGLA was signed into law in October 2006 analysts estimated
that the alue o British Internet gambling stocks declined by >.6 billion.
55
In
the 9 months between January 1, 2006 and Noember 1, 2006, just ater the
passage o UIGLA, three major Luropean online gambling irms lost an
estimated 5 o their alue, totaling approximately 8.3 billion euros.
56

An estimate by the Luropean Commission o the likely eolution o the U.S.
market in the absence o the speciic restrictions imposed in 2006, based on an
assumption o a 3 yearly growth, show U.S. Internet gambling accounting or
about >5.8 billion per year in gross reenue in 2006, and reaching almost >14.5
billion in 2012. lollowing the passage o UIGLA, the annual igure declined to
about >4.0 billion in 2006, and by 2012 was estimated to be at only >4.6
billion.
5
UIGLA reduced thus the size o the U.S. market well below what it
would otherwise hae been.
Internet Gambling Assessment
On equity grounds, it seems that the payment system connection to Internet
gambling is too passie to justiy imposing legal responsibility or blocking
illegal Internet gambling. Payment intermediaries are not to blame when others
use their system or Internet gambling because these intermediaries hae no
speciic connection to the actiity other than operating a general purpose
payment system. 1hey do not reap extra proits through special arrangements
with the Internet gambling merchants. Internet gambling transactions are no
dierent rom any other payment card transaction. On pure equity grounds
alone, then, there is no reason to single out these transactions and impose
special legal responsibilities.
A market analysis indicates that there are still some easible enorcement
arrangements that were not established prior to the passage o the UIGLA.
Although intermediaries may not be responsible or their customers` gambling,
many o them are concerned about the social ills connected with the actiity and
want to reduce its prealence.
58
U.S. inancial intermediaries had already reused
to sign up domestic Internet merchants because these merchants were not

55
Lric Panner and leather 1immons, |.K. ee/. Ctobat Rvte. for Ovtive Cavbtivg,
IN1LRNA1IONAL lLRALD 1RIBUNL, No. 2, 2006, at 14, aailable at
!""#$%%&&&'*4"3-)5'2/-%3!"%IPPN%HH%PI%")2!*/,/14%:AS>PI1.-(,)'!"-,. 1he
basis or this decline in share alue was the withdrawal o these irms rom the lucratie US
market and the perception that they would not be able to recoer the reenue lost rom non-
U.S. customers.
56
LC Cavbtivg Reort, supra note 51, at 83.
5
a. at 19.
58
ee ivavciat ..ect. of vtervet Cavivg earivg, .vra note 39, at 25-26 ,statement o Mark
MacCarthy, Senior Vice President, Public Policy, Visa U.S.A., Inc.,.

authorized to act legally in the United States.
59
Some state attorneys general
requested the intermediaries to block oshore gambling actiities, and many
cooperated.
60
1hese agreements did not extend to all inancial institutions and
did not coer all states, but they could hae been extended without imposing a
legislatie requirement.
A cost-beneit analysis o the UIGLA starts with an estimate o its eect on the
amount o illegal Internet gambling actiity. As we hae seen, the legislation did
not eliminate Internet gambling in the United States, but it did reduce it
substantially below what it would otherwise hae been.
1he costs associated with the payment systems` compliance with the legislation
include the costs o maintaining and enorcing an Internet gambling coding and
blocking scheme, which is entirely manual and cannot be automated, as noted
aboe.
Another cost is the oer-blocking problem created by the way in which
payment intermediaries comply with UIGLA. Perectly legal transactions will
likely be blocked because payment intermediaries cannot distinguish them rom
illegal transactions. 1his example illustrates that intermediaries are usually better
than others at monitoring their own systems or business actiity o a certain
type, but not at detecting the illegality o actiity on their systems.
61
1he point
arises in Internet gambling because the codes used by inancial institutions
relect the business actiity o gambling, not its status as legal or illegal. As a
result, the payment systems` policies and procedures, which were adopted to
comply with the Act and which hae been accepted by the implementing
regulations, oer-block and preent perectly legal actiity rom taking place.
62

59
a. at 26, U.S. GLN. ACCOUN1ING OllICL, .vra note 31, at 20.
60
ee, e.g., JACK GOLDSMI1l & 1IM \U, \lO CON1ROLS 1lL IN1LRNL1:
ILLUSIONS Ol A BORDLRLLSS \ORLD 82 ,2006, ,discussing Spitzer`s eorts to
conince eery major American credit card proider and online payment system to stop
honoring web gambling transactions.`,.
61
ee Mann & Belzley, .vra note 14, at 28 ,Surely eBay is more adept at searching and
monitoring its marketplace than 1iany & Co., while eBay probably is not as eectie as
1iany & Co. in distinguishing bona ide 1iany products rom countereits.`,, .ee at.o
Schruers, .vra note 20, at 252 ,|1|he ISP is not the least-cost aoider when it comes to
discoering |illegal| content, it is only well suited or cost aoidance ater it is apprized o the
problem.`,. Schruers adds that in this case, the wronged party may be better suited to the
task o locating the oending content. a. at 252.
62
Mann & Belzley, .vra note 14, at 294. Mann and Belzley hae a useul discussion o this
oer-blocking issue:
|A| risk always exists that imposing additional burdens on intermediaries will
chill the proision o aluable goods and serices. 1hat will be especially
problematic in cases where considerable risk o chilling legal conduct that is

Alternatives to UIGEA
In light o this diiculty, there might be more eectie ways o assigning
liability. 1he new law creates unnecessary conusion by ailing to deine the
term unlawul Internet gambling.` Congressman Barney lrank has introduced
legislation to license and regulate Internet gambling merchants.
63
1he lack o
clarity about which merchants are legal would be resoled through a licensing
process. At best, the system would rely on a list o approed gambling entities
that the payment networks could check beore approing gambling transactions
rom particular Internet merchants.
64

1he new licensing regime proposed in Congressman lrank`s legislation would
be an improement oer the existing system in the short term. But oer time,
the only way payment systems can operate is through a reduction in the
diersity o the laws they must accommodate. 1he U.S. goernment must either
ind other ways to enorce its laws abroad or begin harmonizing its laws with
those o other countries. One solution is an international agreement that would
recognize licensing arrangements in dierent countries as long as they satisied
certain agreed-upon minimum standards.

adjacent to the targeted conduct exists. As discussed below, that might tend
to make the use o intermediaries less plausible in ile-sharing contexts where
determining whether any particular act o ile-sharing is illegal is diicult, and
much more plausible in the gambling context where in many cases
substantially all traic to a particular site likely inoles illegal conduct.
Requiring intermediaries to make those kind |sic| o subjectie decisions
imposes costs not only on the intermediaries ,that must make those
decisions,, but also on the underlying actors whose conduct might be iltered
incorrectly.
a. at 24. 1he Internet gambling case illustrates that determining when a website is engaged
in illegal gambling is not a simple task. It is raught with the kind o subjectie decisions`
that Mann and Belzley are properly concerned about. Payment systems aced with this
diiculty do not to make these subjectie decisions, instead blocking att gambling actiity,
including legal gambling transactions.
63
Internet Gambling Regulation, Consumer Protection, and Lnorcement Act, l.R. 226,
111th Cong. ,2009,.
64
See text o l.R. 226 and discussion at
!""#$%%83*.*23.,5)0G32)5'!/65)'1/G%#0)55%9^C0"32,)'.5#]LW)&5:;MRUO. 1he louse
linancial Serices Committee approed the measure on July 29, 2010. See Sewell Chan,
Covgre.. Retbiv/. it. av ov vtervet Cavbtivg, NL\ \ORK 1IMLS, July 29, 2010 aailable at
!""#$%%&&&'*4"3-)5'2/-%IPHP%PQ%IU%65%#/,3"325%IU1.-(,)'!"-,. 1he reised
legislation contains a ban on the use o credit cards or any Internet gambling, een the
newly-legalized merchants, but debit cards can be used at the licensed sites. 1he text o the
reised legislation is aailable at
!""#$%%83*.*23.,5)0G32)5'!/65)'1/G%<)+3.%83,)%-.0E6#5%Q7IX7IPHP%C-)*+-)*"5>>
A^[IPIINQ%_0.*EHI'#+8

Online Copyright Infringement
1he ideal copyright enorcement mechanism would be or content owners to
sue direct inringers. But oten, direct inringers are too ubiquitous, too small,
and too diicult to ind. 1he result is well-deeloped notions o secondary
liability or copyright inringement that inole intermediaries-as Paul Szynol
dicusses in another essay in this collection. 1hese doctrines o secondary liability
hae eoled substantially oer the past decades.
Legal Context for Intermediary
Liability in Copyright Infringement
Court cases and ederal statute deine some indirect responsibilities o
intermediaries regarding copyright. 1he 1984 Supreme Court decision in ov,
Cor. of .verica r. |virer.at Cit, tvaio., vc.
65
established a standard or assessing
third party liability. Proiders o a technology that can be used or inringing
actiities are not liable when there are substantial non-inringing uses` o the
technology.
66
1he Digital Millennium Copyright Act o 1998 enabled copyright
owners to enorce their existing rights in the Internet context by enlisting the
help o Internet intermediaries.
6
1he key mechanism or gaining the
cooperation o intermediaries is a sae harbor rom secondary liability. ISPs are
gien an exemption rom secondary liability so long as they act as a pure
conduit, proiding only transitory communications and system caching.
68
\eb
hosts and search engines also receie a sae harbor, proided they comply with a
speciic notice-and-takedown procedure.
69
Upon receiing notiication o
claimed inringement, the proider must expeditiously take down or block
access to the material.
0

Successul litigation against peer-to-peer networks in the digital music area also
increased the ability o copyright owners to use third parties to combat
copyright inringement where the third party is airmatiely inoled in
ostering the inringement. In an early ile-sharing case, the Ninth Circuit ound
that the peer-to-peer serice Napster was liable or secondary inringement
based on its control and acilitation o its users` inringement o music
copyrights,
1
1he company subsequently went out o business in its original

65
464 U.S. 41 ,1984,.
66
a. at 442.
6
1 U.S.C. 512 ,2006,.
68
1 U.S.C. 512,a,.
69
1 U.S.C. 512,b,.
0
a.
1
A&M Records, Inc. . Napster, Inc., 239 l.3d 1004 ,9th Cir. 2001,.

orm.
2
More recently, the Supreme Court ound that another peer-to-peer
serice, Grokster, iolated ederal copyright law when it took airmatie steps
taken to oster inringement . by third parties,` such as adertising an
inringing use or instructing how to engage in an inringing use.
3

Against this background arose a question regarding payment systems: Are they
liable or secondary inringement when they are used or direct inringement In
Perfect 10 r. 1i.a vtervatiovat errice ...`v,
4
a subscription-based adult content
website alleged that numerous websites based in seeral countries had stolen its
proprietary images, altered them, and illegally oered them or sale online.
5
In
response to complaints, Visa did not deny payment serices to the allegedly
inringing sites, and Perect 10 brought a contributory and icarious
inringement action against Visa. 1he Ninth Circuit airmed the district court`s
rejection o liability or Visa.
6

In Perfect 10, the Ninth Circuit dismissed the charge o contributory
inringement by ocusing on whether the credit card companies materially
contributed` to the inringement.
1he court said the credit card companies did

not materially contribute to the inringement because they had no direct
connection` to the inringement.
8
1o hae direct connection to the
inringement they would hae had to reproduce, display, or distribute the
allegedly inringing works, which they did not do.
9
Payment serices might
make it more proitable to inringe, but they are too ar remoed in the causal

2
Benny Langelista, â.ter Rvv. Ovt of ire. - ]vage Rvte. .gaiv.t ate, S.l. ClRONICLL, Sept.
4, 2002, at B1.
3
MGM Studios, Inc. . Grokster, Ltd., 545 U.S. 913, 919 ,2005,.
4
494 l.3d 88 ,9th Cir. 200,, .ee Jonathan Band, 1be Perfect 10 1ritog,, 5 COMPU1LR L. RLV.
IN1`L 142 ,200, ,discussing Perfect 10 r. 1i.a vtervatiovat errice ...`v and its relationship to
similar secondary liability cases,. Band summarizes the Visa case:
lere the Ninth Circuit rejected what would hae represented a signiicant
expansion o secondary liability to actors ar remoed rom the inringing
actiity. loweer, unlike the other cases, this case prooked a strong dissent
by respected jurist Alex Kozinski. 1his dissent suggests that the outer edges
o secondary liability remain to be deined.
a. at 14. Judge Kozinski`s dissent is indeed stinging, but it also underestimates the burden
that secondary liability would place on intermediaries. a.
5
Perfect 10, 494 l.3d at 93.
6
a.
a. at 96.
8
a.
9
a.

chain that leads to the actual inringing acts or them to be described as making
a material contribution.
80

1he court made a similar point about icarious liability, inding that the card
companies had no practical ability or right to preent the inringing actiity.
81

\hile credit card serices can exert inancial pressure on the inringing
websites, they cannot stop the actual reproduction or distribution o the
inringing images.
82

In his dissent, Judge Kozinski rejected both arguments.
83
According to Judge
Kozinski, the card companies were directly connected to the inringement
because they proided payment serices.
84
\ithout these payment serices there
would be no inringement.
85
1he card companies had the contractual right to
terminate illegal actiity on their systems, as well as the practical ability to exert
inancial pressure to stop or limit the inringing actiity.
86

1his dissent apparently played a role in a more recent case in which a district
court ound payment processors liable or trademark inringement or ailing to
take down allegedly inringing content. A key element in this case was the
knowledge imputed to the payment processor o inringing actiity that should
hae been apparent rom an analysis o chargeback claims.
8

Payment System Complaint Program
Len though payment intermediaries may not be required to take steps against
online copyright inringement, they hae chosen to do so.
88
Payment systems
cannot monitor their networks or copyright law iolations. 1hey do not hae
the actual basis to conclude that a particular sale o a product is a iolation o

80
a. at 9.
81
a. at 803.
82
a. at 804.
83
a. at 810-11 ,Kozinski, J., dissenting,.
84
a. at 811-12.
85
a.
86
a. at 816-1.
8
Gucci . lrontline No. 9 Ci.6925 ,lB, U.S. District Court or the Southern District, June
23, 2010
88
See generally International Piracy: 1he Challenges o Protecting Intellectual Property in the
21st Century: learing Beore the Subcomm. on Courts, the Internet, and Intellectual
Property o the l. Comm. on the Judiciary, 110th Cong. 3-82 ,200, |hereinater
International Piracy learing| ,statement o Mark MacCarthy, Senior Vice President or
Global Public Policy, Visa Inc., ,proiding this account o payment intermediaries and
intellectual property,.

someone`s copyright.
89
Many music downloads are perectly legal transactions,
but some are not. Distinguishing the two is oten a complex actual and legal
question which payment intermediaries do not hae the expertise or ability to
resole.
1he payment systems hae no way o knowing whether a transaction inoles
copyright inringement without a complaint. 1he payment networks hae thus
deeloped policies and procedures to handle these complaints.
90

1he complaint process starts when a business entity approaches a payment
system with clear, documented eidence o illegal actiity and adequately
identiies the inringing Internet merchant.
91
1he business entity must proide
substantiation that the actiity is illegal and documentation that payment cards
are actually being used or this illegal actiity.
92

1he next step is to assess legality, which can be complex in cross-border
situations.
93
Ater wrestling with these issues, the payment networks deeloped
a policy or cross-border transactions: I a transaction would be illegal in either
the jurisdiction o the merchant or the jurisdiction o the cardholder, the
transactions should not be in the payment system.
94
In cases like copyright
inringement, this means that merchants are responsible or making sure that
the transactions they submit to the payment system are legal in both their
operating jurisdiction and the jurisdiction in which their customer is located.
1his assessment o legality requires the payment network to determine whether
the type o transaction would be illegal in either jurisdiction.
95
Since the acts
and law inoled are oten complex, the payment networks are willing to take
on only the clearest cases o copyright iolation. Once they determine illegality,
the payment proiders do what they reasonably can to assist the complaining
party. Since payment networks do not work directly with merchants, they
typically try to locate the bank that has the merchant account and proide the
complaint to the bank inoled, which usually resoles the issue.
96
In most
cases, either the bank does not want the business and terminates the merchant

89
a. at 6.
90
a. at .
91
1his Section describes the process at Visa, but other payment networks use a similar process.
ee ia. at 85.
92
a.
93
a. at -8.
94
a. at 8.
95
a.
96
a.

or takes other action to bring the merchant into compliance.
9
I the bank does
not take action, the payment networks can take urther enorcement action
against the bank.
98

Allofmp3.com
In some instances, the merchant resists the enorcement eorts o payment
systems, insists on the legality o the underlying actiity, and goes to a local
court to indicate its perceied rights under local law. 1his is what occurred in
the Allomp3.com case.
In 2005, Visa receied a documented complaint rom International lederation
o the Phonographic Industry ,IlPI,, which represents copyright owners based
in more than seenty countries.
99
1he complaint alleged that Allomp3.com, a
website located in Russia, was inringing on the copyrights o IlPI`s members
by allowing unauthorized downloads o music.
100
Visa assessed the legal
situation, in part by obtaining a reiew by outside counsel, and concluded that
the transactions were illegal under local Russian law.
101
1hey were also illegal
under the laws o the ast majority o the merchant`s customers who were
located primarily in the United Kingdom and the United States.
102
In October
2005, the Italian authorities shut down a localized ersion o Allomp3.com,
allomp3.it, and began a criminal inestigation o the Italian site.
103
In addition,

9
a.
98
a. Payment systems hae a oluntary program such as this in place or countereiting
complaints as well. 1his program includes include haing a process in place to respond to
complaints o the use o a payment brand or sales o countereit goods. 1rademark owners
would proide inormation such as a description o the allegedly countereit transaction and
eidence that the payment system brand was inoled, and the payment system would look
into the allegation and take action in according with a publicly stated policy, which could
include suspension o the merchant inoled. 1rademark owners would agree to indemniy
payment systems or steps taken and or legal risk. 1his system is described by IN1A in
Addressing the Sale o Countereits on the Internet,` September 2009 aailable as
attachment 3 in the IN1A Submission On 1he Request lor Public Comment Regarding 1he
Joint Strategic Plan lor IP Lnorcementc or the Oice o the Intellectual Property
Lnorcement Coordinator ,IPLC, through the Oice o Management and Budget , March
24, 2010 aailable at
!""#$%%&&&'&!3")!/65)'1/G%/-(%:9`d%80*72/--)*"5%:*")0*."3/*.,S0.+)-.0EC
55/23."3/*'#+8
99
a.
100
a. ,discussing IlPI`s role,, Nate Anderson, Mv.ic vav.tr, vcovragea 1i.a to Pvtt tbe Ptvg ov
.ttofMP.cov, ARS1LClNICA, Oct. 19, 2006,
!""#$%%.05")2!*32.'2/-%(653*)55%*)&5%IPPN%HP%XPIU'.05.
101
vtervatiovat Pirac, earivg, .vra note 88, at 9 ,statement o Mark MacCarthy, Senior Vice
President or Global Public Policy, Visa Inc.,.
102
a.

the United States 1rade Representatie interened with the Russian goernment
to urge them to shut down Allomp3.com.
104

At the beginning o September 2006, ater appropriate notice, the Russian bank
working with Allomp3.com stopped processing Visa transactions or
Allomp3.com.
105
At the end o September 2006, the bank also stopped
processing transactions rom an ailiated site called all1unes.
106
Ater these Visa
transactions ended, urther conirmation o the site`s illegality was orthcoming,
a Danish court ordered the Internet proider 1ele2 to block its subscribers`
access to allomp3.com, thereby making it harder or potential customers in
Denmark to access the site.
10
MasterCard also cut o payment serices to
allomp3.com.
108
By May o 200, the site`s popularity had plummeted.
109

1he company was all but out o business, but the legal process was just starting.
1he owner o all1unes sued the bank that had stopped processing its Visa
transactions in a Russian court.
110
Visa was a party to that litigation on the side
o the bank.
111
In June 200, the owner won a judgment that the bank had
iolated its contract with the merchant, and the judgment required the bank to
continue to proide processing serices.
112
In response to the bank`s claim that
the merchant was acting illegally, the court determined that there were no
rulings in Russia establishing that all1unes was making illegal use o exclusie
rights belonging to rights holders.
113

103
Press Release, IlPI, .ttofv.cov: ettivg tbe Recora traigbt, June 2, 2006,
!""#$%%&&&'38#3'/01%2/*")*"%5)2"3/*7*)&5%IPPNPNPH'!"-,.
104
ee vtervatiovat Pirac, earivg, .vra note 88, at 26 ,testimony o Victoria A. Lspinel, Assistant
U.S. Rep. or Intellectual Property and Innoation, Oice o the U.S. 1rade Rep., ,\e will
continue to press Russia to shut down and prosecute the operators o illegal \eb sites
operating in Russia, including the successors to the inamous AllOMP3.com.`,.
105
a. at 9 ,statement o Mark MacCarthy, Senior Vice President or Global Public Policy, Visa
Inc.,.
106
a.
10
Press Release, IlPI, êr Covrt etbac/ for .ttofv.cov, Oct. 26, 2006,
!""#$%%&&&'38#3'/01%2/*")*"%5)2"3/*7*)&5%IPPNHPIN'!"-,.
108
BBC News, MP .ite`. rovcber .,.tev cto.e., May 21, 200,
!""#$%%*)&5'((2'2/'6E%I%!3%)*")0".3*-)*"%NNQQINO'5"-.
109
IlPI reported in May 200 that Allomp3 rated outside the top 2000 websites.` Press
Release, IlPI, Potice Darv Raia to. .ttofv.cov Pirate 1ovcber. cbeve, May 21, 200,
!""#$%%&&&'38#3'/01%2/*")*"%5)2"3/*7*)&5%IPPQPOIH'!"-,.
110
Arbitration Court o Moscow 200, A40-0411,06-6-500.
111
a. at 1.
112
a. at 5.
113
a. 1he court stated:

In August 200, another Russian court issued a ruling in a dierent case,
relating to criminal copyright inringement initiated by IlPI against the owner
o Allomp3.com.
114
1his ruling stated that there had not been suicient
conirmation o any illegal actiity by the site`s owner.
115
Len though the
copyright owners had not gien permission to distribute their recorded material,
a Russian collectie rights society ,the Russian Multimedia and Internet Society,
or ROMS by its initials in Russian, was deemed to be operating legitimately
under Russian law.
116
1he court implied that Allomp3.com and similar sites
would be in compliance with Russian law to the extent that they paid or rights
rom this Russian collectie rights society.
11

1hese court cases created a challenge or Visa because the payment system had
responded to a documented complaint o copyright inringement.
118
Despite an
outside reiew that seemed to establish illegality in the local jurisdiction, a local
court ordered a local bank to continue to proide payment serices.
119
\et these
transactions would still be illegal in irtually eery other country in the world.
1o presere its cross-border policy, Visa decided to allow the local bank to
proide only domestic serice to the site inoled in the court case.
120

1ransactions rom customers in other countries would not be allowed.
121

According to Article 49 o the Russian lederation Law On Copyright and
Allied Rights,` it is only the Court that can execute actions in connection
with illegal use o copyrights and allied rights, i there is a lawsuit iled by
exclusie right holders, which the Deendants, VISA and IlPI are not, while
in this case there are no court rulings with the orce o re. ;vaicata establishing
the Plainti`s illegal use o exclusie rights belonging to some right holders.
a. 1he Deendant was Rosbank, the Russian inancial institution licensed by Visa to
authorize merchants in Russia to accept Visa. a.
114
Cheremushkinsky |District Court o Moscow|, 200, No. 1-151-0.
115
a. at 4.
116
a. at 5.
11
a., .ee at.o vtervatiovat Pirac, earivg, .vra note 88, at 99 ,testimony o Victoria A. Lspinel,
Assistant U.S. Rep. or Intellectual Property and Innoation, Oice o the U.S. 1rade Rep.,
,My understanding o the case is that Media Serices, the company that operated all1unes,
was able to successully argue in Russian court that it was not acting illegally because it was
paying royalties to collecting societies, collecting societies that were not authorized by the
rights holders.`,.
118
a. at 80 ,statement o Mark MacCarthy, Senior Vice President or Global Public Policy, Visa
Inc.,.
119
a. at 80-81.
120
a.
121
a. at 81.

Assessment of Payment System Actions
on Online Copyright Infringement
Are payment systems doing enough on their own to respond to online
copyright inringement Does there need to be a system o legal liability or
them to control online copyright inringement using their payment systems
lirst, Perfect 10 properly rejected indirect liability or payment intermediaries.
122

1he inolement o payment networks in copyright iolations is attenuated and
entirely passie. On control grounds, there is simply no way to draw a line
between payment network inolement in allegedly inringing transactions and
inolement in a wide range o other potentially illegal actiities. I they are
liable in this case, why wouldn`t they be liable or all cases o illegal actiity on
their payment systems Unintentionally, Judge Kozinski`s dissent brought out
this implication.
123

But the actual experience o payment intermediaries reeals that things are
neer as simple as remoing inringing material. At best, there is a well-
documented assertion o inringement under the laws o a particular
jurisdiction. Judge Kozinski appears to aor a notice-and-takedown approach,
so that payment intermediaries are not responsible or illegal conduct o which
they are unaware.
124
But as Visa ound in .ttofv.cov, payment card serices
and their associated inancial serice partners can be liable or wrongul
termination o serices in those jurisdictions i they react to an allegation o
inringement by kick|ing| the pirates o their payment networks.`
125

122
Perect 10, Inc. . Visa Int`l Ser. Ass`n, 494 l.3d 88, 98 ,9th Cir. 200,. lor analysis, see
Band, .vra note 4.
123
ee ia. at 824 ,Kozinski, J., dissenting, ,Credit cards already hae the tools to police the
actiities o their merchants, which is why we don`t see credit card sales o illegal drugs or
child pornography.`,. O course, card companies use dierent tools in the case o illegal
drugs and child pornography, namely, proactie monitoring, but it is hard to see on
Kozinski`s analysis why card companies shouldn`t use whateer tools they can to stop illegal
actiity in all cases. ee ia. ,Plainti is not asking or a huge change in the way credit cards
do business, they ask only that deendants abide by their own rules and stop doing business
with crooks. Granting plainti the relie it seeks would not . be the end o Capitalism as
we know it.`,. But it might be the end o payment systems as we know them i indirect
liability or them means an obligation to stop doing business with eeryone who might be
inoled with illegality anywhere. Kozinski attempts to limit his analysis to those cases where
there are special arrangements between bad actors and the payment system, ia. at 819-20, but
nothing in his analysis turns on these special arrangements. 1hese special arrangements turn
out to be risk-based pricing or adult content websites. \ould he really hae oted with the
majority i the price that adult content merchants ace or accepting cards was the same as
the price set or less risky merchants
124
Perfect 10, 494 l.3d at 824 ,Kozinski, J., dissenting,.
125
a. at 81.

Second, there is no market ailure in this situation that would justiy imposing
intermediary liability on payment systems. 1here are aailable arrangements
between payment intermediaries and copyright owners that can reduce the
amount o copyright inringement on the Internet. 1hese arrangements are
inormal, but expanding. 1hey rely on complaints by copyright owners,
ollowed by inestigation and action by intermediaries. 1hey seem to strike a
cost-based balance by putting the burden o discoering inringement on the
copyright owner and triggering action by the third party only ater notiication.
1he arrangements may inole compensating payment intermediaries or
perorming enorcement serices, but i this enables copyright owners to reduce
the harm o copyright inringement, they might ery well pay. I there are extra
eorts, aboe and beyond standard practices, that a particular copyright owner
would like payment intermediaries to make, those eorts should be open to
negotiation. 1here do not seem to be any transaction costs that would preent
the parties rom negotiating adjustments to these arrangements oer time. And
there appears to be no market ailure that would justiy not relying on priate
sector enorcement arrangements.
1hird, gien the legal risks inoled, copyright owners should be willing to
indemniy payment intermediaries or damages resulting rom enorcement
actions against alleged inringers. .ttofv.cov indicates that these legal risks are
not hypothetical. I the copyright owner beliees in the legal soundness o his
case, he should be prepared to assume the risk. It might be one way to assure
that only strong complaints are brought to the attention o the payment
intermediary. An additional mechanism might be to require the presence o a
court or goernmental agency that holds that the actiity inoled is inringing.
A statute could potentially help proide legal immunity to payment
intermediaries when they take good-aith action against alleged inringers. But
U.S. law cannot proide immunity in other jurisdictions, which is where the aid
o global payment intermediaries is needed.
lourth, this case illustrates the need or greater clarity in the legal enironment
in which intermediaries operate. Intermediaries cannot be in the position o
creating new global law through their own interpretation o current statutes.
Again .ttofv.cov suggests the need or een greater harmonization o local
laws that intermediaries are expected to enorce.
In sum, the experience o payment intermediaries indicates that some eorts on
their part to respond to legitimate complaints would be justiied. It is not
appropriate to do nothing in response to allegations o copyright inringement.
1he current complaint procedure and case-by-case response is reasonable. It
could be improed through urther discussions among the parties, urther
recourse to court judgments o inringement, and harmonization o current
international standards.

Conclusion
1he question remains: Should the goernment place an enorcement burden on
payment intermediaries 1he standard least cost analysis suggests that the
adantages o goernment interention sometimes appear to be substantial, but
nothing in the analysis suggests that Internet intermediaries are always the best
ehicle or goernment control. 1he costs, beneits and equities inoled in
speciic cases hae not been adequately assessed. Intermediaries are oten in a
position to oluntarily police their own communities and hae taken steps to do
this without explicit goernment requirements. 1he equities set out in current
law establish a regime that works tolerably well. Len when goernment
requirements are explicit, as in the case o Internet gambling, they are oten
crated to it the architecture and structure o the intermediaries themseles.
\hile some adjustments would improe these legal regimes, nothing suggests
that more liability imposed unilaterally by local goernments would be an
improement.
Greater goernment coordination on the rules that intermediaries must ollow
on the Internet would be an improement. 1o aoid legal liability and to
comply with local laws, payment intermediaries are moing toward accepting
the laws o all jurisdictions. 1hey also hae wide discretion on what actiities to
allow on their systems. But this situation is problematic. Intermediaries are not
the best- situated to decide which rules to ollow. Also, no laws are sel-
interpreting. 1hey oten apply to particular situations in obscure and heaily
act-dependent ways. Intermediaries` lexibility in adjudication leaes room or
priate, strategic, and unaccountable decisions that aect the shape and
direction o online actiity. Coordinated goernment rules are best or an
additional reason: 1he intermediary role does not scale well in a world o
multiple, oerlapping, and conlicting rules. I goernments are going to use
intermediaries to regulate the Internet, they need to coordinate their own laws
to make that role possible.


Fuzzy Boundaries:
The Potential Impact of Vague
Secondary Liability Doctrines
on Technology Innovation
By Paul Szynol
*

Last year, Ninth Circuit Judge Alex Kozinski, with Josh Goldoot rom the
Department o Justice`s Criminal Diision, published an article in the Columbia
Journal o Law and the Arts entitled A Declaration o the Dependence o
Cyberspace.` Its title is a play on the title o John Perry Barlow`s 1996 A
Declaration o the Independence o Cyberspace`, its content, as the title
suggests, is something o an attack on Barlow`s philosophies, and, more
generally, on the idea that the Internet is a unique entity that requires custom
legal treatment. 1he authors make seeral key claims about the law`s
relationship to the internet, but the central argument ocuses on secondary
liability-the copyright doctrine that makes makers o multi-use technologies
legally liable or other people`s inringing uses o their technology.
Broadly stated, the rationale at the heart o the secondary liability doctrine is
this: An entity that knowingly helps to acilitate the commission o an illegal act
,such as copyright inringement, or example, should be penalized or its
contribution to the illegal actiity.
1
I a technology company induces its
customers to use its product or inringing purposes, or instance, both the
users ava the company should be liable or such inringement-the users or
direct inringement and the company or contributory inringement, which is a
species o secondary liability.
1he doctrine is appealing as a practical solution to widespread inringement
because it targets the entities that enable illegal behaior-e.g., the Napsters and
Groksters o the world-and thus eradicates the distribution mechanism that
enables inringement in the irst place. Judge Kozinski and Mr. Goldoot ,I`ll
generally reer to them as the authors` rom here on,, like the moie and
music industries, certainly beliee that the doctrine o secondary liability should
be readily used as a handy and eectie tool or weeding out copyright

Paul Szynol graduated rom Columbia Uniersity, where he studied history and philosophy,
and \ale Uniersity, where he studied intellectual property law.
1
1he speciic theories o secondary liability hae more nuanced elements, such as the
requirements o materiality or contributory inringement and direct inancial beneit or
icarious inringement. Since these elements are not critical to the essay`s main thesis, Iè
aoided spelling them out in detail.

inringement. According to the authors, people who proide powerul tools
that can be used or good or eil hae some responsibility to make sure that
those tools are used responsibly.` Put more bluntly, howeer, i you outlaw the
tool, you needn`t chase ater the users, so in practice it`s less a question o ethics
and more a question o conenience and eiciency.
One o the principal problems with this approach, howeer, is the act that the
boundaries o secondary liability are not precisely set, and, short o extreme
cases, it is not at all clear under what circumstances a product manuacturer will
be liable or secondary inringement. Such wholesale endorsement or
secondary liability doctrines should thereore gie us some pause. lor example,
at what point does a sotware company that deelops a peer-to-peer application
utilized by end users to exchange copyrighted materials begin to contribute` to
the inringement and become secondarily liable Does the company contribute
simply by writing sotware that is merely capable o inringing uses
2
Or does
the company contribute only i the sotware`s primary use is, by design,
inringing Or, urther yet, does the company contribute only i a substantial
portion o the end-users utilize the technology or inringing purposes I so,
how much o the user base must engage in inringing actiity or it to be a
substantial portion
3
Or, as yet another option, does the company contribute`
only i it promotes inringing uses o its sotware And, i that`s the case, how
much promotion is too much promotion lor example, is the adertising
slogan Rip. Mix. Burn.` too much o an inducement to make inringing
copies o music
4

1hese are undamental, starting-point questions about the secondary liability
doctrine, and one would expect that case law or legislation proides a clear
answer to each. \et the law is ambiguous ,and the authors are altogether silent,
on these points. Outside o extreme cases, no one knows with certainty-
including lawyers, judges, company oicers, engineers and academics-when
secondary liability might attach to a product that acilitates the transmission o
copyrighted materials. 1he legal system`s ailure to proide clear guidelines is

2
An argument that the Supreme Court amously rejected in its 1984 Betamax` decision.
ov, Cor. of .verica r. |virer.at Cit, tvaio., vc., 464 U.S. 41 ,1984,.
3
See, or example, the Napster litigation. According to the District Court`s opinion, 8 o
the content on Napster was copyrighted, and irtually all Napster users` transerred
copyrighted content. . c M Recora., vc. r. â.ter, Inc., 114 l. Supp. 2d 896 ,N.D.Ca.
2000,. A decade later, a critical question remains essentially unanswered: low much lower
would those percentages hae to be or a manuacturer to be sae rom secondary liability
4
1he standard introduced in Cro/.ter is clear expression`, which is not much o a lodestar
or someone seeking to gauge risk with any degree o precision. MetroCotar,vMa,er tvaio.
vc. r. Cro/.ter, ta., 545 U.S. 913, 914 ,2005,. One could persuasiely argue that Apple`s
ery large, ery prominent and ery ubiquitous Rip. Mix. Burn.` billboards amounted to
clear expression.`

the equialent o posting a sign on a reeway that says obey the speed limit`
without giing an actual speed.
1he eect is potentially detrimental to the entire technology sector. A clear rule
is a predictable rule, and a predictable rule is one on which innoators can rely
when deeloping a product. \ithout clear guidance rom the legal system, tech
companies are orced to engage in a ingers crossed` product design process,
and, subsequently, ace a market that can be an explosie landmine o
inringement liability. 1he potential economic damage to a company ound
guilty o secondary liability can be substantial, to say the least. Since statutory
damages or copyright inringement range rom >50 to >150,000 per
inringement, a maker o a multi-use technology may conront liabilities on a
scale that can threaten the iability o een the wealthiest corporations. 1he risk
is urther exacerbated by the recent trend o unpredictable and oten ery
bloated damage awards granted to copyright plaintis. Such risk can dissuade
een the most resolute inestors rom marketing their inention-and it can
literally bankrupt the braer among them. 1he loss o a robust distribution tool
harms the content sector, too, since a powerul method or distributing content
to end users will not be brought to market.
Judge Kozinski and Mr. Goldoot are not concerned with the chilling eect that
the legal system`s ambiguity can hae on technology innoation. In act, they
reject the proposition, and conidently point to the pharmaceutical and auto
industries as counter-examples: Both industries hae to comply with legal
regulation yet manuacturers in both industries neertheless innoate.
It`s not a ery persuasie comparison. lirst, the auto industry is hardly a hotbed
o innoation. \e might really like power windows and power steering, but, as
adancements oer prior art, these innoations are an order o magnitude
smaller than the innoation weè seen on the internet. Second, the players in
the auto and pharmaceutical industries are requently dierent rom the players
in the technology sector. It is rare, ater all, i not unheard o, that a single
person inents aluable medicine-the medical R&D process takes place in the
laboratories o some o the wealthiest companies under the sun. In addition,
medical innoation is subject to reiew and approal by goernment regulatory
agencies, so by the time a medicine reaches the market, it has already been
approed by the goernment. Innoation in inormation technology, in
contrast, is oten the result o the proerbial garage inentor who releases the
technology entirely on its own. 1hink o eBay, Napster, Apple, Google and
Microsot, each o which had a modest start in someone`s home or garage at the
hands o one or two people ,and many subsequently acquired similarly
independent garage innoations,. 1he distinction between a multinational
company and a garage inentor is critical. lirst, there is no goernment
imprimatur or multi-use technologies. Second, in contrast to wealthy
companies that can aord sophisticated legal teams, garage inentors typically

lack the economic resources necessary to pay or a comprehensie legal reiew
o product design prior to the product`s release. 1hat inability increases the
likelihood that the garage inentor will-unwittingly-design its product in a
way that leads to legal liability, or the likelihood that, ater releasing the product
and receiing angry threats o litigation, the garage inentor will hae to
backtrack and redesign the product in order to aoid liability. 1hese are ery
expensie measures. I the inentor can aord them, the inentor will hae
spent money that it would hae saed had the law simply been clearer in the
irst place, i the inentor cannot aord them, the outcome is een worse: the
start-up will simply old, thus wasting its inestment costs, while consumers will
miss out on the product altogether.
1hat outcome is bad enough, but it`s the third reason or the comparison`s
inadequacy that should gie all o us some pause: Because the legal landscape
around copyright secondary liability is so unclear, een i the would-be inentor
did hae the resources to hire outside counsel, lack o clarity in the law means
that, unless the product clearly crosses a line, lawyers-no matter how high
their hourly rates-won`t be able to conidently proide the inentor with a
legal imprimatur. In other words, no matter how much a company tries, lack o
clear standards means that its lawyers might get it wrong,` and the company
may ace inringement liability i it releases the product, or incur the costs o
post-release redesign, or both. 1hat is a ery expensie proposition, and its
corollary is clear: laced with potential liability exposure and potential redesign
costs, each o which could igure in the millions or een billions o dollars,
5

some would-be inentors and inestors will, as rational economic actors, orego
the whole enterprise-not because they analyzed the risk and ound it
potentially too costly, but because the law`s ambiguity meant they simply covtav`t
properly analyze the risk in the irst place. Notably, the oregoing outcome will
apply to garage inentors and big companies alike. 1he garage inentor whose
coers won`t be able to withstand the potential cost will retreat to the sound o
a distant death knell, the big company will retreat because it knows that its deep
pockets makes it an attractie target or a lawsuit and thereore may well decide
that the potential litigation and licensing costs, een i not atal, just aren`t worth
it. Again, consumers will miss out on a new product.
An ambiguous secondary liability doctrine also disadantages American
products in a global market: U.S. companies will hae to worry about drowning
in the unpredictable and poorly charted quicksand o secondary liability, while
their international competitors will hae clear legal rules to guide them. 1he

5
It`s worth emphasizing that the billion dollar igure is not hyperbole-just ask SAP, which
recently lost its legal dispute with Oracle and was ordered to pay >1.3 billion in damages.
See Sam Diaz, ]vr,: .P Ore. Oracte 1. ittiov for Co,rigbt vfrivgevevt, ZDNL1, No. 23,
2010. 1he acts o that case are quite dierent rom the examples gien here, o course, but
the award is a ery conspicuous reminder that such astronomical damage awards are a
startling reality o present day copyright litigation.

domestic market suers as well: By creating barriers to entry ,high and
unreliable due diligence costs as well as post-release redesign costs,, the
ambiguity aors entrenched entities oer newcomers. Adocating secondary
liability without remoing the ambiguity also contradicts the authors` claim that
the same set o laws should apply to oline and online worlds: 1he uzzy
secondary liability doctrine which they so strongly espouse in connection with
technology wouldn`t ly in the physical world. lor example, should a car
company be held liable or driers who speed Ater all, it would be easy
enough to add a speed limit compliance chip.` \et auto manuacturers are not
orced to pay any portion o a speeding drier`s ticket. Oline, in other words,
bad actors-the v.er. o technology-are punished or their own transgressions.
Online, howeer, the law chases the manuacturers-and applies ad-hoc,
ambiguous standards to their products. It would seem that the authors want
Internet-speciic laws ater all.
None o this sounds like wise intellectual property policy. 1he legal system has
a constitutional imperatie to incentiize inentors, ater all, and it achiees this
objectie in part by proiding both content producers
6
ava innoators with a
stable and predictable legal climate,

such as the bright line` rule deised by the
Supreme Court in its 1984 ov, ruling.
In its current state, the law threatens to

punish rather than reward those who hae the courage to release an innoatie
technology i that technology may be misused by its adopters and i that
technology has yet to be contemplated and cleared by the judiciary or
legislature. 1hat is not an enironment that encourages innoation. I the
intent o the judiciary and the Department o Justice is indeed to mightily wield
the secondary liability sword across the technology sector, the doctrine must be
clearly deined, so that the rules o engagement are clearly stated and U.S.
innoators can design their products with conidence-not in ear.

6
In Covvvvit, for Creatire ôv1iotevce r. Reia, the Supreme Court acknowledged Congress`
paramount goal in reising the 196 Act o enhancing predictability and certainty o
copyright ownership.` 490 U.S. 30, 49 ,1989,.
ov, r. |virer.at Cit, tvaio., .vra note 2.


399

CHAPTER 7
IS SEARCH NOW AN
ESSENTIAL FACILITY?
An Essential Cultural & Political Facility 401
!"#$% '#()*#+,
The Problem of Search Engines as Essential Facilities:
An Economic & Legal Assessment 419
-,.//",0 12 3#$$,
Some Skepticism About Search Neutrality 435
4#5,( -"655,+5#$$
Search Engine Bias & the Demise
of Search Engine Utopianism 461
7"68 -.+95#$

400 CHAPTER 7: IS SEARCH NOW AN ESSENTIAL FACILITY?

An Essential Cultural
& Political Facility
By Frank Pasquale
*

Many worry about search engines` growing power. low are worldiews being
biased by them Do search engines hae an interest in getting certain
inormation prioritized or occluded
1
Dominant search engines ,DSLs`,
2
are a
key hub o Internet traic. 1hey proide an eer-expanding array o serices.
Google, or instance, just announced its intention to go into trael shopping. As
they amass inormation about their users, calls or regulation hae ocused on
the threats to priacy they generate. Some o these eorts hae been successul,
others look more doubtul. One thing is certain: 1hey are only the beginning o
a struggle oer the rights and responsibilities o key intermediaries. Some hope
that competition law-and particularly the doctrine o essential acilities`-will
lead policymakers to scrutinize search engines actions.
\hen American lawyers talk about essential acilities,` they are reerring to
antitrust doctrine that has tried, at arious points, to make certain bottlenecks`
in the economy proide access on air and nondiscriminatory terms to all
comers. As robust American competition law ades into a secluded corner o
legal history,
3
essential acilities` doctrine still remains, or some scholars, a ray
o hope or intermediary responsibility.
4
Oren Bracha and I helped uel this

Proessor o Law, Seton lall Law School, Visiting lellow, Princeton Center or Inormation
1echnology Policy.
1
ALLX lALAVAIS, SLARCl LNGINL SOCIL1\ 85 ,Polity 2008, ,In the process o ranking
results, search engines eectiely create winners and losers on the web as a whole. Now that
search engines are moing into other realms, this oten opaque technology o ranking
becomes kingmaker in new enues.`,, Chi-Chu 1schang, 1be qveee at Cbiva`. aiav,
BUSINLSS\LLK, Dec. 31, 2008, at
!!!"#$%&'(%%!(()"*+,-,./.0&'(-*+'1('1-23425-#67782579725:8";1, ,Salespeople
working or Baidu drop sites rom results to bully companies into buying sponsored links |a
orm o paid adertising|, say some who hae been approached.`,.
2
\e can proisionally deine a dominant search engine ,DSL`, as one with more than 40
percent market share. Google clearly satisies this criterion in the United States and Lurope.
ee Daid S. Lans, .vtitrv.t ..ve. Rai.ea b, tbe vergivg Ctobat vtervet covov,, 102 N\. U. L.
RLV. COLLOQU\ 285 ,2008, ,reporting market shares or leading internet intermediaries,.
3
BARR\ L\NN, CORNLRLD: 1lL NL\ MONOPOL\ CAPI1ALISM AND 1lL LCONOMICS Ol
DLS1RUC1ION ,John \iley & Sons, Inc. 2010, ,describing the declining impact o American
antitrust law,.
4
Brett lrischmann & Spencer \eber \aller, Reritatiivg ..evtiat acititie., 5 AN1I1RUS1 L.J. 1,
2 ,2008, ,inrastructure subject to substantial access and nondiscrimination norms |has| .
been heaily regulated.`,.
hope in our 2008 article eaerat earcb Covvi..iov, which compared dominant
search engines to railroads and common carriers in the hope that they would be
recognized as inrastructural oundations o the inormation economy.
5
But I
now see that eaerat earcb Covvi..iov, like many other parts o the search engine
accountability literature, tried too hard to shoehorn a wide ariety o social
concerns about search engines into the economic language o antitrust policy.
6

It is now time or scholars and actiists to moe beyond the crabbed ocabulary
o competition law to deelop a richer normatie critique o search engine
dominance.
1his will not be an easy sell in cyberlaw, which tends to uncritically promote
competition and innoation as the highest aims o Internet policy. I a
dominant search engine is abusing its position, market-oriented scholars say,
market orces will usually sole the problem, and antitrust law can step in when
they ail to do so. Len those who aor net neutrality rules or carriers are
wary o applying them to other intermediaries, like search engines. All tend to
assume that the more innoation` happens on the Internet, the more choices
users will hae and the more eicient the market will become. \et these
scholars hae not paid enough attention to the /iva o innoation that is best
or society, and whether the uncoordinated preerences o millions o web users
or low-cost conenience are likely to address the cultural and political concerns
that dominant search engines raise.
In this article, I hope to demonstrate two points. lirst, antitrust law terms ,like
essential acility`, cannot hope to capture the complexity o concerns raised by
an inormation landscape where one company seres as the predominant map
o the web, and simultaneously attempts to exploit that dominance by endlessly
expanding into adjoining ields. Second, I hope to point the way toward a new
concept o essential cultural and political acility,` which can help policymakers
realize the situations where a bottleneck has become important enough that
special scrutiny is warranted. 1his scrutiny may not always lead to regulation-
which the lirst Amendment renders a dicey enterprise in any corner o the
inormation economy. loweer, it could lead us to recognize the importance
o publicly unded alternaties to the concentrated conduits and content-
proiders colonizing the web.

5
Oren Bracha & lrank Pasquale, eaerat earcb Covvi..iov: airve.., .cce.., ava .ccovvtabitit, iv
tbe ar of earcb, 93 CORNLLL L. RLV. 1193 ,2008,.
6
RIClARD POSNLR, AN1I1RUS1 LA\, at ix ,2d ed. 2001, ,Almost eeryone proessionally
inoled in antitrust today-whether as litigator, prosecutor, judge, academic, or inormed
obserer-not only agrees that the only goal o the antitrust laws should be to promote
economic welare, but also agrees on the essential tenets o economic theory that should be
used to determine the consistency o speciic business practices with that goal.`,.
The Limits of Antitrust as Search Policy
Antitrust cases tend to consume a great deal o time, in part because economic
conduct is subject to many dierent interpretations.
One person`s
anticompetitie conduct is another`s eectie business strategy. 1he same
unending ,and indeterminate, arguments threaten to stall discourse on search
policy. lor example, the lederal 1rade Commission`s ,l1C, reiew o the
Google-DoubleClick merger ocused almost entirely on the economic eects
o the proposed combination, rather than the threats to priacy it posed.
8

Search engines are among the most innoatie serices in the global economy.
1hey proide extraordinary eiciencies or adertisers and consumers by
targeting messages to iewers who are most likely to want to receie them. In
order to attract more users, search engines use reenues rom adertising to
organize and index a great deal o content on the Internet. Like the major
broadcast networks, search engines are now beginning to displace. 1hey proide
opportunities to iew content ,organic search results, in order to sell adertising
,paid search results,.
9
Search engines hae prooked antitrust scrutiny because
proposed deals between major search engines ,and between search engines and
content proiders, suggest undue coordination o competitors in an already
concentrated industry.
10

ee Jonathan Zittrain, 1be |vMicro.oft |vRevea,: ar Cav Prerevt tbe Probtev tbat t Cav`t
Patcb ater, 31 CONN. L. RLV. 1361, 1361-62 ,1999, ,=1he main concern in inding a remedy
or |bad monopolist behaiors`| may be time: 1he technology enironment moes at a
lightning pace, and by the time a ederal case has been made out o a problem, the problem
is proen, a remedy ashioned, and appeals exhausted, the damage may already be
irreersible.`,.
8
News Release, l1C, lederal 1rade Commission Closes Google,DoubleClick Inestigation
,Dec. 20, 200,, araitabte at www.tc.go,opa,200,12,googledc.shtm ,1he Commissioners
... wrote that as the sole purpose o ederal antitrust reiew o mergers and acquisitions is
to identiy and remedy transactions that harm competition,` the l1C lacks the legal authority
to block the transaction on grounds, or require conditions to this transaction, that do not
relate to antitrust. Adding, howeer, that it takes consumer priacy issues ery seriously, the
Commission cross-reerenced its release o a set o proposed behaioral marketing
principles that were also announced today.`,.
9
According to the Google corporate home page, |\|e distinguish ads rom search results or
other content on a page by labeling them as sponsored links` or Ads by Google.` \e don`t
sell ad placement in our search results, nor do we allow people to pay or a higher ranking
there.` Google, Inc., Corporate Inormation: Company Oeriew,
www.google.com,corporate, ,last isited Mar. 12, 2010,.
10
lor example, the deal reached between Microsot and \ahoo! that would hae Microsot`s
Bing search engine delier results or searches on \ahoo! has prooked antitrust concerns
both domestically and internationally. ee Christopher S. Rugaber, Micro.oft-Yaboo Deat to ace
1ovgb .vtitrv.t Probe, ABCNL\S, July 29, 2009,
;11>?--%(.11@(1&,(%"'!%+$A*("*+,-;1,@-@+*.@'(!%-52238:B:864.>$%,&*A+%+C1D.;+
+.'1&1A$%1";1,@.
1hose opposed to regulation oten claim that antitrust law oers a more
targeted and eicient response to abuses. As Justice Breyer explained in his
classic work Regvtatiov ava t. Reforv:
|1|he antitrust laws dier rom classical regulation both in their
aims and in their methods . . |1|hey act negatiely, through a
ew highly general proisions robibitivg certain orms o priate
conduct. 1hey do not airmatiely order irms to behae in
speciied ways, or the most part, they tell priate irms what
not to do . . Only rarely do the antitrust enorcement
agencies create the detailed web o airmatie legal obligations
that characterizes classical regulation.
11

Gien the lack o search engine regulation in the U.S., actual and threatened
antitrust inestigations hae been a primary goernment inluence on Google`s
business practices as its dominance in search grows. Many beliee that the
Department o Justice`s ,DOJ, suspicion o the company`s proposed joint
enture with \ahoo! in the search adertising ield eectiely scuttled the deal
by late 2008.
12
loweer, antitrust enorcement appears less promising in other
aspects o search.
13
1his section discusses the limits o antitrust in addressing
the cultural and political dilemmas raised by Google`s proposed Book Search
deal with publishers,
14
and its dominance o online adertising.

11
S1LPlLN BRL\LR, RLGULA1ION AND I1S RLlORM 156-5 ,larard Uni. Press 1982,. vt
.ee A. Douglas Melamed, .vtitrv.t: 1be êr Regvtatiov, 10 AN1I1RUS1 13, 13 ,1995,
,describing two paradigms,` the law enorcement model and the regulatory model, and the
shit o antitrust law rom the ormer to the latter,.
12
Nicholas 1hompson & lred Vogelstein, 1be Ptot to Kitt Coogte, \IRLD, Jan. 19, 2009, at 88,
araitabte at www.wired.com,techbiz,it,magazine,1-02,_killgoogle ,noting that antitrust
scrutiny culminated in a hearing in which the DOJ threatened to bring an antitrust case
against Google and that one prominent DOJ attorney expressed the iew that Google
already is a monopoly,.
13
Daniel Rubineld, ovvaatiov. of .vtitrv.t ar ava covovic., iv lO\ 1lL ClICAGO SClOOL
OVLRSlO1 1lL MARK: 1lL LllLC1 Ol CONSLRVA1IVL LCONOMIC ANAL\SIS ON U.S.
AN1I1RUS1 51, 5 ,Robert Pitosky ed., 2008, ,describing how conseratie economics has
ostered a tendency to downplay enorcement in dynamic technological industries in which
innoation issues play a signiicant role`,.
14
Despite the DOJ`s interention to aect the terms o the proposed settlement, many leading
antitrust experts hae argued that the settlement would not iolate the antitrust laws. ee, e.g.,
Liner Llhauge, !b, tbe Coogte oo/. etttevevt . ProCovetitire 58 ,larard Law Sch., Law &
Lcon. Discussion Paper No. 646, larard Law Sch., Pub. Law & 1heory Research Paper
No. 09-45, 2009,, araitabte at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG768325H ,1he settlement does
not raise rial barriers to oering |many| books, but to the contrary lowers them. 1he
output expansion is particularly dramatic or out-o-print books, or which there is currently
no new output at all.`,.
Priacy concerns are nearly impossible to address within the economic models
o contemporary competition law. Antitrust scrutiny did little to address the
priacy concerns raised when Google proposed to merge with the web
adertising irm DoubleClick.
15
1he proposed deal prooked a complaint rom
the Llectronic Priacy Inormation Center ,LPIC,. LPIC claimed that Google`s
modus operandi amounts to a deceptie trade practice`:
Upon arriing at the Google homepage, a Google user is not
inormed o Google`s data collection practices until he or she
clicks through our links. Most users will not reach this page
. . Google collects user search terms in connection with his
or her IP address without adequate notice to the user.
1hereore, Google`s representations concerning its data
retention practices were, and are, deceptie practices.
16

One key question raised by the proposed merger was whether priacy and
consumer protection concerns like these can be addressed by traditional
antitrust analysis.
1
Priacy law expert Peter Swire argued that they can, because
priacy harms reduce consumer welare . |and| lead to a reduction in the
quality o a good or serice.`
18
Swire belieed that consumers would be worse
o ater the merger because o the unparalleled digital dossiers the combined
entity could generate:
Google oten has deep` inormation about an indiidual`s
actions, such as detailed inormation about search terms.
Currently, DoubleClick sets one or more cookies on an
indiidual`s computers, and receies detailed inormation about
which sites the person isits while suring. DoubleClick has

15
Dawn Kawamoto & Anne Broache, 1C .ttor. Coogte-DovbteCtic/ Merger to Proceea, CNL1
NL\S, Dec. 20, 200, ;11>?--'(!%"*'(1"*+,-IJKL.@@+!%LM++/@(LN+$#@(K@&*)L
,(A/(AL1+L>A+*((F-5722L72564BL:55B:B7";1,@ ,describing U.S. authorities` blessing o the
proposed deal,.
16
ee Complaint and Request or Injunction, Request or Inestigation and or Other Relie, v
re Google Inc. and DoubleClick, Inc., No. 01-010 ,l1C Apr. 20, 200,, araitabte at
;11>?--(>&*"+A/->A&O.*D-C1*-/++/@(-(>&*4*+,>@.&'1">FC at 9 |hereinater Google, Inc.
and DoubleClick Complaint|.
1
ee Sia Vaidhyanathan, 1he Googlization o Lerything, Google and DoubleClick: A
Bigger Antitrust Problem than I lad Imagined,
www.googlizationoeerything.com,200,10,google_and_doubleclick_a_bigge.php ,Oct.
21, 200, 16:05 LS1,.
18
Peter Swire, Protectivg Cov.vver.: Prirac, Matter. iv .vtitrv.t .vat,.i., C1R. lOR AM. PROGRLSS,
Oct. 19, 200, www.americanprogress.org,issues,200,10,priacy.html ,italics omitted,.
broad` inormation about an indiidual`s actions, with its
leading ability to pinpoint where a person surs.
19

Initial points o contention include ,a, the deinition o the products at issue,
and ,b, how to weigh the costs and beneits o a merger. 1he combined
company would hae dierent segments o customers` in a two-sided
market:
20
,1, searchers trying to ind sites, and ,2, ad buyers trying to reach
searchers. Swire contends that many people care about priacy, and |i|t would
be illogical to count the harms to consumers rom higher prices while excluding
the harms rom priacy inasions-both sorts o harms reduce consumer
surplus and consumer welare in the releant market.`
21

loweer, the web searcher category not only consists o consumers who care
about priacy, but also includes many people who do not highly alue it or who
actiely seek to expose their inormation in order to receie more targeted
solicitations. According to Lric Goldman`s work on personalized search, some
may een consider the gathering o data about them to be a serice.
22
1he more
inormation is gathered about them, the better intermediaries are able to sere
them releant ads. Many economic models o web publication assume that
users pay` or content by iewing ads,
23
they may eectiely pay less i the
adertisements they iew bear some relation to things they want to buy. So
while Swire models adertising and data collection as a cost to be endured,

19
a. According to Swire, |i| the merger is approed, then indiiduals using the market
leader in search may ace a search product that has both deep` and broad` collection o
inormation. lor the many millions o indiiduals with high priacy preerences, this may be
a signiicant reduction in the quality o the search product-search preiously was
conducted without the combined deep and broad tracking, and now the combination will
exist.` a.
20
lor a deinition o two-sided market, see Nicholas Lconomides & Joacim 1ag, êt êvtratit,
ov tbe vtervet: . 1roiaea Mar/et .vat,.i. 1 ,NL1 Inst., \orking Paper No. 0-45, N.\. Uni.
Law and Lcon., Research Paper No. 0-40, 200,,
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG7273757 ,|P|latorms sell
Internet access serices to consumers and may set ees to content and applications proiders
on the other side` o the Internet.`,. In the search engine context, consumers pay` by
attending to ads, and ad-purchasers pay Google or the chance to get ad iewers` attention.
21
Swire, .vra note 18.
22
Lric Goldman, . Coa.eav .vat,.i. of Mar/etivg, 2006 \IS. L. RLV. 1151, 1162-64 ,1hree
components determine an indiidual consumer`s utility rom a marketing exposure: ,1, the
consumer`s substantie interest in the marketing, ,2, the consumer`s nonsubstantie reactions
to the marketing exposure, and ,3, the attention consumed by ealuating and sorting the
marketing. . |A| consumer may derie utility rom the rote act o being contacted by
marketers or exposed to the marketing, regardless o the marketing content.`,.
23
Daid S. Lans, 1be covovic. of tbe Ovtive .arerti.ivg vav.tr,, RLV. NL1\ORK LCON. 359,
359 ,2008,, araitabte at www.bepress.com,rne,ol,iss3,2 ,describing how many o the top
websites hae adopted the ree-t` model where the publisher generates traic by not
charging or readers but then sell that traic to adertisers,.
Google and DoubleClick argue that the resulting personalized ads sere
customers. 1heir arguments preailed, and Google oicially acquired
DoubleClick in 2008.
24

Antitrust law is ill prepared to handle a market` where some percentage o
consumers consider loss o priacy a gain and others consider it a loss.
Lconomic reasoning in general alters in the ace o externalities, but usually we
can all agree that, say, pollution is a harm ,or negatie externality, and lowers
are a boon ,or positie externality,. Priacy preerences are much more
idiosyncratic.
Critics o the merger do hae a response to this problem o dierse
preerences-they can shit rom characterizing lost priacy as a cost o web
searching to describing it as a reduction in the quality o the serices oered by
the merging entities.
25
Douglas Kysar`s work on the product-process
distinction is encouraging here. Kysar has claimed that consumers should hae
a right to make choices o products based on how the products are made, not
just how well they work.
26
Kysar argues in aor o acknowledging and
accommodating |consumer| process preerences within policy analysis, gien
the potential signiicance that such preerences may sere in the uture as
outlets or public-minded behaior.`
2
Neertheless, the aluation problems
here are daunting. low are we to determine how much consumers are willing
to pay to aoid priacy-eroding companies
28

Perhaps, as Lisa leinzerling and lrank Ackerman suggest in their book Pricete..,
we should stop een trying to pretend that these decisions can be made on

24
ee Press Release, Google Inc., Google Closes Acquisition o DoubleClick ,Mar. 11, 2008,,
araitabte at www.google.com,intl,en,press,pressrel,20080311_doubleclick.html.
25
Both Supreme Court precedent and DOJ guidelines support this approach. ee Nat`l Soc`y
o Pro `l Lng`rs . United States, 435 U.S. 69, 695 ,198, ,1he assumption that
competition is the best method o allocating resources in a ree market recognizes that all
elements o a bargain-quality, serice, saety, and durability-and not just the immediate
cost, are aorably aected by the ree opportunity to select among alternatie oers.`,, U.S.
DLP`1 Ol JUS1ICL, lORIZON1AL MLRGLR GUIDLLINLS 4, at 30-32 ,199, ,eicient market
behaior is indicated by lower prices, new products, and improed quality`,.
26
Douglas A. Kysar, Preferevce. for Proce..e.: 1be Proce..,Proavct Di.tivctiov ava tbe Regvtatiov of
Cov.vver Cboice, 118 lARV. L. RLV. 526, 529 ,2004, ,|C|onsumer preerences may be heaily
inluenced by inormation regarding the manner in which goods are produced.`,.
2
a. at 534.
28
Christopher \oo has demanded this kind o accounting in the context o net neutrality. ee
Christopher \oo, e,ova êtror/ êvtratit,, 19 lARV. J.L. & 1LCl. 1, 54 ,2005, ,1here is
nothing incoherent about imposing regulation to promote alues other than economic
welare. . |but| such a theory must proide a basis or quantiying the noneconomic
beneits and or determining when those beneits justiy the economic costs.`,.
anything approaching a purely economic basis.
29
Lngaging in a cost-beneit
analysis diminishes priacy`s status as a right. 1hough many scholars hae
compellingly argued or broader oundations or competition law, the
mainstream o contemporary antitrust policy in the United States cannot
accommodate such concerns. Antitrust`s .vvvvv bovvv is the maximization o
consumer welare,` and this measure o eiciency is notoriously narrow.
30
lor
example, the DOJ was hard pressed to adequately actor in a basic democratic
commitment to dierse communicatie channels during many media mergers.
31

Gien antitrust doctrine`s pronounced tendency to suppress or elide the cultural
and political consequences o concentrated corporate power, the Bureau o
Competition and the Bureau o Lconomics within the l1C are ill-equipped to
respond to the most compelling issues raised by search engines.
32
1he Google-
Doubleclick merger proceedings ultimately ended with an oerwhelming win
or Google at the l1C.
33
1his outcome was all but ineitable gien the
oundations o contemporary antitrust doctrine,
34
and is the logical outgrowth
o oerreliance on legal economic theory that uncritically priileges market

29
lrank Ackerman & Lisa leinzerling, Priceless: On Knowing the Price o Lerything and
the Value o Nothing 8-9 ,1he New Press 2004,.
30
ee Maurice L. Stucke, etter Covetitiov .arocac,, 82 S1. JOlN`S L. RLV. 951, 1001 ,2008,
,obsering the primacy o allocatie eiciency in antitrust analysis,. Stucke notes that
|b|ehind allocatie eiciency`s aade o positiism lie |many| moral questions . .` a. ee
at.o Julie L. Cohen, êtror/ torie., 0 LA\ & CON1LMP. PROBS. 91, 92 ,200, ,\hat makes
the network good can only be deined by generating richly detailed ethnographies o the
experiences the network enables and the actiities it supports, and articulating a normatie
theory to explain what is good, and worth presering, about those experiences and
actiities.`,.
31
ee C. Ldwin Baker, Meaia Covcevtratiov: Cirivg | ov Devocrac,, 54 lLA. L. RLV. 839, 85
,2002, ,|1|he dominant antitrust ocus on orer orer ricivg can be distinguished rom orer
orer tbe covtevt araitabte for cov.vver cboice. In the currently dominant paradigm, a merger that
dramatically reduced the number o independent suppliers o a particular category o
content-say, news or local news or Black actiist news-creates no antitrust problem i, as
likely, it does not lead to power to raise prices.`,.
32
ee S1A1LMLN1 Ol 1lL lLDLRAL 1RADL COMMISSION CONCLRNING
GOOGLL,DOUBLLCLICK, l1C lile No. 01-010 ,l1C Dec. 20, 200,, araitabte at
;11>?--!!!"C1*"/+O-+%-*.%(@&%1-2972792-297552%1.1(,('1">FC |hereinater S1A1LMLN1
Ol l1C CONCLRNING GOOGLL,DOUBLLCLICK| ,Although |priacy concerns| may present
important policy questions or the Nation, the sole purpose o ederal antitrust reiew o
mergers and acquisitions is to identiy and remedy transactions that harm competition.`,.
33
a.
34
Maurice Stucke describes and critiques this bias in some detail. ee Stucke, .vra note 30, at
1031 ,describing a mishmash o neoclassical economic theory, ignettes o zero-sum
competition, and normatie weighing o the anticompetitie ethereal-deadweight welare
loss-against the conjectures o procompetitie eiciencies` at the core o too much
antitrust law and theory,. Among his many important contributions to the literature, Stucke
makes it clear that competition policy includes ar more goals and tactics than antitrust
enorcement alone. a. at 98-1008.
outcomes.
35
As long as contemporary doctrine holds that antitrust is singularly
ocused on the consumer welare` a proposed transaction will generate,
36

antitrust policymakers will be unable to address the cultural and political
consequences o consolidation in the search industry.
Antitrust challenges to the proposed settlement o a copyright lawsuit by
authors and publishers against Google`s Book Search program are likely to be
similarly constrained.
3
As in the Google-Doubleclick merger, the priacy
implications o Google`s proposed deal with publishers are proound.
38
Anyone
who cares about public input into the uture o access to knowledge should
approach the potential deal here warily, een i the prospect o constructing a
digital Library o Alexandria tempts scholars.
39
As larard librarian Robert
Darnton has argued, only a naie optimist could ignore the perils o haing one
proit-drien company eectiely entrusted with a comprehensie collection o
the world`s books.
40

\hen publishers challenged Google`s book scanning in 200, many hoped that
public interest groups could leerage copyright challenges to Google`s book

35
Reza Dibadj, e,ova acite ...vvtiov. ava Raaicat ...ertiov.: . Ca.e for Criticat egat
covovic.,` 2003 U1Al L. RLV. 1155, 1161 ,|1|hree o the most basic assumptions to the
popular |law & economics| enterprise-that people are rational, that ability to pay
determines alue, and that the common law is eicient-while couched in the metaphors o
science, remain unsubstantiated.`,. vt .ee JAMLS R. lACKNL\, JR., UNDLR COVLR Ol
SCILNCL: AMLRICAN LLGAL-LCONOMIC 1lLOR\ AND 1lL QULS1 lOR OBJLC1IVI1\ 164-66
,Duke Uni. Press 200, ,describing the notable moement to broaden the scope o legal-
economic theory under the rubric o socioeconomics`,.
36
ee Leegin Creatie Leather Prods., Inc. . PSKS, Inc., 551 U.S. 8, 906 ,200,
,acknowledging the economic oundations o U.S. antitrust law,.
3
Motoko Rich, Coogte ava .vtbor. !iv tev.iov for oo/ etttevevt, N.\. 1IMLS, No. 9, 2009, at
B3, araitabte at
!!!"'D1&,(%"*+,-5223-77-72-1(*;'+@+/D-*+,>.'&(%-72/#++)%";1,@E4AG7. 1he
DOJ expressed dissatisaction with the parties` most recent proposed settlement, as well. ee
Cecilia Kang, ]vage Pvt. Off Rvtivg ov Coogte`. Proo.ea Digitat oo/ etttevevt, \ASl. POS1, leb.
19, 2010, araitabte at www.washingtonpost.com,wp-
dyn,content,article,2010,02,18,AR2010021800944.htmlhpid~moreheadlines.
38
Llectronic lrontier loundation, Google Book Search Settlement and Reader Priacy,
aailable at !!!"(CC"+A/-&%%$(%->A&O.*D-/++/@(L#++)L%(.A*;L%(11@(,('1 ,last isited
July 11, 2010,. As author Michael Chabon argues, i there is no priacy o thought -
which includes implicitly the right to read what one wants, without the approal, consent or
knowledge o others - then there is no priacy, period.` a.
39
ee, e.g., Diane Leenheer Zimmerman, Cav Ovr Cvttvre e area. 1be vtvre of Digitat
.rcbirivg, 91 MINN. L. RLV. 989, 990-91 ,200, ,looking at the Google Book Search project
as a means o saing culture and explor|ing| whether saing culture and saing copyright
can be made compatible goals`,.
40
Robert Darnton, 1be ibrar, iv tbe êr .ge, 55 N.\. RLV. BOOKS, June 12, 2008, at 39,
araitabte at www.nybooks.com,articles,21514.
search program to promote the public interest. Courts could condition a pro-
Google air use inding on uniersal access to the contents o the resulting
database. Landmark cases like ov, r. |virer.at
41
set a precedent or taking such
broad public interests into account in the course o copyright litigation.
42
1hose
who opt out o the settlement may be able to ight or such concessions, but or
now the battle centers on challenges to the settlement itsel.
Both James Grimmelmann and Pamela Samuelson hae suggested seeral
principles and recommendations to guide judicial deliberations on the proposed
settlement.
43
Grimmelmann`s work has ocused primarily on antitrust issues,
44

while Samuelson has concentrated on the concerns o academic authors.
45

Grimmelmann has succinctly summarized the settlement`s potential threats to
innoation and competition in the market or book indices, and books
themseles:
1he antitrust danger here is that the settlement puts Google in
a highly priileged position or book search and book sales. .
1he authors and publishers settled oluntarily with Google, but
there`s no guarantee they`ll oer similar terms, or any terms at
all, to anyone else. . |1hey| could unilaterally decide only to
talk to Google.
46

41
Sony Corp. o Am. . Uniersal City Studios, Inc., 464 U.S. 41, 442 ,1984,.
42
lrank Pasquale, Breaking the Vicious Circularity: Sony`s Contribution to the lair Use
Doctrine, 55 CASL \. RLS. L. RLV. , 90 ,2005,.
43
ee Pamela Samuelson, Coogte oo/ earcb ava tbe vtvre of oo/. iv C,ber.ace, 94 MINN. L.
RLV. ,orthcoming, 2010,, araitabte at ;11>?--F&/&1.@L
%*;+@.A%;&>"+A/-F&/&1.@)+.'%-5272-27-7B-/++/@(L#++)L%(.A*;L.'FL1;(LC$1$A(L+CL
#++)%L&'L*D#(A%>.*(- ,discussing the six categories o serious reserations that hae
emerged about the settlement . relected in the hundreds o objections and numerous
amicus curiae bries iled with the court responsible or determining whether to approe the
settlement.`,.
44
ee geveratt, James Grimmelmann, or to i tbe Coogte oo/ earcb etttevevt, 12 J. IN1LRNL1
L., Apr. 2009, at 1 ,arguing that the Google Book Search antitrust case settlement should be
approed with additional measures designed to promote competition and protect
consumers, |hereinater Grimmelmann, Coogte oo/ earcb etttevevt|.
45
Letter rom Pamela Samuelson, Richard M. Sherman Distinguished Proessor o Law,
Uniersity o Caliornia, Berkeley School o Law, to lon. Denny Chin, Judge, S.D.N.\.
,Sept. 3, 2009,, araitabte at www.scribd.com,doc,19409346,Academic-Author-Letter-090309
,urging the judge to condition approal o the Settlement Agreement on modiication o
arious terms identiied herein so that the Agreement will be airer and more adequate
toward academic authors.`,.
46
James Grimmelmann, v Coogte !e .vtitrv.t, 1PMCAlL BOOK CLUB, Jan. 15, 2009,
;11>?--1>,*.C("1.@)&'/>+&'1%,(,+"*+,-5223-27-78-&'4/++/@(4!(4.'1&1A$%1.
Grimmelmann proposes seeral methods o assuring that the publishers will
deal with other book search serices.
4
Grimmelmann suggests an |a|ntitrust
consent decree` and |n|ondiscrimination among copyright owners` as
potential responses to the issues raised by the settlement.
48
Most o his
proposal relects a policy consensus that presumes competition is the ideal
solution to abuses o power online.
49

\et there are many reasons why competition is unlikely to arise in book search
serices, een i the settlement is altered in order to promote it.
50
Licensing
costs are likely to be a substantial barrier to entry. A key to competition in the
search market is haing a comprehensie database o searchable materials, the
more these materials need to be licensed, the less likely it is that a second comer
can set up its own book archie. As scholars hae demonstrated, deals like
Google`s proposed settlement help entrench copyright holders` claims or
licensing reenue.
51
Moreoer, innoation in search is heaily dependent on
haing an installed base o users that eectiely train` the search engine to be
responsie.
52
1he more search queries an engine gets, the better able it is to
sharpen and perect its algorithm.
53
Lach additional user tends to decrease the
cost o a better quality serice or all subsequent users by contributing actiity
that helps the search engine dierentiate between high and low quality
organizational strategies.
54
1hus, incumbents with large numbers o users enjoy

4
a.
48
Grimmelmann, Coogte oo/ earcb etttevevt, .vra note 44, at 15.
49
Grimmelmann does also propose some reised terms that would not be primarily designed
to incentiize the deelopment o new alternaties to Google Book Search, or example, he
proposes |l|ibrary and reader representation at the |Book Rights R|egistry` that would
administer many aspects o the settlement. a.
50
ee Bracha & Pasquale, .vra note 5, at 1152 ,1hough the market choices o users and
technological deelopments constrain search engine abuse to some extent, they are unlikely
to indicate |certain social| alues . .`,, lrank Pasquale, erev Rea.ov. to Dovbt Covetitiov iv
tbe Ceverat earcb vgive Mar/et, MADISONIAN, Mar. 18, 2009,
;11>?--,.F&%+'&.'"'(1-5223-2B-7H-%(O('LA(.%+'%L1+LF+$#1L*+,>(1&1&+'L&'L1;(L
/('(A.@L%(.A*;L('/&'(L,.A)(1.
51
ee James Gibson, Ri./ .rer.iov ava Rigbt. .ccretiov iv vtettectvat Proert, ar, 116 \ALL L.J.
882, 884 ,200, ,describing how the decision as to whether to ight or air use or license a
copyrighted work can be diicult because the penalties or inringement typically include
supracompensatory damages and injunctie relie `,.
52
James Pitkow et al., Per.ovatiea earcb, 45 COMMS. ACM, Sept. 2002, at 50 ,discussing
methods o personalizing search systems,.
53
lor example, i 100 people search or alternaties to Microsot \ord sotware` on a search
engine on a gien day and all pick the third-ranked result, the search algorithm may adjust
itsel and put the third-ranked result as the irst result the next day. 1he most-used search
engine will hae more data to tweak its algorithms than its less-used rials.
54
Oren Bracha & lrank Pasquale, eaerat earcb Covvi..iov: airve.., .cce.., ava .ccovvtabitit, iv
tbe ar of earcb, 93 CORNLLL L. RLV. 1141, 1181 ,2008,, Daid A. Vise & Mark Malseed,
substantial adantages oer smaller entrants. Restrictie terms o serice also
deter competitors who aspire to reerse engineer and deelop better ersions o
such serices.
55
In general purpose search, users cannot reproduce, copy, or
resell any Google serice or av, reason, een i the behaior is manual and non-
disruptie.
56
Another section proscribes creat|ing| a deriatie work o . the
Sotware.`
5
Adertisers ace other restrictions, as Google`s Ad\ords
Application Programming Interace ,API, 1erms & Conditions impede
adertisers` eorts to eiciently copy their ad campaigns to other proiders.`
58

All o these actors militate against robust competition in the comprehensie
book search ield.
Quantum leaps in technology capable o oercoming these brute disadantages
are unlikely, particularly because search is as much about personalized serice as
it is about technical principles o inormation organization and retrieal.
59

Current adantage in search is likely to be sel-reinorcing, especially gien that
so many more people are using the serices now than when Google oertook
other search engines in the early 2000s.
60

\hat does an online world eaturing an entrenched Google Book Search as
gatekeeper look like Initially, it will proe a ast improement on the status

1he Google Story 215 ,2005, ,noting that the most-used search engine will hae more data
to tweak its algorithms than its less-used rials. ,. ,
55
1hough the precise terms o serice o Google Book Search hae not been inalized,
Google`s more general terms o serice are not promising. Google`s terms o serice
prohibit any action that intereres with or disrupts` Google`s serices, networks, or
computers. Google Inc., 1erms o Serice 5.4 ,Apr. 16, 200,,
www.google.com,accounts,1OS. Repeated queries to the serice necessary to gather data
on its operations may well iolate these terms.
56
a. 5.5.
5
a. 10.2. Section 5.3 would proscribe both the automatic data collection and the use o a
nonapproed interace` or accessing Google`s database, regardless o the exact means. a.
5.3.
58
Ben Ldelman, PPC Ptatforv Covetitiov ava Coogte`. Ma, ôt Co,` Re.trictiov, June 2, 2008,
;11>?--!!!"#('(F(@,.'"+A/-'(!%-2:592HL7";1,@ ,arguing that Google`s restrictions
on export and copying o adertisers` campaigns . hinder competition in Internet
adertising`,. 1hough the hearing at which Proessor Ldelman was to testiy was cancelled,
he has documented these problems in some detail at his website, www.benedelman.org.
59
John Battelle, 1lL SLARCl: lO\ GOOGLL AND I1S RIVALS RL\RO1L 1lL RULLS Ol
BUSINLSS AND 1RANSlORMLD OUR CUL1URL 8 ,2005,.at 8 ,describing how personalized
search enhances the alue o search engines to both users and adertisers,. Due to trade
secrecy, it is impossible or policymakers to discoer how much o the intermediary`s success
is due to its employees` inentie genius, and how much is due to the collectie contributions
o millions o users to the training o the intermediary`s computers.
60
ee Randall Stross, Planet Google: One Company`s Audacious Plan to Organize Lerything
\e Know 98 ,lree Press 2008, ,describing success o \ou1ube, a subsidiary o Google,.
quo o bulky, hard-to-acquire, physical copies o books. But when we consider
the ways in which knowledge can be rationed or proit, or structured to
promote political ends, some worries arise. Google plans to monetize the book
search corpus, and one predictable way o increasing its alue is to make parts
o it unaailable to those unwilling to pay high licensing ees. I the settlement
allowed Google to charge such ees in an unconstrained manner, unmoored
rom the underlying costs o operating the project, the company would
essentially be exploiting a public easement ,to copy books, or unlimited priate
gain.
61
1he Open Content Alliance has questioned the restrictie terms o the
contracts that Google strikes when it agrees to scan and create a digital database
o a library`s books.
62
1hose restrictie terms oreshadow potential uture
restrictions on book search serices. 1he proposed deal raises undamental
questions about the proper scope o priate initiatie in organizing and
rationing access to knowledge.
\ell-unded libraries may pay a premium to gain access to all sources, lesser
institutions may be granted inerior access. I permitted to become prealent,
such tiered access to inormation could rigidiy and reinorce existing
inequalities in access to knowledge.
63
Inormation tiering inequitably
disadantages many groups, promoting the leeraging o wealth into status,
educational, or other occupational adantage. Inormation is not only
intrinsically aluable, but also can be a positional good, useul or scoring
adantages oer others.
64

:7
\riters` Reps and Richard A. Lpstein Objection iled with the Southern District o New
\ork in re Google Book Search, aailable at
;11>?--!!!"!A&1(A%A(>%"*+,-C(.1$A(".%>PEI(.1$A(QNG795 ,arguing that the Google
Book Search Settlement would accomplish|| orphan legislation-but just or Google. . I
|Google| is to be handed exclusie possession ater stealing the scans to begin with, then it
should be required to share those scans.`,.
62
ee Open Content Alliance, Let`s Not Settle or 1his Settlement,
www.opencontentalliance.org,2008,11,05,lets-not-settle-or-this-settlement ,last isited
Mar. 12, 2010, ,At its heart, the settlement agreement grants Google an eectie monopoly
on an entirely new commercial model or accessing books. It re-conceies reading as a
billable eent. 1his reading eent is thereore controllable and trackable. It also orces
libraries into inancing a ending serice that requires they perpetually buy back what they
hae already paid or oer many years o careul collection.`,.
63
lrank Pasquale, 1ecbvotog,, Covetitiov, ava 1atve., 8 MINN. J. L. SCI. & 1LCl. 60, 608 ,200,
,explaining how much technology is used not just simply to improe its user`s lie, but also
to help its user gain adantage oer others`,. lor example, |t|est-preparation technologies
. creat|e| inequalities, students able to aord test-preparation courses, such as those
oered by Kaplan, hae a deinite adantage oer those who do not hae access to such
courses.` a. at 615 ,internal citation omitted,.
64
larry Brighouse & Adam Swit, qvatit,, Priorit,, ava Po.itiovat Cooa., 116 L1lICS 41, 42
,2006, ,|Positional goods| are goods with the property that one`s relatie place in the
distribution o the good aects one`s absolute position with respect to its alue. 1he ery
act that one is worse o than others with respect to a positional good means that one is
Admittedly, Google Book Search has so ar proen a great resource or
scholars. It has made book learning accessible on a new, worldwide scale,
despite the great digital diide that separates the poor rom the
computerized.`
65
Current access to knowledge is stratiied in many troubling
ways, the works o John \illinsky
66
and Peter Suber
6
identiy many troubling
current orms o tiering that pale beore the present impact o Google Book
Search.
68
Gien the aggressie pricing strategies o many publishers and
content owners, Google Book Search is a ital alternatie or scholars.
Neertheless, there is no guarantee in the current ersion o the settlement that
Google Book Search will presere its public-regarding eatures.
69
It may well
end up like the powerul group purchasing organizations` in the American
health care system that started promisingly, but hae eoled to exploit their
intermediary role in troubling ways.
0
Google is more than just one among
many online serice proiders jostling or a competitie edge on the web. It is
likely to be the key priate entity capable o competing or cooperating with
academic publishers and other content proiders. Dedicated monitoring and
regulation o the settlement terms now could help ensure that book digitization

worse o, in some respect, than one would be i that good were distributed equally. So
while it might indeed be pererse to adocate leeling down all things considered, leeling
down with respect to positional goods beneits absolutely, in some respect, those who would
otherwise hae less than others.
65
Darnton, .vra note 40, at 6.
66
JOlN \ILLINSK\, 1lL ACCLSS PRINCIPLL: 1lL CASL lOR OPLN ACCLSS 1O RLSLARCl AND
SClOLARSlIP 5 ,1he MI1 Press 2005, ,describing extreme digital diide` between those
most connected to inormation resources and those cut o rom them,.
6
ee geveratt, Peter Suber, Open Access News, www.earlham.edu,~peters,os,osblog.html.
Suber is a leader o the open access moement, which aims to |p|ut|| peer-reiewed
scientiic and scholarly literature on the internet|,| |m|ak|e| it aailable ree o charge and
ree o most copyright and licensing restrictions|,| |and| remo|e| the barriers to serious
research.` a.
68
ee ia. ,chronicling on a daily basis news and controersies related to open access to scholarly
materials on the Internet,.
69
Sia Vaidhyanathan, aiav.cov .ccv.ea of Riggivg earcb, 1he Googlization o Lerything,
Global Google, Jan. 2009,
;11>?--!!!"/++/@&0.1&+'+C(O(AD1;&'/"*+,-5223-27-#.&F$*+,4.**$%(F4+C4A&//&'/
4%(">;> ,leb. 19, 2009, 14:20 LS1, ,Public ailure` |is a| phenomenon in which a priate
irm steps into a acuum created by incompetent or gutted public institutions. A irm does
this not or immediate rent seeking or een reenue generation. It does so to enhance
presence, reputation, or to build a platorm on which to generate reenue later or elsewhere.
It`s the opposite o market ailure.` And it explains a lot o what Google does.`,.
0
lor background on group purchasing organizations, .ee S. PRAKASl SL1lI, GROUP
PURClASING ORGANIZA1IONS: AN UNDISCLOSLD SCANDAL IN 1lL U.S. lLAL1l CARL
INDUS1R\ 122 ,Palgrae MacMillan 2009, ,1he beneits o combined purchases would be
greatly reduced in conditions where the middlemen . control the entire process through
restrictie arrangements with suppliers and customers.`,.
protects priacy, dierse stakeholder interests, and air pricing o access to
knowledge. Alliances between Google Book Search and publishers desere
public scrutiny because they permit priate parties to take on what hae oten
been public unctions o determining access to and pricing o inormation.
\here regulatory copyright`
1
has answered such questions with compulsory
licenses,
2
the new alliances aspire to put into place a regime o cross-
subsidization resistant to public scrutiny or input.
3
Gien the ital public
interests at stake in the deelopment o this inormation inrastructure,
monitoring is ital.
4
Lxtant law proides little assurance that it will actually
occur.
A Public Alternative?
In other work, I hae proposed a number o regulations that would permit
either goernment or public accountability groups to monitor search engines to
detect abuses o their dominant position. 1o conclude this piece, I would like
to raise one other alternatie: a publicly unded search engine.
1o the extent that search engines resist monitoring and accountability,
goernments should consider establishing public alternaties to them. lere,
lessons rom recent debates oer health insurance may be instructie. 1here
are structural parallels between the intermediary role o priate health insurers
,which stand as a gatekeeper between patients and proiders o health products
and serices, and that o search engines ,which stand between searchers and

1
ee Joseph P. Liu, Regvtator, Co,rigbt, 83 N.C. L. RLV. 8, 91 ,2004, ,describing the growth
and scope o compulsory licensing statutes that proide or compensation or copyright
holders while denying them the right to eto particular uses o their work,.
2
Marybeth Peters, the U.S. Register o Copyrights, has objected to the proposed Google
Books Settlement on the grounds that it would iolate traditional norms o separation o
powers in copyright policy. ee earivg ov Covetitiov ava Covverce iv Digitat oo/.: 1be
Proo.ea Coogte oo/ etttevevt efore tbe ov.e Covv. ov tbe ]vaiciar,, 111th Cong. ,2009,
,statement o Marybeth Peters, Register o Copyrights,, araitabte at
;11>?--R$F&*&.AD";+$%("/+O-;(.A&'/%->FC-S(1(A%232372">FC, at 2 ,In the iew o the
Copyright Oice, the settlement proposed by the parties would encroach on responsibility
or copyright policy that traditionally has been the domain o Congress. . \e are greatly
concerned by the parties` end run around legislatie process and prerogaties, and we submit
that this Committee should be equally concerned.`,.
3
Google considers its pricing and ranking decisions a closely held trade secret-an assertion
that would seem ery strange i it came rom a public library. ee Pamela Samuelson, Coogte
oo/. . ôt a ibrar,, 1lL lUllING1ON POS1, Oct. 13, 2009,
!!!";$CC&'/1+'>+%1"*+,->.,(@.L%.,$(@%+'-/++/@(L#++)%L&%L'+1L.L
@&#4#4B7987H";1,@ ,Libraries eerywhere are terriied that Google will engage in price-
gouging when setting prices or institutional subscriptions to |Google Book Search|
contents.`,.
4
lrank Pasquale, e,ova Covetitiov ava vvoratiov: 1be êea for Qvatifiea 1rav.arevc, iv vtervet
vterveaiarie., 104 Nw. U. L. RLV. 105 ,2010, ,oering proposals or monitoring internet
intermediaries,.
proiders o inormation,. 1he 1965 decision to establish Medicare as a public
option or an elderly population ill-sered by priate proiders and insurers may
proe a model or an inormation economy plagued by persistent digital diides.
As the United States debated health reorm rom 2009 to 2010, there was a
tension between regulation-ocused approaches ,which would require reelation
and alteration o priate insurers` unair practices, and a public option that
would compete with existing insurers. Democrats ultimately gae up on
pushing the public option, but the debate exposed the many positie aspects a
state-sponsored alternatie can proide in certain markets. A public option
could play a role in search parallel to the role that Medicare plays in the health
system: guaranteeing some baseline o transparency in pricing and ealuation.
5

1he recent Google Book Search settlement negotiations hae led Sia
Vaidhyanathan to characterize Google`s archie project as eidence o a public
ailure.`
6
\hereas goernment interention is oten necessary in cases o
market ailure,` Vaidhyanathan argues that the reerse can occur: market
actors can step into a acuum where goernment should hae been. In the case
o digitized books, the problem is presented starkly: \hy has the Library o
Congress ailed to require aigitat deposit o books, instead o merely accepting
paper copies \e can debate when such a requirement became plausible,
howeer, had the goernment required such deposit as soon as it became
easible, the problematic possibility o a Google monopoly here would be much
less troubling. I digital deposit eer is adopted, the goernment could license
its corpus to alternatie search serices. 1here is no good reason why the
company that is best capable o reproducing books ,and settling lawsuits based
on that reproduction, should hae a monopoly on search technologies used to
organize and distribute them.
More ambitiously, an NGO or quasi-administratie NGO could undertake to
index and archie the web, licensing opportunities to search and organize it to
arious entities that promise to maintain open standards or ranking and rating
websites and other Internet presences.
\ikipedia, Slashdot, and eBay all

5
lor more on the role o public options like Medicare in the modern medical sector, .ee lrank
Pasquale, Ma/ivg tbe Ca.e for tbe Pvbtic Ptav, Part : Pvbtic Otiov a. Prirate evcbvar/, July 15,
2009, aailable at ;11>?--#.@)&'"#@+/%>+1"*+,-5223-2:-,.)&'/L*.%(LC+AL>$#@&*L>@.'L
>.A1L&&";1,@.
6
Vaidhyanathan, .vra note 69. , Public ailure` |is a| phenomenon in which a priate irm
steps into a acuum created by incompetent or gutted public institutions. A irm does this
not or immediate rent seeking or een reenue generation. It does so to enhance presence,
reputation, or to build a platorm on which to generate reenue later or elsewhere. It`s the
opposite o market ailure.` And it explains a lot o what Google does.`,.
lor a cultural case or goernment interention here, see Mario J. Sila, 1be Ca.e for a
Portvgve.e !eb earcb vgive,
;11>?--!!!"/++/@("*+,-$A@E%.G1T%+$A*(G!(#T*FG7TO(FG2KUVWIRXXT$A@G;11
>YBXY5IY5I*&1(%((AP"&%1">%$"(F$Y5IO&(!F+*Y5IF+!'@+.FYBIF+&YBN72"7"7"72:"
suggest methods o ealuating releance and authority that could be employed
by public, open search engines. I such a search engine became at least
somewhat popular ,or popular within a gien niche,, it could proide an
important alternatie source o inormation and metadata on ranking processes.
1he need or a public option in search becomes een more apparent when we
consider the waste and ineiciency caused by opaque intermediaries in other
ields. Like priate health insurers, Google is a middleman, standing between
consumers and producers o knowledge. In programs like Book Search, it will
eectiely collaborate with copyright owners to determine what access people
get, how much they hae to pay, and on what terms. In the health ield,
proiders and priate insurers are both ery concentrated in the U.S., and
consumers ,i.e., the businesses and indiiduals who buy insurance plans, are not.
Insurers and proiders also jealously guard the secrecy o many pricing
decisions.
8
1hat is one key reason why the U.S. spends so much more on
health care than other industrialized nations, without getting consistently better
results, access, or quality.
lealth care reormers oten split into two camps: those who beliee that
regulation o middlemen like insurers can bring about air results, and those
who beliee that only a public option can sere as a benchmark or judging the
behaior o priate insurers. 1he Patient Protection and Aordable Care Act
,PPACA, o 2010 decisiely opted or the regulatory option, and the early stages
o its implementation hae been rocky. 1he constitutional challenges to search
engine regulation would likely proe more serious than the many lawsuits now
attacking PPACA. 1hereore, een i the public option in health care is o the
table now, it should inspire uture proposals in inormation policy, where
regulation o intermediaries may be een more diicult than it has proen to be
in health care. I search engines consistently block or rustrate measures to
increase their accountability, public alternaties could proe to be an
indispensable oundation o a air, just, and open inormation enironment.

8BB6Y5:A(>YBNA(>7Y5:1D>(YBN>FCT(&GQZ[\J]#.K+^KH/.>O\5PK!T$%/GXIW
RK_`FJS>JUV$_`[;Ja[1M.b&c^S:K6/T%&/5G3.+.^de&eaaV$\`f(!+>c
W ,describing the alue o a Portuguese-oriented search engine,, JLAN NOLL JLNNLN\,
GOOGLL AND 1lL M\1l Ol UNIVLRSAL KNO\LLDGL: A VIL\ lROM LUROPL ,Uni. o
Chicago Press 200,. \hereas these authors beliee that Lnglish-language bias is a
particularly problematic aspect o Google`s hegemony in the ield, I argue that the possibility
o many kinds o hidden bias counsel in aor o at least one robust, publicly unded
alternatie.
8
ee, e.g., Uwe Reinhart, 1be Pricivg of |.. o.itat errice.: Cbao. ebiva a 1eit of ecrec,, at
;11>?--;(.@1;.CC";&/;!&A("+A/-*/&-*+'1('1-.#%1A.*1-58-7-89, Annemarie Bridy, 1raae
ecret Price. ava igb1ecb Derice.: or Meaicat Derice Mavvfactvrer. are ee/ivg to v.taiv Profit. b,
Proertiivg Price., 1 1LX. IN1LLL. PROP. L.J. 18 ,2009, ,discussing recent claims by the
medical deice manuacturer Guidant that the actual prices its hospital customers pay or
implantable deices, including cardiac pacemakers and deibrillators, are protectable as trade
secrets under the Uniorm 1rade Secrets Act.`,.

The Problem of Search Engines
as Essential Facilities: An
Economic & Legal Assessment
By Geoffrey A. Manne
*

\hat is wrong with calls or search neutrality, especially those rooted in the
notion o Internet search ,or, more accurately, Google, the policy scolds` bte
noir o the day, as an essential acility,` and necessitating goernment-
mandated access As others hae noted, the basic concept o neutrality in
search is, at root, arcical.
1
1he idea that a search engine, which oers its users
edited access to the most releant websites based on the search engine`s
assessment o the user`s intent,
2
should do so neutrally` implies that the search
engine`s eorts to ensure releance should be cabined by an almost-limitless
range o ancillary concerns.
3

Neertheless, proponents o this iew hae begun to adduce increasingly detail-
laden and complex arguments in aor o their positions, and the Luropean
Commission has een opened a ormal inestigation into Google`s practices,
based largely on arious claims that it has systematically denied access to its top
search results ,in some cases paid results, in others organic results, by
competing serices,
4
especially ertical search engines.
5
1o my knowledge, no

Lxecutie Director, International Center or Law & Lconomics and Lecturer in Law, Lewis
& Clark Law School. !!!"@.!(*+'*('1(A"+A/,
!!!"@*@.A)"(F$-@.!-C.*$@1D-/(+CCA(D4,.''(.
1
ee, e.g., Danny Sullian, 1be vcreaibte tviait, of vre.tigativg Coogte for .ctivg i/e a earcb
vgive, SLARCl LNGINL LAND, ;11>?--%(.A*;('/&'(@.'F"*+,-1;(L&'*A(F&#@(L%1$>&F&1DL
+CL&'O(%1&/.1&'/L/++/@(LC+AL.*1&'/L@&)(L.L%(.A*;L('/&'(L895:H ,A search engine`s job is
to point you to destination sites that hae the inormation you are seeking, not to send you
to other search engines. Getting upset that Google doesn`t point to other search engines is
like getting upset that the New \ork 1imes doesn`t simply hae headlines ollowed by a
single paragraph o text that says read about this story in the \all Street Journal.``,.
2
A remarkable eat, gien that this intent must be inerred rom simple, context-less search
terms.
3
Perectly demonstrated by lrank Pasquale`s call, elsewhere in this olume, or identiying
search engines as essential cultural and political acilities,` thereby mandating incorporation
into their structure whateer cultural` and political` preerences any suiciently-inluential
politician ,or law proessors, happens to deem appropriate.
4
Competing serices include, or example, MapQuest ,!!!",.>g$(%1"*+,, ,competing
with Google Maps,, Veoh ,!!!"O(+;"*+,, ,competing with \ou 1ube, and Bing Shopping
,!!!"#&'/"*+,-%;+>>&'/, ,competing with Google Products,.
5
Vertical search engines are search engines that ocus on a particular category o products, or
on a particular type o search. Lxamples include Kayak ,!!!").D.)"*+,, ,trael search,,
one has yet claimed that Google should oer up links to competing general
search engines as a remedy or its perceied market oreclosure, but Microsot`s
experience with the Browser Choice Screen` it has now agreed to oer as a
consequence o the Luropean Commission`s successul competition case
against the company is not encouraging.
6
1hese more supericially sophisticated
claims are rooted in the notion o Internet search as an essential acility`-a
bottleneck limiting eectie competition.
1hese claims, as well as the more undamental harm-to-competitor claims, are
diicult to sustain on any economically-reasonable grounds. 1o understand this
requires some basic understanding o the economics o essential acilities, o
Internet search, and o the releant product markets in which Internet search
operates.
The Basic Law & Economics
of Essential Facilities
1here are two ways to deal with a problematic bottleneck: Remoe the
bottleneck or regulate access to it. 1he latter is the more common course
adopted in the U.S. and elsewhere. Complex, Byzantine and oten counter-
productie regulatory apparatuses are required to set and monitor the terms o
access. Among other things, this paes the way or either intensely-problematic
judicial oersight o court-imposed remedies or else the creation o sector-
speciic regulatory agencies subject to capture, political inluence, bureaucratic
ineiciency, and ineicient longeity. 1he Interstate Commerce Commission
,and its successor agencies within the Department o 1ransportation, and the
lederal Communications Commission ,and its implementation beginning in
1996 o the monstrous 1elecommunications Act, in the U.S. are paradigmatic
examples o these costly eects, and it is certainly questionable whether the
disease is worse than the cure.

Obiously, an essential acility must be e..evtiat. Lorts oer the years to
shoehorn arious markets into this category hae sometimes strained credulity,
as it has ariously been claimed that Aspen, Colorado ski hills,
8
local oice mail

Source1ool ,!!!"%+$A*(1++@"*+,, ,business input sourcing,, and loundem
,!!!"C+$'F(,"*+,, ,retail product search and price comparison,.
6
ee Luropean Commission, !eb bror.er cboice for vroeav cov.vver.,
;11>?--(*"($A+>."($-*+,>(1&1&+'-*+'%$,(A%-!(#4#A+!%(A%4*;+&*(4('";1,@ ,last
accessed Dec. 8, 2010,.
Oren Bracha and lrank Pasquale`s call or a lederal Search Commission` modeled on the
lederal 1rade Commission is in act an evbrace o the need or a bureaucratic apparatus to
regulate the orced access called or by search neutrality proponents. ee Oren Bracha and
lrank Pasquale, eaerat earcb Covvi..iov: airve.., .cce.., ava .ccovvtabitit, iv tbe ar of earcb,
93 CORNLLL L. RLV. 1193 ,2008,.
8
Aspen lighlands Skiing Corp. . Aspen Skiing Co., 38 l.2d 1509 ,10th Cir. 1984,, a `d on
other grounds, 42 U.S. 585 ,1985,.
serices,
9
sot drinks
10
and direct reight lights between New \ork and San
Juan
11
,among many other things, were essential acilities necessitating
mandated access under the antitrust laws.
12
In these and many other cases,
myriad alternaties to the allegedly-monopolized market exist and it is arguable
that there was nothing whatsoeer essential` about these markets.
In antitrust literature and jurisprudence, a plainti would need to proe the
ollowing to preail in a monopolization case rooted in the essential acilities
doctrine:
1. Control o the essential acility by a monopolist,
2. A competitor`s inability practically or reasonably to duplicate the
essential acility,
3. 1he denial o the use o the acility to a competitor, and
4. 1he easibility o proiding the acility to competitors.
13

Arguably, since the Supreme Court`s 2004 1riv/o decision,
14
a plainti would
also need to demonstrate the absence o ederal regulation goerning access.
1he 1riv/o decision signiicantly circumscribed the area subject to essential
acilities arguments, limiting such claims to instances where, as in the ..ev
/iivg case, a competitor reuses to deal on reasonable terms with another
competitor with whom it has, in act, dealt in the past.
15

A key problem with many essential acilities cases is the non-essentiality o the
releant acility. \hile there can be no doubt that to particular competitors,
particularly those constrained to only one aenue o access to consumers by
geography or natural monopoly, a acility may indeed seem essential, the
touchstone o U.S. antitrust law has long been consumer, not competitor,
welare. So while, indeed, Aspen lighlands may hae had diiculty competing
with the Aspen Ski Company or consumers who had already chosen to ski in
Aspen, consumers nonetheless had unettered access to a wide range o
alternatie ski ,and other acation, destinations, such that the likelihood o the

9
C1C Communications Corp. . Bell Atlantic Corp., l. Supp. 2d 124 ,D. Me. 1999,.
10
Sun Dun . Coca-Cola Co., 40 l. Supp. 381 ,D. Md. 1990,.
11
Century Air lreight, Inc. . American Airlines, Inc., 59 l. Supp. 564 ,S.D.N.\. 1984,.
12
lor a more complete list o essential acilities ,and attempted essential acilities, cases, as
well as an important treatment o the essential acilities doctrine in US antitrust law, see
Abbott B. Lipsky, Jr. & J. Gregory Sidak, ..evtiat acititie., 51 S1AN. L. RLV. 118 ,1999,.
13
MCI Comm`ns Corp. . American 1el. & 1el. Co., 08 l.2d 1081 ,
th
Cir. 1982,.
14
Verizon Comm`ns. . Law Oices o Curtis V. 1rinko LLP, 540 U.S. 398 ,2004,.
15
ee, e.g., PlILLIP L. ARLLDA & lLRBLR1 lOVLNKAMP, AN1I1RUS1 LA\ ,2004 supp., at 199.
monopolization o Aspen`s ski hills aecting oerall consumer welare was
essentially non-existent.
16
In such a circumstance, should it matter i a particular
competitor is harmed Is that a unction o antitrust-releant conduct on the
part o another irm, or an unortunate set o business decisions on the part o
the irst irm
As Phillip Areeda and lerbert loenkamp hae amously said o the essential
acilities doctrine, |it| is both harmul and unnecessary and should be
abandoned.`
1
As another antitrust expert has described it:
At bottom, a plainti making an essential acilities argument is saying
that the deendant has a aluable acility that it would be diicult to
reproduce, and suggesting that is a reason or a court to interene and
impose a sharing duty. But at least in the ast majority o the cases, the
act that the deendant has a highly alued acility is a reason to re;ect
sharing, not to reqvire it, since orced sharing may lessen the incentie
or the monopolist, the rial, or both to inest in those economically
beneicial acilities.`
18

1his perennial problem-antitrust laws being used to protect competitors rather
than consumers-lies at the heart o claims surrounding Internet search as an
essential acility.
1here is much tied up in the argument, and proponents hae oten been careul
to at least go through the motions o drawing the rhetorical line back to
consumers. In its ullest expression, it is claimed that harm to competitors now
will mean the absence o competitors later and thus an unettered monopoly
with the intent and power to harm consumers.
19
It is also oten argued that
consumers ,in this case Internet users searching or certain websites or the
products they sell, are intrinsically harmed by the unaailability o access to the
inormation contained in sites that are denied access to the search engine`s
essential acility.`
20

16
1he courts, howeer, did not agree.
1
3A ARLLDA & lOVLNKAMP, AN1I1RUS1 LA\ 1c, at 13 ,2002,.
18
R. lewitt Pate, Refv.at. to Deat ava ..evtiat acititie., 1estimony Submitted to DOJ,l1C
learings on Single lirm Conduct, Jul. 18, 2006, araitabte at
;11>?--!!!"R$%1&*("/+O-.1A->$#@&*-;(.A&'/%-%&'/@(4C&A,-F+*%-57H:63";1, ,quoting
1riv/o, 540 U.S.. at 408,.
19
ee, e.g., vroeav Covvi..iov avvcbe. .vtitrv.t vre.tigatiov of Coogte, SLARCl
NLU1RALI1\.ORG, No. 30, 2010, ;11>?--!!!"%(.A*;'($1A.@&1D"+A/ ,Google is exploiting
its dominance o search in ways that stile innoation, suppress competition, and erode
consumer choice.`,. Meanwhile, complainants hae gone to Lurope where a showing o
consumer harm is not necessary to preail under its competition laws.
20
As Oren Bracha and lrank Pasquale put it, Search engines, in other words, oten unction
not as mere satisiers o predetermined preerences, but as shapers o preerences,` eaerat
earcb Covvi..iov, 93 CORNLLL L. RLV. at 1185. Bracha and Pasquale also claim that Market
participants need inormation about products and serices to make inormed economic
1he basic essential acilities case against Google is that it controls a bottleneck
or the Internet-it is the access point or most consumers, and search results
on Google determine which websites are successul and which end up in
obliion.
21
More particularly, it is argued that Google has used its control oer
this bottleneck to deny access by competitors to Google`s users. 1o understand
this requires a brie discussion o the economics releant to Internet Search and
its releant market.
The Basic Economics of Internet Search
Implicit in claims that Google controls access to an essential acility is that
access by some releant set o consumers ,or competitors, to releant content is
accessible only ,or irtually only, through Google. It is necessary, then, to
assess whether Google`s search results pages are, in act, without signiicant
competition or the economic actiity at their heart. O course the economic
actiity at their heart is adertising.
22

It is hard to conceie o Internet search-let alone Google`s website-as the
only means o reducing search costs or potential consumers ,Internet
searchers, and prospectie sellers. Leaing aside the incredible range o
alternatie sources to the Internet or commerce,
23
o the top o my head, I can
imagine Google`s competitor websites inding access to users by 1, adertising
in print publications and 1V, 2, using social networking sites to promote their
sites, 3, being linked to by other websites including sites specializing in rating
websites, online magazines, reiew sites, and the like, 4, implementing ailiate
programs or other creatie marketing schemes, 5, purchasing paid adertising,
both in Google`s own paid search results, as well as on other, heaily-traicked
websites, and 6, securing access ia Google`s general search competitors like
\ahoo! and Bing. Competitors denied access to the top ew search results at

decisions. . |A|attaining isibility and access to users is critical to competition and
cooperation online. Centralized control or manipulation by search engines may stile
innoation by irms relegated to obscurity.` a. at 113-4.
21
a. at 113 ,Concentrated control oer the low o inormation, coupled with the ability to
manipulate this low, may reduce economic eiciency by stiling competition.`,.
22
ee KLN AULL11A, GOOGLLD: 1lL LND Ol 1lL \ORLD AS \L KNO\ I1 16 ,2009, ,quoting
Google CLO Lric Schmidt as saying, \e are in the adertising business`,.
23
1here is a tendency or \eb sites to iew their Internet enterprises as dierent than their
oline counterparts`, but, at root, most Internet sites ,other than branded ones attached
directly to oline stores, are ounded by entrepreneurs who made a simple business decision
to ply their trade online rather than o. 1hat this decision may hae oreclosed easy access
to certain oline customers, or put the entrepreneur in a position where access to customers
could be rustrated by certain competitie disadantages speciic to the Internet, does not
conert these competitie disadantages into special problems desering o antitrust
treatment. 1o do so would be to inappropriately and ineiciently insulate the online,oline
business decision rom the healthy eects o Schumpeter`s perennial gale o creatie
destruction.` JOSLPl SClUMPL1LR, 1lL PROCLSS Ol CRLA1IVL DLS1RUC1ION ,1942,.
Google`s site are still able to adertise their existence and attract users through a
wide range o other adertising outlets-extremely wide, in act: According to
one estimate Google was responsible in 200 or only about .5 o the
world`s adertising.
24

lor Google to proit rom its business-whether as a monopolist or not-it
must delier up to its adertisers a set o users. Interestingly, users o Google`s
general search engine are mostly uninterested in the paid results. 1hey click
through the unpaid or organic` search results by a wide margin ahead o paid
results.
25
1here is thus an asymmetry. On one side o its platorm are
adertisers who care about the quantity and quality ,the likelihood that users
who see an ad will click through to adertisers` sites and purchase something
while there, o the users on the other side. Meanwhile, users care ery little
about the quantity o adertisers and care only somewhat about the quality o
adertisers ,preerring greater releance to lesser, but requently ignoring paid
results anyway,. Neertheless, the core o this enterprise is search result
releance. Greater releance improes the quality o searchers rom the
adertisers` point o iew, ensuring that adertisers` paid results are clicked on
by the users most likely to ind the adertiser`s site o interest and to purchase
something there.
But there are problems inherent in the ambiguity o search terms and the ability
to game the system` that preent een the most sophisticated algorithms rom
oering up perect releance. lirst, search terms are oten context-less, and a
user searching or jaguar` may be searching or inormation on the car
company, the operating system, the big cat, or something else.
26
Along a
dierent dimension, a user searching or Nikon camera` might be looking to
bv, a Nikon camera or might be looking or a ictvre o a Nikon camera to post
on his blog. Obiously adertisers care ery much which o these users clicks
on their paid result. At the same time, many undesirable websites ,spam sites
and the like, can and do take adantage o predictable search results to occupy
desirable search result real estate to the detriment o the search engine, its users
and its adertisers. Lorts to keep these sites out o the top results and to
ensure maximum releance rom ambiguous search terms require a host o
algorithm tweaks and een human interentions. 1hat these may ,intentionally
or inadertently, harm some websites` rank in certain search results is consistent
with a well-unctioning search platorm.

24
ee Lrick Schoneld, .tivate. Pvt vtervet .arerti.ivg at 21 ittiov iv |.., 1: ittiov Ctobatt,,
1LClCRUNCl, leb. 26, 2008, ;11>?--1(*;*A$'*;"*+,-522H-25-5:-(%1&,.1(%L>$1L
&'1(A'(1L.FO(A1&%&'/L.1L57L#&@@&+'L&'L$%L68L#&@@&+'L/@+#.@@D-.
25
ee, e.g., Neil \alker, Coogte Orgavic Ctic/ 1brovgb Rate ;C1R), UK SLO CONSUL1AN1, May 11,
2010, ;11>?--!!!"%(+,.F"*+,-hiaU@+/-/++/@(L+A/.'&*L*@&*)L1;A+$/;LA.1(L
*1A";1,@.
26
ee Bill Slawski, . oo/ at Coogte Miaage Qver, Refivevevt., SLO B\ 1lL SLA, Apr. 20, 2006,
;11>?--!!!"%(+#D1;(%(."*+,-E>G796.
Google oers its organic search results and its other serices as a solution to the
two-sided platorm problem mentioned aboe: In order to attract paying
adertisers, Google also has to attract ,and match up, the adertisers` target
audience. Google oers eerything it does to its users in an eort to attract
these users and to glean inormation rom them that acilitates its all-important
matching ,releance, unction. In the process, Google generates reenue rom
adertisers eager to sell` to this audience. lor a host o reasons, Google ,like
all search engines, does not charge searchers to access its arious serices, but it
does charge adertisers. Just because search is an ancillary business to Google`s
true adertising business does not necessarily mean it is not a releant market
or purposes o antitrust analysis, neertheless it is essential to aoid the pitall
o examining one side o a two-sided market in isolation. As Daid Lans
notes, |t|he analysis o either side o a two-sided platorm in isolation yields a
distorted picture o the business.`
2
1wo-sided market deinition is complex,
and little understood-especially by non-experts throwing around arious
alleged markets in which companies like Google are said to be dominant.`
1here is actually substantial reason to doubt the propriety o a narrow market
deinition limited to online search adertising.
28
Len where there are dierent
purposes or dierent types o adertising-e.g. brand recognition or display
ads and eorts to sell or search ads and other outlets like coupons-this is
merely a dierence in degree. Both are undamentally orms o reducing the
costs o a user`s search or a product, as we hae understood since George
Stigler`s seminal work on the subject in 1968,
29
and the releant question is
whether the dierence is signiicant enough to render decisions in one market
essentially unaected by decisions or prices in the other.
1here is eidence that adertisers iew online and oline adertising as
substitutes, and this applies not only to traditional adertisers but also Internet
companies. 1hus, in 2009, Pepsi decided not to adertise during the 2010 Super
Bowl, in order to ocus instead on a particular type o online campaign. 1his
year or the irst time in 23 years, Pepsi will not hae ads in the Super Bowl
telecast .. Instead it is redirecting the millions it has spent annually to the

2
Daid S. Lans, 1roiaea Mar/et Defivitiov, ABA SLC1ION Ol AN1I1RUS1 LA\, MARKL1
DLlINI1ION IN AN1I1RUS1: 1lLOR\ AND CASL S1UDILS ,orthcoming,, aailable at
;11>?--%%A'"*+,-.#%1A.*1G7B3:987 at p. 9.
28
Readers interested in a uller treatment o the market deinition question surrounding
Google are directed toward Georey A. Manne & Joshua D. \right, Coogte ava tbe ivit. of
.vtitrv.t: 1be Ca.e .gaiv.t tbe Ca.e .gaiv.t Coogte, 34 lARV. J. L. & PUB. POL`\ 1 ,2011,
,orthcoming,, rom which much o the discussion o Google`s markets and economics in
this essay is drawn.
29
GLORGL JOSLPl S1IGLLR, 1lL ORGANIZA1ION Ol INDUS1R\ 201 ,Uni. o Chi. Press 1983,
,1968,.
Internet.`
30
And een Google itsel adertises oline.
31
Another study suggests
that there is indeed a trade-o between online and more traditional types o
adertising: Aid Goldarb and Catherine 1ucker hae demonstrated that
display adertising pricing is sensitie to the aailability o oline alternaties.
32

And o course companies hae limited adertising budgets, distributed across a
broad range o media and promotional eorts. As one commentator notes: By
2011 web adertising in the United States was expected to climb to sixty billion
dollars, or 13 percent o all ad dollars. 1his meant more dollars siphoned rom
traditional media, with the largest slice probably going to Google.`
33

Adertising reenue on the Internet is drien initially by the size o the
audience, with a signiicant multiplier or the likelihood that those consumers
will purchase the adertisers` products
34
,based on a iewer`s propensity to
click through` to the adertiser`s site,. Google`s competition in selling ads thus
comes, in arying degrees, not only rom other search sites, but also rom any
other site that oers a serice, product, or experience that consumers might
otherwise ind in Google`s organic` search results, or which Google is not
paid. lor Google`s competitors, this means seeking orced access to its users.
But access to eyeballs can be had rom a large range o access points around the
\eb.
Social media sites like 1witter and lacebook are thereore signiicant access
points, occupying, as they do, a considerable amount o Internet eyeball` time.
1he Pepsi deiation o adertising reenue rom the Super Bowl to the Internet
is not likely to hae inured much to Google`s beneit as the strategy was a
social media play,` building on the expressed brand loyalties and peer
communications that propel social media.
35
In a world o scarce adertising
dollars and eectie marketing ia social media sites, Google and all other
adertisers, online and o, must compete with the growing threat to their
reenue rom these still-noel marketing outlets. I lacebook`s community o

30
Larry D. \oodard, Pe.i`. ig Cavbte: Ditcbivg ver ort for ociat Meaia, ABC NL\S, Dec. 23,
2009, ;11>?--.#*'(!%"/+"*+,->A&'1E&FG3625876.
31
ee Danny Sullian, Coogte Pv.be. Cbrove ror.er 1ia êr.aer .a., SLARCl LNGINL LAND,
No. 21, 2010, ;11>?--%(.A*;('/&'(@.'F"*+,-/++/@(L>$%;(%L*;A+,(L#A+!%(ALO&.L
'(!%>.>(AL.F%L8::22.
32
Ai Goldarb & Catherine 1ucker, earcb vgive .arerti.ivg: Pricivg .a. to Covtet 96 ,NL1
Institute \orking Paper No. 0-23, 200, araitabte at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG7257687TA(*G7T%A*.#%G722H6
33
KLN AULL11A, GOOGLLD: 1lL LND Ol 1lL \ORLD AS \L KNO\ I1 16 ,2009,.
34
Daid S. Lans, 1be covovic. of tbe Ovtive .arerti.ivg vav.tr,, RLV. Ol NL1\ORK LCON.
359, 359-60 ,2008,.
35
ee Larry D. \oodard, Pe.i`. ig Cavbte: Ditcbivg ver ort for ociat Meaia, ABC NL\S, Dec.
23, 2009, ;11>?--.#*'(!%"/+"*+,->A&'1E&FG3625876.
users got more o their inormation through |the lacebook| network, their
Internet search engine and naigator might become lacebook, not Google.`
36

1he upshot: 1o the extent that inclusion in Google search results is about
Stiglerian` search-cost reduction or websites ,and it can hardly be anything
else,, the range o alternate acilities or this unction is nearly limitless.
linally, Google competes not only with other general search engines ,and
possibly all other orms o adertising, but also with so-called ertical search
engines. 1hese are search engines and e-commerce websites with search
unctionality that specializes in speciic content: Amazon in books, music, and
other consumer goods, Kayak in trael serices, eBay in consumer auctions,
\ebMD in medical inormation and products, Source1ool in business-to-
business supplies, \elp in local businesses, and many others. 1o the extent that
Internet users bypass Google and begin their searches at one o these
specialized sites ,as is increasingly the case,, the alue to these heaily-traicked
websites rom access to Google`s users decreases.
3

Competition rom ertical search engines is important because ad click-through
rates likely are higher when consumers are actiely searching or something to
buy-just as search adertising targets consumers who express some interest in
a particular search term, the eect is magniied i the searcher can be identiied
as an immediate consumer. 1hus online retailers like CDnow that can establish
their own brands and their own naigation channels
38
hae a signiicant
adantage in drawing searchers-and adertisers-away rom Google: 1he act
that a consumer is perorming a search on a retail site itsel coneys important
and aluable inormation to adertisers that is not otherwise aailable rom
most undierentiated Google searches-it certainly increases the chance that
the searcher is searching to buy a CD rather than learn something about the
singer. Because this ready-to-buy` traic is the most aluable, there is a
possibility o two separate search markets, with most high-alue traic
bypassing general-purpose search engines or product search sites like eBay and
Amazon.com, and with Google and other general-purpose search engines
sering primarily non-targeted, lower-alue traic. 1he implication is that, while
een relatiely small-scale competition may present a potentially signiicant
threat to Google`s search business, this threat does not depend on links to these
sites rom Google`s search results. And thus these competitors hae a strong,

36
KLN AULL11A, GOOGLLD: 1lL LND Ol 1lL \ORLD AS \L KNO\ I1 12-3 ,2009,.
3
lor example, in the thirty days ending on lebruary 23, 2010, less than ten percent o isits
to eBay.com originated rom a search engine. ee ALLXA, ea,.cov ite vfo,
;11>?--!!!".@(P."*+,-%&1(&'C+-(#.D"*+,.
38
ee Donna L. loman & 1homas P. Noak, or to .cqvire Cv.tover. ov tbe !eb, lARV. BUS.
RLV., May-June 2000, at 3, 5, . ,CDnow was acquired by Amazon.com in 2001.,
independent incentie to deelop marketing programs outside o Google`s
search pages-and there is good reason not to deputize Google in the process.
Is Google an Essential Facility?
Recall that the basic claim is that Google`s competitors are oreclosed rom
access to Google`s desirable ,essential, marketing platorm and thereby suer
signiicant harm. O course rom the outset, this has it backwards ,and this is a
core problem with the essential acilities doctrine as a whole,.
I there is a problem, it should be the problem o limited access b, Coogte`. v.er.
to Google`s competitors. Sometimes the absence o access by competitors to
consumers is the same thing as the absence o access by consumers to
competitors, but it depends on how well the market has been deined. In the
most undamental sense Google has precisely zero control oer access by
consumers ,meaning users who use Google to search the Internet, to
competitors: Anyone with access to a browser can access any site on the
Internet simply by typing its URL into the browser. Perhaps understanding
this, proponents o the Internet search is an essential acility` claim argue that
mere access is insuicient, and that consumers are essentially ignorant about the
aluable content on the web except by search engines, which are subject to the
search engine`s editorial control oer that access. 1o the typical Google user,
according to this iew, Google`s competitors are eectiely non-existent unless
they appear in the top ew search results.
Now we are dangerously close to the sort o arbitrary market deinition exercise,
deoid o the discipline imposed by economics, that identiies an
anticompetitie problem by narrowing the market until eery company is a
monopolist oer some small group o consumers. Indeed, one can always
deine a market by ocusing on idiosyncratic preerences or product ariations.
Justice lortas decried this type o analysis in his dissent in Crivvett ,regarding
home security systems,, and it merits quoting at length:
1he trial court`s deinition o the product` market een more
dramatically demonstrates that its action has been
Procrustean-that it has tailored the market to the dimensions
o the deendants. It recognizes that a person seeking
protectie serices has many alternatie sources. It lists
watchmen, watchdogs, automatic proprietary systems
conined to one site, ,oten, but not always,, alarm systems
connected with some local police or ire station, oten
unaccredited CSPS |central station protectie serices|, and
oten accredited CSPS.` 1he court inds that een in the same
city a single customer seeking protection or seeral premises
may exercise its option` dierently or dierent locations. It
may choose accredited CSPS or one o its locations and a
dierent type o serice or another.
But the court isolates rom all o these alternaties only those
serices in which deendants engage. It eliminates all o the
alternatie sources despite its conscientious enumeration o
them. Its deinition o the releant market` is not merely
conined to central station` protectie serices, but to those
central station protectie serices which are accredited` by
insurance companies.
1here is no pretense that these urnish peculiar serices or
which there is no alternatie in the market place, on either a
price or a unctional basis. 1be covrt retie. .otet, vov it. fivaivg tbat
tbe .errice. offerea b, accreaitea cevtrat .tatiov. are of better qvatit,, and
upon its conclusion that the insurance companies tend to gie
noticeably larger` discounts to policyholders who use
accredited central station protectie serices. 1his Court now
approes this strange red-haired, bearded, one-eyed man-with-
a-limp classiication.
39

In Internet search as well, complainants imply a market based on the act that
Google oers better quality` access to a larger set o Internet users than the
myriad existing alternaties. But claiming essentiality based on a competitor`s
relatie high quality is deeply problematic.
1his point is o great importance in assessing the economics o the essential
acilities doctrine generally and its application to Internet search in particular. It
is clear, een under a airly expansie reading o the essential acilities doctrine,
that een a monopolist has no duty to subsidize the eorts o a less-eectie
rial.
40
Arguably the Aspen Skiing case should hae been tossed out on this
basis. As a practical matter, the Aspen Ski Company, by entering into a joint
marketing agreement with its smaller rial, Aspen lighlands, allowed lighlands
to take adantage o its markedly larger productiity ,both in deeloping ski
terrain and amenities, as well as marketing Aspen as a ski destination,. Its
subsequent decision to drop lighlands rom its marketing program or ailing
to oer suicient return on its inestment should hae been unobjectionable.
41

Similarly, the explicit claim in cases brought against Google by its allegedly-
oreclosed rials is that these ,relatiely miniscule, sites should hae access to
Google`s eectie and inexpensie marketing tool. But it is by no means clear
that Google does or should hae this duty to promote its rials ,without
compensation to Google, as it happens,. 1his is particularly true when, as
discussed aboe, other modes o access exist or competitors` actiities, een i

39
|.. r. Crivvett Cor., 384 U.S. 563, 590-91 ,1966, ,emphasis added,.
40
ee Olympia Lquip. Leasing Co. . \estern Union 1el. Co., 9 l.2d 30 ,
th
Cir. 1986,.
41
ee KLI1l N. l\L1ON, AN1I1RUS1 LA\: LCONOMIC 1lLOR\ AND COMMON LA\
LVOLU1ION 205-06 ,2003,.
these modes o access are o lower quality or higher cost. Particularly where, as
here, the alleged bottleneck arises not out o a combination with another irm
or irms but out o unilateral conduct ,success in the marketplace,, the claim
that a superior access point among many ,inerior, access points should be pried
open or the beneit o its competitors is specious.
It is worth noting that an alleged Google competitor, Source1ool, in the
1radeComet complaint,
42
has made a ersion o this argument, alleging that
Google once engaged in proitable commerce with Source1ool ,by selling
Source1ool ads next to Google search results, and then penalized Source1ool
to its ,Google`s, economic detriment.
43
1he shape o this argument is a
transparent eort to remain under what is let o the essential acilities doctrine
ollowing 1riv/o. But notice that een i it is true that Google intentionally
ended a proitable arrangement with Source1ool ,which is by no means clear,,
the claim still doesn`t pass muster. It is almost impossible that Google could be
receiing less reenue rom whateer site has replaced Source1ool in the paid
search result spots Source1ool once paid or. As a result, een i Google were
oregoing a preiously-proitable relationship ritb ovrce1oot, it is not, in act,
suering any economic harm because another adertiser has stepped into
Source1ool`s shoes.
O course the argument that Google`s competitors are eectiely absent
without ,guaranteed, access to Google`s top ew search results proes too
much. 1here is a scarcity o top ew search results,` and any eectie search
engine must hae the ability to ensure that those results are the most releant
possible, as well as that they do not iolate arious quality, saety, moral or
other standards that the search engine chooses to promote. lorcing |owners
o essential acilities| to share access may not enhance consumer welare.`
44

Pure neutrality` is neither possible nor desirable, and the exclusion o certain
websites rom these coeted positions should be deemed utterly unpersuasie in
making out een a riva facie monopolization case against a search engine.
And it is not een the case that Source1ool, loundem,
45
and other competing
websites are absent rom Google, it is, howeer, sometimes the case that these

42
Complaint , 1raaeCovet.Cov C r. Coogte, vc., 693 l. Supp. 2d 30 ,S.D.N.\. 2009, ,No.
09Ci.1400,SlS,,.
43
a. 8.
44
KLI1l N. l\L1ON, AN1I1RUS1 LA\: LCONOMIC 1lLOR\ AND COMMON LA\ LVOLU1ION
208 ,2003,.
45
loundem is a ertical search and price comparison` site in the UK. ee
!!!"C+$'F(,"*+"$). 1he company is at the heart o the search neutrality` debate in
Internet search. It has created a website to adocate its iews on the neutrality issue at
!!!"%(.A*;'($1A.@&1D"+A/, and its claims are at the heart o the Luropean Commission`s
inestigation o Google. See loundem`s discussion o the LU action and its relationship to
loundem`s claims in vroeav Covvi..iov avvcbe. .vtitrv.t vre.tigatiov of Coogte, SLARCl
NLU1RALI1\.ORG, No. 30, 2010, ;11>?--!!!"%(.A*;'($1A.@&1D"+A/.
sites do not show up in the top ew organic search results ,and, oten at the
same time, Google`s own competing product search results do,. But i access to
the top ew search results is required to ensure the requisite access sought by
Google`s competitors, the releant market has been narrowed considerably,
creating a standard that can`t possibly be met, no matter how neutral` a search
engine`s results.
Meanwhile, i loundem were to disappear rom the ace o the Larth, who,
other than its inestors and employees ,and perhaps their landlords,, would be
harmed 1he implicit claim ,i an antitrust case is to be made, is that websites
like loundem apply a constraint on Google`s ability to extract monopoly rents
,presumably rom adertisers,. But this is a curious claim to make while
simultaneously arguing that Google itsel is made better` ,as in, searchers are
indeed looking or loundem in searches rom which the site may be excluded,
by the inclusion o loundem in its search results ,thus, presumably, increasing
Google`s attractieness to its users and thus its adertisers,, while also claiming
that loundem would cease to exist without access to the top ew Google search
results.
Google does not sell retail goods, and does not proit directly rom its own
product search oerings ,which compete with loundem,, instead receiing
beneit by increasing its customer base and the eicacy ,presumably, o paid
adertisements on its search pages that include a link to its own price
comparison results. It is a remarkably tenuous claim to make that Google
proits more by degrading its search results than by improing them. I the
contrary claim is really true-i, that is, Google harms itsel or its adertisers by
intentionally penalizing competing sites like loundem-then that argument and
any eidence or it is absent rom the current debate. And, o course, i Google
is, as it claims, actually improing its product by applying qualitatie decisions to
demote sites like loundem and others that, Google claims, merely re-publish
inormation rom elsewhere on the web with precious little original content,
then Google`s eorts should be seen as a eature and not a bug.
Moreoer, the extension o the essential acilities logic to competition between
Google and competitors like loundem, MapQuest or Kayak is extremely
problematic. 1o the extent that Google and loundem, or example, are
competitors, they are competitors not in the adertising space but rather in the
inormation dissemination and retail distribution channel` space. I`m not sure
what else to call it. loundem earns reenue by directing customers to retail
sites to purchase goods. In this sense, loundem acts like a shopping mall.
Google does the same, only instead o receiing a cut rom the sale, as
loundem does, Google sells adertisements. 1hus, when loundem complains
about access to Google`s site, it is a competing channel o distribution,
complaining that it needs access to its competitor`s distribution channel in order
to compete.
It`s a weird sort o complaint. It isn`t the same as the classic essential acilities
sort o complaint where, to simpliy, the owner o a ertically-integrated railroad
and rail transport company preents access by other transport companies to its
railroad line. Instead this would be like railroad company A arguing that
railroad B must gie A access to B`s tracks so A can sell access to those tracks
to other rail transport companies.
But een this doesn`t completely capture the audacity o the complaint, because
or the analogy to hold, railroad A would actually be asking the court to orce
railroad B to put up a sign at the head o its tracks allowing railroad A to oer
to trains already on B`s railroad the opportunity to jump o B`s railroad and
start oer again on A`s railroad that ollows another route-but without
knowing or sure i the route is better or worse until you jump onto A`s tracks.
Something like that. Again, it`s weird.
And note, o course, the problem that at the head o the tracks` ,as in
something like the irst, second or third organic result`, is a problematic
requirement as only three sites at any gien time can occupy those spots-but
there may be many more than three irms complaining o Google`s conduct
and,or aected by the agaries o its product design decisions. Or to keep with
the shopping mall analogy, it`s like the owner o any o a number o small, new
shopping malls requiring the owner o a large, established shopping mall to
permit each o the new mall`s owners to set up a bus line to erry shoppers to
the new mall as they enter the established mall. Len where the established
mall has a geographic, reputational and resource adantage, no one would argue
that this access was essential to eicient commerce, and the cost to the
successul incumbent would be maniestly too high.
As discussed aboe, sites like loundem do indeed hae access to Google`s end
users ia any number o keywords on Google`s site. 1ype UK price
comparison site` into Google and a number o Google competitors come up
including loundem ,and Google`s own price comparison site is seemingly
absent,. 1he claim thus becomes one that is either inappropriately aggregated
,or all search terms on aerage that may direct users to loundem, loundem is
eectiely denied access to the top search results`, or else oerly narrow ,we
preer customers to ind us by typing Nikon camera` into Google, not by typing
price comparison Nikon camera` into Google`,. In any case, access is in act
aailable or these competitors, and the indispensable requirement or
inoking the |essential acilities| doctrine is the unaailability o access to the
essential acilities`, where access exists, the doctrine seres no purpose.`
46

Meanwhile, it is diicult to see how releance ,and thus eiciency, could be
well-sered by a neutrality principle that required a tool that reavce. search costs
to inherently ivcrea.e those costs by directing searchers to a duplicate search on
another site. I one is searching or a speciic product and hoping to ind price
comparisons on Google, why on earth would that person be hoping to ind not
Google`s own eorts at price comparison, built right into its search engine, but

46
1riv/o, 540 U.S. at 411.
instead a link to another site that requires another seeral steps beore inding
the inormation
Seen this way, Google`s decision to promote its own price comparison results is
a simple product pricing and design decision, protected by good sense and the
1riv/o decision ,at least in the U.S.,. Unlike the majority o its ertical search
competitors and by design, Google makes no direct reenue rom users clicking
through to purchase anything rom its shopping search results, and this allows it
to oer a dierent ,and, to many consumers, a signiicantly better, set o
results. 1he page has paid search results only in small boxes at the top and
bottom, the inormation is all algorithmically generated, and retailers do not pay
to hae their inormation on the page. lor this product design-by deinition
o great alue to users ,in eect lowering the price to them o their product
search,-to merit Google`s inestment, it is necessary that its own, more-
releant and less-expensie results receie priority. I this is generating
something o alue or Google it is doing so only in the most salutary ashion:
by oering additional resources or users to improe their search experience`
and thus induce them to use Google`s search engine. 1o require neutrality` in
this setting is to impair the site`s ability to design and price its own product.
Len the ..ev /iivg decision didn`t go that ar, requiring access to a joint
marketing arrangement but not obligating Aspen Ski Company to alter its prices
or skiers seeking to access only its own slopes.
And the same analysis holds or assessments o Google`s other oerings ,maps
and ideos, or example, that compete with other sites. Look or the nearest
McDonalds in Google and a Google Map is bound to top the list ,but not be
the exclusie result, o course,. But why should it be any other way In eect,
what Google does is gie you the \eb`s content in as accessible and appropriate
a orm as it can-design decisions that, Google must beliee, increase quality
and reduce eectie price or its users. By oering not only a link to
McDonalds` web site, as well as arious other links, but also a map showing the
locations o the nearest restaurants, Google is oering up results in dierent
orms, hoping that one is what the user is looking or. 1here is no economic
justiication or requiring a search engine in this setting to oer another site`s
rather than its own simply because there happen to be other sites that do,
indeed, oer such content ,and would like cheaper access to consumers,.
Conclusion
Search neutrality and orced access to Google`s results pages is based on the
proposition that-Google`s users` interests be damned-i Google is the easiest
way competitors can get to potential users, Google must proide that access.
1he essential acilities doctrine, dealt a near-death blow by the Supreme Court
in 1riv/o, has long been on the ropes. It should remain moribund here. On
the one hand Google does not preclude, nor does it hae the power to preclude,
users rom accessing competitors` sites, all users need do is type
oundem.com` into their web browser-which works een i it`s Google`s
own Chrome browser! 1o the extent that Google can and does limit
competitors` access to its search results page, it is not controlling access to an
essential acility` in any sense other than \al-Mart controls access to its own
stores. Google search results generated by its proprietary algorithm and ound
on its own web pages` do not constitute a market to which access should be
orcibly granted by the courts or legislature.
1he set o claims that are adduced under the rubric o search neutrality` or the
essential acilities doctrine` against Internet search engines in general and, as a
practical matter, Google in particular, are deeply problematic. 1hey risk
encouraging courts and other decision makers to ind antitrust iolations where
none actually exist, threatening to chill innoation and eiciency-enhancing
conduct. In part or this reason, the essential acilities doctrine has been
relegated by most antitrust experts to the dustbin o history. As Joshua \right
and I conclude elsewhere:
Indeed, it is our iew that in light o the antitrust claims arising
out o innoatie contractual and pricing conduct, and the
apparent lack o any concrete eidence o anticompetitie
eects or harm to competition, an enorcement action against
Google on these grounds creates substantial risk or a alse
positie` which would chill innoation and competition
currently proiding immense beneits to consumers.
4

4
Georey A. Manne & Joshua D. \right, Coogte ava tbe ivit. of .vtitrv.t: 1be Ca.e .gaiv.t tbe
Ca.e .gaiv.t Coogte, 34 lARV. J. L. & PUB. POL`\ at 62.
Some Skepticism
About Search Neutrality
By James Grimmelmann
*

1be erfect .earcb evgive rovta be ti/e tbe viva of Coa.
1

1be Coa tbat bota. ,ov orer tbe it of bett, vvcb a. ove bota. a .iaer, or
.ove toatb.ove iv.ect, orer tbe fire, abbor. ,ov, ava i. areaafvtt,
roro/ea; bi. rratb torara. ,ov bvrv. ti/e fire; be too/. vov ,ov a.
rortb, of votbivg et.e, bvt to be ca.t ivto tbe fire .
2

f Coa aia vot ei.t, it rovta be vece..ar, to ivrevt biv.
3

Search engines are attention lenses, they bring the online world into ocus.
1hey can redirect, reeal, magniy, and distort. 1hey hae immense power to
help and to hide. \e use them, to some extent, always at our own peril. And
out o the many ways that search engines can cause harm, the thorniest
problems o all stem rom their ranking decisions.
4

\hat makes ranking so problematic Consider an example. 1he U.K.
technology company loundem oers ertical search`
5
-it helps users
compare prices or electronics, books, and other goods. 1hat makes it a Google
competitor.
6
But in June 2006, Google applied a penalty` to loundem`s

Associate Proessor o Law, New \ork Law School. I would like to thank Aislinn Black and
lrank Pasquale or their comments. 1his essay is aailable or reuse under the Creatie
Commons Attribution 3.0 United States license,
;11>?--*A(.1&O(*+,,+'%"+A/-j@&*('%(%-#D-B"2-$%-.
1
Charles lerguson, !bat`. êt for Coogte, 1LCl. RLV., Jan. 1, 2005, at 38, aailable at
;11>?--!!!"1(*;'+@+/DA(O&(!"*+,-!(#-762:8- ,quoting Sergey Brin, co-ounder o
Google,.
2
Jonathan Ldwards, ivver. iv tbe ava. of av .vgr, Coa ,sermon deliered July 8, 141 in
Lnield, Connecticut,, araitabte iv 22 \ORKS Ol JONA1lAN LD\ARDS 411 ,larry S. Stout &
Nathan O. latch eds., \ale Uniersity Press 2003,.
3
Voltaire, itre a tàvtevr av tirre ae. 1roi. ivo.tevr. |etter to tbe .vtbor of 1he 1hree
Impostors| ,168,, araitabte at ;11>?--!!!"!;&1,.'"(F$-chX-1A+&%"&,>+%1($A%";1,@.
4
ee James Grimmelmann, 1be trvctvre of earcb vgive ar, 93 IO\A L. RLV. 1, 1-44 ,200,
,identiying nine distinct types o harm search engines can cause to users, inormation
proiders, and third parties,.
5
ee geveratt, JOlN BA11LLLL, 1lL SLARCl: lO\ GOOGLL AND I1S RIVALS RL\RO1L 1lL
RULLS Ol BUSINLSS AND 1RANSlORMLD OUR CUL1URL 24-6 ,2005, ,discussing domain-
speciic search`,.
6
ee Coogte Proavct earcb eta, GOOGLL, ;11>?--!!!"/++/@("*+,->AF;>.
website, causing all o its pages to drop dramatically in Google`s rankings.
It
took more than three years or Google to remoe the penalty and restore
loundem to the irst ew pages o results or searches like compare prices
shoei xr-1000.`
8
loundem`s traic, and hence its business, dropped o
dramatically as a result. 1he experience led loundem`s co-ounder, Adam Ra,
to become an outspoken adocate: creating the site searchneutrality.org,
9
iling
comments with the lederal Communications Commission ,lCC,,
10
and taking
his story to the op-ed pages o 1be êr Yor/ 1ive.,
11
calling or legal protection
or the loundems o the world.
O course, the goernment doesn`t get inoled eery time a business is harmed
by a bad ranking-or Cov.vver Reort. would be out o business.
12
Instead,
search-engine critics base their case or regulation on the immense power o
search engines, which can break the business o a \eb site that is pushed
down the rankings.`
13
1hey hae the power to shape what millions o users,
carrying out billions o searches a day, see.
14
At that scale, search engines are
the new mass media
15
-or perhaps the new veta media-capable o shaping
public discourse itsel. And while power itsel may not be an eil, abuse o
power is.
Search-engine critics thus aim to keep search engines-although in the U.S. and
much o the Lnglish-speaking world, it might be more accurate to say simply
Google`
16
-rom abusing their dominant position. 1he hard part comes in
deining abuse.` Ater a decade o arious attempts, critics hae hit on the

ee ovvaev`. Coogte tor,, SLARClNLU1RALI1\.ORG ,Aug. 18, 2009,,

;11>?--!!!"%(.A*;'($1A.@&1D"+A/-C+$'F(,L/++/@(L%1+AD.
8
1he Shoei XR-1000 is a motorcycle helmet-according to loundem, it`s 149.99 plus 10
deliery rom lelmet City.
9
.bovt, SLARCl NLU1RALI1\.ORG, Oct. 9, 2009, ;11>?--!!!"%(.A*;'($1A.@&1D"+A/-.#+$1.

10
Reply Comments o loundem, In the Matter o Presering the Open Internet Broadband
Industry Practices, GN Docket No. 09-191 ,l..C.C,.
11
Adam Ra, earcb, vt Yov Ma, ôt iva, N.\. 1IMLS, Dec. 2, 2009, at A2.
12
Cf. Bose Corp. . Consumers Union, 466 U.S. 485 ,1984, ,holding Cov.vver Reort. not
subject to product disparagement liability or negatie reiew o Bose speaker,.
13
1be Coogte .tgoritbv, N.\. 1IMLS, July 14, 2010, at A30.
14
ee GRAN1 LSKLLSLN L1 AL, 1lL DIGI1AL LCONOM\ lAC1 BOOK 12-13 ,10th ed. 2009,,
;11>?-->CC"+A/-&%%$(%L>$#%-#++)%-C.*1#++)4721;4iF">FC.
15
ee geveratt, KLN AULL11A, GOOGLLD: 1lL LND Ol 1lL \ORLD AS \L KNO\ I1 ,2009,
,trying to understand Google by adopting the perspectie o the media industry,. Cf. Aaron
Swartz, Coogtivg for ocioatb., RA\ 1lOUGl1 ,Dec. 14, 2009,,
;11>?--!!!"..A+'%!"*+,-!(#@+/-/++/@(F ,describing Coogtea as a history o |Google|
as told by the incumbent sociopaths`,.
16
ee LSKLLSLN L1 AL., lAC1 BOOK, .vra note 14.
idea o neutrality` as a goerning principle. 1he idea is explicitly modeled on
network neutrality, which would orbid operators o broadband networks to
discriminate against third-party applications, content or portals.`
1
Like
broadband Internet serice proiders ,ISPs,, search engines accumulate great
power oer the structure o online lie.`
18
1hus, perhaps search engines should
similarly be required not to discriminate among websites.
lor some academics, this idea is a thought experiment: a way to explore the
implications o network neutrality ideas.
19
lor others, it is a real proposal: a
preliminary agenda or action.
20
Lawyers or ISPs ighting back against network
neutrality hae seized on it, either as a reavctio aa ab.vravv or a way to kneecap
their bitter rial Google.
21
Len the êr Yor/ 1ive. has gotten into the game,
running an editorial calling or scrutiny o Google`s editorial policy.`
22
Since
êr Yor/ 1ive. editorials, as a rule, relect no independent thought but only a
kind o preailing conentional wisdom, it is clear that search neutrality has
truly arried on the policy scene.
Notwithstanding its sudden popularity, the case or search neutrality is a
muddle. 1here is a undamental misit between its aowed policy goal o
protecting users and most o the tests it proposes to protect them. Scratch
beneath the surace o search neutrality and you will ind that it would protect

1
Barbara an Schewick, 1orara. av covovic raveror/ for êtror/ êvtratit, Regvtatiov, 5 J.
1LLLCOMM. & lIGl 1LCl. L. 329, 333 ,200,, aailable at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FGH75337.
18
lrank Pasquale, vtervet ôvai.crivivatiov Privcite.: Covverciat tbic. for Carrie. ava earcb
vgive., 2008 U. ClI. LLGAL. lORUM 263, 298, aailable at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG77B6783 |hereinater Pasquale,
vtervet ôvai.crivivatiov Privcite.|.
19
ee Mark R. Patterson, ôvêtror/ arrier. to êtror/ êvtratit,, 8 lORDlAM L. RLV. 2843
,2010,, Andrew Odlyzko, êtror/ êvtratit,, earcb êvtratit,, ava tbe êrerevaivg Covftict
etreev fficievc, ava airve.. iv Mar/et., 8 RLV. NL1\ORK LCON. 40 ,2009,.
20
ee DA\N C. NUNZIA1O, VIR1UAL lRLLDOM: NL1 NLU1RALI1\ AND lRLL SPLLCl IN 1lL
IN1LRNL1 AGL ,2009, |hereinater NUNZIA1O, VIR1UAL lRLLDOM|, Pasquale, vtervet
ôvai.crivivatiov Privcite., .vra note 18, Oren Bracha & lrank Pasquale, eaerat earcb
Covvi..iov: .cce.., airve.., ava .ccovvtabitit, iv tbe ar of eecb, 93 CORNLLL L. RLV. 1149
,2008, |hereinater Bracha & Pasquale, eaerat earcb Covvi..iov|, Jennier A. Chandler, .
Rigbt to Reacb av .vaievce: .v .roacb to vterveaiar, ia. ov tbe vtervet, 35 lOlS1RA L. RLV.
1095 ,200,.
21
Letter rom Robert \. Quinn, Jr., Senior Vice President, A1&1, to Sharon Gillett, Chie,
\ireline Competition Bureau lederal Communications Commission ,Sept. 25, 2009,,
aailable at
;11>?--/A.>;&*%H"'D1&,(%"*+,->.*)./(%->FC-1(*;'+@+/D-522323584XJJL
d(11(A">FC.
22
1be Coogte .tgoritbv, .vra note 13. But see Danny Sullian, 1be êr Yor/ 1ive. .tgoritbv c
!b, t êea. Corervvevt Regvtatiov, SLARCl LNGINL LAND ,July 15, 2010, ,parodying New
\ork 1imes editorial on Google,.
not search users, but websites. In the search space, howeer, websites are as
oten users` evevie. as not, the whole point o search is to help users aoid the
sites they don`t want to see.
In short, search neutrality`s ends and means don`t match. 1o explain why, I will
deconstruct eight proposed search-neutrality principles:
1. qvatit,: Search engines shouldn`t dierentiate at all among websites.
2. Ob;ectirit,: 1here are correct search results and incorrect ones, so search
engines should return only the correct ones.
3. ia.: Search engines should not distort the inormation landscape.
4. 1raffic: \ebsites that depend on a low o isitors shouldn`t be cut o
by search engines.
5. Reteravce: Search engines should maximize users` satisaction with
search results.
6. etfivtere.t: Search engines shouldn`t trade on their own account.
. 1rav.arevc,: Search engines should disclose the algorithms they use to
rank web pages.
8. Mavivtatiov: Search engines should rank sites only according to general
rules, rather than promoting and demoting sites on an indiidual basis.
As we shall see, all eight o these principles are unusable as bases or sound
search regulation.
I would like to be clear up ront about the limits o my argument. Just because
search neutrality is incoherent, it doesn`t ollow that search engines desere a
ree pass under antitrust, intellectual property, priacy, or other well-established
bodies o law.
23
Nor is search-speciic legal oersight out o the question.
Search engines are capable o doing dastardly things: According to
Business\eek, the Chinese search engine Baidu explicitly shakes down
websites, demoting them in its rankings unless they buy ads.
24
It`s easy to tell
horror stories about what search engines vigbt do that are just plausible enough
to be genuinely scary.
25
My argument is just that search neutrality, as currently
proposed, is unlikely to be workable and quite likely to make things worse. It
ails at its own goals, on its own deinition o the problem.

23
1his essay is not the place or a ull discussion o these issues ,although we will meet
antitrust and consumer protection law in passing,. Grimmelmann, 1be trvctvre of earcb
vgive ar, .vra note 4, proides a more detailed map.
24
Chi-Chu 1schang, 1be qveee at Cbiva`. aiav, BUSINLSS\LLK, Dec. 31, 2008,
;11>?--!!!"#$%&'(%%!(()"*+,-,./.0&'(-*+'1('1-23425-#67782579725:8";1,
,alleging that Baidu directly retaliates against sites that reuse to buy sponsored links by
demoting them in its organic rankings,.
25
ee, e.g., Cory Doctorow, croogtea, ;11>?--*A.>;+$'F"*+,-%*A++/@(F";1,@, 1om Slee, Mr.
Coogte`. Cviaeboo/, \lIMSLL\ ,Mar. , 2008,,
;11>?--!;&,%@(D"1D>(>.F"*+,-!;&,%@(D-522H-2B-,AL/++/@(%L/$&F";1,@.
Theory
Beore deling into the speciics o search-neutrality proposals, it will help to
understand the principles said to justiy them. 1here are two broad types o
arguments made to support search neutrality, one each ocusing on users and
on websites. A search engine that misuses its ranking power might be seen
either as vi.teaaivg v.er. about what`s aailable online, or as btoc/ivg websites
rom reaching users.
26
Consider the arguments in turn.
|.er.: Search helps people ind the things they want and need. Good search
results are better or them. And since search is both subjectie and personal,
users themseles are the ones who should deine what makes search results
good. 1he usual term or this goal is releance`: releant results are the ones
that users themseles are most satisied with.
2
All else being equal, good search
policy should try to maximize releance.
A libertarian might say that this goal is triial.
28
Users are ree to pick and
choose among search engines and other inormational tools.
29
1hey will
naturally lock to the search engine that oers them the most releant results,
the market will proide just as much releance as it is eicient to proide.
30

1here is no need or regulation, releance, being demanded by users, will be

26
Other arguments or search neutrality reduce to these two. Bracha and Pasquale, or
example, are concerned about democracy. 1hey want an open and relatiely equal chance
to all members o society or participation in the cultural sphere.` Bracha & Pasquale, eaerat
earcb Covvi..iov, .vra note 20, at 1183-84. Search engines proide that chance i
indiiduals can both ind ,as users, and be ound ,as websites, when they participate in
politics and culture. Similarly, Bracha and Pasquale`s economic eiciency argument turns on
users` ability to ind market inormation, ia. at 113-5. and their airness concern speaks to
websites` losses o audience or business,` ia. at 115-6. \hateer interest society has in
search neutrality arises rom users` and websites` interests in it-so we are justiied in
ocusing our attention on users and websites.
2
ee BA11LLLL, 1lL SLARCl, .vra note 5, at 19-25.
28
lor a clear statement o a libertarian perspectie on search neutrality, see Mike Masnick`s
posts at 1echdirt on the subject, collected at
;11>?-!!!"1(*;F&A1"*+,-#@+/">;>E1./G%(.A*;k'($1A.@&1D. Lric Goldman`s earcb
vgive ia. ava tbe Devi.e of earcb vgive |toiavi.v, 8 \ALL J. L. & 1LCl. 188 ,2006,, makes a
general case against the regulation o releance on similar grounds.
29
In Google`s words, Competition is just one click away.` Adam Koaceich, Coogte`.
.roacb to Covetitiov, GOOGLL POLIC\ BLOG ,May 8, 2009,,
;11>?--/++/@(>$#@&*>+@&*D"#@+/%>+1"*+,-5223-28-/++/@(%L.>>A+.*;L1+L
*+,>(1&1&+'";1,@.
30
ee Lric Goldman, . Coa.eav .vat,.i. of Mar/etivg, 2006 \IS. L. RLV. 1151.
supplied by search engines. And this is exactly what search engines themseles
say: releance is their principal, or only, goal.
31

1he response to this point o iew-most careully argued by lrank
Pasquale
32
-is best described as liberal.` It ocuses on maximizing the
eectie autonomy o search users, but questions whether market orces
actually enable users to demand optimal releance. lor one thing, it questions
whether users can actually detect deiations rom releance.
33
1he user who
turns to a search engine, by deinition, doesn`t yet know what she`s looking or
or where it is. ler own knowledge, thereore, doesn`t proide a ully reliable
check on what the search engine shows her. 1he inormation she would need
to know that the search engine is hiding something rom her may be precisely
the inormation it`s hiding rom her-a releant site that she didn`t know
existed.
34

Perhaps just as importantly, structural eatures o the search market can make it
hard or users to discipline search engines by switching. Search-neutrality
adocates hae argued that search exhibits substantial barriers to entry.
35
1he
web is so big, and search algorithms so complex and reined, that there are
substantial ixed costs to competing at all.
36
Moreoer, the rise o personalized
search both creates switching costs or indiidual users
3
and also makes it
harder or them to share inormation about their experiences with multiple
search engines.
38

!eb.ite.: 1he case or protecting websites reaches back into ree speech theory.
Jerome Barron`s 196 article, .cce.. to tbe Pre..-. êr ir.t .vevavevt Rigbt,
39

31
ee, e.g., 1ecbvotog, Orerrier, GOOGLL, www.google.com,corporate,tech.html, or !eb
Docvvevt. .re Rav/ea, \AlOO!,
;11>?--;(@>"D.;++"*+,-@-$%-D.;++-%(.A*;-&'F(P&'/-A.')&'/L27";1,@, ../ earcb
1ecbvotog,, ASK, ;11>?--%>".%)"*+,-('-F+*%-.#+$1-.%)41(*;'+@+/D"%;1,@.
32
ee Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, Bracha & Pasquale, eaerat
earcb Covvi..iov, .vra note 20, lrank Pasquale, ..teri./ Reri.itea: Debativg a Rigbt of Ret, ov
earcb Re.vtt., 3 J. BUS. & 1LCl. L. 61 ,2008,, lrank Pasquale, Rav/ivg., Reavctiovi.v, ava
Re.ov.ibitit,, 54 CLLV. S1. L. RLV. 115 ,2006,.
33
ee Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1116, Patterson, ôvêtror/
arrier., .vra note 19, at 2860-62.
34
ee Bracha & Pasquale, eaerat earcb Covvi..iov, .vra note 20, at 1183-84.
35
ee ia. at 1181-82.
36
ee ia. at 1181.
3
ee Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 265.
38
ee lrank Pasquale, Covta Per.ovatiea earcb Rviv Yovr ife., CONCURRING OPINIONS ,leb. ,
2008,,
;11>?--!!!"*+'*$AA&'/+>&'&+'%"*+,-.A*;&O(%-522H-25->(A%+'.@&0(F4%(";1,@.
39
80 lARV. L. RLV. 1641 ,196,.
argued that reedom o speech is an empty right in a mass-media society unless
one also has access to the mass media themseles. le thus argued that
newspapers should be required to open their letters to the editor and their
adertising to all points o iew.
40
Although his proposed right o access is
basically a dead letter as ar as lirst Amendment doctrine goes,
41
it captured the
imaginations o media-law scholars and media adocates.
42

Scholars hae begun to adapt Barron`s ideas to online intermediaries, including
search engines. Dawn Nunziato`s book 1irtvat reeaov draws extensiely on
Barron to argue that Congress may need to authorize the regulation o
dominant search engines to require that they proide meaningul access to
content.`
43
Jennier Chandler applies Barron`s ideas to propose a right to
reach an audience`
44
that would gie website owners arious protections against
exclusion
45
and demotion by search engines.
46
Similarly, lrank Pasquale
suggests bringing uniersal serice` oer into the search space,
4
perhaps
through a goernment-proided search engine.
48

1he Barronian argument or access, howeer, needs to be qualiied. 1he ree-
speech interest in access to search engine ranking placement is really avaievce.`
ree speech interest, the real harm is that search users hae been depried o
access to the speech o websites, not that websites hae been depried o access
to users. Put another way, websites` access interest is deriatie o users`
interests. In the Supreme Court`s words, 1he lirst Amendment protects the
right o eery citizen to reach the minds o rittivg listeners.``
49
Or, in Jerome

40
a. at 166.
41
ee Miami lerald Pub`g Co. . 1ornillo, 418 U.S. 241 ,194, ,striking down llorida law
requiring newspapers to proide equal space or political responses,.
42
ee, e.g., Rectaivivg tbe ir.t .vevavevt: Cov.titvtiovat 1beorie. of Meaia Reforv, 35 lOlS1RA L.
RLV. 91-1582 ,symposium issue collecting papers rom conerence honoring the 40
th

anniersary o publication o .cce.. to tbe Pre..,.
43
NUNZIA1O, VIR1UAL lRLLDOM, .vra note 20, at 150.
44
Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1103-1 ,search engines,, 1124-30
,proposed right,.
45
Lxclusion rom a search index may sound like a bright-line category o abuse, but note that
a demotion rom, say, 1 to 58,610 will hae the same eect. No one eer clicks through
5861 pages o results. 1hus, in practice, any rule against exclusion would also need to come
with a-more problematic-rule against substantial demotions.
46
a. at 111-18.
4
Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 289-92. lis example, which
ocuses on Google`s scans o books or its Book Search project, is interesting, but is
uniersal access` only in a loose, metaphorical sense.
48
ee lrank Pasquale, Dovivavt earcb vgive.: .v ..evtiat Cvttvrat ava Potiticat acitit,, ivfra 258.
49
leron . Int`l Soc. or Krishna Consciousness, Inc., 452 U.S. 640, 655 ,1981, ,quoting
Koacs . Cooper, 336 U.S. , 8 ,1949,, ,emphasis added,.
Barron`s, |1|he point o ultimate interest is not the words o the speakers but
the minds o the hearers.`
50
\ith these purposes in mind, let us turn to actual
search-neutrality proposals.
Equality
Scott Cleland obseres that Google`s algorithm reportedly has oer 1,000
ariables,discrimination biases which decide which content gets suraced.`
51

le concludes that Google is not neutral` and thus should be subject to any
lCC network-neutrality regulation.
52
On this iew, a search engine does
something wrong i it treats websites dierently, surac|ing|` some, rather than
others. 1his is a theory o neutrality as equality, it comes rom the network-
neutrality debates, and it is nonsensical as applied to search.
Lquality has a long pedigree in telecommunications. lor years, common-carrier
regulations required the A1&1 system to oer its serices on equal terms to
anyone who wanted a phone.
53
1his kind o equality is at the heart o proposed
network neutrality regulations: treating all packets identically once they arrie at
an ISP`s router, regardless o source or contents.
54
\hether or not equality in
packet routing is a good idea as a technical matter, the rule itsel is simple
enough and relatiely clear. One can, without diiculty, identiy Comcast`s
orging o packets to terminate Bit1orrent connections as a iolation o the
principle.
55
As long as an ISP isn`t oerloaded to the point o losing too many
packets, equality does what it`s supposed to: ensures that eery website enjoys
access to the ISP`s network and customers.
1ry to apply this orm o equality to search and the results are absurd. O
course Google dierentiates among sites-that`s why we use it. Systematically
aoring certain types o content oer others isn`t a deect or a search engine-
it`s the oivt.
56
I I search or Machu Picchu pictures,` I want to see llamas in a

50
Barron, .cce.. to tbe Pre.., .vra note 39, at 1653.
51
Scott Cleland, !b, Coogte . ôt êvtrat, PRLCURSOR BLOG ,No. 4, 2009,,
;11>?-->A(*$A%+A#@+/"*+,-*+'1('1-!;DL/++/@(L&%L'+1L'($1A.@.
52
a.
53
ee geveratt, JONA1lAN L. NLUCl1LRLLIN & PlILIP J. \LISLR, DIGI1AL CROSSROADS 45-68
,2005,.
54
lor an accessible introduction to the technical issues, see Ldward \. lelten, 1be ^vt. ava
ott. of êtror/ êvtratit, ,2006,, ;11>?--&1>+@&*D">A&'*(1+'"(F$->$#-'($1A.@&1D">FC.
55
ee v re orvat Covt. of ree Pre.. c Pvbtic Kvorteage .gaiv.t Covca.t Cor. for ecrett, Degraaivg
PeertoPeer .ticatiov., \C Docket No. 0-52, Order, 23 l.C.C. Rcd. 13,028, 13,029-32
,discussing blocking,, 13,050-58 ,inding that blocking iolated ederal policy, ,2008,, racatea,
Comcast . lCC, 600 l.3d 642 ,D.C. Cir. 2010,.
56
ee Karl Bode, Coogte Migbt to 1iotativg earcb êvtratit,f .v,boa, Kver !bat 1bat
.ctvatt, Meavt, 1LClDIR1 ,May , 2010,,
ruined city on a cloud-orest mountaintop, not horny housewies who whiten
your teeth while you wait or them to reinance your mortgage. Search
ineitably requires some orm o editorial control.
5
A search engine cannot
possibly treat all websites equally, not without turning into the phone book. But
or that matter, een the phone book is not neutral in the sense o giing ully
equal access to all comers, as the prolieration o AAA Locksmiths and Aabco
Plumbers attests. Dierentiating among websites, without something more, is
not wrongul.
Objectivity
I search engines must make distinctions, perhaps we should insist that they
make correct distinctions. loundem, or example, argues that the Google
penalty was unair by pointing to positie write-ups o loundem rom the
UK`s leading technology teleision programme` and the UK`s leading
consumer body,` and to its high search ranks on \ahoo! and Bing.
58
1he
unoiced assumption here is that search queries can hae objectiely right and
wrong answers. A search on James Grimmelmann blog` should come back
with my weblog at ;11>?--@.#+A.1+A&$,"'(1, anything else is a wrong answer.
But this iew o what search is and does is wrong. A search or apple` could
be looking or inormation about liji apples, Apple computers, or liona Apple.
bbs` could reer to airgun pellets, bulletin-board systems, or bed-and-
breakasts. Dierent people will hae dierent intentions in mind, een the
same person will hae dierent intentions at dierent times. Sergey Brin`s
theological comparison o perect search to the mind o God`
59
shows us why
perect search is impossible. Not een Google is-or eer could be-
omniscient. 1he search query itsel is necessarily an incomplete basis on which
to guess at possible results.
60

1he objectie iew o search, then, ails or two related reasons. lirst, search
users are prooundly dierse. 1hey hae highly personal, highly contextual
goals. One size cannot it all. And second, a search engine`s job always
inoles guesswork.
61
Some guesses are better than others, but the search

;11>?--!!!"1(*;F&A1"*+,-.A1&*@(%-52722826-7B56593B22"%;1,@ ,|1|he entire purpose
o search is to discriminate and point the user toward more pertinent results.`,.
5
ee Goldman, earcb vgive ia., .vra note 28, at 115-18.
58
ovvaev`. Coogte tor,, .vra note .
59
vra note 1.
60
ee geveratt, ALLX lALAVAIS, 1lL SLARCl-LNGINL SOCIL1\ 32-55 ,2009, ,discussing
diiculties o ascertaining meaning in search process,.
61
ee Lric Goldman, Deregvtativg Reteravc, iv vtervet 1raaevar/ ar, 54 LMOR\ L.J. 50, 521-28
,2005,.
engine will always hae to guess. James Grimmelmann blog` shouldn`t take
users to 1oyota`s corporate page-but perhaps they were interested in my
guest-blogging at Covcvrrivg Oiviov., or in blogs about me, or they hae me
mixed up with Lric Goldman and were actually looking or bi. blog. 1ime
\arner Cable`s complaint that signiicant components o |Google`s| Ad Rank
scheme are subjectie`
62
is beside the point. earcb it.etf i. .vb;ectire.
63

lew scholars go so ar as to adocate explicit re-ranking to correct search
results.
64
But een those who acknowledge that search is subjectie sometimes
write as though it were not. lrank Pasquale gies a hypothetical in which
\ou1ube`s results always appear as the irst thirty |Google| results in response
to certain ideo queries or which |a rial ideo site| has demonstrably more
releant content.`
65
One might ask, demonstrably more releant` b, rbat
.tavaara Oten the answer will be contentious.
In loundem`s case, what dierence should it make that \ahoo! and others liked
loundem So 1hat`s their opinion. Google had a dierent one. \ho is to
say that \ahoo! was right and Google was wrong
66
One could equally well
argue that Google`s low ranking was correct and \ahoo!`s high ranking was the
mistake. compare prices shoei xr-1000` is not the sort o question that admits

62
Comments o 1ime \arner Cable Inc. , In the Matter o Presering the Open Internet
Broadband Industry Practices, GN Docket No. 09-191 ,l.C.C. comments iled Jan. 14,
2010,.
63
ee Goldman, earcb vgive ia., supra note 28, at 112-13. 1his point should not be
conused with a considered opinion on the question o how the lirst Amendment applies to
search-ranking decisions. Search engines make editorial judgments about releance, but they
also present inormation that can only be described as actual ,such as maps and addresses,,
extol their objectiity in marketing statements, and are perceied by users as haing an aura
o reliability. It is possible to make alse statements een when speaking subjectiely-or
example, I would be lying to you i I said that I enjoy eating scallops. 1he act that search
engines` judgments are expressed algorithmically, including in ways not contemplated by
their programmers, complicates the analysis een urther. 1he deinitie lirst Amendment
analysis o search-engine speech has yet to be written. Academic contributions to that
conersation include Goldman, earcb vgive ia., .vra note 28, at 112-15, Bracha &
Pasquale, eaerat earcb Covvi..iov, .vra note 20, at 1188-1201, Pasquale, ..teri./ Reri.itea,
.vra note 32, at 68-85, NUNZIA1O, VIR1UAL lRLLDOM, .vra note 20, a..iv ,and
particularly pages 149-51,, Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1124-29,
James Grimmelmann, 1be Coogte Ditevva, 53 N.\.L.S. L. RLV. 939, 946 ,2009,,
Grimmelmann, 1be trvctvre of earcb vgive ar, .vra note 4, at 58-60. Some leading cases
are listed in note 85, ivfra.
64
vt .ee Sandeep Pandey et al., bvfftivg a tac/ea Dec/: 1be Ca.e for Partiatt, Ravaoviea earcb
Re.vtt., PROC. 31S1 VLR\ LARGL DA1ABASLS CONl. 81 ,2005, ,arguing or randomization in
search results to promote obscure websites,.
65
Pasquale, vtervet ôvai.crivivaiov Privcite., .vra note 18, at 296.
66
Cf. Rebecca 1ushnet, t Deeva. ov !bat tbe Meavivg of at.e ., at.it, ava Mi.teaaivgve.. iv
Covverciat eecb Doctrive, 41 LO\OLA L.A. L. RLV. 101 ,2008, ,arguing that judgments about
alsity requently embody contested social policies,.
o a right answer. 1his is why it doesn`t help to say that the loundem ote is
our-to-one against Google. I deiation rom the majority opinion makes a
search engine wrong, then so much or search engine innoation-and so much
or unpopular iews.
6

Bias
Ironically, it is the goal o protecting unpopular iews that dries the concern
with search engine bias.` Lucas Introna and lelen Nissenbaum, or example,
are concerned that search engines will direct users to sites that are already
popular and away rom obscure sites.
68
Alex lalaais calls or resistance to
the homogenizing process o major search engines,`
69
including goernmental
interentions.
0
1hese are structural concerns with popularity-based search.
Others worry about more particular biases. A1&1 complains that Google`s
algorithms unquestionably ao aor some companies or sites.`
1
Scott Cleland
objects that Google demotes content rom other countries in its country-
speciic search pages.
2

1he point that a technological system can display bias is one o those proound
obserations that is at once both startling and obious.
3
It naturally leads to
the question o whether, when, and how one could correct or the bias search
engines introduce.
4
But to pull that o, one must hae a working
understanding o what constitutes search-engine bias. Batya lriedman and
lelen Nissenbaum deine a computer system to be biased` i it .,.tevaticatt,
and vvfairt, discriminates against certain indiiduals or groups o indiiduals in
aor o others.`
5
Since search engines systematically discriminate by design,

6
1his last point should be especially troubling to Barron-inspired adocates o access,` since
the point o such a regime is to rovote opinions that are not widely shared.
68
Lucas D. Introna & lelen Nissenbaum, baivg tbe !eb: !b, tbe Potitic. of earcb vgive.
Matter., 16 INlO. SOC. 169, 15 ,2000,.
69
lALAVAIS, SLARCl LNGINL SOCIL1\, .vra note 60, at 106.
0
a. at 132-38.
1
Comments o A1&1 Inc. 102, In the Matter o Presering the Open Internet Broadband
Industry Practices, GN Docket No. 09-191 ,l.C.C. comments iled Jan. 14, 2010,.
2
Cleland, !b, Coogte . ôt êvtrat, .vra note 51.
3
In Landgon \inner`s phrase, artiacts hae politics.` LANGDON \INNLR, 1lL \lALL
AND 1lL RLAC1OR: A SLARCl lOR LIMI1S IN AN AGL Ol lIGl 1LClNOLOG\19 ,Uniersity
o Chicago Press 1986,.
4
ee, e.g., Pandey et al, bvfftivg a tac/ea Dec/, .vra note 64.
5
Batya lriedman & lelen Nissenbaum, ia. iv Covvter ,.tev., 14 ACM 1RANS. ON
COMPU1LR S\S. 330, 332 ,1996,. ee at.o Alejandro M. Diaz, 1hrough the Google Goggles:
Sociopolitical Bias in Search Lngine Design ,May 23, 2005, ,unpublished B.A. thesis,
Stanord Uniersity,, araitabte at
;11>?--(>@"%*$"(F$-l%1%O.@$(%-A(.F&'/%-N&.041;(%&%4C&'.@">FC.
all o the heay liting in the deinition is done by the word unair.` But this
just kicks the problem down the road. One still must explain when
discrimination is unair` and when it is not. lriedman and Nissenbaum`s
discussion is enlightening, but does not by itsel help us identiy which practices
are abusie.
6

1he point that socio-technical systems hae embedded biases also cuts against
search neutrality. \e should not assume that i only the search evgive could be
made properly neutral, the search re.vtt. would be ree o bias. Lery search
result requires both a user to contribute a search query, and websites to
contribute the content to be ranked. Neither users nor websites are passie
participants, both can be wildly, prooundly biased.
On the website side, the web is anything but neutral.
\ebsites compete
iercely, and not always ethically, or readers.
8
It doesn`t matter rbat the search
engine algorithm is, websites will try to game it. Search-engine optimization, or
SLO, is as much a ixture o the Internet as spam. Link arms,
9
spam blog
comments, hacked websites-you name it, and they`ll try it, all in the name o
improing their search rankings. A ully inisible search engine, one that
introduced no new alues or biases o its own, would merely replicate the
underlying biases o the web itsel:
80
heaily commercial, and subject to a truly
mindboggling quantity o spam. Ra says that search algorithms should be
comprehensie.`
81
But should users be subjected to a comprehensie
presentation o discount Canadian pharmaceutical sites
On the user side, sometimes the bias is between the keyboard and the chair.
lully de-biasing search results would also require de-biasing search queries-
and users` ability to pick which results they click on. 1ake a search or jew,`
or example. Google has been criticized both or returning anti-Semitic sites ,to

6
I one ears, with Bracha and Pasquale, that a handul o powerul gatekeepers` wield
disproportionate inluence, then the solution is simple: break up the bastards. I they
reassemble or reacquire too much power, do it again. Neutrality will always be an imperect
hal-measure i power itsel is the problem.
ee Clay Shirky, Porer ar., !ebtog., ava veqvatit,, SlIRK\.COM ,leb. 8, 2003,,
;11>?--!!!"%;&A)D"*+,-!A&1&'/%->+!(A@.!4!(#@+/";1,@ ,discussing ast
disproportion o prominence between amous and obscure weblogs,.
8
ee IAN l. \I11LN L1 AL., \LB DRAGONS: INSIDL 1lL M\1lS Ol SLARCl 1LClNOLOG\
145-5 ,Morgan Kaumann Publishers 200,.
9
A link arm is a group o automatically generated web sites that heaily link to each other.
1he point is to trick a popularity-based search engine into belieing that all o the sites in the
group are popular. ee Grimmelmann, 1be Coogte Ditevva, .vra note 63, at 946,
80
ee Patterson, ôvêtror/ arrier., .vra note 19, at 2854-55.
81
Ra, earcb, vt Yov Ma, ôt iva, .vra note 11.
American users, and or vot returning such sites ,to German users,.
82
1he
inescapable issue is that Google has users who want to read anti-Semitic web
pages and users who don`t. One might call some o those users biased,` but i
they are, it`s not Google`s ault.
Some bias is going to leak through as long as search engines help users ind
what they want. And helping users ind what they want is such a proound
social good that one should be skeptical o trying to inhibit it.
83
1elling users
what they .bovta see is a serious intrusion on personal autonomy, and thus
deeply inconsistent with the liberal argument or search neutrality. I you want
Google to steer users to websites with iews that dier rom their own,
84
your
goal is not properly described as search vevtratit,. In eect, you hae gone back
to asserting the objectie correctness o search results: Certain sites are good or
users, like whole grains.
Traffic
1he most common trope in the search debates is the website whose traic
anishes oernight when it disappears rom Google`s search results.
85
Because
so much traic lows through Google, it holds websites oer the lames o
website hell, ready at any instant to let them all in the rankings. Chandler`s
proposed right to reach an audience and loundem`s proposed eectie,
accessible, and transparent appeal process`
86
attempt to protect websites rom

82
ee Grimmelmann, 1be Coogte Ditevva, .vra note 63, at 943-45.
83
ee James Grimmelmann, Dov`t Cev.or earcb, 11 \ALL L.J. POCKL1 PAR1 48 ,200,.
84
ee geveratt, CASS SUNS1LIN, RLPUBLIC.COM 2.0 ,Princeton Uniersity Press 200,.
85
ee, e.g., BA11LLLL, 1lL SLARCl, .vra note 5, at 153-59 ,2bigeet.com, main index,,
NUNZIA1O, VIR1UAL lRLLDOM, .vra note 20, at 14-1 ,arious sites, Ad\ords and Google
News,, Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1110 ,BM\ Germany and
Ricoh Germany, main index,, Michael \. Park, ]ovrvati.t !bo o.e. |.^. Corrvtiov
Di.aear. frov Coogte, lOX NL\S, leb. 18, 2008,
;11>?--!!!"C+P'(!%"*+,-%1+AD-2m53BBmBB772:m22";1,@ ,Inner City Press, Google
News,, Cleland, !b, Coogte . ôt êvtrat, .vra note 51 ,Lxtreme1ech.com and
lotolog.com, Ad\ords,, Dan Mcsai, CRaitea: !b, Dia Coogte vr, tbe !eb`. Otae.t
vtertaivvevt Pvbticatiov., lAS1COMPAN\.COM ,Dec. 2, 2009,,
;11>?--!!!"C.%1*+,>.'D"*+,-#@+/-F.'L,.*%.&->+>!&%(-!;DLF&FL'($1A.@L/++/@(L
F(L@&%1L!(#%L+@F(%1L('1(A1.&',('1L>$#@&*.1&+' ,Studio Brieing, Ad\ords and main
index,, ovvaev`. Coogte tor,, .vra note ,loundem, main index,. Opinions in lawsuits
challenging demotions or exclusions include Langdon . Google Inc., 44 l. Supp. 622 ,D.
Del. 200, ,NCJusticelraud.com and ChinaIsLil.com, Ad\ords,, Kinderstart.com LLC .
Google, Inc., No. C 06-205 Jl ,RS,, 2006 U.S. Dist. LLXIS 82481 ,N.D. Cal. July 13, 2006,
,Kinderstart.com, main index, and Search King, Inc. . Google 1ech., Inc., No. CIV-02-
145-M, 2003 U.S. Dist. LLXIS 2193 ,\.D. Okla. 2003, ,SearcchKing.com, main index,.
86
being dropped. Dawn Nunzatio, or her part, would require search engines to
open their sponsored links to political candidates.
8

A right to continued customer traic would be a legal anomaly, oline
businesses enjoy no such right. Some Manhattanites who take the ree IKLA
erry to its store in Brooklyn eat at the nearby ood trucks in the Red look Ball
lields.
88
1he ood truck owners would hae no right to complain i IKLA
discontinued the erry or moed its store. Search neutrality adocates, howeer,
would say that Redlooklood1ruck.com has a Jerome Barron-style ree-speech
interest in haing access to the search engine`s result pages, and thus has more
right to complain i the Google erry no longer comes to its neighborhood.
89

But, as we saw aboe, this is really an argument that v.er. hae a reteravce interest
in seeing the site. I no one actually wants to isit Redlooklood1ruck.com,
then its owner shouldn`t be heard to complain about her poor search ranking.
\hen push comes to shoe, search neutrality adocates recognize that websites
must plead their case in terms o users` needs. Chandler`s modern right o
access is a right to reach a rittivg audience,`
90
which she describes as the right
to be ree o the imposition o discriminatory ilters tbat tbe ti.tever rovta vot
otberri.e bare v.ea.`
91
Len loundem`s Adam Ra presents his actual search-
neutrality principle in user-protectie terms: search engines should hae no
editorial policies other than that their results be comprehensie, impartial and
ba.ea .otet, ov reteravce.`
92
Releance is, o course, the touchstone o users`
interests, not websites`.
Indeed, looking at the rankings rom a website`s perspectie, rather than rom
users`, can be counterproductie to ree-speech alues. I users really ind other
websites more releant, then making them isit Redlooklood1ruck.com
impinges on their autonomy and on tbeir ree speech interests as listeners. lor
any gien search query, there may be dozens, hundreds, thousands o
competing websites. 1he ra.t va;orit, o them will thus hae interests that
dierge rom users`-and eery incentie to oerride users` wishes.

8
NUNZIA1O, VIR1UAL lRLLDOM, .vra note 20, at 150-51.
88
ee Adam Kuban, Rea oo/ 1evaor.: . Qvic/ Cviae for tbe |vivitiatea, SLRIOUS LA1S ,July 18,
2008,, ;11>?--'(!D+A)"%(A&+$%(.1%"*+,-522H-29-A(FL;++)LO('F+A%L%+**(AL1.*+%L
/$&F(L;+!L1+L/(1L1;(A(L!;.1L1+L(.1";1,@.
89
ee Chandler, Rigbt to Reacb av .vaievce, .vra note 20, NUNZIA1O, VIR1UAL lRLLDOM, .vra
note 20.
90
Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1099 ,emphasis added,.
91
a. at 1103 ,emphasis added,.
92
Ra, earcb, vt Yov Ma, ôt iva, .vra note 11 ,emphasis added,.
Len when users are genuinely indierent among arious websites, some search
neutrality adocates think websites should be protected rom arbitrary` or
unaccountable` ranking changes as a matter o airness.
93
\e should call the
websites that currently sit at the top o search engine rankings by their proper
name-ivcvvbevt.-and we should look as skeptically on their demands to
remain in power as we would on any other incumbent`s. 1he search engine that
ranks a site highly has conerred a beneit on it, turning that gratuitous beneit
into a permanent entitlement gets the ethics o the situation exactly backwards.
Indeed, giing highly-ranked websites what is in eect a property right in search
rankings runs counter to eerything we know about how to hand out property
rights. \ebsites don`t create the rankings, search engines do. Similarly, search
engines are in a better position to manage rankings and preent waste. And i
each indiidual search ranking came with a right to placement, eery search-
results page would be an anti-commons in the making.
94

1hus, it is irreteravt that loundem had a prominent search placement on Google
beore it landed in the doghouse. Just as the subjectiity o search means that
search engines will requently disagree with each other, it also means that a
search engine will disagree with itsel oer time. lrom the outside looking in,
we hae no basis to say whether the initial high ranking or the subsequent low
ranking made more sense. 1o gie loundem-and eery other website
currently enjoying a good search ranking-the right to continue where it is
would lock in search results or all time, obliterating search-engine
experimentation and improement.
Relevance
Gien the importance o user autonomy to search-neutrality theory, releance is
a natural choice or a neutrality principle. In loundem`s words, search results
should be based solely on releance.`
95
Chandler proposes a rule against
discrimination that listeners would not hae chosen.`
96
Bracha and Pasquale
decry search engines |that| highlight or suppress critical inormation` and
thereby shape and constrain |users`| choices`-that is, hide inormation that
users would hae ound releant.
9

93
Bracha & Pasquale, eaerat earcb Covvi..iov, .vra note 20, at 115-6.
94
ee geveratt, Michael leller, 1be 1ragea, of tbe .vticovvov., 111 lARV. L. RLV. 621 ,1998,
,arguing that when too many owners hae exclusion rights oer a resource, it is prone to
underuse,.
95
earcb êvtratit,, SLARCl NLU1RALI1\.ORG ,Oct. 11, 2009,,
;11>?--!!!"%(.A*;'($1A.@&1D"+A/-.
96
Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 1098.
9
Bracha & Pasquale, eaerat earcb Covvi..iov, .vra note 20, at 11.
Releance, howeer, is such an obious good that its irtue erges on the
tautological. Search engines covete to gie users releant results, they exist at all
only because they do. 1elling a search engine to be more releant is like telling
a boxer to punch harder. O course, sometimes boxers do throw ights, so it
isn`t out o the question that a search engine might underplay its hand. low,
though, could regulators tell Regulators can`t declare a result releant`
without expressing a iew as to why other possibilities are irreleant,` and that
is almost always going to be contested.
lere`s an example: loundem. Recall that loundem is a ertical search site`
that specializes in consumer goods. \ell, a great many ertical search sites are
worthless. ,I you don`t beliee me, please try using a ew or a bit., Like other
kinds o sites that simply roll up existing content and slap some o their own
ads on it-\ikipedia clones and local business directories also come to mind-
they supericially resemble legitimate sites that proide something o alue to
users.
98
But only supericially. 1he penalties` that reduce ertical search sites`
Google ranks aren`t an attempt to reduce competition at the expense o
releance, they`re an attempt to ivtevevt releance.
99
1here are a ew relatiely
good, usable product-search sites, but most o them are junk and good riddance
to them. \ou`re welcome to disagree-search is subjectie-but I`d rather hae
the anti-ertical penalty in place than not. 1hose who would argue that
Google`s rankings don`t relect releance hae a heay burden o proo, in the
ace o ample, easily eriied eidence to the contrary.
In act, behind almost eery well-known story o search engine caprice, there is
a more persuasie releance-enhancing counter-story. lor example,
Source1ool, another ertical search engine, has sued Google under antitrust law
or, in eect, demoting it in Google`s rankings or search ads.
100
Source1ool,
though, is a directory` with a taxonomic logic o dubious utility-the United
Nations Standard Products and Serices Code-and almost no content o its
own. It`s the rare user indeed who will ind Source1ool releant. I you care
about releance and user autonomy, you should applaud Google`s decision to
demote Source1ool.

98
ee Chris Lake, ovvaev r. Coogte: . Ca.e tva, iv O ait, LCONSUL1ANC\ ,Aug. 18, 2009,,
;11>?--(*+'%$@1.'*D"*+,-#@+/-668:LC+$'F(,LO%L/++/@(L.L*.%(L%1$FDL&'L%(+LC.&@,
ittte or ô Origivat Covtevt, GOOGLL \LBMAS1LR CLN1RAL ,updated June 10, 2009,,
www.google.com,support,webmasters,bin,answer.pyanswer~66361.
99
ee John Lettice, !bev .tgoritbv. .ttac/, Doe. Coogte ear Yov creav., 1lL RLGIS1LR ,No.
19, 2009,, ;11>?--!!!"1;(A(/&%1(A"*+"$)-5223-77-73-/++/@(4;.'F4+C4/+F-.
100
ee 1radeComet.com LLC . Google Inc., No. 09-CIV-1400 ,S.D.N.\. complaint iled leb.
1, 2009,. 1he District Court dismissed the case on the basis o the orum-selection clause
in Google`s adertiser agreement, without reaching the merits o the case. ee
1radeComet.com LLC . Google Inc., 693 l. Supp. 2d 30 ,S.D.N.\. 2010,.
Self-Interest
In practice, een as search-neutrality adocates claim releance` as their goal,
they rely on proxies or it. 1he most common is sel-interest. A Consumer
\atchdog report accuses Google o an abandonment o |its| pledge to proide
neutral search capability` by steering Internet searchers to its own serices` to
muscle its way into new markets.`
101
loundem alleges that Google demotes it
and other ertical search sites to end o competition, and alleges that Google`s
links to itsel gie it an unassailable competitie adantage.`
102
Bracha and
Pasquale worry that search engines can change their rankings in response to
positie or negatie inducements rom other parties.`
103

Bad motie may lead to bad releance, but it`s also a bad proxy or it. 1he irst
problem is eidentiary. By deinition, motiations are interior, personal.
104
O
course, the law has to guess at moties all the time, but the task is by its nature
harder than looking to extrinsic eidence. People get it wrong all the time. In
2009, an Amazon employee with a at inger hit a wrong button and categorized
tens o thousands o gay-themed books as adult.`
105
An angry mob o
Netizens assumed the company had deliberately pulled the books rom its
search engine out o anti-gay animus, and used the 1witter hashtag amazonail
to express their ery public outrage.
106
Amazon`s reclassiication was a mistake
,a quickly corrected one,, and a iid demonstration o the power o search
algorithms-but not a case o bad moties.
10

In all but the most blatant o cases, in act, a search engine will be able to tell a
plausible releance story about its ranking decisions.Proing that a releance
story is pretextual will be extraordinarily diicult, in iew o the complexity and
subjectiity o search. But it would also be disastrous to adopt the opposite
point o iew and presume pretext. 1he absence o bad motie is a negatie
that it will oten be impossible or the search engine to proe. low can it

101
1RAllIC RLPOR1: lO\ GOOGLL IS SQULLZING OU1 COMPL1I1ORS AND MUSCLING IN1O
NL\ MARKL1S ,Consumer \atchdog 2010,,
;11>?--!!!"*+'%$,(A!.1*;F+/"+A/-A(%+$A*(%-JA.CC&*h1$FDLM++/@(">FC.
102
Ret, Covvevt. of ovvaev, .vra note 10, at 1.
103
104
As an artiicial corporate entity, a search engine may not een hae moties other than the
ones the law attributes to it.
105
ee Nick Laton, .vaovait: .v v.iae oo/ at !bat aevea, AMAZON & 1lL ONLINL
RL1AIL BLOG ,Apr. 13, 2009,, ;11>?--#@+/"%(.11@(>&"*+,-.,.0+'-.A*;&O(%-7::BH6".%>.
106
ee Clay Shirky, 1be aitvre of =avaovfait, SlIRK\.COM ,Apr. 15, 2009,,
;11>?--!!!"%;&A)D"*+,-!(#@+/-5223-26-1;(LC.&@$A(L+CL.,.0+'C.&@-.
10
vt .ee Mary lodder, !b, .vaov Diav`t ]v.t are a Ctitcb, 1LClCRUNCl ,Apr. 14, 2009,,
;11>?--1(*;*A$'*;"*+,-5223-26-76-/$(%1L>+%1L!;DL.,.0+'LF&F'1LR$%1L;.O(L.L
/@&1*;-.
establish, or example, that the engineer who added the anti-ertical penalty
didn`t hae a lunchroom conersation with an executie who played up the
competition angle 1his is not to say that serious cases o abuse are
implausible,
108
just that inestigation will be unusually hard and that alse
posities will be dangerously requent.
1here i. a nontriial antitrust issue lurking here. In the United States, Google
has a dominant market share in both search and search adertising, and one
could argue that Google has started to leerage its position in anticompetitie
ways.
109
Antitrust, howeer approaches such questions with a well-deeloped
analytical toolkit: releant markets, market power, pro-competitie and anti-
competitie eects, and so on.
110
Antitrust rightly ocuses on the eects o
business practices on consumers, search neutrality should not short-circuit that
consumer-centric analysis by oeremphasizing the role o a search engine`s
moties. Some things can be good or Google ava good or its users.
1hus, when Google links to its own products, not only can there be substantial
technical beneits rom integration, but oten Google is helping users by
pointing them to serices that really are better than the competition. Consumer
\atchdog, or example, cries oul that Google put its own |map| serice atop
all others or generic address searches,`
111
and that Google Maps has taken hal
o the local search market at the expense o preiously dominant MapQuest and
\ahoo! Maps.
112
But perhaps MapQuest and \ahoo! Maps desered to lose.
Google Maps was groundbreaking when launched, and years later, it remains
one o the best-implemented serices on the Internet, with astonishingly cleer
scripting, lexible route-inding, and a powerul application programming
interace ,API,.
113

108
Baidu`s alleged shakedown ,.ee .vra note 24 and accompanying text,, i true, would be an
example. \illingness to buy Baidu search ads is not in itsel a reliable indicator o releance
to Baidu searchers. But then again, een pay-or-placement was once considered a plausible
model or main-column search results-and willingness to pay is not inherently a crazy
proxy or releance. ee BA11LLLL, 1lL SLARCl, .vra note 5, at 104-14 ,discussing Go1o`s
pay-or-placement model,. ee at.o Goldman, Coa.eav .vat,.i., .vra note 30 ,enisioning a
uture in which adertisers and users negotiate oer access to users` attention,. Indeed,
search ads today are sold on an auction-based basis. 1hey`re oten as releant as main-
column search results, sometimes more so. It might be better to say that Baidu`s real
problems are monopoly pricing and ,compulsory, stealth marketing.
109
ee, e.g., Brad Stone, vre, t`. ig, vt . 1bat aa., N.\. 1IMLS, May 21, 2010, at BU1.
110
See generally Georey A. Manne & Joshua D. \right, Coogte ava tbe ivit. of .vtitrv.t: 1be
Ca.e .gaiv.t tbe .vtitrv.t Ca.e .gaiv.t Coogte, lARV. J. L. & PUB. POL`\. ,orthcoming,.
111
1RAllIC RLPOR1, .vra note 101, at 5.
112
a. at 5-.
113
ee, e.g., John Carroll, Coogte Ma. ava vvoratiov, A DLVLLOPLR`S VIL\ ,Oct. 12, 2005,,
;11>?--!!!"0F'(1"*+,-#@+/-*.AA+@@-/++/@(L,.>%L.'FL&''+O.1&+'-7633.
One orm o sel-interest that may be well-enough deined to justiy regulatory
scrutiny is the straightorward bribe: a payment rom a website to change its
ranking, or a competitor`s. Search-engine critics argue that search engines
should disclose commercial relationships that bear on their ranking decisions.
114

1his is a standard, sensible policy response to the ear o stealth marketing.
115

Indeed, the lederal 1rade Commission ,l1C, has speciically warned search
engines not to mix their organic and paid search results.
116
More generally, the
l1C endorsement guidelines proide that endorsements must relect the
honest opinions, indings, belies, or experience o the endorser`
11
and that any
connections between endorser and seller that might materially aect the weight
or credibility o the endorsement`
118
must be ully disclosed. 1hese policies
hae a natural application to search engines. A search engine that actors
payments rom sponsors into its ranking decisions is lying to its users unless it
discloses those relationships, and this sort o lie would trigger the l1C`s
jurisdiction.
119
1his isn`t a neutrality principle, or een unique to search, it`s just
a natural application o a well-established legal norm.
Transparency
Search-engine critics generally go urther and argue that search engines should
also be required to disclose their atgoritbv. in detail:
Introna and Nissenbaum: As a irst step we would demand ull and
truthul disclosure o the underlying rules ,or algorithms, goerning
indexing, searching, and prioritizing, stated in a way that is meaningul
to the majority o web users.`
120

loundem: Search Neutrality can be deined as the principle that
search engines should be open and transparent about their editorial
policies . .`
121

114
ee, e.g., Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 286.
115
See generally Lllen P. Goodman, Stealth Marketing and Lditorial Integrity, 85 1LX. L. RLV.
83 ,2006,.
116
ee Letter rom leather lippsley, Acting Assoc. Dir., Di. o Ader. Practices, led. 1rade
Comm., to Gary Ruskin, Lxecutie Dir. at Commercial Alert ,June 2, 200,, aailable at
;11>?--!!!"C1*"/+O-+%-*@+%&'/%-%1.CC-*+,,(A*&.@.@(A1@(11(A"%;1,.
11
16 ClR 255.1,a,.
118
16 ClR 255.5.
119
Disclosure in common cases need not be onerous. \here, or example, a search engine
auctions o sponsored links on its results pages, telling users that those links are auctioned
o should generally suice. ee geveratt, Letter rom leather lippsley, .vra note 116.
120
Introna and Nissenbaum, .vra note 68, at 181.
121
earcb êvtratit,, SLARClNLU1RALI1\.ORG ,Oct. 11, 2009,,
;11>?--!!!"%(.A*;'($1A.@&1D"+A/-%(.A*;L'($1A.@&1D.
Pasquale: |Dominant search engines| should submit to regulation that
bans stealth marketing ava retiabt, rerifie. tbe ab.evce of tbe ractice.`
122

1hese disclosures are meant to inorm users about what they`re getting rom a
search engine ,Introna and Nissenbaum,, to inorm websites about the
standards they`re being judged by ,loundem,,
123
or to inorm regulators about
what the search engine is actually doing ,Pasquale,.
124

Algorithmic transparency is a delicate business. lull disclosure o the algorithm
itsel runs up against critical interests o the search engine. A ully public
algorithm is one that the search engine`s competitors can copy wholesale.
125

\orse, it is one that websites can use to create highly optimized search-engine
spam.
126
\riting in 2000, long beore the ull extent o search-engine spam was
as clear as it is today, Introna and Nissenbaum thought that the impact o
these unethical practices would be seerely dampened i both seekers and those
wishing to be ound were aware o the particular biases inherent in any gien
search engine.`
12
1hat underestimates the scale o the problem. Imagine
instead your inbox without a spam ilter. \ou would doubtless be aware o the
particular biases` o the people trying to sell you ancy watches and penis
pills-but that will do you little good i your inbox contains a thousand pieces
o spam or eery email you want to read. 1hat is what will happen to search
results i search algorithms are ully public, the spammers will win.
lor this reason, search-neutrality adocates now acknowledge the danger o
SLO and thus propose only limited transparency.
128
Pasquale suggests, or
example, that Google could respond to a question about its rankings with a list
o a ew actors that principally aected a particular result.
129
But search is
immensely complicated-so complicated that it may not be possible to boil a
ranking down to a simple explanation. \hen the law demands disclosure o
complex matters in simple terms, we get pro orma statements and boilerplate.

122
Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 299 ,emphasis added,.
123
ee at.o ia. at 285 ,arguing that search engines beneit rom hidden algorithms because
websites, lacking clear inormation about how to achiee high organic search rankings, must
resort to buying paid search ads,.
124
ee at.o 1be Coogte .tgoritbv, .vra note 13 ,recommending required disclosure,.
125
ee Grimmelmann, trvctvre of earcb vgive ar, .vra note 4, at 49, 55.
126
a. at 44-46, 56.
12
Introna and Nissenbaum, .vra note 68, at 181.
128
ee Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 29, Bracha & Pasquale,
eaerat earcb Covvi..iov, .vra note 20, at 1201-02, Chandler, Rigbt to Reacb av .vaievce, .vra
note 20, at 111.
129
Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 296-9.
Consumer credit disclosures and securities prospectuses hae brought
important inormation into the open, but they haen`t done much to aid the
understanding o their aerage recipient.
Google`s algorithm depends on more than 200 dierent actors.
130
Google
makes about 500 changes to it a year,
131
based on ten times as many
experiments.
132
One sixth o the hundreds o millions o queries the algorithm
handles daily are queries it has neer seen beore.
133
1he PageRank o any
webpage depends, in part, on eery other page on the Internet.
134
And een
with all the computational power Google can muster, a ull PageRank
recomputation takes weeks.
135
PageRank is, as algorithms go, elegantly
simple-but I certainly wouldn`t want to hae the job o making Marko chains
and eigenectors meaningul to the majority o \eb users.`
136
In practice, any
simpliied disclosure is likely to leae room or the search engine to bury plenty
o bodies.
Some scholars hae suggested that concerns about transparency could be
handled through regulatory opacity: 1he search engine discloses its algorithm to
the goernment, which then keeps the details rom the public.
13
1his is a
promising way o dealing with search engines` operational needs or secrecy, but
it sharpens the question o regulators` technical competence. I the record is
sealed, they won`t hae third-party experts and interested amici to walk them
through noel technical issues. Lerything will hinge on their own ability to
ealuate the implications o small details in search algorithms. 1he track record
o agencies and courts in dealing with other digital technologies does not
proide grounds or optimism on this score.
138
Pasquale makes an important

130
ee, e.g., 1ecbvotog, Orerrier, GOOGLL, ;11>?--!!!"/++/@("*+,-*+A>+A.1(-1(*;";1,@.
131
ee Steen Ley, v.iae tbe o, \IRLD, Mar. 2010, at 96.
132
ee Rob lo, Coogte`. |ai Mavber: earcb . .bovt Peote, ôt ]v.t Data, 1lL 1LCl BLA1 ,Oct.
1, 2009,,
;11>?--!!!"#$%&'(%%!(()"*+,-1;(41;A(.F-1(*;#(.1-.A*;&O(%-5223-72-/++/@(%4$F
&4,.'#(A4%(.A*;4&%4.#+$14>(+>@(4'+14R$%14F.1.";1,@.
133
a.
134
ee AM\ N. LANGVILLL & CARL D. ML\LR, GOOGLL`S PAGLRANK AND BL\OND: 1lL
SCILNCL Ol SLARCl ,Princeton Uniersity Press 2006,.
135
a.
136
In 1be Coogte Ditevva, .vra note 63, I didn`t een try to explain the math to law proessors.
13
Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18 at 29-98, Bracha & Pasquale,
eaerat earcb Covvi..iov, .vra note 20, at 1294-96. ee geveratt, Via R. Moat, Regvtativg
earcb, 22 lARV. J. L. & 1LCl. 45 ,2009, ,discussing institutional choice issues in search
regulation,.
138
vt .ee lrank Pasquale, 1rv.tivg ;ava 1erif,ivg) Ovtive vterveaiarie.` Poticivg, .vra at 258
,proposing Internet Intermediary Regulatory Council` and arguing that it could deelop
point that it is essential that .oveove has the power to look under the hood,``
139

but it is also important that algorithmic disclosure remain connected to a
workable theory o what regulators are looking or and what they would do i
they ound it.
Manipulation
Perhaps the most interesting idea in the entire search neutrality debate is the
manipulation` o search results. It`s a slippery term, and used inconsistently in
the search-engine debates-including by me.
140
In the dictionary sense o
process, organize, or operate on mentally or logically, to handle with mental or
intellectual skill,`
141
all search results are manipulated and the more skillully the
better. But in the dictionary sense o manage, control, or inluence in a subtle,
deious, or underhand manner,`
142
it`s a bad thing indeed: no one likes to be
manipulated.
143

In practice-although this is rarely made explicit-the concern is with what I
hae described elsewhere as hand manipulation.`
144
1his idea imagines the
search engine as haing both an automatic, general-purpose ranking algorithm
and a human-created list o exceptions. Consumer \atchdog, or example,
derides Google`s claim to rank results automatically by algorithms,` saying, It
is hard to see how this can still be true, gien the increasingly pronounced tilt
toward its own serices in Google`s search results.`
145
loundem calls it
manual interention,` special treatment,` and manual bias,` and documents
how Google`s public statements hae quietly backed away rom claims that its
rankings are objectie` and automatic.`
146

Put this way, the distinction between objectie algorithm and subjectie
manipulation is incoherent. Both kinds o decisions come rom the same

suicient technical expertise to generate oicial and een public understanding o |search
engines`| practices`,.
139
Pasquale, vtervet ôvai.crivivatiov Privcite., .vra note 18, at 286.
140
Covare Grimmelmann, trvctvre of earcb vgive ar, .vra note 4, at 44 ,technical arms
race between engines and manipulators`, ritb ia., at 59-60 ,hand manipulation o results
|by search engines|`,.
141
Oxord Lnglish Dictionary ,June 2010 drat,.
142
a.
143
ee Bracha & Pasquale, eaerat earcb Covvi..iov, .vra note 20, at 116-9 ,discussing eects
o manipulation on user autonomy,.
144
Grimmelmann, trvctvre of earcb vgive ar, .vra note 4, at 59.
145
1RAllIC RLPOR1, .vra note 101, at 8.
146
source: the search engine`s programmers.
14
Nor can the algorithm proide a
stable baseline against which to measure manipulation, since each
manipulation` is a change to the algorithm itsel. It`s not like Bing has rooms
ull o employees looking oer search results pages and making last-minute
tweaks beore the pages are deliered to users.
Academics, being more careul with concepts, hae ocused on intentionality:
does the search engine intend the promotions and demotions that will result
rom an algorithmic change Mark Patterson, or example, reers to
intentional manipulation o results.`
148
Bracha and Pasquale sharpen this idea
to speak o highly speciic or local manipulations,` such as singling out
websites or special treatment.
149
Chandler argues that search engines should
not manipulate ivairiavat search results except to address instances o suspected
abuse.`
150
Google itsel is remarkably coy about whether and when it changes
rankings on an indiidual basis.
151

Surprisingly, no one has explained why special-casing in and o itsel is a
problem. One possibility is that it captures the distinction between indiidual
adjudication and general rulemaking: changes that only aect a ew websites
trigger a kind o due process interest in indiidualized procedural protections.
152

1here is also a kind o Rawlsian argument
153
here, that algorithmic decisions
should be made rom behind a eil o ignorance, not knowing which websites
they will aor. lor whateer reason, local manipulations make people nerous,
nerous enough that most o the stories told to instill ear o search engines
inole what is or looks like manipulation.
154

Local manipulation, howeer, is a distraction. 1he real goal is releance. lrom
that point o iew, most local manipulations aren`t wrongul at all. loundem
should know, it beneited rom a local manipulation. 1he penalty that alicted

14
ee Goldman, earcb vgive ia., .vra note 28, at 112-15, Grimmelmann, trvctvre of earcb
vgive ar, .vra note 4, at 59-60.
148
Patterson, ôvêtror/ arrier., .vra note 19, at 2854.
149
150
Chandler, Rigbt to Reacb av .vaievce, .vra note 20, at 111 ,emphasis added,.
151
ee, e.g., Lettice, !bev .tgoritbv. .ttac/, .vra note 99, James Grimmelmann, Coogte Retie. to
earcbKivg ar.vit, LA\MLML ,Jan. 9, 2003,,
;11>?--@.!,(,("A(%(.A*;"D.@("(F$-,+F$@(%">;>E'.,(G_(!%TC&@(G.A1&*@(T%&FGH2
9.
152
Covare Londoner . Dener, 210 U.S. 33, 386 ,1908, ,hearing required when tax
assessment aects only a ew people, ritb Bi-Metallic In. Co. . Colorado, 239 U.S. 441,
445-46 ,hearing not required when tax assessment aects all citizens equally,.
153
ee JOlN RA\LS, A 1lLOR\ Ol JUS1ICL ,191,.
154
ee, e.g., Lettice, !bev .tgoritbv. .ttac/, .vra note 99.
it or three years appears to hae been a relatiely general change to Google`s
algorithm, one designed to aect a great many low-alue ertical search sites.
155

\hen loundem was promoted back to prominent search placement, tbat was
actually the manipulation, since it aected loundem and loundem alone.
Google thus manipulated` its search results to exempt loundem rom what
would otherwise hae been a generally applicable rule. 1o condemn
manipulation on the basis o its speciicity is to say that Google acted more
rightully when it demoted loundem in 2006 than when it promoted it back in
2009.
156

1he point is that local manipulations, being quick and easy to implement, are
oten a useul part o a search engine`s toolkit or deliering releance. Search-
engine-optimization is an endless game o loopholing. Regulators who attempt
to prohibit unair manipulations will hae to wade quite ar into the swamp o
white-hat and black-hat SLO.
15
Prohibiting local manipulation altogether
would keep the search engine rom closing loopholes quickly and punishing the
loopholers-giing them a substantial leg up in the SLO wars. Search results
pages would ill up with spam, and users would be the real losers.
Conclusion
Search neutrality gets one thing ery right: Search is about user autonomy. A
good search engine is more exquisitely sensitie to a user`s interests than av,
otber covvvvicatiov. tecbvotog,.
158
Search helps her ind whateer she wants,
whateer she needs to lie a sel-directed lie. It turns passie media recipients
into actie seekers and participants. I search did not exist, then or the sake o
human reedom it would be necessary to inent it. Search neutrality properly
seeks to make sure that search is liing up to its liberating potential.
laing asked the right question-are .trvctvrat force. tbrartivg .earcb`. abitit, to
rovote v.er avtovov,.-search neutrality adocates gie answers concerned with
protecting websites rather than users. \ith disturbing requency, though,
websites are not users` riends. Sometimes they are, but oten, the websites
want isitors, and will be willing to do what it takes to grab them.

155
a.
156
I you are bothered more by demotions than promotions, remember that search rankings are
zero-sum. loundem`s 50-place rise is balanced out by 50 one-place alls or other websites.
15
On the distinction between ethical, permitted white-hat` SLO and unethical, orbidden
black-hat` SLO, see lrank Pasquale, 1rv.tivg ;ava 1erif,ivg) Ovtive vterveaiarie.` Poticivg, .vra
at 258. I beliee that what Pasquale calls the intermediate grey-hat` zone between the two
is generally less grey than he and his sources perceie it to be.
158
Lxcept, perhaps, the library reerence desk. Unortunately, librarians don`t scale.
I llowers by Irene sells a bouquet or 30 that Bob`s llowers sells or 50,
then Bob`s interest in being ound is in direct conlict with users` interest in
being directed to Irene. 1he last thing that Bob wants is or the search engine
to maximize releance. Search-neutrality adocates ear that Bob will pay o
the search engine to point users at his site. But that`s not the only way the story
can play out. Bob could also engage in sel-help SLO to try to boost his
ranking. In that case, the search engine may respond by demoting his site. And
i that happens, then Bob has another card to play: search-neutrality itsel.
Regulators bearing search neutrality can inadertently preent search engines
rom helping users ind the websites they want. 1he typical model assumed by
search neutrality is o a website and a search engine corruptly conspiring to put
one oer on users. But much, indeed most, o the time, the real alliance is
between search engines and users, together trying to sort through the clamor o
millions o websites` sales pitches. Giing websites search-neutrality rights gies
them a powerul weapon in their wars with each other-one that need not be
wielded with users` interests in mind.
159
Search neutrality will be born with one
oot already in the grae o regulatory capture.
1here is a proound irony at the heart o the liberal case or search neutrality.
Requiring search evgive. to behae neutrally` will not produce the desired goal
o neutral search re.vtt.. 1he web is a place where site owners compete iercely,
sometimes iciously, or iewers and users turn to intermediaries to deend
them rom the sometimes-abusie tactics o inormation proiders. 1aking the
search engine out o the equation leaes users ulnerable to precisely the sorts
o manipulation search neutrality aims to protect them rom. \hether it ranks
sites by popularity, by personalization, or een by the idiosyncratic whims o its
operator, a search engine proides an attervatire to the lobbesian world o the
unmediated Internet, in which the richest oices are the loudest, and the
greatest authority on any subject is the spammer with the astest serer. Search
neutrality is cynical about the Internet-but perhaps not cynical enough.

159
1his has already happened in trademark law, which is supposed to preent consumer
conusion, but just as oten is a orm o oensie warare among companies, consumer
interests be damned. ee Mark A. Lemley & Mark P. McKenna, Orvivg Mar/;et). ,Stanord
Law and Lconomics Olin \orking Paper No. 395, May 2010,, araitabte at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG7:26H68. lor an exploration o
the competitie dynamics o trademark in the search-engine context, .ee Goldman,
Deregvtativg Reteravc,, .vra note 61.

Search Engine Bias & the
Demise of Search Engine
Utopianism
By Eric Goldman
*

In the past ew years, search engines hae emerged as a major orce in our
inormation economy, helping searchers perorm hundreds o millions ,or een
billions, o searches per day.
1
\ith this broad reach, search engines hae
signiicant power to shape searcher behaior and perceptions. In turn, the
choices that search engines make about how to collect and present data can
hae signiicant social implications.
1ypically, search engines automate their core operations, including the
processes that search engines use to aggregate their databases and then
sort,rank the data or presentation to searchers. 1his automation gies search
engines a eneer o objectiity and credibility.
2
Machines, not humans, appear
to make the crucial judgments, creating the impression that search engines
bypass the structural biases and skewed data presentations inherent in any
human-edited media.
3
Search engines` marketing disclosures typically reinorce
this perception o objectiity.
Unortunately, this romanticized iew o search engines does not match reality.
Search engines are media companies. Like other media companies, search

Associate Proessor, Santa Clara Uniersity School o Law and Director, ligh 1ech Law
Institute. lome page: ;11>?--!!!"(A&*/+@F,.'"+A/. Lmail: (/+@F,.'n/,.&@"*+,. I
appreciate the comments o Nico Brooks, Soumen Chakrabarti, Ben Ldelman, Llizabeth
Van Couering and the participants at the \ale Law School Regulating Search Symposium
and the 2005 Association o Internet Researchers ,AoIR, Annual Meeting. 1his essay
ocuses principally on American law and consumer behaior. Consumer behaior and
marketplace oerings ary by country, so this discussion may not be readily generalizable to
other jurisdictions.
1
In 2003, search engines perormed oer a hal-billion searches a day. ee Danny Sullian,
earcbe. Per Da,, SLARCl LNGINL \A1Cl, leb. 25, 2003,
;11>?--%(.A*;('/&'(!.1*;"*+,-A(>+A1%-.A1&*@(">;>-578:6:7.
2
ee Jason Lee Miller, eft, Rigbt, or Cevter. Cav a earcb vgive e ia.ea., \LBPRONL\S.COM,
May 10, 2005, ;11>?--!!!"!(#>A+'(!%"*+,-&'%&F(%(.A*;-&'%&F(%(.A*;-!>'L8:L
52282872d(C1b&/;1+AK('1(AK.'.h(.A*;i'/&'(U(U&.%(F";1,@.
3
1here is a broad perception that search engines present search results passiely and neutrally.
ee Leslie Marable, at.e Oracte.: Cov.vver Reactiov to earvivg tbe 1rvtb .bovt or earcb
vgive. !or/, CONSUMLR RLPOR1S \LB\A1Cl, June 30, 2003,
;11>?--!!!"*+'%$,(A!(#!.1*;"+A/-FD'.,&*-%(.A*;LA(>+A1LC.@%(L+A.*@(%L
.#%1A.*1"*C,, Maureen O`Rourke, Defivivg tbe ivit. of reeRiaivg iv C,ber.ace: 1raaevar/
iabitit, for Metataggivg, 33 GONZ. L. RLV. 2 ,1998,.
engines make editorial choices designed to satisy their audience.
4
1hese
choices systematically aor certain types o content oer others, producing a
phenomenon called search engine bias.`
Search engine bia. sounds scary, but this essay explains why such bias is both
necessary and desirable. 1he essay also explains how emerging personalization
technology will soon ameliorate many concerns about search engine bias.
Search Engines Make Editorial Choices
Search engines requently claim that their core operations are completely
automated and ree rom human interention,
5
but this characterization is alse.
Instead, humans make numerous editorial judgments about what data to collect
and how to present that data.
6

vaeivg. Search engines do not index eery scrap o data aailable on the
Internet. Search engines omit ,deliberately or accidentally, some web pages
entirely
or may incorporate only part o a web page.

8

4
ee, e.g., C. LD\IN BAKLR, ADVLR1ISING AND A DLMOCRA1IC PRLSS ,1994,.
5
ee, e.g., Doe. Coogte rer Mavivtate t. earcb Re.vtt.., GOOGLL.COM,
;11>?--!!!"/++/@("*+,-%$>>+A1-#&'-.'%!(A">DE.'%!(AG6778T1+>&*GB:H ,1he
order and contents o Google search results are completely automated. No one hand picks a
particular result or a gien search query, nor does Google eer insert jokes or send messages
by changing the order o results.`,, Doe. Coogte Cev.or earcb Re.vtt.., GOOGLL.COM,
;11>?--!!!"/++/@("*+,-%$>>+A1-#&'-.'%!(A">DE.'%!(AG79938T1+>&*GB:H ,Google
does not censor results or any search terms. 1he order and content o our results are
completely automated, we do not manipulate our search results by hand.`,, 1ecbvotog,
Orerrier, GOOGLL.COM, ;11>?--!!!"/++/@("*+,-*+A>+A.1(-1(*;";1,@ ,1here is no
human inolement or manipulation o results..`,, or Cav vrore M, ite`. Rav/ivg.,
GOOGLL.COM, ;11>?--!!!"/++/@("*+,-%$>>+A1-!(#,.%1(A%-#&'-.'%!(A">DE.'%!(A
GB66B5T1+>&*GH856 ,Sites` positions in our search results are determined automatically
based on a number o actors, which are explained in more detail at
;11>?--!!!"/++/@("*+,-1(*;'+@+/D-&'F(P";1,@. \e don`t manually assign keywords to
sites, nor do we manipulate the ranking o any site in our search results.`, .ee at.o Complaint
at 3-38, 52-56, KinderStart.com LLC . Google, Inc., Case No. C 06-205 RS ,N.D. Cal.
Mar. 1, 2006, ,giing other examples o Google`s claims to be passie,. Note that Google
has subsequently reised some o these cited pages ater its censorship controersy in China.
6
ee geveratt, Abbe Mowshowitz & Akira Kawaguchi, ia. ov tbe !eb, COMM. ACM, Sept. 2002,
at 56 ,distinguishing indexical bias` and content bias`,.
ee Judit Bar-Ilan, ectatiov. 1er.v. Reatit, - earcb vgive eatvre. êeaea for !eb Re.earcb at
Mia200:, 9 C\BLRML1RICS 2 ,2005,,
;11>?--!!!"*&'F+*"*%&*"(%-*D#(A,(1A&*%-.A1&*@(%-O3&7>5";1,@.
8
lor example, many search engines ignore metatags. ee Lric Goldman, Deregvtativg Reteravc,
iv vtervet 1raaevar/ ar, 54 LMOR\ L.J. 50, 56-68 ,2005,. Search engines also incorporate
only portions o ery large iles. ee Bar-Ilan, .vra note , !b, Doe.v`t M, ite are a Cacbea
Co, or a De.critiov., GOOGLL.COM, ;11>?--!!!"/++/@("*+,-%$>>+A1-#&'-
.'%!(A">DE.'%!(AG878T1+>&*GB:8 ,describing how some pages are partially indexed`,,
During indexing, search engines are designed to associate third party
metadata` ,data about data, with the indexed web page. lor example, search
engines may use and display third party descriptions o the website in the search
results.
9
Search engines may also index anchor text` ,the text that third parties
use in hyperlinking to a website,,
10
which can cause a website to appear in
search results or a term the website neer used ,and may object to,.
11

linally, once indexed, search engines may choose to exclude web pages rom
their indexes or a ariety o reasons, ranging rom iolations o quasi-objectie
search engine technical requirements
12
to simple capriciousness.
13

Rav/ivg. 1o determine the order o search results, search engines use complex
proprietary ranking algorithms.` Ranking algorithms obiate the need or
humans to make indiidualized ranking decisions or the millions o search

a. Coogte Droea 1beir 101K Cacbe ivit., RLSLARClBUZZ!, Jan. 31, 2005,
;11>?--!!!"A(%(.A*;#$00"+A/-5228-27-;.%4/++/@(4FA+>>(F41;(&A4727)"%;1,@
,discussing how historically Google indexed only the irst 101k o a document,.
9
ee M, ite`. i.tivg . vcorrect ava êea it Cbavgea, GOOGLL.COM,
;11>?--!!!"/++/@("*+,-!(#,.%1(A%-B";1,@. Google`s automated descriptions hae
spawned at least one lawsuit by a web publisher who belieed the compilation created a alse
characterization. ee Seth lineberg, Catif. CP. ve. Coogte Orer Mi.teaaivg earcb Re.vtt.,
ACC1. 1ODA\, Apr. 19, 2004, at 5, araitabte at
;11>?--!!!"!(#*>."*+,-.A1&*@("*C,E.A1&*@(&FG73BT>/L.**1+F.DT>A&'1GD(%.
10
ee Jagdeep S. Pannu, .vcbor 1et Otiviatiov, \LBPRONL\S.COM, Apr. 8, 2004,
;11>?--!!!"!(#>A+'(!%"*+,-(#$%&'(%%-%(+-!>'L6L
5226262HX'*;+AJ(P1a>1&,&0.1&+'";1,@.
11
lor example, the irst search result in Google and \ahoo! or the keyword miserable
ailure` is President George \. Bush`s home page because so many websites hae linked to
the biography using the term miserable ailure.` ee 1om McNichol, Yovr Me..age ere, N.\.
1IMLS, Jan. 22, 2004, at G1. 1his algorithmic ulnerability has spawned a phenomenon
called Google bombing,` where websites coordinate an anchor text attack to intentionally
distort search results. ee John liler, Coogte 1ive ovb, MICROCON1LN1 NL\S, Mar. 3, 2002,
;11>?--!!!",&*A+*+'1('1'(!%"*+,-.A1&*@(%-/++/@(#+,#%";1,.
12
ee, e.g., Steanie Olsen, earcb vgive. Detete .arare Covav,, CNL1 NL\S.COM, May 13,
2004, ;11>?--'(!%"*+,"*+,-5725L72564BL8575693";1,@E1./G%1"$1&@">A&'1 ,Google and
\ahoo kicked \henU.com out o their indexes or allegedly displaying dierent web pages
to searchers and search engine robots, a process called cloaking`,.
13
1his is the heart o KinderStart`s allegations against Google. ee Complaint,
KinderStart.com LLC . Google, Inc., Case No. C 06-205 ,N.D. Cal. Mar. 1, 2006,.
Although the complaint`s allegations about Google`s core algorithmic search may not be
proen, Google does liberally excise sources rom Google News. lor example, Google
claims that news sources are selected without regard to political iewpoint or ideology,` .ee
Coogte êr. ;eta), GOOGLL.COM,
;11>?--'(!%"/++/@("*+,-&'1@-('4$%-.#+$14/++/@(4'(!%";1,@o58, but Google
dropped a white supremacist news source rom Google News because it allegedly
promulgated hate content.` ee Susan Kuchinskas, Coogte .e. ate êr.,
IN1LRNL1NL\S.COM, Mar. 23, 2005,
;11>?--!!!"&'1(A'(1'(!%"*+,-PhS-.A1&*@(">;>-B635B:7.
terms used by searchers, but they do not lessen the role o human editorial
judgment in the process. Instead, the choice o which actors to include in the
ranking algorithm ,and how to weight them, relects the search engine
operator`s editorial judgments about what makes content aluable. Indeed, to
ensure that these judgments are producing the desired results, search engines
manually inspect search results
14
and make adjustments accordingly.
Additionally, search engines claim they do not modiy algorithmically-generated
search results, but there is some eidence to the contrary. Search engines
allegedly make manual adjustments to a web publisher`s oerall ranking.
15
Also,
search engines occasionally modiy search results presented in response to
particular keyword searches. Consider the ollowing:
Some search engines blocked certain search terms containing the
keyword phpBB.`
16

In response to the search term Jew,` or a period o time ,including, at
minimum Noember 2005 when the author obsered the
phenomenon,, Google displayed a special result in the sponsored link,
saying Oensie Search Results: \e`re disturbed about these results
as well. Please read our note here.` 1he link led to a page explaining
the results.
1

Reportedly, Ask.com blocked search results or certain terms like
pedophile,` bestiality,` sex with children` and child sex.`
18

Google remoed some websites rom its index in response to a Digital
Millenium Copyright Act ,DMCA, take-down demand rom the
Church o Scientology. loweer, Google displayed the ollowing
legend at the bottom o aected search results pages ,such as search
results or scientology site:xenu.net`,: In response to a complaint we
receied under the US Digital Millennium Copyright Act, we hae

14
ee Posting o Lric Goldman to 1echnology & Marketing Law Blog, Coogte`. vvav
.tgoritbv, ;11>?--#@+/"(A&*/+@F,.'"+A/-.A*;&O(%-5228-2:-/++/@(%4;$,.'4.";1,
,June 5, 2005, 14:11 LS1, ,Google hires students to manually reiew search results or
quality purposes,.
15
ee Search King, Inc. . Google 1ech., Inc., No. CIV-02-145-M, at 4 ,\.D. Okla. Jan. 13,
2003, ,Google knowingly and intentionally decreased the PageRanks assigned to both
SearchKing and PRAN.`,. 1his manual adjustment has also been alleged in the recent
Kivaertart lawsuit. ee Complaint, KinderStart.com L.L.C. . Google, Inc., Case No. C 06-
205 RS ,N.D. Cal. Mar. 1, 2006,.
16
ee M^ toc/aae. b earcber., 1RIMMAIL`S LMAIL BA11LLS, Jan. 18, 2006,
;11>?--!!!"(,.&@#.11@(%"*+,-.A*;&O(-#.11@(%-O$@'4..*/C#/F*#4RF-.
1
ee ;11>?--!!!"/++/@("*+,-(P>@.'.1&+'";1,@.
18
ee Jennier Laycock, ../.cov .ctiret, Cev.orivg ove earcb Pbra.e., SLARCl LNGINL GUIDL,
June 25, 2006, ;11>?--!!!"%(.A*;('/&'(/$&F("*+,-%(.A*;#A&(C-%('(!%-229HB9";1,@.
On Aug. 1, 2006, I was unable to replicate these results.
remoed 2 result,s, rom this page. I you wish, you may read the
DMCA complaint that caused the remoal,s, at ChillingLects.org.`
19

Covctv.iov. Search engines hae some duality in their sel-perceptions, and this
duality creates much conusion.
20
Search engines perceie themseles as
objectie and neutral because they let automated technology do most o the
hard work. loweer, in practice, search engines make editorial judgments just
like any other media company. Principally, these editorial judgments are
instantiated in the parameters set or the automated operations, but search
engines also make indiidualized judgments about what data to collect and how
to present it. 1hese manual interentions may be the exception and not the
rule, but these exceptions only reinorce that search engines play an actie role
in shaping their users` experiences when necessary to accomplish their editorial
goals.
Search Engine Editorial Choices
Create Biases
Search results ordering has a signiicant eect on searchers and web publishers.
Searchers usually consider only the top ew search results, the top-ranked search
result gets a high percentage o searcher clicks, and click-through rates quickly
decline rom there.
21
1hereore, een i a search engine deliers hundreds or
een thousands o search results in response to a searcher`s query, searchers

19
ee Chris Sherman, Coogte Ma/e. cievtotog, vfrivgevevt Devava Pvbtic, SLARCl LNGINL
\A1Cl, Apr. 15, 2002,
;11>?--%(.A*;('/&'(!.1*;"*+,-%(.A*;F.D-.A1&*@(">;>-5783:37.
20
ee Danny Sullian, Kivaertart ecove. Kivaertoea v Rav/ivg ar.vit .gaiv.t Coogte, SLARCl
LNGINL \A1Cl, July 14, 2006, ;11>?--#@+/"%(.A*;('/&'(!.1*;"*+,-#@+/-2:2976L
2H6H65. 1his duality, i it ends up leading to the dissemination o alse inormation, could
also create some legal liability. ee KinderStart . Google, No. 5:06-c-0205-Jl ,N.D. Cal.
motion to dismiss granted July 13, 2006, ,pointing out the potential inconsistency o
Google`s position that PageRank is both Google`s subjectie opinion but an objectie
relection o its algorithmic determinations,.
21
ee iPro.ect earcb vgive |.er ebarior tva,, IPROSPLC1, Apr. 2006,
;11>?--!!!"&>A+%>(*1"*+,->A(,&$,SNI%-Z;&1(S.>(A4522:4h(.A*;i'/&'(V%(AU(
;.O&+A">FC ,62 o searchers click on a search result on the irst results page,, Jakob
Nielsen, 1be Porer of Defavtt., JAKOB NILLSLN`S ALLR1BOX, Sept. 26, 2005,
;11>?--!!!"$%(&1"*+,-.@(A1#+P-F(C.$@1%";1,@ ,citing a study by Cornell proessor
1horsten Joachims that the irst search result gets 42 o clicks and the second search result
gets 8, urther, when the irst two search results are switched, the irst search result gets
34-meaning that positioning dictated searcher behaior,, Nico Brooks, 1be .tta. Rav/
Reort: or earcb vgive Rav/ vact. 1raffic, A1LAS INS1I1U1L DIGI1AL MARKL1ING
INSIGl1S, June 2004, ;11>?--.>>".1@.%+'(>+&'1"*+,->FC-X1@.%b.')b(>+A1">FC ,the
irst-ranked search result may get ten times the quantity o clicks as the tenth-ranked search
result,.
eectiely ignore the ast majority o those search results. Accordingly, web
publishers desperately want to be listed among the top ew search results.
22

lor search engines, results placement determines how the searcher perceies the
search experience. I the top ew search results do not satisy the searcher`s
objecties, the searcher may deem the search a ailure. 1hereore, to maximize
searcher perceptions o search success, search engines generally tune their
ranking algorithms to support majority interests.
23
In turn, minority interests
,and the websites catering to them, oten receie marginal exposure in search
results.
1o gauge majority interests, search engines requently include a popularity
metric in their ranking algorithm. Google`s popularity metric, PageRank, treats
inbound links to a website as popularity otes, but otes are not counted
equally, links rom more popular websites count more than links rom lesser-
known websites.
24

Beyond promoting search results designed to satisy majority interests,
PageRank`s non-egalitarian oting structure causes search results to be biased
towards websites with economic power
25
because these websites get more links
due to their marketing expenditures and general prominence.
Indeed, popularity-based ranking algorithms may reinorce and perpetuate
existing power structures.
26
\ebsites that are part o the current power elite get
better search result placement, which leads to greater consideration o their
messages and iews. lurthermore, the increased exposure attributable to better
placement means that these websites are likely to get more otes in the uture,

22
ee Michael 1otty & Mylene Mangalindan, !eb ite. 1r, rer,tbivg 1o Ctivb Coogte Rav/ivg.,
\ALL S1. J. ONLINL, leb. 26, 2003,
;11>?--+'@&'("!%R"*+,-.A1&*@(-hU726:55:7:2HH63:B36B";1,@E(,.&@CGD(%.
23
ee Lucas D. Introna & lelen Nissenbaum, baivg tbe !eb: !b, tbe Potitic. of earcb vgive.
Matter., INlO. SOC`\, July-Sept. 2000, at 169.
24
ee Ovr earcb: Coogte 1ecbvotog,, GOOGLL.COM, ;11>?--!!!"/++/@("*+,-1(*;'+@+/D-.
25
ee Nia Llkin-Koren, et tbe Crarter. Crart: Ov 1irtvat Cate/eeer. ava tbe Rigbt to ctvae
vaeivg, 26 U. DA\1ON L. RLV. 19, 188 ,2001,, lrank Pasquale, Rav/ivg., Reavctiovi.v, ava
Re.ov.ibitit,, SL1ON lALL PUBLIC LA\ RLSLARCl PAPLR NO. 88832, at 25, leb. 25, 2006,
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FGHHHB59, 1rystan Upstill et al.,
Preaictivg ave ava ortvve: PageRav/ or vaegree., PROC. Ol 1lL 81l AUS1RALASIAN
DOCUMLN1 COMPU1ING S\MP., Dec. 15, 2003,
;11>?--A(%(.A*;",&*A+%+C1"*+,-$%(A%-'&*)*A->$#%-$>%1&@@4.F*%2B">FC ,showing that
Business\eek 1op Brand, lortune 500 and lortune Most Admired companies get
disproportionately high PageRank,.
26
ee Introna & Nissenbaum, .vra note 23, Matthew lindman et al., Coogtearcb,: or a er
earit,iv/ea ite. Dovivate Potitic. ov tbe !eb, Mar. 31, 2003,
;11>?--!!!">A&'*(1+'"(F$-l,;&'F,.'-/++/@(.A*;DLL;&'F,.'">FC.
leading to a sel-reinorcing process.
2
In contrast, minority-interest and
disenranchised websites may hae a diicult time cracking through the
popularity contest, potentially leaing them perpetually relegated to the search
results hinterlands.
28

A number o commentators hae lamented these eects and oered some
proposals in response:
vrore earcb vgive 1rav.arevc,. Search engines keep their ranking
algorithms secret.
29
1his secrecy hinders search engine spammers rom
gaining more prominence than search engines want them to hae, but
the secrecy also preents searchers and commentators rom accurately
assessing any bias. 1o enlighten searchers, search engines could be
required to disclose more about their practices and their algorithms.
30

1his additional inormation has two putatie beneits. lirst, it may
improe market mechanisms by helping searchers make inormed
choices among search engine competitors. Second, it may help
searchers determine the appropriate leel o cognitie authority to
assign to their search results.
Pvbtict, vva earcb vgive.. Arguably, search engines hae public
good`-like attributes, such as reducing the social costs o search
behaior. I so, priate actors will not incorporate these social beneits
into their decision-making. In that case, public unding o search
engines may be required to produce socially-optimal search results.
31

2
ee gatitariav vgive., LCONOMIS1, No. 1, 2005 ,there is a widespread belie among
computer, social and political scientists that search engines create a icious circle that
ampliies the dominance o established and already popular websites`,, .ee at.o Junghoo Cho
& Sourashis Roy, vact of earcb vgive. ov Page Povtarit,, \\\ 2004, May 2004,
;11>?--+.)"*%"$*@."(F$-l*;+->.>(A%-*;+L#&.%">FC, Upstill, .vra note 25. vt .ee Santo
lortunato et al., 1be gatitariav ffect of earcb vgive., No. 2005,
;11>?--.AP&O"+A/->FC-*%"K\-2877228 ,questioning the consequences o the rich-gets-
richer` eect,.
28
ee Cho & Roy, .vra note 2, bvt .ee lilippo Menczer et al., Coogtearcb, or Coogtocrac,., ILLL
SPLC1RUM, leb. 2006 ,proiding empirical eidence suggesting that search engines direct
more traic than expected to less popular sites`,.
29
ee Search King Inc. . Google 1ech., Inc., No. CIV-02-145-M, at 3 n.2 ,\.D. Okla. Jan. 13,
2003, ,Google`s mathematical algorithm is a trade secret, and it has been characterized by
the company as one o Google`s most aluable assets.``,, Steanie Olsen, Pro;ect earcbe. for
Oevovrce îcbe, CNL1 NL\S.COM, Aug. 18, 2003, ;11>?--'(!%"*+,"*+,-5725L72B54BL
82:637B";1,@E1./G%14$1&@4>A&'1.
30
ee Introna & Nissenbaum, .vra note 23.
31
ee ia., Lszter largittai, Oev Portat. or Cto.ea Cate.. Cbavvetivg Covtevt ov tbe !orta !iae !eb,
2 POL1ICS 233 ,2000,, cf. CASS SUNS1LIN, RLPUBLIC.COM 10-2 ,2001, ,adocating
publicly unded deliberatie domains`,.
Indeed, there hae been seeral proposals to create goernment-unded
search engines.
32

Mavaate Cbavge. to Rav/ivg,ortivg Practice.. Search engines could be
orced to increase the exposure o otherwise-marginalized websites. At
least ie lawsuits
33
hae requested judges to orce search engines to
reorder search results to increase the plainti`s isibility.
34

In addition to plaintis, some academics hae supported mandatory reordering
o search results. lor example, Pandey et al. adocate a randomized rank
promotion` scheme where obscure websites randomly should get extra credit in
ranking algorithms, appearing higher in the search results on occasion and
getting additional exposure to searchers accordingly.
35
In another essay in this
collection, lrank Pasquale proposes that, when people think the search engines
are proiding alse or misleading inormation, search engines should be orced
to include a link to correctie inormation.
36

Search Engine Bias Is
Necessary and Desirable
Beore trying to sole the problem o search engine bias, we should be clear
how search engine bias creates a problem that requires correction. lrom my
perspectie, search engine bias is the unaoidable consequence o search

32
ee Kein J. O`Brien, vroeav. !eigb Ptav ov Coogte Cbattevge, IN1`L lLRALD 1RIB., Jan. 18,
2006 ,discussing a Luropean initiatie called Quaero, which is intended to break the
American hegemony implicit in Google`s dominant market position,, Graeme \earden,
]aav Ma, Create t. Orv earcb vgive, CNL1 NL\S.COM, Dec. 21, 2005,
;11>?--'(!%"*+,"*+,-].>.'k,.Dk*A(.1(k&1%k+!'k%(.A*;k('/&'(-5722L72584BL
2262B9";1,@.
33
ee Search King, Inc. . Google 1ech., Inc., No. CIV-02-145-M ,\.D. Okla. Jan. 13, 2003,,
KinderStart.com LLC . Google, Inc., No. C 06-205 RS ,N.D. Cal. dismissed July 13, 2006,,
Langdon . Google, Inc., No. 1:06-c-00319-JJl ,D. Del. complaint iled May 1, 2006,,
Roberts . Google, No. 1-06-CV-06304 ,Cal. Superior Ct. complaint iled May 5, 2006,,
Datner . \ahoo! Inc, Case No. BC35521 ,Cal. Superior Ct. complaint iled July 11, 2006,
|note: this list updated as o July 24, 2006|.
34
As Google said in its response to the KinderStart lawsuit, Plainti KinderStart contends
that the judiciary should hae the inal say oer |search engines`| editorial process. It has
brought this litigation in the hopes that the Court will second-guess Google`s search rankings
and order Google to iew KinderStart`s site more aorably.` Motion to Dismiss at 1,
KinderStart.com LLC . Google, Inc., No. C 06-205 RS ,N.D. Cal. May 2, 2006,.
35
ee Sandeep Pandey et al., bvfftivg a tac/ea Dec/: tbe Ca.e for Partiatt, Ravaoviea Rav/ivg of
earcb vgive Re.vtt.,
;11>?--!!!"*%"*,$"(F$-l+@%1+'->$#@&*.1&+'%-A.'F+,b.')&'/">FC, cf. SUNS1LIN,
.vra note 31 ,explaining that websites should be orced to link to contrary iews as a way o
increasing exposure to alternatie iewpoints,.
36
ee Pasquale, .vra at 401, .ee at.o Pasquale, .vra note 25, at 28-30 ,proposing that the link be
displayed as an asterisk to the search results,.
engines exercising editorial control oer their databases. Like any other media
company, search engines simply cannot passiely and neutrally redistribute third
party content ,in this case, web publisher content,. I a search engine does not
attempt to organize web content, its system quickly and ineitably will be
oertaken by spammers, raudsters and malcontents.
3
At that point, the search
engine becomes worthless to searchers.
Instead, searchers ,like other media consumers, expect search engines to create
order rom the inormation glut. 1o preent anarchy and presere credibility,
search engines must exercise some editorial control oer their systems. In turn,
this editorial control necessarily will create some bias.
lortunately, market orces limit the scope o search engine bias.
38
Searchers
hae high expectations or search engines: they expect search engines to read
their minds
39
and iner their intent based solely on a small number o search
keywords.
40
Search engines that disappoint ,either by ailing to delier releant
results, or by burying releant results under too many unhelpul results, are held

3
Lery Internet enue accepting user-submitted content ineitably gets attacked by unwanted
content. I let untended, the enue inexorably degrades into anarchy. ee, e.g., teb,te:
or to Cet O^ of Page. vaeea b, Coogte, MONL1IZL BLOG, June 1, 2006,
;11>?--,(A/(F"*.-,+'(1&0(-C@.1-;+!L1+L/(1L#&@@&+'%L+CL>./(%L&'F(P(FL#DL
M++/@(";1,@ ,Google indexed oer ie billion spam` pages rom a single spammer beore
manually de-indexing the sites,, Alorie Gilbert, Coogte ie. Ctitcb 1bat |vtea.bea tooa of Porv,
CNL1 NL\S.COM, No. 28, 2005, ;11>?--'(!%"*+,"*+,-5725L72584BL
83:3933";1,@E1./G%1"$1&@">A&'1 ,describing how Google Base, a enue or user-submitted
content, was oertaken by pornographers: the amount o adult content on Google Base
was staggering considering Google only launched the tool a week ago.`,, Josh Quittner, 1be
!ar etreev att.ta.tete.. ava rec.et..cat., \IRLD, May 1994, at 46 ,describing how a group o
anarchists, or un, took oer a USLNL1 newsgroup about pets,.
38
ee Mowshowitz & Kawaguchi, .vra note 6, at 60 ,market orces are the best way to counter
aderse eects o search engine bias,.
39
ee Ovr Pbito.ob,, GOOGLL.COM, ;11>?--!!!"/++/@("*+,-*+A>+A.1(-1('1;&'/%";1,@
,1he perect search engine . would understand exactly what you mean and gie back
exactly what you want.`,, Chris Sherman, f earcb vgive. Covta Reaa Yovr Miva, SLARCl
LNGINL \A1Cl, May 11, 2005,
;11>?--%(.A*;('/&'(!.1*;"*+,-%(.A*;F.D-.A1&*@(">;>-B82B3B7.
40
Searchers routinely use a ery small number o keywords to express their search interests.
ee iProspect.com, Inc., iPro.ect âtvrat O Ke,rora evgtb tva,, No. 2004,
;11>?--!!!"&>A+%>(*1"*+,->A(,&$,SNI%-)(D!+AF4@('/1;4%1$FD">FC ,eighty-eight
percent o search engine reerrals are based on only one or two keywords,, .ee at.o Declan
Butler, ovea| earcb vgive., NA1URL, May 11, 2000, at 112, 115 ,citing an NLC Research
Institute study showing that up to 0 o searchers use only a single keyword as a search
term,, Bernard J. Jansen et al., Reat ife vforvatiov Retrierat: . tva, of |.er Qverie. ov tbe !eb,
32 SIGIR lORUM 5, 15 ,1998, ,stating that the aerage keyword length was 2.35 words, one-
third o searches used one keyword and 80 used three keywords or ewer,, Jakob Nielsen,
JAKOB NILLSLN`S ALLR1BOX, earcb: 1i.ibte ava ivte, May 13, 2001,
;11>?--!!!"$%(&1"*+,-.@(A1#+P-5227287B";1,@ ,stating that the aerage keyword length
was 2.0 words,.
accountable by ickle searchers.
41
1here are multiple search engines aailable to
searchers,
42
and ew barriers to switching between them.
43

As a result, searchers will shop around i they do not get the results they want,
44

and this competitie pressure constrains search engine bias. I a search engine`s
bias degrades the releancy o search results, searchers will explore alternaties
een i searchers do not realize that the results are biased. Meanwhile, search
engine prolieration means that niche search engines can segment the market
and cater to undersered minority interests.
45
Admittedly, these market orces

41
ee Kim Peterson, Micro.oft earv. to Crart, SLA11LL 1IMLS, May 2, 2005 ,MSN Search
learned that the arcane searches were the make-or-break moments or \eb searchers.
People weren`t just happy when a search engine could ind answers to their most bizarre,
obscure and diicult queries. 1hey would switch loyalties.`,, Bob 1edeschi, rer, Ctic/ Yov
Ma/e, 1be,`tt e !atcbivg Yov, N.\. 1IMLS, Apr. 3, 2006,
;11>?--!!!"'D1&,(%"*+,-522:-26-2B-#$%&'(%%-2B(*+,";1,@E(&G8232T('G3(88.(:
6C:356BB.T(PG7B2797:H22T>.A1'(AGA%%$%(A@.'FT(,*GA%%T>./(!.'1(FG>A&'1.
42
In addition to the recent launch o major new search engines by proiders like MSN, the
open-source sotware community is deeloping Nutch to allow anyone to build and
customize his or her own web search engine. ;11>?--'$1*;".>.*;("+A/-, .ee at.o Olsen,
Oevovrce îcbe, .vra note 29.\hile there are multiple major search engines, the market
may still resemble an oligopoly, a ew major players ,Google, \ahoo, MSN, Ask Jeees, hae
the lion`s share o the search engine market. loweer, this may construe the search engine
market too narrowly. Many types o search proiders compete with the big mass-market
search engines, ranging rom specialty search engines ,e.g., 1echnorati, to alternatie types o
search technology ,e.g., adware, to non-search inormation retrieal processes ,e.g., link
naigation,. Ultimately, eery search engine competes against other search engines and these
other search,retrieal options.
43
ee Rahul 1elang et al., .v viricat .vat,.i. of vtervet earcb vgive Cboice, Aug. 2002 ,on ile
with author,. lor example, search engines use the same basic interace ,a white search box,,
and searchers rarely use adanced search eatures that might require additional learning time
at other search engines.
44
ee Grant Crowell, |vaer.tavaivg earcber ebarior, SLARCl LNGINL \A1Cl, June 14, 2006,
;11>?--%(.A*;('/&'(!.1*;"*+,-%;+!S./(";1,@E>./(GB:7B537 ,citing a Kelsey
Research study that 63 o searchers used two or more search engines,, Press Release,
Viidence, Inc., Coogte !iv. |.er.` eart., vt ôt 1beir .a Ctic/. ,May 25, 2004,,
;11>?--!!!"O&O&F('*("*+,->$#@&*-*+,>.'D-'(!%k.'Fk(O('1%->A(%%kA(@(.%(%-52
26L28L58k*(kA.')&'/%k%(.A*;";1, ,stating that up to 4 o searchers try another
search engine when their search expectations are not met,.
45
ee Rahul 1elang et al., 1be Mar/et trvctvre for vtervet earcb vgive., 21 J. MGM1. INlO. S\S.
13 ,2004,, araitabte at ;11>?--!!!";(&'0"*,$"(F$-lA1(@.'/-('/&'(4R,&%4C&'.@">FC
,describing how searchers sample heterogeneous ranking algorithms, which support a
diersity o search engines,, Mario J. Sila, 1be Ca.e for a Portvgve.e !eb earcb vgive,
;11>?--P@F#"C*"$@">1-F.1.-S$#@&*.1&+'%4.11.*;-1$,#.L&*!&522BLC&'.@">FC ,describing
the alue o a Portuguese-oriented search engine,, Jerey McMurray, ociat earcb Provi.e.
etter vtettigevce, ASSOCIA1LD PRLSS, July 9, 2006 ,discussing niche search engines that draw
on social networking,, cf. Jakob Nielsen, Direr.it, i. Porer for eciatiea ite., JAKOB NILLSLN`S
ALLR1BOX, June 16, 2003, ;11>?--!!!"$%(&1"*+,-.@(A1#+P-522B2:7:";1,@ ,describing
how specialized sites will lourish on the Internet,.
are incomplete-searchers may neer consider what results they are not
seeing-but they are powerul nonetheless.
In contrast, it is hard to imagine how regulatory interention will improe the
situation. lirst, regulatory solutions become a ehicle or normatie iews
about what searchers should see-or should ravt to see.
46
low should we
select among these normatie iews \hat makes one bias better than another
Second, regulatory interention that promotes some search results oer others
does not ensure that searchers will ind the promoted search results useul.
Determining releancy based on ery limited data ,such as decontextualized
keywords, is a challenging process, and search engines struggle with this
challenge daily. Due to the complexity o the releancy matching process,
goernment regulation rarely can do better than market orces at deliering
results that searchers ind releant. As a result, searchers likely will ind some o
the promoted results irreleant.
1he clutter o unhelpul results may hinder searchers` ability to satisy their
search objecties, undermining searchers` conidence in search engines` mind-
reading abilities.
4
In this case, regulatory interention could counter-
productiely degrade search engines` alue to searchers. \hateer the aderse
consequences o search engine bias, the consequences o regulatory correction
are probably worse.
48

Technological Evolution Will
Moot Search Engine Bias
Currently, search engines principally use one-size-its-all` ranking algorithms to
delier homogeneous search results to searchers with heterogeneous search
objecties.
49
One-size-its-all algorithms exacerbate the consequences o search
engine bias in two ways: ,1, they create winners ,websites listed high in the

46
ee, e.g., Susan L. Gerhart, Do !eb earcb vgive. vre.. Covtrorer.,., lIRS1 MONDA\, Jan.
2004, ;11>?--!!!"C&A%1,+'F.D"+A/-&%%$(%-&%%$(347-/(A;.A1-. Gerhart argues that
search engines do not adequately prioritize search results that expose controersies about the
search topic. loweer, her argument assumes that controersy-related inormation has
alue to consumers, an assumption that deseres careul ealuation.
4
ee Lric Goldman, . Coa.eav .vat,.i. of Mar/etivg, 2006 \IS. L. RLV. 1151, araitabte at
;11>?-->.>(A%"%%A'"*+,-%+@B->.>(A%"*C,E.#%1A.*14&FG375856.
48
ee Susan P. Craword, bortve.. of 1i.iov: Regvtator, .vbitiov iv tbe Digitat .ge, 4 lORDlAM
L. RLV. 695 ,2005, ,discussing the shortcomings o regulatory interention in organic
inormation systems,.
49
ee James Pitkow et al., Per.ovatiea earcb, COMM. ACM, Vol. 45:9 ,Sept. 2002, at 50-1.
search results, and losers ,those with marginal placement,, and ,2, they delier
suboptimal results or searchers with minority interests.
50

1hese consequences will abate when search engines migrate away rom one-
size-its-all algorithms towards personalized` ranking algorithms.
51

Personalized algorithms produce search results that are custom-tailored to each
searcher`s interests, so searchers will see dierent results in response to the
same search query. lor example, Google oers searchers an option that orders
your search results based on your past searches, as well as the search results and
news headlines youè clicked on.`
52

Personalized ranking algorithms represent the next major adance in search
releancy. One-size-its-all ranking algorithms hae inherent limits on their
maximum releancy potential, and urther improements in one-size-its
algorithms will yield progressiely smaller releancy beneits. Personalized
algorithms transcend those limits, optimizing releancy or each searcher and
thus implicitly doing a better job o searcher mind-reading.
53

Personalized ranking algorithms also reduce the eects o search engine bias.
Personalized algorithms mean that there are multiple top` search results or a
particular search term instead o a single winner,`
54
so web publishers will not
compete against each other in a zero-sum game. In turn, searchers will get
results more inluenced by their idiosyncratic preerences and less inluenced by
the embedded preerences o the algorithm-writers. Also, personalized
algorithms necessarily will diminish the weight gien to popularity-based metrics
,to gie more weight or searcher-speciic actors,, reducing the structural biases
due to popularity.

50
ee Michael Kanellos, Micro.oft .iv. for earcb ov t. Orv 1erv., CNL1 NL\S.COM, No. 24,
2003, ;11>?--'(!%"*+,"*+,-5725L722H4BL8772372";1,@E1./G%1"$1&@">A&'1 ,quoting a
Microsot researcher as saying I the two o us type a query |into a search engine|, we get
the same thing back, and that is just brain dead. 1here is no way an intelligent human being
would tell us the same thing about the same topic.`,, Daid l. lreedman, !b, Prirac, !ov`t
Matter, NL\S\LLK, Apr. 3, 2006, Personalization o Placed Content Ordering in Search
Results, U.S. Patent App. 0050240580 ,iled July 13, 2004,.
51
ee Pitkow, .vra note 49, at 50.
52
!bat`. Per.ovatiea earcb., GOOGLL.COM,
;11>?--!!!"/++/@("*+,-%$>>+A1-#&'-.'%!(A">DE.'%!(AG5::87T1+>&*G783B.
53
ee Jaime 1eean et al., Per.ovatiivg earcb ria .vtovatea .vat,.i. of vtere.t. ava .ctiritie.,
SIGIR 05, ;11>?--;.D%1.*)"@*%",&1"(F$->.>(A%-1((O.'"%&/&A28">FC, 1erry McCarthy, Ov
tbe rovtier of earcb, 1IML, Aug. 28, 2005 ,Search will ultimately be as good as haing 1,000
human experts who know your tastes scanning billions o documents within a split
second.`, ,quoting Gary llake, Microsot Distinguished Lngineer,.
54
ee Kein Lee, earcb Per.ovatiatiov ava PPC earcb Mar/etivg, CLICKZ NL\S, July 15, 2005,
;11>?--!!!"*@&*)0"*+,-(P>(A1%-%(.A*;-%1A.1->A&'1">;>-B873H9:.
Personalized ranking algorithms are not a panacea-any process where humans
select and weight algorithmic actors will produce some bias
55
-but
personalized algorithms will eliminate many o the current concerns about
search engine bias.
Conclusion
Complaints about search engine bias implicitly relect some disappointed
expectations. In theory, search engines can transcend the deiciencies o
predecessor media to produce a type o utopian media. In practice, search
engines are just like eery other medium-heaily reliant on editorial control
and susceptible to human biases. 1his act shatters any illusions o search
engine utopianism.
lortunately, search engine bias may be largely temporal. In this respect, I see
strong parallels between search engine bias and the late 1990s keyword metatag
problem.`
56
\eb publishers used keyword metatags to distort search results,
but these techniques worked only so long as search engines considered keyword
metatags in their ranking algorithms. \hen search engines recognized the
distortie eects o keyword metatags, they changed their algorithms to ignore
keyword metatags.
5
Search result releancy improed, and the problem was
soled without regulatory interention.
Similarly, search engines naturally will continue to eole their ranking
algorithms and improe search result releancy-a process that, organically, will
cause the most problematic aspects o search engine bias to largely disappear.
1o aoid undercutting search engines` quest or releance, this eort should
proceed without regulatory distortion.

55
Personalized algorithms hae other potentially aderse consequences, such as creating sel-
reinorcing inormation lows. ee SUNS1LIN, .vra note 31. lor a critique o these
consequences, see Goldman, Coa.eav .vat,.i., .vra note 4.
56
ee geveratt, Goldman, Deregvtativg Reteravc,, supra note 8.
5
ee Danny Sullian, Deatb of a Meta 1ag, SLARCl LNGINL \A1Cl, Oct. 1, 2002,
;11>?--!!!"%(.A*;('/&'(!.1*;"*+,-%(A(>+A1->A&'1">;>-B6957457:82:7.

475

CHAPTER 8
WHAT FUTURE FOR PRIVACY?
Privacy Protection in the Next Digital Decade:
Trading Up or a Race to the Bottom? 477
Michael Zimmer
The Privacy Problem: Whats Wrong with Privacy? 483
Stewart Baker
A Market Approach to Privacy Policy 509
Larry Downes

476 CHAPTER 8: WHAT FUTURE FOR PRIVACY?


Privacy Protection in the Next
Digital Decade: Trading Up or
a Race to the Bottom?
By Michael Zimmer
*

Apparent to most citizens o contemporary, industrialized society, people no
longer exist and lie in ixed locations and spaces. Instead people are on the
moe in their personal, proessional, intellectual, and social spheres. \ithin and
across these spheres, mobility, rather than permanence, is likely to be the norm.
Manuel Castells captures this eature o modern lie in his theory o the space o
lows, arguing that our society is constructed around lows: lows o capital,
lows o inormation, lows o technology, lows o organizational interaction,
lows o images, sounds, and symbols.`
1
1hese lows-particularly ivforvatiov
lows-constitute what Castells describes as the network society,` where
networks constitute the new social morphology o our societies, and the
diusion o networking logic substantially modiies the operation and outcomes
in processes o production, experience, power and culture.`
2

Nowhere is Castells network society` more apparent than in our contemporary
global digital inormation network, with the Internet as its backbone.
Originating rom a handul o uniersities and research laboratories in the
1960s, the Internet began to take shape as a ubiquitous inormation network
with the emergence o the dot-com` economy in the 1990s. Dot-com business
models aried-and met aried leels o success-but most relied on the rapid
deliery o serices and exchange o inormation. \hile much o the dot-com
economy burst with the dot-com bubble in 2000, the Internet remained a
powerul network enabling robust lows o inormation, continually modiying
experience, power and culture,` just as Castells described.
In the past digital decade, the Internet has proided new linkages and spaces or
inormation lows, and has particularly emerged as a potent inrastructure or
the low and capture o er.ovat inormation. 1hese lows take many orms and
stem rom arious motiations. Large-scale web adertising platorms and
search engines utilize robust inrastructures to collect data about web browsing
and search actiities in order to proide releant adertising. Users`
consumption habits are captured by online serice proiders like Amazon and
Netlix, ueling powerul recommendation systems meant to improe user

School o Inormation Studies, Uniersity o \isconsin-Milwaukee

1
MANULL CAS1LLLS, 1lL RISL Ol 1lL NL1\ORK SOCIL1\ 412 ,1996,.
2
a. at 469.

satisaction. Indiiduals openly share personal inormation with riends and
colleagues on social networking serices such as lacebook and LinkedIn, and
their thoughts with the world on platorms like Blogger and 1witter. Looking
back at the past decade, the Internet has become a platorm or the open low
o personal inormation-lows that are largely oluntarily proided by users-
and as such, appear to hae alidated Scott McNealy`s ,in,amous 1999 remark
that \ou hae zero priacy anyway . get oer it.`
3

Notwithstanding McNealy`s iew, priacy has remained a central concern amid
the open inormation lows in our contemporary network society, including
worries about the growing size and role o networked databases,
4
the possibility
o tracking and sureillance by Internet serice proiders
5
and \eb search
engines,
6
priacy threats rom digital rights management technologies,
and
growing concerns about protecting the priacy o users o social networking
sites and related \eb 2.0 serices.
8

\hile scholars continue to detail possible threats to priacy spawned by the last
decade o innoations on the Internet, goernments hae struggled with
whether-and how-to regulate inormation lows across these global networks

3
Polly Sprenger, vv ov Prirac,: Cet Orer t,` \IRLD, March 31, 200,
!""#$%%&&&'&()*+',-.%#-/("(,0%/1&%2*&0%3444%53%36789.
4
SIMSON GARlINKLL, DA1ABASL NA1ION: 1lL DLA1l Ol PRIVAC\ IN 1lL 21S1 CLN1UR\
,2000,.
5
Colin J. Bennett, Cookies, !eb vg., !ebcav. ava Cve Cat.: Patterv. of vrreittavce ov tbe !orta
!iae !eb, 3,3, L1lICS AND INlORMA1ION 1LClNOLOG\ 195, 19-210 ,2001,, Paul Ohm,
1be Ri.e ava att of vra.ire P vrreittavce, 2009 UNIVLRSI1\ Ol ILLINOIS LA\ RLVIL\ 141-
1496.
6
M. Goldberg, 1be Coogtivg of Ovtive Prirac,: Cvait, earcbvgive i.torie., ava tbe êr rovtier
of Protectivg Prirate vforvatiov ov tbe !eb, 9 Lewis & Clark Law Reiew 249-22 ,2005,,
Michael Zimmer, 1be Cae of tbe Perfect earcb vgive: Coogte a. av vfra.trvctvre of Datareittavce
in \LB SLARClING: MUL1IDISCIPLINAR\ PLRSPLC1IVLS -99 ,Amanda Spink & Michael
Zimmer, eds., 2008,.
Julie L. Cohen, . Rigbt to Reaa .vov,vov.t,: . Cto.er oo/ at Co,rigbt Mavagevevt` iv

C,ber.ace, 28,4, CONNLC1ICU1 LA\ RLVIL\ 981-1039 ,1996,, Julie L. Cohen, DRM ava
rirac,, 18 BLRKLLL\ 1LClNOLOG\ LA\ JOURNAL 55-61 ,2003,.
8
Ralph Gross & Alessandro Acquisti, vforvatiov Reretatiov ava Prirac, iv Ovtive ociat êtror/.
,ACM \orkshop on Priacy in the Llectronic Society, Alexandria, VA, 2005,, Michael
Zimmer, 1be tervatitie. of earcb 2.0: 1be vergivg Prirac, 1breat. !bev tbe Drire for tbe Perfect
earcb vgive Meet. !eb 2.0, lIRS1 MONDA\, Mar. 3, 2010,
!""#$%%:()0".-2+1;'-)<%!"=(2%,<(&)1#%=(2%->0%(2+*?'#!#%:.%1)"(,/*%@(*&%A38B%3
4CC, Joseph Bonneau & Soren Preibusch, 1be Prirac, ]vvgte: Ov tbe Mar/et for Data Protectiov iv
ociat êtror/. ,1he Lighth \orkshop on the Lconomics o Inormation Security ,\LIS
2009,,, James Grimmelmann, aceboo/ ava tbe ociat D,vavic. of Prirac,, 95,4, IO\A LA\
RLVIL\ 113 ,2009,, Marc Parry, ibrar, of Covgre.., acivg Prirac, Covcerv., Ctarifie. 1ritter
.rcbire Ptav, 1lL ClRONICLL Ol lIGlLR LDUCA1ION, June 1, 2010,
!""#$%%,!)-2(,/*',-.%=/-<D-0"%E(=)1);F-:FG-2<)*00FH1,(2<%A8939%.

to protect the priacy o their citizens. Gien the diersity o interests, histories,
and cultural contexts, a complicated terrain o trans-national laws and policies
or the protection o priacy and personal data lows across networks has
emerged across the globe. Some jurisdictions hae opted or broad, and
relatiely strict, laws regulating the collection, use and disclosure o personal
inormation, such as Canada`s Personal Inormation Protection and Llectronic
Documents Act ,PIPLDA,
9
or the Luropean Union`s Data Protection
Directie.
10
1he United States, howeer, maintains a more sectoral approach to
priacy legislation, with laws addressing only speciic types o personal
inormation. lor example, the lealth Insurance Portability and Accountability
Act ,lIPAA,
11
oers protection o personal medical inormation, the lair
Credit Reporting Act
12
regulates the collection and low o personal inancial
data, and the Video Priacy Protection Act
13
makes the wrongul disclosure o
ideo rental records illegal.
1he dierences between Canadian,LU approaches to priacy and that o the
United States hae been well documented and analyzed.
14
Put bluntly, the
Canadian,LU regulators can be described as embracing a more paternalist
approach to data protection policy, aiming to presere a undamental human
right o its citizens through preemptie goernmental action. In contrast, the
goernance o priacy in the U.S. typically emerges only ater some
inormational harm has occurred, oten taking the orm o industry sel-
regulation or ery targeted legislation, with the responsibility o initiating
enorcement resting on the harmed data subject hersel. As Dorothee
leisenberg summarizes, In practical terms, the LU and the US reached ery
dierent conclusions about the rights o businesses and indiiduals related to
personal data.`
15
\hile the LU and Canada ocus on direct and preemptie

9
R.S., 1985, c. P-21, !""#$%%/1&0'>I0"(,*'<,',1%*2%DFA3%(2+*?'!"./.
10
Directie 95,46,LC o the Luropean Parliament and o the Council o 24 October 1995
on the protection o indiiduals with regard to the processing o personal data and on the
ree moement o such data, !""#$%%*I)F
/*?'*I)-#1'*I%E*?J)(K*)@%E*?J)(K*)@'+-LI)(MGNENO$83447E55CB$NP$QRSE.
11
lealth Insurance Portability and Accountability Act o 1996, l. Rept. 104-36,
!""#$%%&&&'<#-'<-@%:+0;0%0*1),!%#1<*+*"1(/0'1,"(-2L<)12I/*T+MGUDRF
35C!)#"68BV#1,W1<*T+MGUDRF35C!)#"68B.
12
lair Credit Reporting Act ,lCRA,, 15 U.S.C. 1681 et seq.,
!""#$%%&&&':",'<-@%-0%0"1"I"*0%583AAC:,)1'#+:.
13
Video Priacy Protection Act o 1988, Pub. L. 100-618 ,codiied at 18 U.S.C. 210,,
!""#$%%&&&'/1&',-)2*//'*+I%I0,-+*%39%A635'!"./.
14
ee, e.g., DORO1lLL lLISLNBLRG, NLGO1IA1ING PRIVAC\: 1lL LUROPLAN UNION, 1lL
UNI1LD S1A1LS, AND PLRSONAL DA1A PRO1LC1ION ,2005,, COLIN J. BLNNL11 & ClARLLS
D. RAAB, 1lL GOVLRNANCL Ol PRIVAC\: POLIC\ INS1RUMLN1S IN GLOBAL PLRSPLC1IVL
,2003,.
15
leisenberg, .vra note 14 at 2.

regulation o the collection and use o personal data, prohibiting excess` data
collection and restricting use to the original and stated purposes o the
collection, the U.S, ramework begins with the assumption that most data
collection and use is both acceptable and beneicial, that guidelines should be
primarily oluntary and non-inasie, and that regulation should only address
documented instances o abuse.
1his dierence in regulatory approaches to priacy-and the underpinning
tensions between dierent jurisdictions` iews towards the rights o data
subjects-becomes complicated urther gien the increasing lows o personal
inormation across and between transnational networks, and thus, across
jurisdictions. Internet companies like Google hae customers accessing their
products and serices rom across the globe, with data processing and storage
acilities equally scattered. A Canadian citizen, or example, might be accessing a
Google product in the United States, while the record o the particular
inormation exchange might be stored on a serer in Ireland. Lach jurisdiction
has its own complex set o regulations and rights assigned to the treatment o
any personal inormation shared and stored.
1hese kinds o scenarios hae prompted growing concerns about whether the
global diersity o priacy goernance will result in a race to the bottom`
where corporate interests in processing personal data will migrate to
jurisdictions where there is little or no control oer the circulation and capture
o personal inormation lows, or a race to the top` where the ashioning o
priacy policy to the highest possible standards in order to be perceied as the
best` protector o personal inormation lows. Ater considering the aailable
eidence, political scientists Colin Bennett and Charles Raab hae suggested that
priacy protection is actually improing globally-a trading up` o the
goernance o priacy.
16
Companies are, on the whole, not moing around in
order to aoid strict priacy regulations, such as those deeloped in the LU,
instead, there has been a gradual increase in awareness and action on the issue
o priacy. Lxamples o this trading up` include lacebook`s strengthening o
its priacy policies and practices in reaction to an inestigation by the Oice o
the Priacy Commissioner o Canada, or Google`s modiying its \eb cookie
and partially anonymizing search logs in response to Norwegian priacy
regulators.
1
In each case, large multi-national Internet companies reacted to
strong regional priacy laws in ways that beneited all users across the globe.

16
Bennett & Raab, .vra note 14.
1
Oice o the Priacy Commissioner o Canada, aceboo/ .gree. to .aare.. Prirac,
Covvi..iover`. Covcerv., Aug. 20, 2010, !""#$%%&&&'#)(@'<,',1%.*+(1%2)F,%A554%2)F
,X5459A6X*',:., Nate Anderson, Coogte 1o .vov,vie og. v . ôa 1o Prirac, .arocate.,
ARS 1LClNICA, Aug. 20, 2010, !""#$%%1)0"*,!2(,1',-.%=I0(2*00%2*&0%A556%58%
<--</*F"-F12-2;.(Z*F/-<0F(2F1F2-+F"-F#)(@1,;F1+@-,1"*0'1)0.

Osetting this positie note, howeer, is the realization that priacy protection
may not be trading up` as rapidly as other global actors, such as the extensie,
intensie processing o personal data across borders and platorms, the
increased ocus on economic growth through the use o electronic
communications and inormation inrastructures, and the harmonization o law
enorcement and security objecties. Bennett and Raab go to some length to
expose the limitations o relying solely on indiidual countries to impose
isolated priacy policies in the ace o a globally-networked computer system
permitting-indeed encouraging-transnational inormation lows.
18
\hile
state-speciic data protection goernance might hae been suicient in the past,
they argue, today`s digitally networked society demands that any country`s
eorts to protect its citizens will inescapably be linked with ,as well as
dependent on, the actions and laws o other, oten disparate, jurisdictions.
1his leads to obious problems, when, or example, a legal approach like that o
the United States, with an emphasis on sel-regulation and public-sector
enorcement, meets a dierent philosophy, such as the more top-down,
paternalistic approach to data protection held by Canada and the Luropean
Union. 1his clash between U.S. and non-U.S. standards or goerning personal
inormation lows has prompted large, multi-national companies dependent on
the relatiely unettered low o inormation across global digital networks to
lobby or some middle ground to be reached. In the case o the U.S. and the
Luropean Union, the result was the 2000 Sae larbor agreement
19
between the
two global economic powers to aoid the most egregious misuse o Luropeans`
priate data, while at the same time creating a semi-permanent cease ire` that
would allow transatlantic data ,and hence commerce, to low, despite ailing to
meet the letter, and perhaps not een the intent, o the L.U. Data Protection
Directie. In the end, while U.S. based companies are orced to proide more
priacy protections than U.S. law demands, the Sae larbor proisions are
weaker than the ull Luropean Directie on Data Protection. As leisenberg
explains, the eolution . o the |Luropean Union| Commission`s stance on
data protection seems to hae been one o sotening a bit` during the Sae
larbor negotiations, as the Commission began to accommodate the US as
priacy legislation clashed with irst commercial, and then security concerns.`
20

So, while there has been no clear race to the bottom` in global priacy
protections, the trading up` to an increased leel o protection o personal

18
Bennett & Raab, .vra note 14.
19
2000,520,LC: Commission Decision o 26 July 2000 pursuant to Directie 95,46,LC o
the Luropean Parliament and o the Council on the adequacy o the protection proided by
the sae harbour priacy principles and related requently asked questions issued by the US
Department o Commerce ,notiied under document number C,2000, 2441,, !""#$%%*I)F
/*?'*I)-#1'*I%E*?J)(K*)@%E*?J)(K*)@'+-LI)(MGNENO$8A555[57A5$NP$QRSE.
20
leisenberg, .vra note 14 at 136.

inormation lows on our transnational digital networks has not materialized as
quickly or clearly as one might expect. leisenberg correctly notes that with the
Sae larbor Agreement, the LU was able to orce the U.S. to deal with the
priacy issues that might hae otherwise been ignored, orce some minor
concessions, and show that the LU`s priacy standard was signiicant, granting
the LU something like a irst-moer adantage` in uture trans-border priacy
disputes.
21
\et, beyond isolated examples o Internet companies` hesitant
acquiescence to non-U.S. regulatory bodies-like the lacebook and Google
examples proided aboe-new norms o personal data protection are unlikely
to emerge in the next digital decade, as data protection oicials in Lurope hae
begun to publicly question the appropriateness o the current leels o
protections.
22

Recalling Castells` warning that networks constitute the new social morphology
o our societies, and the diusion o networking logic substantially modiies the
operation and outcomes in processes o production, experience, power and
culture,` we are let to consider the status o priacy protections in the next
digital decade. Our network society will continue to grow in size and density, as
well as in its global importance and interconnectedness. \ithout concerted
eorts to ensure a trading up` in global priacy protections-a renewed
commitment to the rights o data subjects embodied in the Canadian and
Luropean Union approach to data protection-those caught within the
inescapable diusion o networking logic` may hae little control oer how
the increased lows o their personal inormation will modiy experience,
power and culture` oer the next digital decade.

21
a. at 10.
22
\. Scott Blackmer, 1he Inormation Law Group, vroeav Re.erratiov.., Aug. 26, 2010,
!""#$%%&&&'(2:-/1&<)-I#',-.%A535%59%1)"(,/*0%*IF3%*I)-#*12F)*0*)@1"(-20%.

The Privacy Problem:
Whats Wrong with Privacy?
By Stewart Baker
*

\hy are priacy groups so iscerally opposed to goernment action that could
reduce the risks posed by exponential technologies 1he cost o their stance was
made clear on September 11, 2001. 1hat tragedy might not hae occurred i not
or the aggressie priacy and ciil liberties protection imposed by the loreign
Intelligence Sureillance Court and the Department o Justice`s Oice o
Intelligence, and it might hae been aoided i border authorities had been able
to use airline reseration data to screen the hijackers as they entered the United
States.
But een ater 9,11, priacy campaigners tried to rebuild the wall and to keep
the Department o lomeland Security ,DlS, rom using airline reseration
data eectiely. 1hey ailed, too much blood had been spilled.
But in the ields where disaster has not yet struck-computer security and
biotechnology-priacy groups hae blocked the goernment rom taking een
modest steps to head o danger.
I like to think that I care about priacy, too. But I had no sympathy or priacy
crusaders` erocious objection to any new goernment use o technology and
data. \here, I wondered, did their objection come rom
So I looked into the history o priacy crusading. And that`s where I ound the
answer.
The Birth of the Right of Privacy
In the 1880s, Samuel Dennis \arren was near the top o the Boston
aristocracy. le had inished second in his class at larard Law School. le
ounded a law irm with the man who inished just ahead o him, Louis
Brandeis, and they prospered mightily. Brandeis was a brilliant, creatie lawyer
and social reormer who would eentually become a great Supreme Court
justice.
But Samuel Dennis \arren was haunted. 1here was a canker in the rose o his
lie. lis wie was a great hostess, and her parties were careully planned. \hen

K"*&1)" \' ]1W*) is a partner in the \ashington oice o Steptoe & Johnson LLP. le
returned to the irm ollowing 3' years at the Department o lomeland Security as its irst
Assistant Secretary or Policy.

\arren`s cousin married, Mabel \arren held a wedding breakast and illed her
house with lowers or the eent. 1he papers described her home as a eritable
loral bower.`
No one should hae to put up with this. Surely you see the problem. No \ell,
Brandeis did.
le and \arren both thought that, by coering a priate social eent, the
newspapers had reached new heights o impertinence and intrusieness. 1he
parties and guest lists o a Boston Brahmin and his wie were no one`s business
but their own, he thought. And so was born the right to priacy.
Angered by the press coerage o these priate eents, Brandeis and \arren
wrote one o the most requently cited law reiew articles eer published. In
act, 1he Right to Priacy,` which appeared in the 1890 larard Law Reiew,
is more oten cited than read-or good reason, as we`ll see.
1
But a close
reading o the article actually tells us a lot about the modern concept o priacy.
Brandeis,
2
also the ather o the policy-oriented legal brie, begins the article
with a candid exposition o the policy reasons why courts should recognize a
new right to priacy. lis argument is uncompromising:
1he press is oerstepping in eery direction the obious
bounds o propriety and o decency. Gossip is no longer the
resource o the idle and o the icious, but has become a trade,
which is pursued with industry as well as erontery . 1o
occupy the indolent, column upon column is illed with idle
gossip, which can only be procured by intrusion upon the
domestic circle. 1he intensity and complexity o lie, attendant
upon adancing ciilization, hae rendered necessary some
retreat rom the world, and man, under the reining inluence
o culture, has become more sensitie to publicity, so that
solitude and priacy hae become more essential to the
indiidual, but modern enterprise and inention hae, through
inasions upon his priacy, subjected him to mental pain and
distress, ar greater than could be inlicted by mere bodily
injury . Len gossip apparently harmless, when widely and
persistently circulated, is potent or eil . \hen personal
gossip attains the dignity o print, and crowds the space
aailable or matters o real interest to the community, what

1
Samuel \arren & Louis D. Brandeis, 1be Rigbt to Prirac,, 4 lARVARD L. RLV. 193 ,1890,.
2
Because the article owes much o its current ame to Brandeis`s later career, I will rom this
point on discuss only his iews without each time laboriously giing credit, i that is the right
word, to his coauthor.

wonder that the ignorant and thoughtless mistake its relatie
importance . 1riiality destroys at once robustness o
thought and delicacy o eeling.
3

\hat does Brandeis mean by this 1o be brie, he thinks it should be illegal or
the newspapers to publish harmless inormation about himsel and his amily.
1hat, he says, is idle gossip, and it distracts ignorant and thoughtless`
newspaper readers rom more high-minded subjects. It also alicts the reined
and cultured members o society-like, say, Samuel Dennis \arren and his
wie-who need solitude but who are instead harassed by the ruits o modern
enterprise and inention.`
\hat`s remarkable about 1he Right to Priacy` is that the article`s title still
inokes reerence, een though its substance is, well, laughable.
Is there anyone alie who thinks it should be illegal or the media to reeal the
guest-list at a prominent socialite`s dinner party or to describe how elaborate the
loral arrangements were 1oday, it`s more likely that the hostess o a prominent
dinner party will blog it in adance, and that the guests will send 1witter updates
while it`s under way. lor most socialites, what would really hurt is a lack o
media coerage. 1o be blunt, when he complains so bitterly about media
interest in a dinner party, Brandeis sounds to modern ears like a wuss.
Lqually peculiar is the suggestion that we should keep such inormation rom
the inerior classes lest they abandon sel-improement and wallow instead in
gossip about their betters. 1hat makes Brandeis sound like a wuss and a snob.
le does sound quite up-to-date when he complains that modern enterprise
and inention` are inading our solitude. 1hat is a amiliar complaint. It`s what
priacy adocates are saying today about Google, not to mention the National
Security Agency ,NSA,. Until you realize that he`s complaining about the
scourge o instantaneous photographs and newspaper enterprise.`
4
luh
Brandeis eidently thinks that publishing a priate citizen`s photo in the
newspaper causes mental pain and distress, ar greater than could be inlicted
by mere bodily injury.`
5

I we agreed today, o course, we probably wouldn`t hae posted 5 billion
photographs o ourseles and our riends on llickr.
6

3
1be Rigbt to Prirac,, at 196 ,1890,.
4
a. at 195.
5
a. at 196.
6
Zack Sheppard, :,000,000,000, llickr Blog, Sept. 19, 2010,
!""#$%%=/-<':/(,W)'2*"%*2%A535%54%34%7555555555%.

Spirit of the Privacy Movement Today
Anachronistic as it seems, the spirit o Brandeis`s article is still the spirit o the
priacy moement. 1he right to priacy was born as a reactionary deense o the
status quo, and so it remains. 1hen, as now, new technology suddenly made it
possible to spread inormation more cheaply and more easily. 1his was new,
and uncomortable. But apart rom a howl o pain-pain ar greater than .
mere bodily injury`-Brandeis doesn`t tell us why it`s so bad. I guess you had to
be there-literally. Unless you were an adult when photography came to
newspapers, you`ll probably neer really understand what the uss was about.
\eè all been photographed, and most o us aren`t happy with the results, at
least not all the time. But that`s lie, and weè learned to lie with it. Most o us
can`t imagine suing to preent the distribution o our photographs-which was
the tort Brandeis wanted the courts to create.
\e should not mock Brandeis too harshly. lis article clearly coneys a heartelt
sense o inasion. But it is a sense o inasion we can neer share. 1he
sensitiity about being photographed or mentioned in the newspapers, a raw
spot that rubbed Brandeis so painully, has calloused oer. So thick is the
callous that most o us would be tickled, not appalled, to hae our dinner parties
make the local paper, and especially so i it included our photos.
And that`s the second thing that Brandeis`s article can tell us about more
contemporary priacy laps. lis brand o resistance to change is still alie and
well in priacy circles, een i the targets hae been updated. Lach new priacy
kerule inspires strong eelings precisely because we are reacting against the
eects o a new technology. \et as time goes on, the new technology becomes
commonplace. Our reaction dwindles away. 1he raw spot grows a callous. And
once the initial reaction has passed, so does the sense that our priacy has been
inaded. In short, we get used to it.
At the beginning, o course, we don`t want to get used to it. \e want to keep
on liing the way we did beore, except with a ew more amenities. And so, like
Brandeis, we are tempted to ask the law to stop the changes we see coming.
1here`s nothing more natural, or more reactionary, than that.
Most priacy adocates don`t see themseles as reactionaries or adocates or
the status quo, o course. Right and let, they cast themseles as underdogs
battling or change against the entrenched orces o big goernment. But
irtually all o their actiism is actually deoted to stopping change-keeping
the goernment ,and sometimes industry, rom taking adantage o new
technology to process and use inormation.
But simply opposing change, especially technological change, is a losing battle.
At heart, the priacy groups know it, which may explain some o their shrillness
and lack o perspectie. Inormation really does want to be ree`-or at least

cheap. And the spread o cheap inormation about all o us will change our
relationship to the world. \e will hae ewer secrets. Crippling goernment by
preenting it rom using inormation that eeryone else can get will not gie us
back our secrets.
In the 190s, well beore the personal computer and the Internet, priacy
campaigners persuaded the country that the lBI`s newspaper clipping iles
about U.S. citizens were a threat to priacy. Sure, the inormation was public,
they acknowledged, but gathering it all in one ile was iewed as aguely sinister.
1he attorney general banned the practice in the absence o some legal reason
or doing so, usually called an inestigatie predicate.`
So, in 2001, when Google had made it possible or anyone to assemble a clips
ile about anyone in seconds, the one institution in the country that could not
print out the results o its Google searches about Americans was the lBI. 1his
was bad or our security, and it didn`t protect anyone`s priacy either.
1he priacy campaigners are ighting the ineitable. 1he permanent record`
our high school principals threatened us with is already here-in lacebook.
Anonymity, its thrills and its reedom, has been characteristic o big cities or
centuries. But anonymity will also grow scarce as data becomes easier and easier
to gather and correlate. \e will lose something as a result, no question about it.
1he priacy groups` response is prooundly conseratie in the \illiam l.
Buckley sense-standing athwart history yelling, Stop!`

I`m all or conseratism, een in unlikely quarters. But using laws to ight the
ineitable looks a lot like Prohibition. Prohibition was put in place by an Anglo-
Saxon Protestant majority that was sure o its moral superiority but not o its
uture. \hat the priacy community wants is a kind o data Prohibition or
goernment, while the rest o us get to spend more and more time in the corner
bar.
1hat might work i goernments didn`t need the data or important goals such
as preenting terrorists rom entering the country. Ater September 11, though,
we can no longer aord the orced ineiciency o denying modern inormation
technology to goernment. In the long run, any eectie method o ensuring
priacy is going to hae to ocus on using technology in a smart way, not just
trying to make goernment slow and stupid.

ee \illiam l. Buckley Jr., Pvbti.ber`. tatevevt, NA1IONAL RLVIL\, No. 19, 1955, at 5,
araitabte at &&&'21"(-21/)*@(*&',-.%1)"(,/*0%AA87C4%-I)F.(00(-2F0"1"*.*2"%&(//(1.F
:F=I,W/*;F>).

The Evolution of Technology
& the Zone of Privacy
1hat doesn`t mean we hae to gie up all priacy protection. It just means that
we hae to look or protections that work with technology instead o against it.
\e can`t stop technology rom making inormation cheap and reducing
anonymity, but we can deploy that same technology to make sure that
goernment oicials can`t misuse data and hide their tracks. 1his new priacy
model is partially procedural-greater oersight and transparency-and partly
substantie-protecting indiiduals rom actual aderse consequences rather
than hypothetical inormational injuries.
Under this approach, the irst people who should lose their priacy are the
goernment workers with access to personal data. 1hey should be subject to
audit, to challenge, and to punishment i they use the data or improper
purposes. 1hat`s an approach that works with emerging technology to build the
world we want to lie in. In contrast, it is simple Luddism to keep goernment
rom doing with inormation technology what eery other part o society can
do.
1he problem is that Luddism always has appeal. Change is bad` is a slogan
that has neer lacked or adherents, and priacy adocates sounded alarm ater
alarm with that slogan as the backdrop when we tried to put in place a data-
based border screening system.
But would we really thank our ancestors i they`d taken the substance o
Brandeis`s article as seriously as its title I, without a legislature eer
considering the question, judges had declared that no one could publish true
acts about a man`s nonpolitical lie, or een his photograph, without his
permission
I don`t think so. 1hings change. Americans grow less priate about their sex
lies but more priate about inancial matters. 1oday, ew o us are willing to
hae strangers liing in our homes, listening to our amily conersations, and
then gossiping about us oer the back ence with the strangers who lie in our
riends` homes. \et I`ll bet that both Brandeis and \arren tolerated without a
second thought the limits that haing serants put on their priacy.
\hy does our concept o priacy ary rom time to time lere`s one theory:
Priacy is allied with shame. \e are all ashamed o something about ourseles,
something we would preer that no one, or just a ew people, know about. \e
want to keep it priate. Sometimes, o course, we should be ashamed. Criminals
always want priacy or their acts. But we`re also ashamed-or at least eel
embarrassment, the irst cousin o shame-about a lot o things that aren`t
crimes.

\e may be ashamed o our bodies, at least until we`re sure we won`t be mocked
or our physical shortcomings. Priacy is similar, we are oten quite willing to
share inormation about ourseles, including what we look like without our
clothes, when we trust our audience, or when the context makes us beliee that
our shortcomings will go unnoticed. Most o us would rather be naked with our
spouse than a random stranger. And we would not appear at the oice in our
underwear, een i it coers more than the bathing suit we wore at the beach on
the weekend.
lor that reason, enorced nudity oten eels like a proound inasion o our
priacy. At least at irst. In act, though, we can get used to it pretty quickly, as
anyone who has played high school sports or sered in the army can attest.
1hat`s because the ear o mockery is usually worse than the experience. So
when we discoer that being naked in a crowd o other naked people doesn`t
lead to mockery and shame, we begin to adapt. \e deelop a callous where we
once were tender.
1he things that Brandeis considered priacy inasions are similar. Very ew o
us are happy the irst time we see our photograph or an interiew in the
newspaper. But pretty soon we realize it`s just not that big a deal. Our nose and
our style o speech are things that the people we know hae already accepted,
and no one else cares enough to embarrass us about them. 1he same is true
when we Google ourseles and see that a bad reiew o our dinner-theater
perormance is number three on the list. Our irst reaction is embarrassment
and unhappiness, but the reaction is oddly eanescent.
I this is so, then the zone o priacy` is going to ary rom time to time and
place to place-just as our concept o physical modesty does. 1he zone o
priacy has boundaries on two sides. \e don`t care about some inormation
that might be reealed about us, probably because the reelation causes us no
harm-or weè gotten used to it. I the inormation is still embarrassing, we
want to keep it priate, and society may agree. But we can`t expect priacy or
inormation that society iews as truly shameul or criminal.
Oer time, inormation will moe into and out o the zone o priacy on both
sides. Some inormation will simply become so unthreatening that we`ll laugh at
the idea that it is part o the priacy zone. Photographs long ago entered that
category, despite Brandeis`s campaigning. Some inormation will moe rom
criminal eidence into the zone o priacy, as sexual preerence has. Conersely,
it may moe in the other direction: inormation that a man beats his wie is no
longer protected by a zone o amilial priacy, as it once was, now it`s iewed as
eidence o a crime.
1he biggest priacy battles will oten be in circumstances where the rules are
changing. 1he subtext o many Internet priacy ights, or example, is whether
some new measure will expose the identities o people who download

pornography or copyrighted music and moies. Society is diided about how
shameul it is to download these items, and it displaces that moral and legal
debate into a ight about priacy.
Diorce litigation, or instance, is brutal in part because inormation shared in a
context o loe and conidence ends up being disclosed to the world in a
deliberately harmul way. Oten the actiity in question ,like making a telephone
call or a credit card purchase, is something that the indiidual does reely, with
clear knowledge that some other people ,his bank or his phone company, know
what he is doing. Sometimes the actiities are proudly public in nature-
protests against goernment policy, or example.
In those cases, the priacy concern is not that the bank or the phone company
,or our spouse, actually has the inormation, but rather what they will do with
the inormation they hae-whether they will use the data in ways we didn`t
expect or gie the data to someone who can harm us. \e want to make sure the
data will not be used to harm us in unexpected ways.
And that helps explain why priacy adocates are so oten Luddite in
inclination. Modern technology keeps changing the ways in which inormation
is used. Once, we could count on practical obscurity-the diiculty o inding
bits o data rom our past-to protect us rom unexpected disclosures. Now,
storage costs are irtually nil, and processing power is increasing exponentially.
It is no longer possible to assume that your data, een though technically public,
will neer actually be used. It is dirt cheap or data processors to compile
dossiers on indiiduals, and to use the data in ways we didn`t expect.
Some would argue that this isn`t really priacy` so much as a concern about
abuse o inormation. loweer it`s deined, though, the real question is what
kind o protection is it reasonable or us to expect. Can we really write a
detailed legislatie or contractual pre-nup or each disclosure, setting orth
exactly how our data will be used beore we hand it oer I doubt it. Maybe we
can orbid obious misuses, but the more detailed we try to get, the more we
run into the problem that our notions o what is priate, and indeed o what is
embarrassing, are certain to change oer time. I so, does it make sense to reeze
today`s priacy preerences into law
In act, that`s the mistake that Brandeis made-and the last lesson we can learn
rom the odd mix o eneration and guawing that his article prookes.
Brandeis wanted to extend common law copyright until it coered eerything
that can be recorded about an indiidual. 1he purpose was to protect the
indiidual rom all the new technologies and businesses that had suddenly made
it easy to gather and disseminate personal inormation: the too enterprising

press, the photographer, or the possessor o any other modern deice or
rewording or reproducing scenes or sounds.`
8

1his proposal is wacky in two ways. lirst it tries to reeze in 1890 our sense o
what is priate and what is not. Second, it tries to dey the graitational orce o
technology.
Lery year, inormation gets cheaper to store and to duplicate. Computers,
iPods, and the Internet are all modern deices` or reproducing scenes or
sounds,` which means that any eort to control reproduction o pictures,
sounds, and scenes becomes extraordinarily diicult i not impossible. In act, it
can`t be done.
1here is a deep irony here. Brandeis thought that the way to ensure the strength
o his new right to priacy was to enorce it just like state copyright law. I you
don`t like the way your` priate inormation is distributed, you can sue
eeryone who publishes it. One hundred years later, the owners o ederal
statutory copyrights in popular music and moies ollowed this prescription to a
1. 1hey began to use litigation to protect their data rights against the
possessor|s| o any other modern deice or . reproducing scenes or
sounds,`
9
a class that now included many o their customers. 1he Recording
Industry Association o America ,RIAA, sued consumers by the tens o
thousands or using their deices to copy and distribute songs.
Unwittingly, the RIAA gae a thorough test to Brandeis`s notion that the law
could simply stand in ront o new technology and bring it to a halt through
litigation. 1here aren`t a lot o people who think that that has worked out well
or the RIAA`s members, or or their rights.
Brandeis wanted to protect priacy by outlawing the use o a common new
technology to distribute priate` acts. lis approach has ared no better than
the RIAA`s. Inormation that is easy to gather, copy and distribute will be
gathered, copied, and distributed, no matter what the law says.
It may seem a little bit odd or me to criticize Brandeis and other priacy
campaigners or resisting the spread o technology. Ater all, we can`t simply
accept the world that technology and commerce sere up.
It`s one thing to redirect the path o technological change by a ew degrees. It`s
another to insist that it take a right angle. Brandeis wanted it to take a right

8
\arren & Brandeis, .vra note 1 at 206.
9
a.

angle, he wanted to dey the changes that technology was pressing upon him. So
did the RIAA.
Both were embracing a kind o Luddism-a reactionary spasm in the ace o
technological change. 1hey were doomed to ail. 1he new technologies, ater all,
empowered ordinary citizens and consumers in ways that could not be resisted.
I the law tries to keep people rom enjoying the new technologies, in the end it
is the law that will suer.
But just because technologies are irresistible does not mean that they cannot be
guided, or cannot hae their worst eects oset by other technologies. 1he
solutions I`m adocating will only work i they allow the world to keep
practically all the beneits o the exponential empowerment that new technology
makes possible.
Privacy for the Real World:
Proposed Solutions
So what`s my solution to the tension between inormation technology and our
current sense o priacy 1he short answer is that we should protect priacy,
but not by deying the course o technology or by crippling goernment when it
inestigates crimes. \e can do it by working ritb technology, not against it. In
particular, we can use inormation technology to make sure that goernment
oicials lose their priacy when they misuse data that has been gathered or
legitimate reasons. Inormation technology now makes it easier to track eery
database search made by eery user, and then to ollow any distribution o that
data outside the system. In other words, it can make misuse o the data in
goernment iles much more diicult and much more dangerous.
But beore talking about what vigbt work, let`s take a closer look at some o the
ideas that don`t.
Ownership of Personal Data
1he irst priacy solution is one weè already seen. It`s the Brandeisian notion
that we should all own` our personal data. 1hat has some appeal, o course. I
I hae a secret, it eels a lot like property. I can choose to keep it to mysel, or I
can share it with a ew people whom I trust. And I would like to beliee that
sharing a secret with a ew trusted riends doesn`t turn it into public property.
It`s like my home. Just because Iè inited one guest home doesn`t mean the
public is welcome.
But in the end, inormation is not really like property. Property can only be held
by one person at a time, or at most by a ew people. But inormation can be
shared and kept at the same time. And those with whom it is shared can pass it
on to others at little or no cost. I you eer told a riend about your secret crush

in junior high, youè already learned that inormation cannot be controlled like
property. As Ben lranklin is credited with saying, 1hree may keep a secret i
two o them are dead.`
10
1he redistribution o inormation cannot be easily
controlled in the best o times, and Moore`s Law is making the control o
inormation nearly impossible.
11

1he recording and moie industries discoered the same thing. I these
industries with their enormous lobbying and litigation budgets cannot control
inormation that they own as a matter o law, the rest o us are unlikely to be
able to control inormation about ourseles. Gossip is not going to become
illegal simply because technology ampliies it.
1hat`s why Brandeis`s proposal neer really got o the ground, at least not as he
enisioned it. Buoyed by Brandeis`s prestige, the idea that priate acts are
priate property lingered on in the courts or years, but what suried o his
proposal is scarcely recognizable today.
In act, so transormed is Brandeis`s priacy doctrine that it is now described,
accurately, as a right o publicity,` which surely would hae him turning in his
grae. Currently, most states honor Brandeis by allowing lawsuits or
unauthorized commercial use o a person`s likeness, either by statute or judge-
made law.
Oer time, courts lost sight o Brandeis`s purpose. 1hey began to take the
analogy to property literally. Brandeis wanted to treat priate inormation like
property because that was the only way to gie a remedy or the mental pain
and distress, ar greater than could be inlicted by mere bodily injury,` that he
thought a man suered when his photo was published without permission. But
as people got used to haing their pictures taken, the mental pain and distress
slowly drained out o the experience.
All that was let was the property analogy. And so judges began shrinking the
right until it only had bite in the one set o circumstances where the right to
control one`s image actually eels like a property right-when the image is
worth real bucks. 1hus, the courts require disgorgement o proits made when a
celebrity`s name, ace, oice, or een personal style is used without permission
to sell or endorse products. As a result, the right to exploit a celebrity`s image
really is property today, it can be sold, transerred, and een inherited.

10
Benjamin lranklin, POOR RIClARD`S ALMANAC, July 135.
11
Moore`s Law describes the long-term trend that the number o transistors that can be placed
inexpensiely on an integrated circuit has doubled approximately eery two years. It is named
ater Intel`s co-ounder Gordon L. Moore, who described the trend in the essay Cravvivg
More Covovevt. Ovto vtegratea Circvit., LLLC1RONICS MAGAZINL 4, 1965.

1here`s only one problem with this eort to turn priacy into property: it hasn`t
done much or priacy. It simply protects the right o celebrities to make money
o their ame. In act, by monetizing things like celebrity images, it rewards
those who hae most relentlessly sacriiced their priacy to gain ame.
1he right o publicity is well named. It is the right to put your priacy up or
sale. Not surprisingly, a lot o people hae been inspired to do just that.
Ironically, Brandeis`s doctrine has helped to destroy the essence o what he
hoped to presere.
Oh, and in the process, Brandeis`s approach has stiled creatiity and restricted
ree speech-muzzling artists, social commentators, and businesspeople who
want to make creatie use o images that are an essential part o our cultural
enironment. It`s a disaster. Slowly, courts are waking up to the irony and
limiting the right o publicity.
1he same priate inormation as property` approach has also made a modest
appearance in some consumer priacy laws, and it`s worked out just as badly. At
bottom, consumer priacy protection laws like the Right to linancial Priacy
Act
12
treat a consumer`s data like a consumer`s money: \ou can gie your data
,or your money, to a company in exchange or some beneit, but only i youè
been told the terms o the transaction and hae consented. Similarly, the Cable
Communications Policy Act o 1984 preents cable proiders rom using or
releasing personal inormation in most cases unless the proiders get the
customer`s consent. 1he ruit o this approach is clear to anyone with a bank
account or an Internet connection. Lerywhere you turn, you`re conronted
with inormed consent` and terms o serice` disclosures, these are uniormly
impenetrable and non-negotiable. No one reads them beore clicking the box,
so the consent` is more iction than reality, certainly it does little to protect
priacy. Indeed, it`s turning out a lot like the right o publicity. By treating
priacy as property, consumer priacy protection law inites all o us to sell our
priacy.
And we do. Only or most o us, the going price turns out to be disconcertingly
cheap.
Mandatory Predicates for Information Access
1he second way o protecting priacy is to require what`s called a predicate`
or access to inormation. 1hat`s a name only a lawyer could loe. In act, the
whole concept is one that only lawyers loe.

12
1he Right to linancial Priacy Act o 198, Pub. L. No. 95-630, 92 Stat. 3695 ,198,
,codiied as amended at 12 U.S.C. 3401 et .eq.,.

Simply put, the notion is that goernment shouldn`t get certain priate
inormation unless it satisies a threshold requirement-a predicate` or access
to the data. Lawyers hae played a huge role in shaping American thinking
about priacy, and the predicate approach has been widely adopted as a priacy
protection. But its alue or that purpose is quite doubtul.
1he predicate approach to priacy can be traced to the lourth Amendment,
which guarantees that no \arrants shall issue, but upon probable cause.`
1ranslated rom legalese, this means that the goernment may not search your
home unless it has a good reason to do so. \hen the goernment asks or a
search warrant, it must show the judge probable cause`-eidence that the
search will likely turn up criminal eidence or contraband. Probable cause is the
predicate or the search.
\hen a lap arose in the 190s oer the lBI practice o assembling domestic
security dossiers on Americans who had not broken the law, the attorney
general stepped in to protect their priacy. le issued new guidelines or the
lBI. le was a lawyer, so he declared that the lBI could not do domestic
security inestigations o Americans without a predicate.
1he predicate wasn`t probable cause, that was too high a standard. Instead, the
attorney general allowed the launching o a domestic security inestigation only
i the bureau presented speciic and articulable acts giing reason to beliee`
that the subject o the inestigation may be inoled in iolence.
13

Actually, the story o the lBI guidelines shows why the predicate approach
oten ails. 1he dossiers being assembled by the lBI were oten just clippings
and other public inormation. 1hey usually weren`t the product o a search in
the classic sense, no ederal agents had entered priate property to obtain the
inormation. Nonetheless, the lBI guidelines treated the gathering o the
inormation itsel as though it were a kind o search.
In so doing, the guidelines were ollowing in Brandeis`s ootsteps-treating
inormation as though it were physical property. 1he collection o the
inormation was equated to a physical intrusion into the home or oice o the
indiidual. Implicitly, it assumes that data can be locked up like property.
But that analogy has already ailed. It ailed or Brandeis and it ailed or the
RIAA. It ailed or the lBI guidelines, too. As clippings became easier to

13
1he Right to linancial Priacy Act o 198, Pub. L. No. 95-630, 92 Stat. 3695 ,198,
,codiied at 12 U.S.C. 3414,a,,5,,A,,, amended by the Uniting and Strengthening America
by Proiding Appropriate 1ools Required to Intercept and Obstruct 1errorism ,USA
PA1RIO1, Act o 2001, Pub. L. 10-56, 115 S1A1. 22, 505,b,,
&&&'/1&',-)2*//'*+I%I0,-+*%#+:%I0,-+*3A%/((XI0,XRTX3AXGQX87XKNX8C53'#+:.

retriee, clippings iles became easier to assemble. 1hen Google made it
possible or anyone to assemble an electronic clips ile on anyone. 1here was
nothing secret about the clippings then. 1hey were about as priate as a bus
terminal.
But the law was stuck in another era. Under the guidelines, only the lBI and
CIA needed a predicate to do Google searches. \ou hae to be a pretty resilient
society to decide that you want to deny to your law enorcement agencies a tool
that is reely aailable to nine-year-old girls and terrorist gangs. Resilient, but
stupid. ,Not surprisingly, the guidelines were reised ater 9,11.,
1hat`s one reason we shouldn`t treat the assembling o data as though it were a
search o physical property. As technology makes it easier and easier to collect
data, the analogy between doing that and conducting a search o a truly priate
space will become less and less persuasie. No one thinks goernment agencies
should hae a predicate to use the \hite Pages. Soon, predicates that keep law
enorcement rom collecting inormation in other ways will become equally
anachronistic, leaing law enorcement stuck in the 1950s while eeryone else
gets to lie in the twenty-irst century.
I saw this lawyerly ainity or predicates up close at DlS. 1he issue was laptop
searches at the border. 1he goernment has always had the right to search
anything crossing the border without probable cause. Smugglers are smart and
highly motiated, they would ind a way to exploit any limitations on the
authority to conduct searches. 1he irst Congress knew that quite well, and in
189, two months beore it sent the lourth Amendment to the states or
approal, Congress gae the customs serice ull power and authority` to
search any ship or essel, in which they shall hae reason to suspect any goods,
wares or merchandise subject to duty shall be concealed.`
14

Obiously, DlS and its border predecessors didn`t search laptops in 189. But
they aia search books, papers, correspondence, and anything else that could
store inormation. 1hat was the law or two hundred years, with one exception.
1he Supreme Court has ruled that a ew extraordinarily intrusie techniques-
body caity searches and orced x-rays-require a reasonable suspicion.`
15

Laptops are treated like books and papers. 1hey are searched wheneer border
oicials think that such a search is likely to be productie. Len the amously

14
An Act to Regulate the Collection o Duties, 1st Cong. 1st Sess., Stat.1, Ch. V, Sec. 24 at 43
,July 31, 189,.
15
U.S. . llores-Montano, 541 U.S. 149 ,2004,.

liberal Ninth Circuit, the court o appeal that includes Caliornia, has had no
trouble approing that practice,
16
and or good reason-laptop searches pay o.
In 2006, or example, border oicials at the Minneapolis-St. Paul airport
reerred a suspect traeler to secondary inspection. 1here they ound that his
computer contained ideo clips o ILDs being used to kill soldiers and destroy
ehicles and a ideo on martyrdom. le was also carrying a manual on how to
make improised explosie deices, or ILDs-a weapon o choice or terrorists
in Aghanistan and Iraq.
Despite two hundred years o history and precedent, as well as the proen alue
o searching electronic media, priacy groups launched a campaign against
laptop searches toward the end o the Bush administration. 1his was a strange
and unhappy era in the debate oer priacy. By 2005, priacy adocates had
ound a growing audience or claims that the Bush administration had
abandoned all limits in pursuing terrorism-that it had swung the pendulum
iolently away rom priacy and in aor o goernment authority.
1he priacy adocates` solution to the laptop issue was the lawyer`s aorite-a
predicate requirement. Laptops should not be searched at the border, they
argued, unless the border oicial could articulate some speciic reason or
conducting the search. 1hat argument was rejected by both the Bush and the
Obama administrations ater careul consideration.
\e rejected it or two reasons. It wouldn`t hae protected priacy in any
meaningul way, and it would hae helped criminals like pedophiles and
terrorists deeat our border deenses. Other than that, it was jim-dandy.
\hy wouldn`t it help protect priacy Because, as a practical matter, no border
oicial today searches a laptop without some reasonable suspicion about the
traeler. 1he exponential increase in commercial jet trael and the unorgiing
thirty-second rule mean that only one traeler in two hundred is sent to
secondary inspection or a closer look. Once there, many traelers quickly
satisy the oicials that they don`t desere more detailed inspection.
Leryone at the border is busy, border oicers don`t hae the luxury o
hooking up the laptops o random traelers or inspection without a good
reason. Oicers who waste their time and DlS`s resources that way are going
to hear rom their superisors long beore they hear rom the traelers` lawyers.
I border oicials only search laptops today when they hae a good reason to
do so, why not make that a requirement \hat harm can it do to make
reasonable suspicion a predicate or laptop searches at the border Plenty.

16
U.S. . Arnold, 523 l.3d 941 ,9th Cir. 2008,.

Requiring reasonable suspicion beore a laptop search will open eery border
search to litigation. And in court, it may be hard to justiy een some ery
reasonable judgments.
Ineitably, enorcement o a predicate requirement or border searches will
produce litigation. 1he litigation will ocus on the moties o the border
oicials. 1he courts will tell those oicials that some reasons are not good
enough. Deense lawyers will want to see the personnel records o border
oicials, hoping to show that theyè inspected a disproportionate number o
laptops belonging to minorities, or to Saudis, or to men, or any other pattern
that might get the case thrown out. Border oicials will hae to start keeping
detailed records justiying each laptop search. New paperwork and new
procedures will clog the inspection process, backing up traelers and penalizing
any inspector who searches a laptop.
\ait a minute, you might ask, what i those oicials are racists or sexists
Let`s assume that this concern is legitimate, at least sometimes, and that there
are biased oicials at work on the border. Surely there`s a better way to ind
them and get them o the job than to count on criminal deense lawyers
exposing them on the witness stand years ater the eent.
By now, notice, we`re not een talking about priacy anymore. 1he predicate`
solution has, in eect, changed the subject. \e`re talking about the moties o
border oicials, or ethnic proiling, or something-but it isn`t priacy. \e`re
also moing the whole discussion into territory that lawyers ind comortable
but that ordinary people might question.
1he lourth Amendment approach to priacy assumes that priacy is best
protected by letting criminals challenge the search that produced the eidence
against them, but beore adopting that solution, we ought to be pretty sure that
we`re going to get beneits that match the cost o letting guilty deendants go
ree, something that isn`t obious here.
Limits on Information Use
1hat leaes the third approach to priacy, one weè already seen in action. I
requiring a predicate is the lawyer`s solution, this third approach is the
bureaucrat`s solution. It is at heart the approach adopted by the Luropean
Union: Instead o putting limits on when inormation may be collected, it sets
limits on how the inormation is used.
1he Luropean Union`s data protection principles coer a lot o ground, but
their uniying theme is imposing limits on how priate data is used. Under those
principles, personal data may only be used in ways that are consistent with the

purposes or which the data were gathered. Any data that is retained must be
releant to the original purposes and must be stored securely to preent misuse.
1he LU`s negotiating position in the passenger name records conlict was
largely deried rom this set o principles. 1he principles also explain Lurope`s
enthusiasm or a wall between law enorcement and intelligence. I DlS
gathered reseration data or the purpose o screening traelers when they cross
the border, why should any other agency be gien access to the data 1his also
explains the LU`s insistence on short deadlines or the destruction o PNR data.
Once it had been used to screen passengers, it had sered the purpose or which
it was gathered and should be promptly discarded.
1here is a core o sense in this solution. It ocuses mainly on the consequences
o collecting inormation, and not on the act o collection. It doesn`t try to insist
that inormation is property. It recognizes that when we gie inormation to
others, we usually hae an expectation about how it will be used, and as long as
the use its our expectations, we aren`t too ussy about who exactly gets to see
it. By concentrating on how personal inormation is used, this solution may get
closer to the core o priacy than one that ocuses on how personal inormation
is collected.
It has another adantage, too. In the case o goernment databases, ocusing on
use also allows us to acknowledge the oerriding importance o some
goernment data systems while still protecting against petty uses o highly
personal inormation.
Call it the deadbeat-dad problem, or call it mission creep, but there`s an
uncomortable pattern to the use o data by goernments. Oten, personal data
must be gathered or a pressing reason-the preention o crime or terrorism,
perhaps, or the administration o a social security system. 1hen, as time goes on,
it becomes attractie to use the data or other, less pressing purposes-
collecting child support, perhaps, or enorcing parking tickets. No one would
support the gathering o a large personal database simply to collect unpaid
parking ines, but mission creep` can easily carry the database well beyond its
original purpose. A limitation on use preents mission creep, or at least orces a
debate about each step in the expansion.
1hat`s all ine. But in the end, this solution is also lawed.
It, too, is ighting technology, though less obiously than the predicate and
property approaches. Data that has already been gathered is easier to use or
other purposes. It`s oolish to pretend otherwise. Indeed, deelopments in
inormation technology in recent years hae produced real strides in searching
unstructured data or in inding relationships in data without knowing or sure
that the data will actually produce anything useul. In short, there are now good

reasons to collate data gathered or widely diering purposes, just to see the
patterns that emerge.
1his new technical capability is hard to square with use limitations or with early
destruction o data. lor i collating data in the goernment`s hands could hae
preented a successul terrorist attack, no one will congratulate the agency that
reused to allow the collation because the data was collected or tax or
regulatory purposes, say, and not to catch terrorists.
\hat`s more, use limitations hae caused great harm when applied too
aggressiely. 1he notorious wall` between law enorcement and intelligence
was at heart a use limitation. It assumed that law enorcement agencies would
gather inormation using their authority, and then would use the inormation
only or law enorcement purposes. Intelligence agencies would do the same.
Or so the theory went. But strict enorcement o this use limitation was
unimaginably costly. In August 2001, two terrorists were known to hae
entered the United States. As the search or them began, the goernment`s top
priority was enorcing the wall -- keeping intelligence about the terrorists rom
being used by the wrong` part o the lBI. Goernment lawyers insisted that
law enorcement resources could not be used to pursue intelligence that two
known al Qaeda agents were in the United States in August 2001.
1his was a atal blunder. 1he criminal inestigators were well-resourced and
eager. 1hey might hae ound the men. 1he intelligence inestigators, in
contrast, had ew resources and did not locate the terrorists, at least not until
September 11, when the terrorists` names were discoered on the maniests o
the hijacked planes. It was a high price to pay or the modest comort o use`
limitations.
Like all use limitations, the wall` between law enorcement sounded reasonable
enough in the abstract. \hile no one could point to a real priacy abuse arising
rom cooperation between the intelligence and law enorcement agencies in the
United States, it was easy to point to the Gestapo and other totalitarian
organizations where there had been too much cooperation among agencies.
\hat was the harm in a little organizational insurance against misuse o
personal data, the argument ran. 1he rules allowed cooperation where that was
strictly necessary, and we could count on the agencies to crowd right up to the
line in doing their jobs. Or so we thought. In act, we couldn`t. As the pressure
and the risk ratcheted up, agents were discouraged rom pushing or greater
communication and cooperation across the wall. All the \ashington-wise knew
that the way to bureaucratic glory and a good press lay in deending priacy.
Actually, more to the point, they knew that bad press and bureaucratic disgrace
were the likely result i your actions could be characterized as hurting priacy.
Congress would hold hearings, appropriators would zero out your oice, the
second-guessing arms o the Justice Department, rom the inspectors general to

the Oice o Proessional Responsibility, would east on eery detail o your
misstep. So, what might hae been a sensible, modest use restriction preenting
the dissemination o inormation without a good reason became an
impermeable barrier.
1hat`s why the bureaucratic system or protecting priacy so oten ails. 1he use
restrictions and related limits are abstract. 1hey make a kind o modest sense,
but i they are enorced too strictly, they preent new uses o inormation that
may be critically important.
And oten they are enorced too strictly. \ou don`t hae to tell a bureaucrat
twice to withhold inormation rom a rial agency. Lawsuits, bad press, and
Congressional inestigations all seem to push against a lexible reading o the
rules. I a use or inormation is not identiied at the outset, it can be nearly
impossible to add the use later, no matter how sensible the change may seem.
1his leads agencies to try to drat broad uses or the data they collect, which
deeats the original point o setting use restrictions.
It`s like wearing someone else`s dress. Oer time, use restrictions end up tight
where they should be roomy-and loose where they should be tight. No one is
let satisied.
The Audit Approach: Enforced Accountability
So what will work Simple: accountability, especially electronically-enorced
accountability.
1he best way to understand this solution is to begin with Barack Obama`s
passport records-and with Joe the Plumber.` 1hese were two minor laps
that punctuated the 2008 presidential campaign. But both tell us something
about how priacy is really protected these days.
In March o 2008, Barack Obama and lillary Clinton were dueling across the
country in weekly primary showdowns. Suddenly, the campaign took an odd
turn. 1he Bush administration`s State Department announced that it had ired
or disciplined seeral contractors or examining Obama`s passport records.
Democrats erupted. It wasn`t hard to jump to the conclusion that the
candidate`s iles had been searched or partisan purposes.
1
Ater an
inestigation, the lap slowly delated. It soon emerged that all three o the main
presidential candidates` passport iles had been improperly accessed.
Inestigators reported that the State Department was able to quickly identiy

1
Karen 1umulty, vooivg vto Obava`. Pa..ort, 1IML, Mar. 21, 2008,
!""#$%%&&&'"(.*',-.%"(.*%#-/("(,0%1)"(,/*%5^9744^36AC7A5^55'!"./.

who had examined the iles by using its computer audit system. 1his system
lagged any unusual requests or access to the iles o prominent Americans.
1he ired contractors did not deny the computer record. Seeral o them were
charged with crimes and pleaded guilty. All, it turned out, had acted purely out
o curiosity.`
Six months later, it was the Republicans` turn to howl about priacy iolations
in the campaign. Samuel Joe` \urzelbacher, a plumber, became an oernight
hero to Republicans in October 2008 ater he was practically the only person
who laid a gloe on Barack Obama during the campaign. 1he candidate made
an impromptu stop in \urzelbacher`s Ohio neighborhood and was surprised
when the plumber orced him into a detailed on-camera deense o his tax plan.
1hree days later, Joe the Plumber` and his taxes were inoked dozens o times
in the presidential debates.
1he price o ame was high. A media renzy quickly stripped \urzelbacher o
anonymity. Scouring the public record, reporters ound that the plumber had
been hit with a tax lien, they also ound goernment data that raised doubts
about the status o his plumbing license.
Reporters weren`t the only ones digging. Ohio state employees also queried
conidential state records about \urzelbacher. In all, they conducted eighteen
state records checks on \urzelbacher. 1hey asked whether the plumber owed
child support, whether he`d eer receied welare or unemployment beneits,
and whether he was in any Ohio law enorcement databases. Some o these
searches were proper responses to media requests under Ohio open records
laws, others looked more like an eort to dig dirt on the man.
Ohio`s inspector general launched an inestigation and in less than a month was
able to classiy all but one o the eighteen records searches as either legitimate
or improper.
18
1hirteen searches were traced and deemed proper, but three
particularly intrusie searches were ound improper, they had been carried out
at the request o a high-ranking state employee who was also a strong Obama
supporter. She was suspended rom her job and soon stepped down. A ourth
search was traced to a ormer inormation technology contractor who had not
been authorized to search the system he accessed, he was placed under criminal
inestigation.
\hat do these two laps hae in common 1hey were inestigated within weeks
o the improper access, and practically eeryone inoled was immediately
caught. 1hat`s itally important. Inormation technology isn`t just taking away
your priacy or mine. It`s taking away the priacy o goernment workers een

18
ee State o Ohio, Oice o Inspector General, Reort of vre.tigatiov, ite D ^vvber
2002, No. 20, 2008, &&&'>I+(,(1/&1",!'-)<%+-,I.*2"0%A554%T_U*#-)"'#+:.

aster. Data is cheap to gather and cheap to store. It`s een getting cheap to
analyze.
So it isn`t hard to identiy eery oicial who accessed a particular ile on a
particular day. 1hat`s what happened here. And the consequences or priacy
are proound.
I the lawyer`s solution is to put a predicate between goernment and the data
and the bureaucrat`s solution is to put use restrictions on the data, then this is
the auditor`s solution. Goernment access to personal data need not be
restricted by speed bumps or walls. Instead, it can be protected by rules, so long
as the rules are enorced.
\hat`s new is that network security and audit tools now make it easy to enorce
the rules. 1hat`s important because it takes the proit motie out o misuse o
goernment data. No proit-motiated oicial is going to take the risk o
stealing personal data i it`s obious that he`ll be caught as soon as people start
to complain about identity thet. Systematic misuse o goernment databases is
a lot harder and more dangerous i good auditing is in place.
1ake another look at why goernment oicials accessed these iles. It wasn`t to
steal identities. 1he reason most o these people accessed the data was simple
curiosity. Len the one access that may hae been or more reprehensible
reasons-the woman who checked conidential child support and welare
records or Joe the Plumber-was quickly caught and the data neer leaked.
1he speed and nearly complete eectieness o the audit process in these cases
tells us that network auditing tools can transorm the way we enorce the rules
or handling data in goernment. lor example, i we catch eery error, we can
improe compliance and at the same time reduce the penalties or mistakes.
larsh penalties are not the most eectie way to enorce rules. In act, they`re
usually a conession o ailure.
\hen we catch eery oender, we can aord to lower the penalty. Lighter,
more certain penalties or priacy iolations sere another purpose, too. \eè
talked a lot about the oddly protean nature o priacy. Not causing harm in
unexpected ways is at the core o the concept, but it`s nearly impossible to write
detailed rules spelling out what is and is not a iolation o priacy. Indeed, the
eort to write such rules and stick to them is what gae us the wall, and
thousands o American dead. So something must be let to discretion.
Goernment employees must use good sense in handling personal data. I they
don`t, they should be punished. But i we are conident that we can identiy any
questionable use o personal data and correct it quickly, the punishments can be
smaller. 1hey can be learning experiences rather than penological experiences.

So why did we criminally prosecute the poor schlubs whose hobby was looking
at the passport pictures o amous people 1he election happened. Lerything
that touched on the election was put under a microscope. Lil moties were
always ascribed to the other side. 1he State Department had to make a blood
sacriice to show that accessing the data was not part o an eil plot by one
party against the other. Opening a criminal inestigation was a way o
condemning the access in the clearest possible ashion. 1hat the poor schlubs
probably only desered demotions counted or little in the super-heated
atmosphere o a presidential campaign.
1hat shows one o the problems with the audit approach. It is too easily turned
into a phony priacy scandal. In both the \urzelbacher and Obama cases, the
audits did their job. \ith one possible exception, they caught the goernment
sta that broke the rules. 1hey preented any harm to either \urzelbacher or
Obama. And they made sure that the oicials who were responsible would
neer repeat their errors again.
1he system worked. Priacy was protected. But that`s certainly not the
impression that was let by coerage o the aairs. Indeed, the chairman o the
Senate Judiciary Committee, Senator Leahy, used the passport lap to tout new
legislation strengthening priacy protections on goernment databases. lrom a
political point o iew, then, the system ailed. 1here were no thanks or the
goernment oicials who put the system in place, who checked the audit logs,
who conronted and disciplined the wrongdoers, and who brought the soled
problem to public attention. 1o the contrary, they were pilloried or allowing
the access in the irst place-een though preenting such access is an
impossible task unless we intend to re-erect walls all across goernment.
low`s that or irony Audits work. But they work too well. Lery time they
catch someone and put a stop to misuse o personal data they also proide an
opening or political grandstanding. In the end, the inger pointing will
discourage audits. And that will mean less priacy enorcement. So, the more we
turn eery successul audit into a priacy scandal, the less real priacy we`re
likely to hae.
1hat would be a shame, because the auditor`s solution to the problem is the
only priacy solution that will get more eectie as technology adances. And
we`re going to need more solutions that allow lexible, easy access to sensitie
databases while still protecting priacy.
I the plight o goernment inestigators trying to preent terrorist attacks
doesn`t moe you, think about the plight o medical technicians trying to keep
you alie ater a bad traic accident.
1he Obama administration has launched a long-oerdue eort to bring
electronic medical records into common use. But the priacy problem in this

area is seere. lew o us want our medical records to be aailable to casual
browsers. At the same time, we can`t personally eriy the bona ides o the
people accessing our records, especially i we`re lying by the side o the road
suering rom what looks like brain or spine damage.
But the electronic record system won`t work i it can`t tell the irst responders
that you hae unusual allergies or a pacemaker. It has to do that quickly and
without a lot o ormalities. Auditing access ater the act is likely to be our best
answer to this problem, as it is to the ery similar problem o how to let law
enorcement and intelligence agencies share inormation smoothly and quickly
in response to changing and urgent circumstances. 1he Markle loundation has
done pioneering work in this area, and its path-breaking 2003 report on priacy
and security in the war on terror recommends embracing technologies that
watch the watchers.
19
A unique mix o security, priacy, and technology experts
managed to reach agreement in that report, they ound that one key to
protecting priacy without sacriicing security was a network that included
access control, authentication, and ull auditing capability.`
20

1he Markle report urges that large databases with personal inormation use
emerging technologies that can identiy all users o the system with certainty
and then gie them access that depends on their roles at any particular time.
1his includes the ability to restrict access priileges so that data can be used
only or a particular purpose, or a inite period o time, and by people with the
necessary permissions.`
21
1he technologies they cited are not pie in the sky.
1hey exist today: smart cards with embedded chips, tokens, biometrics, and
security circuits` as well as |i|normation rights management technologies.`
22

1he Markle task orce later did a thoughtul paper on one o those technologies,
which would presere audit logs een i high-ranking oicials seek to destroy or
modiy them later.
23

1hese technologies can be ery lexible. 1his makes them especially suitable or
cases where outright denial o data access could hae atal results. 1he tools can
be set to gie some people immediate access, or to open the databases in certain
situations, with an audit to ollow. 1hey can monitor each person with access to

19
Markle loundation 1ask lorce, Creativg a 1rv.tea êtror/ for ovetava ecvrit,, Dec. 2003,
!""#$%%&&&'.1)W/*'-)<%+-&2/-1+1=/*X100*"0%20":X)*#-)"AX:I//X)*#-)"'#+:.
20
a. at 15.
21
a.
22
a.
23
MARKLL lOUNDA1ION 1ASK lORCL, IMPLLMLN1ING A 1RUS1LD INlORMA1ION SlARING
LNVIRONMLN1: USING IMMU1ABLL AUDI1 LOGS 1O INCRLASL SLCURI1\, 1RUS1, AND
ACCOUN1ABILI1\ ,2006,, aailable at
!""#$%%&&&'.1)W/*'-)<%+-&2/-1+1=/*X100*"0%20":XT\EX5A545B'#+:.

the data and learn that person`s access patterns-what kinds o data, at what
time, or how long, with or without copying, and the like. Deiations rom the
established pattern can hae many consequences. Perhaps access will be granted
but the person will be alerted that an explanation must be oered within
twenty-our hours. Or access could be granted while a silent alarm sounds,
allowing systems administrators to begin a real-time inestigation.
1here`s a kind o paradox at the heart o this solution. \e can protect people
rom misuse o their data, but only by stripping network users o any priacy or
anonymity when they look at the data. 1he priacy campaigners aren`t likely to
complain, though. In our experience, their interest in presering the priacy o
intelligence and law enorcement oicers is pretty limited.
\hen I was general counsel o the National Security Agency, a well-known
priacy group headed by Marc Rotenberg iled a lreedom o Inormation Act
request asking the NSA to assemble all documents and emails sent to or rom
Stewart Baker.` 1hen as now, the NSA was orbidden to assemble iles on
American citizens who were not agents o a oreign power. Len so, Rotenberg
was asking NSA to assemble a dossier on me. Since NSA and I were locked in a
battle with Rotenberg oer encryption policy at the time, the purpose o the
dossier was almost certainly to look or embarrassing inormation that might
help Rotenberg in his political ight. Indeed, Rotenberg claimed when I
conronted him that he was planning to scrutinize my dossier or eidence o
misconduct.
lad the lBI or NSA assembled a dossier on their political adersaries, it would
hae been a iolation o law. In act, it would hae caused a priacy scandal. But
Rotenberg saw no irony in his request. It wasn`t a priacy problem, in his iew,
because goernment oicials desere no priacy.
I still think Rotenberg`s tactics were reprehensible: le had singled me out or a
selectie loss o priacy because he didn`t like my iews. But Iè come to
appreciate that there`s a core o truth to his iew o goernment. Anyone who
has access to goernment iles containing personal data has special
responsibilities. le should not expect the same priacy when he searches that
data as he has while he`s suring the net at home. And now that technology
makes it easy to authenticate and track eery person, eery deice, and eery
action on a network, perhaps it`s time to use that technology to presere
eeryone else`s priacy.
In the end, that`s the dierence between a priacy policy that makes sense and
one that doesn`t. \e can`t lock up data that is getting cheaper eery day.
Pretending that it`s property won`t work. Putting predicates` between
goernment and the data it needs won`t work, and neither will insisting that they
may only be used or purposes oreseen when it was collected.

\hat we cav do is use new inormation technology tools to deter goernment
oicials rom misusing their access to that data.
As you know by now, I think that some technology poses extraordinary risks.
But we can aoid the worst risks i we take action early. \e shouldn`t try to
stop the trajectory o new technology. But we can bend it just a little. Call it a
course correction on an exponential cure.
1hat`s also true or priacy. 1he uture is coming-like it or not. Our data will
be eerywhere. But we can bend the cure o technology to make those who
hold the data more accountable. Bending the exponential cure a bit-that`s a
priacy policy that could work. And a technology policy that makes sense.


A Market Approach
to Privacy Policy
By Larry Downes
*

Privacy: The Problem
\hat happens when the cost o deleting inormation is higher than the cost o
retaining it
1he answer is that nothing gets deleted. In the age o cloud computing, mobile
deices, and social networking, what that really means is that more and more
data-some o it enterprise data, some o it personal inormation, and more and
more o it something that merges the two-is being saed.
Soon, perhaps already, much o it will be consolidated, aggregated, reorganized,
and mined or aluable patterns, behaiors, and insights. Priacy has become
an unintended casualty o Moore`s Law-collateral damage rom riendly ire.
1hat, at least, is one way o thinking about priacy in the digital age, one that
has been on my mind or the last seeral months. I wrote about the priacy
problem in my recent book, 1be ar. of Di.rvtiov, in which I argued that the
real solution to concerns about priacy in the digital age would be the
emergence o robust markets or priate inormation, where consumers would
be able to trade personal inormation with other indiiduals and enterprises
when doing so generated mutual beneit.
1

1he priacy problem has morphed since then into the latest terror o the digital
age, surpassing earlier shibboleths, such as copyright piracy, identity thet, cyber
war and net neutrality.
2
Daily media coerage o the latest priacy policy

E1)); [-&2*0 is an Internet analyst and consultant, helping clients deelop business
strategies in an age o constant disruption caused by inormation technology. le is the
author o JPEN\KQTP_ RQN `TEENU \DD$ [T_TR\E KRU\RN_TNK HaU S\U`NR
[aSTP\PGN ,larard Business School Press 1998, and, most recently, o RQN E\bK aH
[TKUJDRTaP$ Q\UPNKKTP_ RQN PNb HaUGNK RQ\R _acNUP ETN \P[ ]JKTPNKK TP RQN
[T_TR\E \_N ,Basic Books 2009, |hereinater 1lL LA\S Ol DISRUP1ION|.
1
ee LARR\ DO\NLS, 1lL LA\S Ol DISRUP1ION, ar 1ro: Prirac,.
2
As Iè written elsewhere, all o these problems share a common core. Lach raises the
undamental question about the nature o digital lie and by whom and how its basic
inrastructure is to be goerned. In some sense, each is another iew o the same regulatory
problem, seen through lenses that are equally unocused, but in dierent ways. ee Larry
Downes, .fter tbe Detvge, More Detvge, 1lL 1LClNOLOG\ LIBLRA1ION lRON1, July 22, 2010,
!""#$%%"*,!/(=*)1"(-2',-.%A535%56%AA%1:"*)F"!*F+*/I<*F.-)*F+*/I<*%.

change, hacking incident, stolen goernment laptop or inadertent disclosure
has raised the stakes and the tension in a problem that, it seems, people react to
with such strong emotions that rational discussion o any solution is now
impossible.
3

1he priacy crisis is ery much on the mind o regulators around the world,
who see the emergence o priacy ears among consumers as the latest and
perhaps the best opportunity to gain a toehold in regulating ,and perhaps
taxing, content on the web. Nearly all o the earlier eorts-including outright
censorship, imposition o protectionist laws on global e-commerce, and
enorcement o strict copyright, trademark and patent regimes onto the
eoling collaboratie ethos o digital lie-hae ailed utterly.
4
By aligning
themseles with consumer interests ,and perhaps helping to stoke the ires o
anxiety,, regulators may hae at last ound their point o entry into the market
or Internet regulation.
1hat certainly seems to be the attitude adopted by the once-moribund U.S.
lederal 1rade Commission ,l1C,, which began a series o workshops in late
2009 aimed at exploring the priacy challenges posed by the ast array o 21st
century technology and business practices that collect and use consumer data.`
5

On January 28th, 2010, which was also dubbed Data Priacy Day by the non-
proit group 1he Priacy Projects,
6
the second workshop in the l1C`s three-
part
series took place at the Uniersity o Caliornia, Berkeley campus.

Attendees heard rom goernment, business, and public interest speakers on

3
Recent examples include the Google Maps drie-by, see Robert Graham, 1ecbvicat Detait. of
tbe treet 1ier !ii Pa,toaa Covtrorer.,, LRRA1A SLCURI1\, May 19, 2010,
!""#$%%*))1"10*,'=/-<0#-"',-.%A535%57%"*,!2(,1/F+*"1(/0F-:F0")**"F@(*&F&(:('!"./,
lacebook`s on-going changes to its priacy policy and user options, 1witter`s l1C
settlement, see Press Release, lederal 1rade Commission, 1witter Settles Charges that it
lailed to Protect Consumers` Personal Inormation, Company \ill Lstablish Independently
Audited Inormation Security Program, June 24, 2010,
!""#$%%&&&':",'<-@%-#1%A535%5B%"&(""*)'0!"., the botched launch o Google Buzz, see
Danny Goodwin, Google to Pay >8.5 Million in Buzz Priacy Class Action Settlement,
SLARClLNGINL\A1Cl.COM, No. 3, 2010, !""#$%%=/-<'0*1),!*2<(2*&1",!',-.%353358F
593689, shocking behaior on Chatroulette, the coniction o Google executies in an Italian
court oer a ideo showing the bullying o a minor with disabilities by a Google Video user,
see Reuters, Italy Conicts Google Lxecs or Down Syndrome Video, leb. 24, 2010,
!""#$%%&&&'&()*+',-.%*#(,*2"*)%A535%5A%<--</*F*?*,I"(@*F,-2@(,"*+F(2F("1/;F:-)F
+-&20F@(+*-%, and the launch o social shopping` websites such as Blippy and Swipely.
4
ee geveratt,, 1lL LA\S Ol DISRUP1ION, .vra note 1.
5
ee torivg Prirac,: . Rovvatabte erie., led. 1rade Comm`n,
&&&':",'<-@%=,#%&-)W0!-#0%#)(@1,;)-I2+"1=/*0% |hereinater 1C Rovvatabte erie.|.
6
lor more inormation on Data Priacy Day, isit !""#$%%+1"1#)(@1,;+1;A535'-)<.
ee 1C Rovvatabte erie., .vra note 5.


whether and how the l1C should regulate the priate collection and use o data
to protect consumer priacy interests. 1he conersation that day, characterized
by histrionic rhetoric, sel-congratulatory moralizing, and an utter lack o ocus,
relects well the current state o the so-called priacy problem.`
\hy is the l1C holding such hearings in the irst place 1he agency`s charter,
which has eoled oer its long history, includes policing anticompetitie
behaior
8
and enorcing a Congressional ban on unair and deceptie acts or
practices.` So ar, the l1C`s main contribution to the debate about digital
priacy has been the drating o non-binding guidelines or consumer notice o
online serices` priacy policies, the so-called lair Inormation Practice
Principles ,lIPs,.`
9
In the United States, the adoption o the lIPs is oluntary,
but ailure to abide by them can lead to l1C enorcement.
10

1he limits o the so-called notice` regime are pretty obious. Consumers
don`t read priacy policies. Len i they did, they would ind them to be
absurdly long, most o them written in some o the worst legalese Iè eer
seen.
Notice is also diicult to achiee in practice. During the last ew years,
lacebook has repeatedly landed in trouble or its mostly-admirable eorts to
crat a working priacy regime or its now 500 million users. 1he generally poor
response to these eorts, I think, stems rom a growing priacy paranoia ueled
by the media and goernments, kindled by the growing pains o a company that
by its nature deals with ery personal, een intimate, inormation and whose
growth rate challenges pretty much eerything.
At the same time, the company`s ounder, Mark Zuckerberg, has demonstrated
remarkably poor timing and nearly perect political tone deaness. Len as the
company dug itsel out o criticism o a new set o priacy tools in the all,
Zuckerberg told an audience in January 2010 that the social norm` or sharing
priate inormation had eoled.`
11
\ell, he is only 25 years old, and many o

8
1he l1C recently reached a settlement with Intel on a broad regulatory action brought
against the company. ee Press Release, led. 1rade Comm., 1C ettte. Cbarge. of
.vticovetitire Covavct .gaiv.t vtet, Aug. 4, 2010,
!""#$%%&&&':",'<-@%-#1%A535%59%(2"*/'0!"..
9
lor more inormation on the lair Inormation Practice Principles, isit
!""#$%%&&&':",'<-@%)*#-)"0%#)(@1,;8%:1()(2:-'0!"..
10
1he l1C can take action i a website iolates its own policy on the basis o its jurisdiction
oer deceptie practices. But there is no requirement or a website to hae a priacy policy,
and no explicit priacy protection in U.S. law or most categories o personally-identiiable
inormation.
11
ee Marshall Kirkpatrick, aceboo/`. Zvc/erberg a,. .ge of Prirac, i. Orer, RLAD\RI1L\LB,
Jan. 9, 2010, !""#$%%&&&')*1+&)("*&*=',-.%1),!(@*0%
:1,*=--W0XZI,W*)=*)<X01;0X"!*X1<*X-:X#)(@1,;X(0X-@'#!#.

his comments, including these, hae been unairly taken out o context by
mainstream, business, and technology media companies.
12

1he limits o notice become acute when the data collection deice is not a
computer. Cell phones, in addition to taking on more and more data unctions,
collect a great deal o inormation about the location and moements o their
users-they hae to in order to unction. But many applications such as
Loopt
13
and loursquare
14
take adantage o GPS data to oer serices not
possible on a ixed computing deice, including locating riends or proiding
location-speciic ads. On the smaller screen o a cell phone, reading any
document is diicult enough. \hat consumer in any case is going to read a
separate priacy policy or eery application they download
Len as it continues to reine and promote lIPs, the l1C has held hearings,
workshops and other inormation-gathering sessions regarding emerging
technologies that seem to raise new and worrisome priacy concerns. 1hese
hae included Radio lrequency ID tags, targeted or behaioral`
adertisements, cookies and now llash-based super cookies.`
15

1here was plenty o eidence at the Berkeley session o these and more o what
Microsot`s Peter Cullen called anxiety-based conersations.`
16
1his year`s
themes include the dangers o mobile computing, social networking, and cloud-
based computing, as well as continuing hand-wringing oer targeted or
contextual adertising.
1he structure o these conersations doesn`t change much oer time. 1he new
technology is discussed by law proessors, company representaties, and l1C

12
Caroline McCarthy, aceboo/ ottie.: . rief i.tor,, CNL1 NL\S.COM, May 13, 2010,
!""#$%%2*&0',2*"',-.%9853F38766X8FA555C978F8B'!"./, Guilbert Gates, aceboo/ Prirac,:
. eritaerivg 1avgte of Otiov., N.\. 1IMLS, May 12, 2010,
!""#$%%&&&'2;"(.*0',-.%(2"*)1,"(@*%A535%57%3A%=I0(2*00%:1,*=--WF#)(@1,;'!"./.
13
Loopt is a mobile mapping serice that allows users to ind local inormation using the
geographic location o their mobile phones. lor more inormation, isit
!""#$%%&&&'/--#"',-..
14
loursquare is a location-based social networking application that allows users to connect
with riends in their geographic location. lor more inormation, isit
!""#$%%:-I)0dI1)*',-..
15
1he super cookie problem,` piled on by many speakers at the Berkeley workshop, turned
out to be a red herring, easily controlled by engineered ixes to major browsers. It`s hard not
to read too much into that example. ee Berin Szoka, Prirac, vvoratiov: .aobe ta.b vort.
Prirate ror.ivg c Detete. ta.b Coo/ie., 1lL 1LCl. LIBLRA1ION lRON1, leb. 1, 2010,
!""#$%%"*,!/(=*)1"(-2',-.%A535%5A%36%#)(@1,;F(22-@1"(-2F1+-=*F:/10!F0I##-)"0F
#)(@1"*F=)-&0(2<F+*/*"*0F:/10!F,--W(*0%.
16
Priate Conersation with Peter Cullen, Jan. 28, 2010 ,notes on ile with author,.

sta ,eerybody but the engineers who know how the technology works,, who
try to point out the beneits and mostly dangers o the technology.
Lxtremists on either end o the spectrum call or the l1C to either ban or
ignore the deelopment. In typical ashion, satisied that all sides hae been
heard, the agency takes the problem under adisement, and then waits or the
next crisis du jour to emerge. No legislation or regulations are enacted.
\ell, that`s probably just as well-assuming there is no real priacy crisis that
needs to be addressed, or rather that needs to be addressed by an organization
with the l1C`s institutional limitations. I mean no disrespect to the l1C`s hard-
working sta. By institutional limits, I am thinking o the inherent constraints
on a U.S. regulatory agency. 1hese include the mismatch o a national regulator
superising behaior that is natiely global, problems in reising rules and
jurisdictions or an enironment that is eoling at accelerating speeds using a
process designed to be slow and deliberatie, and the dangers o soling what
are largely engineering problems with sta whose expertise is policy-and
oline policy at that.
It`s probably clear that I don`t think there is a crisis. But I admit that it`s
diicult to know. Both sides in this non-debate hae an unortunate habit o
relying on unscientiic surey data and anecdotal eidence, the latter o which,
on closer inspection, turns out to be highly incomplete i not urban myth. Pam
Dixon o the \orld Priacy lorum, or example, told a story about grocer
\hole loods using acial recognition sotware in stores to collect data rom the
tomato aisle or what Dixon called direct marketing purposes.` But according
to the lorum`s own report, 1be Ove !a, Mirror ociet,, ,whose inestigation o
the \hole loods story was limited to parsing company press releases, the
sotware could at best distinguish gender, not speciically identiy customers.
1

Direct marketing is targeted to an indiidual ,or their likely interests, rather than
demographic characteristics such as gender. So whateer \hole loods was
doing, it wasn`t direct marketing.
1he sureys that purportedly show a priacy panic are, or the most part, poorly
constructed and unscientiically executed. lor some reason law proessors and
their students, who hae at best a casual acquaintance with the methods and
rigors o any social science, are the ones called on by priate and public actors
to conduct these studies.
Consider, or example, two questions about the same technical eature o cell
phones. I you ask or an emotional response to the statement, My cell phone

1
Pam Dixon, 1be Ove!a,Mirror ociet,: Prirac, vticatiov. of tbe êr Digitat igvage êtror/.,
\ORLD PRIVAC\ lORUM, Jan. 2, 2010,
!""#$%%&&&'&-)/+#)(@1,;:-)I.'-)<%#+:%-2*&1;.())-)0-,(*";:0'#+:.

tracks where I go` you`ll get one answer, but i you phrase it, My cell phone
can tell me where I am,` you`ll get a ery dierent result. It`s the same eature.
1he indings` are useless. 1he choice and wording o the questions are the
only aluable inormation, in that it reeals a great deal about the authors o the
surey. But the sureys tell us nothing about the respondents or the choices
they would make when aced with real-world tradeos between restricting data
and the beneits that low rom it-such as getting more releant ads or a
greater quality and quantity o ree` online content and serices supported by
adertising reenue.
Privacy: Defining the Problem
1here`s a bigger problem here, and that is with the terms o the debate. In most
conersations, no one knows what anyone else means by priacy,` or what
inormation is included in the term personally-identiiable inormation,` which
dries much o the priacy regulations in the Luropean Union. 1he discussion
at the l1C`s Berkeley roundtable, as with all priacy discussions, conlated
seeral dierent inormation problems into one, reely mixing and matching
issues and regulatory solutions that don`t actually go together. Until we separate
the problems and sole them indiidually, the chances or meaningul policy
solutions are nil.
1o start with, it`s essential to understand the unique properties o inormation
as an economic good. Inormation has ery dierent properties rom
traditional commodities such as arm products, timber, and oil. Inormation
can be used simultaneously by eeryone, or one thing, and when we`re done
using it, it`s still there, potentially more aluable because o the use. 1hese
remarkable eatures make inormation what economists call a non-rialrous or
public` good, and they are the main reason that inormation now dries
economic actiity in much o the deeloped world. ,1he other is the continued
decline in the cost o computing power.,
So rather than talking about who owns` priate inormation or who is
stealing` data ,words that make more sense when talking about traditional
commodities,, I ind it much more constructie to talk about whether any
particular use o inormation is productie` or destructie` A productie
use o inormation is one that makes it more aluable, including collaboration,
remixing, and alidation. Destructie uses leae the inormation less aluable,
and include misrepresentation, misidentiication, and dilution.
18

18
Some inormation uses include both productie and destructie elements. Arguably,
remixing and other inormation sampling adds alue to inormation protected by copyright
and trademark law while potentially diluting markets the law protects on behal o the
inormation producer. ee Larry Downes, 1iacov r. Yov1vbe: 1be Privcite of ea.t Co.t
.roiaavce, 1he 1ech. Liberation lront, June 26, 2010,

As I argue in my book, 1be ar. of Di.rvtiov, priacy laws are best understood
as legal protections against destructie uses o inormation by dierent
categories o users. Seen that way, there is not one oerarching and
oerwhelming priacy problem, but seeral ery dierent priacy problems,
each desering o particular analysis and, one hopes, its own resolution. lere
are the main categories o destructie inormation uses:
T2:-).1"(-2 J0*) [*0")I,"(@* J0*0
Criminals Identity thet, phishing, malware and other orms o
raud
Commercial
enterprises
Surreptitious collection o consumer inormation or sale
or use in marketing, oten without adequate
compensation or reenue sharing with the consumer
Other consumers,
riends, amily
Stalking, bullying, accidental or intentional disclosure o
embarrassing or secret inormation
Goernment and
other state actors
Unlawul search and seizure, accidental disclosure
News media Publication o deamatory or erroneous inormation that
damages reputation
Lmployers and
business associates
Laesdropping and other monitoring to identiy poor
perormance, iolation o employer rules, or business
secrets
Insurers and health
care proessionals
Collection and use o known and potential risks to
determine coerage or the danger o accidental
disclosure

1hough many o these destructie uses were discussed at the l1C hearing, it
should be noted at the outset that the agency`s charter only extends to the irst
and second category.
19
,1o be air, l1C sta requently reminded the speakers
to limit their discussion to topics oer which the agency had jurisdiction., \hy,
then, did the speakers repeatedly bring up all the others lor one thing, some

!""#$%%"*,!/(=*)1"(-2',-.%A535%5B%AB%@(1,-.F@F;-I"I=*F"!*F#)(2,(#/*F-:F/*10"F
,-0"F1@-(+12,*%. lor now, I`ll stick to the easier` problems o uses that are almost purely
destructie.
19
ee led. 1rade Comm`n, Prirac, vitiatire., vtroavctiov, http:,,&&&':",'<-@%#)(@1,;. As the
agency explains the scope o its priacy initiaties, 1he lederal 1rade Commission is
educating consumers and businesses about the importance o personal inormation priacy,
including the security o personal inormation. Under the l1C Act, the Commission guards
against I2:1()2*00 12+ +*,*#"(-2 by enorcing companies` priacy promises about how
they collect, use and secure consumers` personal inormation. Under the _)1..FE*1,!F
]/(/*; \,", the Commission has implemented rules concerning :(212,(1/ #)(@1,; notices
and the administratie, technical and physical 01:*<I1)+(2< o personal inormation, and it
aggressiely enorces against #)*"*?"(2<. 1he Commission also protects consumer priacy
under the H1() G)*+(" U*#-)"(2< \," and the G!(/+)*2e0 a2/(2* D)(@1,; D)-"*,"(-2 \,".`

o the most lurid stories suggesting a priacy crisis come rom the other
categories, making them irresistible to those arguing or a crisis response.
Unortunately, most parties in the priacy debate` so ar hae shown little
interest in cabining the discussion to manageable and discrete problems when
an emotional point is there to be scored.
Indeed, one important reason to ealuate the categories o destructie
inormation use separately is to help us see that some goals o the priacy
moement are mutually exclusie. In the abstract, or example, most people are
uncomortable with the prolieration o sureillance cameras in urban locations.
But listen to the indignation that erupts the minute a serious crime or terrorist
act occurs and the police turn out not to hae caught it on ilm.
Or take the oten-repeated example o the ictim o domestic iolence, used as
a stalking horse or the proposition that search engines, cell phone carriers, and
other serice proiders, who collect bits and pieces o inormation that might be
used to identiy and locate an indiidual, should immediately purge their
databases, lest they all into the wrong hands.
1urn the problem around, howeer, and you can make the exact opposite case.
lor the ictim, it`s important to erase all traces o their online actiity. But or
the perpetrator, eectie law enorcement requires as much inormation as
possible.
20
Optimally, we`d like to tell inormation collectors to purge data
about ictims but retain it or criminals, but o course, we don`t know who is
who until ater the act. \hat`s a data collector to do
21

1his isn`t a hypothetical problem. Lawmakers in the United States and the
Luropean Union are simultaneously putting pressure on phone companies,
search engines and social networking sites to both purge and retain the same
data, a kind o whipsaw that has led these proiders uncharacteristically to call
or new laws-laws that would gie them a straight answer on what is expected
o them.
It`s also worth noting that in the United States in particular, most o the existing
legal protections or priate inormation are squarely aimed at deterring
destructie uses by criminals and by goernments themseles. Lery year I
hae to conince another batch o students that the right to priacy recognized
by U.S. courts and grounded in the U.S. Constitution does not apply to conlicts

20
Declan McCullagh, !eb earcbe. eaa to Mvraer Covrictiov, CNL1 NL\S.COM, leb. 12, 2010,
!""#$%%2*&0',2*"',-.%9853F38769X8F35C7AC63F89'!"./.
21
ee Miguel lelt, or Data, 1vg Cror. Orer Prirac, r.. ecvrit,, N. \. 1IMLS, Aug. 2, 2010,
!""#$%%&&&'2;"(.*0',-.%A535%59%58%"*,!2-/-<;%58=/1,W=*));'!"./. ee at.o Lance
\hitney, Cervav Covrt Rvte. .gaiv.t Data Retevtiov Potic,, CNL1 NL\S.COM, March 2, 2010,
!""#$%%2*&0',2*"',-.%9853F38769X8F35CBA336F89'!"./.

with commercial enterprises, parents, riends, or the news media. ,Indeed, the
latter are strongly protected by the lirst Amendment against such regulation.,
\ith at least a whi o lirst Amendment rationale, een the common law torts
that deal with conlicts between indiiduals oer inormation use-including
rights o publicity, deamation, and alse light` claims-hae allen into
disrepute during the last ity years.
22

lere there are also economic orces at work: inormation technology has
erased the temporary mask o anonymity created by 19
th
century urban lie,
reiing a social goal o transparency as old as lawthorne`s 1be cartet etter.
23

American lie, built o equal parts rontier necessity and Puritan aspiration, calls
or complete and accurate inormation about each other. 1hat goal increasingly
weighs more heaily in priate disputes oer what \arren and Brandeis in 1890
amously termed the right to be let alone.`
24
More about that in a moment.
\hy the ocus on state actors 1he principal ear o the draters o the
Constitution, obiously inormed by the experience o the colonies, was with
potential tyranny rom goernment. 1he goernment, ater all, practically holds
a monopoly on the coercie power o the military and the ability to incarcerate.
lor historical reasons, the ocus is ery dierent in Lurope and many parts o
Asia, which hae enacted strong priacy laws that align citizens and democratic
goernments against eerybody else. 1he dierence between U.S. and
Luropean law, in particular here, is perhaps the broadest eort to apply
terrestrial laws to digital lie generally.
25

In the United States, distrust o goernment eoled to include a ear that
priate inormation could be misused to achiee the same ends as more oert
repression. 1hese ears were underscored by the late 20
th
century scandals,

22
ee laynes . Alred A. Knop, Inc., 8 l.3d 1222 ,th Cir. 1993, ,opinion by Posner,.
23
NA1lANILL lA\1lORNL, 1lL SCARLL1 LL11LR ,1icknor, Reed & lields 1850,.
24
Samuel \arren & Louis D. Brandeis, 1be Rigbt to Prirac,, 4 lARV. L. RLV. 193 ,1890,,
,discussed below,. ee at.o Peggy Noonan, 1be ,e. are t, \ALL S1. J., May 21, 2010,
!""#$%%-2/(2*'&0>',-.%1)"(,/*%P\XbKfXDJ]$K]35553CAC57A6C965877455C767A7B68A
5CA997B89'!"./.
25
ee Adam Liptak, !bev .vericav ava vroeav aea. of Prirac, Cottiae, N.\. 1IMLS, leb. 26,
2010, !""#$%%&&&'2;"(.*0',-.%A535%5A%A9%&**W(2)*@(*&%A9/(#"1W'!"./, Kein J.
O`Brien, vroe a,. earcb irv. .re 1iotativg Data Rvte., N.\. 1IMLS, May 26, 2010,
!""#$%%&&&'2;"(.*0',-.%A535%57%A6%"*,!2-/-<;%A6+1"1'!"./, Jessica L. Vascellaro,
1ev Covvtrie. ../ Coogte to ao More to Protect Prirac,, \ALL S1. J., April 20, 2010,
!""#$%%-2/(2*'&0>',-.%1)"(,/*%P\XbKfXDJ]$K]35553CAC57A6C965CB6345C76734C44A
964764B9A'!"./.

including 1he Pentagon Papers and later \atergate, which led to the irst
comprehensie anti-wiretapping law at the ederal leel.
26

So, with notable exceptions, all existing U.S. priacy laws aord limited-and
perhaps inadequate-protection to citizens agaiv.t tbeir gorervvevt.. 1hat act is
oten whitewashed in the uror o the priacy debate. 1he ACLU o Northern
Caliornia, or example, recently published a white paper called Ctova Covvtivg:
torv !arvivg. for Prirac,.
2
1he paper points out the mismatch between
existing priacy law and the reality o cloud computing, where personal
inormation is turned oer or storage and processing to a ariety o third
parties and oten to their unnamed and changing business partners.
1he report neer quite says so directly, but all the proposed reorms are aimed
at curbing the ability o state actors, not priate parties, to gain access and make
use o data in the cloud. It is not consumers` as the report characterizes them,
who need to be worried about their priacy protections` in the cloud. It is
citizens. I think the ACLU is right to be worried about goernment access to
cloud data sources, but I wish it wouldn`t pretend to hae a broader agenda
than it does.
1hough the right to priacy against goernment is now irmly established, it`s
also worth remembering that this is a relatiely new right. Despite all the talk o
one`s right to priacy, you will scour the Bill o Rights in ain or any reerence
to priacy een against state actors.
It wasn`t until the 1960s that the Supreme Court began to interpret the lirst
Amendment`s ree speech proisions, along with the lourth Amendment`s
prohibition against unreasonable searches and seizures o people and their
property, as implying a more general right to priacy enorceable against state
actors.
28
In key cases, including Cri.rota r. Covvecticvt ,birth control,, Roe r. !aae
,abortion,, and more recently arrevce r. 1ea. ,homosexuality,, the Court
struggled to reign in goernment intrusions into the priate lies o citizens,
intrusions the lounders would neer hae imagined possible. Lacking a

26
ee 1he Omnibus Crime Control and Sae Streets Act o 1968, 18 U.S.C. 2510 et. seq.
2
Am. Ciil Liberties Union o N. Cal., Ctova Covvtivg: torv !arvivg. for Prirac,., Jan. 2010,
!""#$%%&&&'+-")(<!"0'-)<%0("*0%+*:1I/"%:(/*0%G/-I+gA5G-.#I"(2<gA5T00I*gA5D1
#*)'#+:.
28
ee, e.g., Griswold . Connecticut, 381 U.S. 49, 483 ,1965, ,In other words, the lirst
Amendment has a penumbra where priacy is protected rom goernmental intrusion.`,,
Roe . \ade, 410 U.S. 113, 152 ,193, ,1he Constitution does not explicitly mention any
right o priacy. In a line o decisions, howeer, going back perhaps as ar as |1891|, the
Court has recognized that a right o personal priacy, or a guarantee o certain areas or
zones o priacy, does exist under the Constitution.`,.

Constitutional proision that guaranteed the right to priacy or basic human
actiities, the Court more-or-less inented one.
At the same time, as 20
th
century society came to recognize that inormation
was a kind o property ,something with alue, in any case,, the perception o a
right to priacy emerged. It was ound, to use Justice Douglas`s amous but
unortunate phrase, in the penumbras and emanations` o the Bill o Rights.
29

But to reiterate, the right to priacy, as it is currently understood, is a right to be
ree o unreasonable intererence rom goernment, not rom each other or the
businesses with which we interact.
Some o today`s most ocal priacy adocates are calling or a broader right o
priacy, one that could be asserted against any or all o the destructie
inormation uses and perhaps against many o the productie uses as well.
An earlier eort to create such a right, it is worth noting, ailed. In 1890,
Samuel \arren and Louis Brandeis wrote a amous article or the arrara ar
Rerier in which they called or the creation o a general right o priacy, deined
as the right to be let alone.`
30
1he article was inspired by \arren`s personal
experience. \arren`s wie was the daughter o a U.S. Senator, and \arren was
appalled to discoer that their daughter`s wedding was reported, with
photographs, in 1be !a.bivgtov Po.t.
31

\arren and Brandeis`s proposal led to some experimentation, mostly at the
state leel and mostly through judge-made law. Common law courts inented
new tort injuries or damage to reputation, rights o publicity, and more
expanded orms o deamation, including the portrayal o someone in a alse
light.` Many o these rights are no longer recognized, or hae become nearly
impossible to enorce.
Under the Supreme Court`s 1964 decision in êr Yor/ 1ive. r. vttirav,
32
or
example, public oicials must proe actual malice to recoer damages or
deamation. Lxtending that decision, the Court held in the toriaa tar
33
case
that a rape ictim could not preent a newspaper rom publishing inormation
rom a police report about the crime committed against her. More recently,

29
Cri.rota, 381 U.S. at 484. Justice 1homas, a strict constructionist, has a sign on his desk
asking isitors to kindly keep their penumbras o his emanations.
30
\arren & Brandeis, .vra note 24.
31
\hat would \arren think o today`s celebrity media, or reality 1V, in which non-celebrities
olunteer to gie themseles the celebrity treatment
32
36 U.S. 254 ,1964,.
33
llorida Star . B.J.l., 491 U.S. 524 ,1989,.

homeowners hae been rebued in eorts to prohibit Google maps rom
photographing their homes.
34

1he \arren and Brandeis experiment ailed or a good reason, at least rom an
economic standpoint. Many o the uses o priate inormation that \arren and
Brandeis sought to outlaw are in act productie uses, generating substantial
social alue that outweighs the costs to the indiiduals o keeping such
inormation public.
It may be ery important or you to keep secret the act that you hae a criminal
record, a communicable disease, a Swiss bank account or a secret liaison with an
employee. But it is more aluable to the rest o us to know these things about
you, i only to know how many others hae the same attribute so we can take
appropriate actions-quarantine the sick, pass stricter banking laws, etc. 1he
beneits o disclosure, the courts hae determined, generally outweigh the costs
o secrecy.
\arren and Brandeis weren`t entirely wrong, howeer. It`s important to note
that the same priate` inormation can also be used destructiely. I might
oerestimate the risks o hiring a ormer elon, or example, or een exclude
potential tenants or my apartments based on an irrational reliance on personal
traits and associated stereotypes. 1hat`s why we hae anti-discrimination laws,
one o the notable exceptions where protections or misuse o personally-
identiiable inormation extend to commercial and other non-goernmental
users. But it is an exception, narrowly ocused on a particular abuse.
Generally speaking, in act, priacy legislation in the U.S. has only been enacted
when legislators ind particular and persistent market ailures-ailures, that is,
to use personally-identiiable inormation in rational ways. Along with anti-
discrimination laws, we hae laws that control priate inormation use in credit
reporting, medical records, and identity thet and other orms o inancial
raud.
35
In each o these cases, the law is ocused on a particular destructie use
in a particular context, with remedies or consumers narrowly-tailored to leae
as much inormation unprotected as possible.
Privacy: The Solution
Particular solutions to particular inormation use ailures are, I beliee, a model
lawmakers ought to be encouraged to continue ollowing.

34
ee Steen Musil, Coogte !iv. treet 1ier Prirac, vit, CNL1 NL\S.COM, leb. 18, 2009,
!""#$%%2*&0',2*"',-.%9853F35A8X8F353BB78AF48'!"./.
35
ee Larry Downes, f ea. ait, !bat Cav to aevtit, 1beft. CIO INSIGl1, July 2005,
http:,,&&&',(-(20(<!"',-.%,%1%D10"Fa#(2(-20%T:FH*+0FH1(/Fb!1"F,12FK"-#F
T+*2"(";FR!*:"%.

But maybe it`s time to reisit \arren and Brandeis`s call or broader priacy
protections against non-state actors. 1he argument in aor would go
something like this: Now that the cost o inormation retention is less than the
cost o inormation deletion, commercial enterprises may soon wield the same
kind o coercie power that until now has usually been the domain o
goernments.
Perhaps the same kinds o risks to society that led the courts to recognize a
zone o priacy` or inormation collection and use now justiy extending that
zone to the complicated web o enterprises that collect, consolidate, and resell
inormation about a wide range o consumer behaiors-as well as to
companies that by design collect intimate inormation, including social
networking sites and anything mobile.
But note that een i this argument carries the day-that is, i the beneits o
reducing destructie uses o priate inormation in some o the other categories
exceed the costs ,including inadertent limits on some productie uses, such as
the adertising that supports so much ree` media,, it doesn`t necessarily
ollow that the l1C or any other goernment entity is the right institution to
deine, enact, and enorce those new legal rights.
Again, the l1C`s authority is limited to inestigating anticompetitie behaior
and unair` or deceptie` trade practices-terms that hae clear meanings
under the agency`s statute, its preious decisions and policy statements, and
court cases interpreting the law.
Moreoer, there is the agency`s tendency simply to react with dismay to new
technologies and then to moe on when the technology soon ater resoles its
own problems. 1his habit-not limited to the l1C by any means-leads me to
doubt its institutional capacity to deine what kind o new priacy rights
consumers should hae, independent o always-changing technological
capacities. 1he l1C`s sta inokes the phrase priacy by design` like a mantra
in conersation with policy representaties and consumer adocates. 1he hope
behind that phrase is that uture technologies can be engineered to protect
priacy interests. But the phrase is meaningless without irst deining what
interests are to be protected.
1here are bigger reasons to question whether ederal or state goernment is
well-suited to the role o priacy cop. lor one thing, competing goernment
interests in eectie law enorcement create a kind o regulatory schizophrenia
oer the use o priacy-enhancing technologies. As an example, consider
encryption. Gien that one common theme o destructie inormation use in
seeral categories, as noted in the chart aboe, inoles accidental disclosure
,including access gained illegally by hackers or by raud,, it might be thought
that more widespread use o encryption technology would reduce those risks.
But the ederal goernment has been ambiguous at best about encouraging

enterprises to adopt encryption. In the 1990s, basic encryption methods were
classiied as a munition` by the Department o State ,and later, the
Department o Commerce,. Phil Zimmerman, the inentor o an open source
encryption protocol known as Pretty Good Priacy, was the subject o a lengthy
criminal inestigation or exporting weapons` without a license. Only under
intense pressure rom the courts in a series o lirst Amendment decisions did
the Commerce Department inally liberalize these export controls.
1here was also the memorable ight in the early 1990s oer the Clipper Chip, a
goernment-deeloped encryption technology or use in cell phones.
36
lad the
chip been made mandatory or een widely deployed ,opponents, including the
Llectronic lrontier loundation ,Lll,, successully preented that outcome,, it
would hae gien law enorcement a back door into mobile phone calls and
might hae led to the prohibition o other encryption technologies.
3

lederal and state goernments also receie ailing grades at adopting and using
the kinds o sae data handling practices some think these goernments should
enorce against eeryone else. 1hat ailure is eidenced by numerous
embarrassing breaches and stolen unsecured laptops containing millions o
records o sensitie citizen data. Goernment in this sense oers a good
example ,seeral, actually, o how vot to manage priacy and the worst, not best,
sae handling practices. linally, a general problem o state, as opposed to
ederal, legislation is the potential or ity dierent deinitions o priacy, all
imposed on what is a global inormation economy.
More broadly, there`s a disturbing irony to handing priacy enorcement oer to
goernments, an argument I hae also made with regard to enorcement o net
neutrality principles. Lnorcement powers would require the agency to examine
the inormation and how it was disclosed. lor agencies to inestigate and
punish banned uses o priate inormation, they would necessarily need to hae
access to the data sources in question.
I I complain, to the l1C, or example, that a social networking enterprise is
selling my inormation in iolation o some uture priacy protection law, how
else will the agency ealuate my claim without looking deeply into the
company`s inormation practices Like any good audit, they will need to ollow
the low o inormation rom beginning to end to determine what, i any,
iolations o law occurred.

36
lor more inormation on the Clipper Chip, including goernment documents and public
response, isit !""#$%%*#(,'-)<%,);#"-%,/(##*).
3
1he Clipper Chip ight was instrumental in the ormation o Lll, proing once again the
law o unintended consequences. 1he goernment lost the battle, but more importantly, it
inadertently helped to organize a permanent and eectie opposition.

\et state actors, as noted aboe, are currently the most restricted rom seeing
this inormation. In a bitter irony, enorcement o priacy rights against
enterprises would hae the eect o exposing data to the goernment that it
otherwise would be orbidden to see. As a result, there is the potential or
abuse o that priilege in the name o law enorcement or other goernment
priorities ,e.g., terrorism or tax raud,.
38

\ho else can deine, let alone enorce, new priacy protections against
enterprises and other non-state actors One obious alternatie is sel-
regulation by the enterprises themseles, perhaps encouraged by the threat o
goernment interention i the market ails.
1here`s already a good deal o sel-regulation, including oluntary adoption o
the lIPs and certiication rom third parties such as 1RUS1e
39
and the Better
Business Bureau.
40
Much o this sel-regulation, howeer, presumes the
useulness o the notice regime which, as noted, no one presumes any more.
1he potential is there, howeer, or more eectie sel-regulation. Len
companies who make money rom inormation collection and its use are
worried about the priacy problem, or at least the perception o one. Some are
een calling or new legislation, in part to sole the whipsaw problem described
aboe.
Microsot and other companies who are counting on cloud computing as the
next major computer architecture are eager or regulatory rameworks that will
both allay consumer anxiety and gie cloud-based serice proiders sae harbors
in which to deelop their oerings.
An emerging consensus among a wide range o technology companies and
inormation serice proiders would moe the discussion rom guidelines about
priate inormation notice to rules about inormation use. 1he use` approach
would deelop acceptable principles or how collected and retained data ,priate
or otherwise, can be used, by whom, and with what controls resered to
consumers to limit or block those uses.
As part o the Business lorum or Consumer Priacy ,BlCP,, Microsot and
others are calling or a legislated ramework or use-based rules.
41
\hat does

38
ee Declan McCullaugh, .vaov igbt. Devava for Cv.tover Recora., CNL1 NL\S.COM, April
19, 2010, !""#$%%2*&0',2*"',-.%9853F38769X8FA555A965F89'!"./.
39
ee !""#$%%&&&'")I0"*',-..
40
ee !""#$%%&&&'==='-)<%I0%]I0(2*00F\,,)*+("1"(-2.
41
1he Business lorum or Consumer Priacy, . |.e ava Obtigatiov. .roacb to Protectivg Prirac,:
. Di.cv..iov Docvvevt, Dec. , 2009,

that mean According to a 2009 BlCP white paper, a useandobligations`
model requires all organizations to be transparent, oer and honor appropriate
choice, ensure that risks to consumers related to use are assessed and managed,
and implement security or the inormation they maintain.
42
1he white paper
describes a taxonomy o use types ,or example, marketing, internal operations,
within the enterprise category, and tries to deine an appropriate set o deault
rules that should apply or each.
1he BlCP approach is certainly more productie than the anxiety-based
conersations. locusing on use, or one thing, moes the conersation away
rom the emotional subject o priacy to the more rational subject o propriety,
by which I mean the recognition that both enterprises and consumers
participate in the creation o aluable new sources o data and both should hae
rights to monetize that alue. Consumers ought to be able to buy out
enterprises or productie uses they don`t want, and ice-ersa.
In this scenario, priacy policy eoles rom sel-regulation by the market to
actually becoming a market or priate and other data. Assuming this market
works, data will be put to the most highly-alued productie uses, and those
uses that are not alued or are destructie will not be implemented.
43
As with
any market, the goernment will stand in the background to ensure the rules are
obeyed and to interene when market ailures occur.
One immediate concern about the creation o a priacy market is that
consumers will hae no oice in its creation or operation. Consumers, ater all,
are indiiduals, easily oerwhelmed by the economic might o large
corporations, who can be expected to deelop rules that gie consumers an
unair disadantage in the inormation marketplace.
1hat would hae been an entirely reasonable concern in the pre-digital age, and
one o the principal justiications or the creation o priate and public
consumer watchdog groups such as the l1C in the irst place ,the agency is
nearly 100 years old,. Such groups, at least in theory, lobby and sue on behal

!""#$%%&&&'!I2"-2:(/*0',-.%:(/*0%&*=I#/-1+%GTDEXJ0*X12+Xa=/(<1"(-20Xb!("*X
D1#*)'#+:.
42
ee lunton & \illiams LLP, v.ive.. orvv for Cov.vver Prirac, vtroavce. êr Data Protectiov
Moaet, Dec. 21, 2009,
!""#$%%&&&'!I2"-2#)(@1,;=/-<',-.%A554%3A%1)"(,/*0%*@*2"0%=I0(2*00F:-)I.F:-)F
,-20I.*)F#)(@1,;F(2")-+I,*0F2*&F+1"1F#)-"*,"(-2F.-+*/%.
43
1his emerging priacy market, operating with minimal transaction costs or riction,` could
present a wonderul opportunity to test Ronald Coase`s theory that absent transaction costs,
stakeholders will necessarily bargain to the most productie use o a gien resource. See
Ronald Coase, 1he Problem o Social Cost,` 3 Journal o Law & Lconomics 1 ,1960, ,the
so-called Coase 1heorem`,.

o large groups o consumers to ensure their interests are airly represented in a
wide ariety o market actiities. ,1he class-action mechanism is another
example., 1o use the economic terminology, these legal constructs oercome
the collectie action problem o indiidual users whose indiidual losses may be
too small to justiy the expense o enorcing or een negotiating rights.
But the same technologies that create the priacy problem are also proing to
be the source o its solution. Len without goernment interention,
consumers increasingly hae the ability to organize, identiy their common
demands, and enorce their will on enterprises.
44

lacebook`s journey to a air priacy policy continues to be instructie here.
1he company`s ongoing priacy crisis actually started in early 2009, when
lacebook announced what was in act a modest change to its terms o serice.
Some users who read the modiication objected to it, and used the ery tools
lacebook proides or group ormation to create a People Against the New
1erms o Serice` page, which quickly signed up 100,000 ans.
45
,Still a
relatiely small number gien lacebook`s size: now oer 500 million users
worldwide.,
1he reolt and ,perhaps more inluential, the ensuing bad publicity led
lacebook to withdraw the changes-no lawsuit or legislation necessary. Len
more, the company soon announced that it was changing its entire approach to
goernance. In the uture, the company said, it would rewrite its user
agreement to be understandable to lay readers, and circulate uture changes as
proposals to the entire population o users.
I enough users objected to a proposed change, the changes would be put to a
ote. ,1he changes to priacy settings that set o the irestorm in the all o
2009 were circulated ahead o time, but didn`t qualiy or a ote, perhaps
suggesting that lacebook`s constitution` still needs some tweaking.,
So assuming that buyers and sellers hae equal access and power, how would a
market or priate data work in practice
46
1here are already some good

44
Stee Lohr, Yov !avt M, Per.ovat Data. Rerara Me for t, N.\. 1IMLS, July 16, 2010,
!""#$%%&&&'2;"(.*0',-.%A535%56%39%=I0(2*00%39I2=-?*+'!"./, L. Gordon Croitz,
Prirac, .v`t rer,tbivg ov tbe !eb,` \ALL S1. J., May 24, 2010,
!""#$%%-2/(2*'&0>',-.%1)"(,/*%P\XbKfXDJ]$K]35553CAC57A6C965C7CB85C767AB5C65
57C8AB85C'!"./.
45
1o iew the lacebook page, isit
!""#$%%&&&':1,*=--W',-.%<)-I#'#!#L<(+M665B4356C8A.
46
Venture capitalists hae begun to recognize the potential o priacy markets and are
inesting accordingly. ee Pui-\ing 1am & Ben \orthen, vva. vre.t iv Prirac, tartv.,
\ALL S1. J., June 20, 2010,

prototypes in place in the orm o consumer loyalty and ainity programs that
hae been around or decades. 1he grocery store scanner data can already
determine what items hae been bought at the same time, but in order to tie
that collection to particular demographic characteristics ,gender and zip code
are the most important or marketing purposes,, it requires the cooperation o
the customer.
So, in exchange or using a club card and giing the store permission to connect
purchases to a particular customer, the consumer gets special discounts. In
eect, the store recognizes the alue o the identiying inormation and pays the
consumer or it.
1hat approach is attractie or seeral reasons. lor one thing, it is entirely
oluntary-consumers make the decision to sign up or the programs or not, or
een whether or not to participate in any indiidual transaction. 1his proto-
market also operates under ery low transaction costs. Instead o negotiating
or the purchase o each indiidual data element with each indiidual consumer,
a ramework is established that allows both parties to opt out at will.
Discounted prices become a new orm o currency requiring no oersight. No
lawyers and no regulators are inoled.
1he loyalty program also makes explicit what I think is the most important
eature o the priacy debate, one that is usually lost in the extreme rhetoric o
both sides. Inormation only attains its true monetizable alue when eery
participant in the transaction has an incentie to proide it, authenticate it and
protect it. Retail data is o limited use to the supply chain without demographic
linkages. Indiidual data is not marketable unless someone is able to collect and
analyze large olumes o it. As with all inormation exchanges, the whole is
greater than the sum o its parts.
1oday, een ity years into the computer reolution, most aspects o retail
engineering, including production planning, pricing and promotions, product
design, orecasting, and marketing, operate with precious little real inormation.
Most are more black arts than science. \ith the potential capture o complete
lie-cycle transaction data, there is at last the hope o discipline, and with it great
increases in eiciency.
And in the next digital decade, we now hae the potential or inormation to be
collected post-sale: low did you like the product Did it perorm as expected
\hat other products did you use it with It becomes een clearer that without

!""#$%%-2/(2*'&0>',-.%1)"(,/*%P\XbKfXDJ]$K]35553CAC57A6C9658C89B5C76783739A
5A76A3769'!"./.

cooperation rom the consumer, no real alue will come rom ast data
warehouses that enterprises hae built but rarely use eectiely.
4

1he priacy marketplace is already here, and there is eery indication that as
technology continues to eole, an increasingly robust and aluable set o
institutions will deelop alongside o it. As new orms o data enter the digital
domain, new tools and techniques will emerge to harness their alue and
allocate it among those who deelop it.
In order or the technology o inormation processing to reach its potential,
howeer, the histrionics o the priacy debate must stop. Instead, consumers
must be encouraged and educated to think about inormation use in terms o
productie and destructie costs and beneits. In short, we must learn to think
rationally about inormation alue. I the l1C and other regulators are looking
or a role in soling the problems o online priacy, a good starting point would
be to contribute constructiely to the emergence o this organic, elegant
solution.

4
ee LARR\ DO\NLS, 1lL S1RA1LG\ MAClINL: BUILDING \OUR BUSINLSS ONL IDLA A1 A
1IML, ,larper Business 2002,, Chapter 3, 1be vforvatiov vt, Cbaiv. ee at.o Jules
Polontesky & Christopher \ol, otrivg tbe Prirac, Ditevva, 1lL lUllING1ON POS1, July 2,
2010, !""#$%%&&&'!I::(2<"-2#-0"',-.%>I/*0F#-/-2*"0W;%0-/@(2<F"!*F#)(@1,;F
+(/*.X=XBB5B94'!"./.

529

CHAPTER 9
CAN SPEECH BE POLICED
IN A BORDERLESS WORLD?
The Global Problem of State Censorship
& the Need to Confront It 531
!"#$ &' ()*+,-./ !,'
The Role of the Internet Community
in Combating Hate Speech 547
0#,123"4#-, 5"*+

530 CHAPTER 9: CAN SPEECH BE POLICED IN A BORDERLESS WORLD?


The Global Problem of State
Censorship & the Need
to Confront It
By John G. Palfrey, Jr.
*

Speech is policed through technical Internet iltering in more than three dozen
states around the world. 1his practice is increasingly widespread. States
including China, Iran, Syria, 1unisia, and Uzbekistan hae extensie Internet
iltering regimes in place. Censorship using technological ilters is oten
combined with restrictie laws related to what the press can publish, opaque
sureillance practices, and seere penalties or people who break the state`s rules
o using the Internet. 1his trend has been emerging, and documented with
precision, or nearly a decade.
1

An empirical study o technical Internet iltering tells only part o the story,
howeer. Speech is policed actiely in parts o the world with regimes that are
substantially more democratic than China or Iran. 1hrough mechanisms that
include sureillance, encouraging` sel-censorship, intellectual property
restrictions, and deamation laws, irtually eery state in the world polices
online speech through multiple means.
2
As more and more o eeryday lie
moes onto the Internet, so has regulation o that actiity. 1hese orms o
regulation are drien by the same types o concerns that animate the regulation
o speech in traditional enironments.

John Palrey is the lenry N. Lss III Proessor o Law at larard Law School and laculty
Co-director o the Berkman Center or Internet & Society at larard Uniersity. 1his
chapter draws upon research by the OpenNet Initiatie, which is a collaboration that joins
researchers at the Citizen Lab at the Munk Centre, Uniersity o 1oronto ,Pro. Ron
Deibert, principal inestigator,, the SecDe Group ,ormerly the Uniersity o Cambridge
where Raal Rohozinski is principal inestigator,, and the Berkman Center ,where the author
and Jonathan Zittrain are co-principal inestigators,. 1he author is grateul to the large
number o researchers who hae participated in gathering, oer nearly a decade, the data on
which this chapter draws. Parts, though not all, o this argument hae been published in
other olumes.
1
ee !!!"#$%&&%'"&%' or the results o the OpenNet Initiatie`s research since 2002. ee
at.o Ronald Deibert, John Palrey, Raal Rohozinski, & Jonathan Zittrain, eds., ACCLSS
DLNILD: 1lL PRAC1ICL AND POLIC\ Ol GLOBAL IN1LRNL1 lIL1LRING ,MI1 Press, 2008,
and Ronald Deibert, John Palrey, Raal Rohozinski, & Jonathan Zittrain, eds., ACCLSS
CON1ROLLLD: 1lL SlAPING Ol PO\LR, RIGl1S, AND RULL IN C\BLRSPACL ,MI1 Press,
2010,, in which ariations on these themes appear throughout.
2
Like technical Internet iltering, this is not a new phenomenon. ee Adam D. 1hierer, 10
vtervet Cev.or.. Ri.ivg Ctobat 1breat. to Ovtive eecb, 38 1echKnowledge, July 26, 2002,
!!!"()'#"#*+,$-./012$3)4"$5$6$-./107889:9.

1he social, political, and cultural issues that gie rise to online speech
restrictions are amiliar. Child saety is one o the primary driers o online
speech regulation, most commonly related to pornography. Police use
sureillance to track the misdeeds o wrong-doers, in turn causing a chilling
eect on online actiities. Business people, authors, musicians, and others want
their intellectual property rights indicated to the ullest extent, prompting
extensie intellectual property restrictions online. 1hose who perceie harm to
their reputations seek the extension o deamation laws to the digital world. In
each o these instances, speech restrictions cross traditional and digital
enironments more or less seamlessly.
Is It True that All Politics Is Local
in the Digitally-Mediated World?
1he problem o policing speech on the Internet arises at eery leel o
goernment, rom local political conersations about community norms to the
international debate oer Internet goernance. 1he principal dierence at these
multiple leels o goernance is the willingness to address the diicult problems
o when to regulate speech online and the ramiications that low rom doing
so. State legislatures ight oer whether to constrain speech in schools when
young people engage in cyber-bullying.
3
In the United States, Congress takes up
the same issue rom the perspectie o ederal law enorcement and unding or
schools and education. At the same time, the topic o speech-based controls is
an important aspect o global Internet regulation. loweer, it rarely makes it
into the agenda at the international leel as a genuine, openly-discussed matter.
At the local leel, there is oten substantial appetite to regulate speech online,
primarily as a response to ears about child saety. 1he clearest example at the
local leel in the United States is the eort to curb cyber-bullying through
school enironments, which states, regional, and municipal goernments seek to
regulate.
4
Speech, largely online in social network enironments, is policed by
local regulation that bars young people rom expression that may harm their
peers psychologically. \hile the debate rages as to the appropriateness o these
regulations, dozens o states in the United States hae enacted, or hae
considered enacting, these types o laws at the local leel. Courts hae split as
to the constitutionality o such proisions. In short, intense local concerns,

3
1he lirst Amendment Center published a ine series o essays and research materials related
to the speech restrictions associated with the anti-bullying eorts, with special attention paid
to state legislatie actiities. ee lirst Amendment Ctr., Ovtive ,vo.ivv: C,berbvtt,ivg c
Pvbtic cboot.;
!!!"=1*2')>%&0>%&'(%&'%*"#*+,(#33%('1#&")2$?61'%>7(4.%*.-3341&+/$-.31(/2(5##32.
4
Sameer linduja, & Justin \. Patchin, Cyberbullying Research Center, tate C,berbvtt,ivg ar.:
. rief Rerier of tate C,berbvtt,ivg ar. ava Poticie., July 2010,
!!!"(4.%*.-3341&+"-2,@-3341&+/)&0/A4.%*.-3341&+/B)!2/CD8DDED8"$0=.

such as about how young people treat one another, are driing legislatie
attempts to regulate online speech through school-based enorcement.
5

At the ederal leel, online speech restrictions hae arisen repeatedly as
proposals and as enacted law, in the United States and around the world.
1wenty-ie years ater the creation o the .COM top-leel domain, it is
apparent that national goernments can, and oten do, assert soereignty oer
the acts o their citizens in the online enironment, including limitations on
their speech. At a basic leel, in the United States and in many other
jurisdictions, speech that is deemed to be harmul in the oline world is
considered equally unlawul in the online enironment. lor the speaker, there is
no ree pass simply because the utterance in question appears online.
1he primary dierence between the policing o speech online and oline lies in
terms o how intermediaries are treated. Under United States ederal law, and
the national law o many other jurisdictions, intermediaries that enable people
to publish speech online are exempt rom liability in most cases, or example,
rom claims o deamation under Section 230 o the Communications Decency
Act.
6
1here are exceptions to this rule, een in the United States: Criminal
matters and copyright iolations all outside o the statute`s sae harbor
proisions.
In other jurisdictions, regulation o intermediaries may soon be the

law. lor instance, in Sweden, the Data Inspection Board issued a report in July
2010 asserting that companies oering social media serices, such as blogs,
lacebook, or 1witter, hae a legal obligation to monitor personal data posted to
the pages on their site.
8

At the international leel, we obsere extensie policing o speech online, but
discussion o the issues inoled is largely inisible-or else not happening at
all. \hile there is extensie and healthy debate about many aspects o the
problem o Internet goernance, the discussion does not reach the hard
problems o when online speech regulation should be permitted. 1here are
many issues worthy o the attention o the many capable minds ocused on
Internet goernance, but the topic o speech regulation rarely makes the list o
what is, in act, publicly discussed and etted seriously.
9
1he primary ocus or

5
ee Ronald Collins, . oo/ at C,berbvtt,ivg ava Pvbtic cboot.,` March 31, 2009,
!!!"=1*2')>%&0>%&'(%&'%*"#*+,)&)34212")2$?6107C8F8D.
6
4 U.S.C. 230.
ee ia.
8
ee Swedish Data Inspection Board Report, July 5, 2010, !!!"0)')1&2$%G'1#&%&"2%,1&H
%&+3125, ,on intermediary liability,, .ee at.o Covavie. Re.ov.ibte for ociat Meaia Covtevt, 1lL
LOCAL: S\LDLN`S NL\S IN LNGLISl, July 5, 2010, !!!"'5%3#()3"2%,CEIDI,CD8DDED9,.
9
1he International 1elecommunication Union ,I1U,, oicial host o the \orld Summit on
the Inormation Society ,\SIS, in Genea, has held seeral eents designed to reine the
debate urther. 1hrough these eents, the I1U has coninced dozens o obserers to

Internet goernance discussions continues to be issues related to the
management o Internet resources, including the domain name system and
related policy issues. Discussion o the non-proit Internet Corporation or
Assigned Names and Numbers ,ICANN, continues to play a central role.
ICANN occupies an arcane bit o tur-essentially, the port allocation business.
1hat is important in some respects but does not appear to concern most users
o the Internet, particularly in a world in which most people ind Internet
resources through search engines and, increasingly, mobile deices and
applications.
10
As an example, within the context o the Internet Goernance
lorum 2009 meeting in Lgypt, the irst substantie panel o the eent was
deoted to traditional ICANN-related matters such as the transition rom
Internet Protocol ersion 4 ,IP4, to IP6 and the addition o new top-leel
domains ,1LDs,.
11
Possible topics or consideration, other than ICANN
reorm and these highly speciic technical issues, each more important to the
end-users o the Internet and their soereigns, hae included a und or
deeloping countries to build Internet inrastructure, the quandary o what to
do about spam, and a cluster o intellectual property concerns.
Internet speech restrictions should sere as the ocal point or the world`s heads
o state and their designees when Internet goernance is on the table. \hile
online speech restrictions raise a wide array o issues, a discussion o Internet
iltering would hone in on whether states actually ravt their citizens to hae ull
access to the Internet or not. It would help guide a public conersation about
what is truly most important about haing access to the Internet and the extent
to which states place a premium, i at all, on the global low o inormation.
\ithout collectie action, the Internet will likely continue to become balkanized
into a series o local networks, each goerned by local laws, technologies,
markets, and norms. As Jonathan Zittrain argued in !bo Rvte tbe êt., the
predecessor o this collection, we may be headed toward a localized ersion o
the Internet, goerned in each instance by local laws.
12
I such a ersion o the
Internet is ineitably part o our uture, there ought to be open and transparent

publish what comprises an extensie body o work on this topic on the I1U website. In
addition, long-time experts in this ield, such as Pro. Milton Mueller o Syracuse, and
others, hae constructed helpul models to structure the conersation. lor suggestions on
urther inormation o this general nature, please see !!!"&%'01)3#+-%"#*+, a joint project
o larard Law School and Stanord Law School.
10
\itness the abysmal turnout or ICANN`s election o 2000, in which a ree and open
election or ie ICANN directors attracted ewer than 100,000 otes globally.
11
Internet Goernance lorum o 2009, Managing Critical Internet Resources, 1ranscript, No.
16, 2009,
!!!"1&'+#J=#*->"#*+,(>2,CDDK,25)*>/%3/L5%1G5,M*)&2(*1$'2,L5)*>NCDO3NCDL
5%1G5NCD8INCDP#J%>.%*NCDCDDKNCDQ)&)+1&+NCDA*1'1()3NCDR&'%*&%'NCDS%2#-*
(%2"$0=.
12
Jonathan Zittrain, e Carefvt !bat Yov ../ or, iv \lO RULLS 1lL NL1 IN1LRNL1
GOVLRNANCL AND JURISDIC1ION 13-30 ,Adam 1hierer et al. eds., Cato Inst. 2003,.

consideration o ways to embrace it that can presere elements o the network
that are the most important.
The Internet Filtering Problem
1he world may appear borderless when iewed rom cyberspace, but
geopolitical lines are, in act, well-established online. 1he act that extensie
Internet iltering occurs at a national leel around the world is clearly
documented. 1hrough a collaboratie research eort called the OpenNet
Initiatie,
13
the Citizen Lab at the Uniersity o 1oronto, the Berkman Center
or Internet and Society at larard Uniersity, and the SecDe Group are
working together to compare Internet iltering practices o states in a
systematic, methodologically rigorous ashion. In the past seeral years,
OpenNet Initiatie has sought to reach substantie conclusions about the
nature and extent o Internet iltering in roughly 0 states and to compare
practices across regions o the world. 1he OpenNet Initiatie has released
extensie reports that document and proide context or Internet iltering,
preiously reported anecdotally, in each o the states that it has studied closely.
Reports released to date hae ocused on states in the Middle Last and North
Arica, Asia, and Central Asia, where the world`s most extensie iltering takes
place. OpenNet Initiatie`s research also coers states in eery region o the
world, including North America and \estern Lurope, where orms o speech
regulation other than technical Internet iltering at the state leel are the norm.
liltering implementations ,and their respectie scopes and leels o
eectieness, ary widely among the countries OpenNet Initiatie has studied.
China continues to institute by ar the most intricate iltering regime in the
world, with blocking occurring at multiple leels o the network and coering
content that spans a wide range o topic areas. 1hough its iltering program is
widely discussed, Singapore, by contrast, blocks access to only a handul o
sites, each pornographic in nature. Most other states that OpenNet Initiatie is
studying implement iltering regimes that all between the poles o China and
Singapore, each with signiicant ariation rom one to the next. 1hese iltering
regimes are properly understood only in the political, legal, religious and social
context in which they arise.
Internet iltering occurs in dierent ways in dierent parts o the world. Some
states implement a sotware application deeloped by one o a small handul o
U.S.-based technology proiders. Burma, in the irst incarnation o its iltering
regime, used an open source product or iltering, called DansGuardian.
14

Others rely less heaily on technology solutions and more extensiely on sot
controls.` Sometimes the iltering regime is supported explicitly by the state`s

13
lor more inormation, see !!!"#$%&&%'1&1'1)'1J%"&%',.
14
lor more inormation on the DansGuardian iltering product, see 0)&2+-)*01)&"#*+,.

legal code, in other cases, the iltering regime is carried out through a national
security authority. In yet other instances, the regulation is simply presumed to
be permissible. 1he content blocked spans a wide range o social, religious, and
political inormation. Studies by OpenNet Initiatie hae reiewed whether
indiidual citizens could access sites in a global basket` o bellwether sites,
testing eery jurisdiction across a ariety o sensitie areas-akin to a stock
index sorted by sector-as well as a list o websites likely to be sensitie in some
categories, but only in some countries.
Extent, Character & Locus of Filtering
More than three dozen states around the world practice technical Internet
iltering o arious sorts.
15
1hat number has grown oer time. 1hose states
that do ilter the Internet hae established a network o laws and technical
measures to carry out substantial amounts o iltering that could allow the
practice to become urther embedded in their political and cultural
enironments. \eb content is constantly changing, which poses a problem or
the censors. Mobile deices and social networks hae urther complicated the
task o speech regulation online. No state yet studied, een China, seems able
to carry out its \eb iltering in a comprehensie manner, i.e., consistently
blocking access to a range o sites meeting speciied criteria. China appears to
be the most nimble at responding to the shiting \eb, likely relecting a
deotion o the most resources, not to mention political will, to the iltering
enterprise.
A state wishing to ilter its citizens` access to the Internet has seeral initial
options: Domain Name System ,DNS, iltering, Internet Protocol ,IP, iltering,
or Uniorm Resource Locator ,URL, iltering.
16
Most states with adanced
iltering regimes implement URL iltering, as it can aoid een more drastic
oer-iltering or under-iltering situations presented by the other choices,
discussed below.
1
1o implement URL iltering, a state must irst identiy
where to place the ilters, i the state directly controls the Internet serice
proider,s, ,ISP,, the answer is clear. Otherwise, the state may require priate
or semi-priate ISPs to implement the blocking as part o their serice. 1he
technical complexities presented by URL iltering become non-triial as the

15
ee Deibert et al., ACCLSS DLNILD, .vra note 1.
16
Nart Villeneue, !b, toc/ P .aare..e.., NAR1 VILLLNLUVL: IN1LRNL1 CLNSORSlIP
LXPLORLR, leb. 14, 2005, !!!"&)*'J"#*+,CDD9,DC,8F,!54H.3#(GH.4H1$H)00*%22,.
1
lor instance, IP iltering orces the choice o blocking all sites sharing an IP address. A
recent OpenNet Initiatie bulletin ound more than 3,000 web sites blocked in an attempt to
preent access to only 31 sites. ee Collateral Blocking: liltering by South Korean
Goernment o Pro-North Korean \ebsites; OpenNet Initiatie Bulletin 009, Jan. 31, 2005,
www.opennetinitiatie.net,bulletins,009,. DNS blocking requires an entire domain and all
subdomains to be either wholly blocked or wholly unblocked. ee Villeneue, .vra note 16.

number o users grows to millions rather than tens o thousands. Some states
appear to hae limited oerall access to the Internet in order to keep URL
iltering manageable. 1he goernment o Saudi Arabia, or example, made
iltering a pre-requisite or public Internet access, delaying any such access or a
period o seeral years until the resources to ilter were set in place.
Citizens with technical knowledge can generally circument ilters that a state
has put in place. Some states acknowledge as much: 1he oerseer o Saudi
Arabia`s iltering program, ia the state-run Internet Serices Unit, admits that
technically say users can simply not be stopped rom accessing blocked
content. Lxpatriates in China, as well as those citizens who resist the state`s
control, requently ind up-to-date proxy serers through which they can
connect to the Internet while eading ilters. \hile no state will ultimately win
a game o cat-and-mouse with those citizens who are resourceul and dedicated
enough to employ circumention measures, many users will neer do so-
rendering iltering regimes at least partially eectie despite the obious
workarounds.
Some o the earliest theorizing about control in the online enironment, as
discussed in !bo Rvte. tbe êt.,
18
suggested that such state-run control o
Internet actiity would not work. It is important to note that states such as
China hae proen that an ambitious state can, by deoting substantial
technical, inancial, and human resources, exert a large measure o control oer
what their citizens do online. States, i they want, can erect certain orms o
gates at their borders, een in cyberspace, and can render them eectie
through a wide ariety o modes o control.
19
1hese controls hae proen the
claims o Jack L. Goldsmith and others who hae emphasized the extent to
which the online enironment can be regulated and the ways in which
traditional international relations theory will goern in cyberspace the same as
they do in real-space.
20

1hat does not mean that the issue is simple. lor starters, states ordinarily need
a great deal o help in carrying out iltering and sureillance regimes. Lnter
ISPs, many o which require a license rom the goernment to lawully proide
Internet access to citizens. Much Internet iltering is eected by these priate
ISPs under respectie states` jurisdictions, though some goernments partially
centralize the iltering operation at priate Internet Lxchange
Pointstopological crossroads or network traicor through explicit state-

18
vra note 12.
19
ee Jack L. Goldsmith and 1im \u, \lO CON1ROLS 1lL IN1LRNL1: ILLUSIONS Ol A
BORDLRLLSS \ORLD 65-86 ,Oxord Uniersity Press 2006,.
20
ee Jack L. Goldsmith, .gaiv.t C,beravarcb,, iv \lO RULLS 1lL NL1: IN1LRNL1
GOVLRNANCL AND JURISDIC1ION ,1hierer et al., eds., Cato Inst. 2003,.

run clearing points established to sere as gatekeepers or Internet traic. Some
goernments implement iltering at public Internet access points such as the
computers ound within cybercas. Such iltering can take the orm o
sotware used in many American libraries and schools or iltering purposes, or
normatie` iltering-goernment-encouraged interentions by shop-owners
and others as citizens sur the Internet in a public place.
Sometimes technical controls are not enough to constrain speech in the manner
that the censors want. 1he exercise o more traditional state powers can hae a
meaningul impact on Internet usage that does not require the complete
technical inaccessibility o particular categories o content. China, Vietnam,
Syria, and Iran hae each jailed cyber-dissidents.`
21
Against this backdrop, the
blocking o \eb pages may be intended to delier a message to users that state
oicials monitor Internet usage-in other words, making it clear to citizens that
someone is watching what you do online.` 1his message is reinorced by
methods to gather what sites a particular user has isited ater the act, such as
the requirement o passports to set up accounts with ISPs and tighter controls
o users at cybercas.
As we learn more and more about how Internet iltering takes place, the
problems o goerning` the Internet come more sharply into relie-about
how control is exerted, about how citizens in one state can or cannot connect to
others in another state, about the relationship between each state and its
citizens, and about the relationships between states.
Types of Content Filtered
Around the world, states are blocking access to inormation online based upon
its content-or what applications hosted at certain sites can do-or political,
religious, cultural, security, and social reasons. Sensitiities within these
categories ary greatly rom country to country. 1hese sensitiities oten track,
to a large extent, local conlicts. 1he Internet content blocked or social
reasons-commonly child saety, pornography, inormation about gay and
lesbian issues, and sex education-is more likely to be the same across countries
than the political and religious inormation to which access is blocked.

21
Reporters Sans lrontieres, vtervet vevie.: Cbiva, %&"*2="#*+,1&'%*&%'H%&%>1%H
(51&);:IIEE"5'>3 ,last accessed Aug. 25, 2010, ,1hirty journalists and seenty-two
netizens are now behind bars or reely expressing their iews.`,, Reporters Sans lrontieres,
vtervet vevie.: 1iet âv, %&"*2="#*+,1&'%*&%'H%&%>1%HJ1%'H&)>;:IIKF"5'>3 ,last accessed
Aug. 25, 2010, ,Vietnam is the world`s second biggest prison or netizens: it now has
seenteen o them behind bars.`,, Reporters Sans lrontieres, vtervet vevie.: ,ria,
%&"*2="#*+,1&'%*&%'H%&%>1%H24*1);:IITK"5'>3 ,last accessed Aug. 25, 2010, ,At least our
netizens are currently behind bars.`,, Reporters Sans lrontieres, vtervet vevie.:rav,
5''$U,,%&"*2="#*+,1&'%*&%'H%&%>1%H1*)&;:IITF"5'>3 ,last accessed Aug. 25, 2010, ,Some
thirty netizens hae been arrested since June 2009, and a dozen are still being detained.`,.

Seeral states carry out extensie iltering on certain topics. OpenNet Initiatie
testing has shown that 50 or more o the sites tested on a gien topic ,like sex
education, or application ,such as anonymization tools, are inaccessible. Very
rarely does any state manage to achiee complete iltering on any
topic,application. 1he only areas in which 100 iltering is approached are
pornography and anonymizers ,sites that, i let uniltered, would deeat iltering
o other sites by allowing a user to access any Internet destination through the
anonymizers` gateways,. States like Burma, which reportedly monitors e-mail
traic, also block a high percentage o ree e-mail serice proiders. Such
complete, or near-complete, iltering is additionally only ound in countries that
hae outsourced the task o identiying pornographic sites to one o seeral or-
proit American companies, and is ineitably accompanied by oer-blocking.
Outside o these three areas, OpenNet Initiatie testers are consistently able to
access some material o a similar nature to the sites that were blocked.
Filtering & Over-breadth
Internet iltering is almost impossible to accomplish with any degree o
precision. 1here is no way to stem the global low o inormation in a
consistently accurate ashion. A country that is deciding to ilter the Internet
must make an oer-broad` or under-broad` decision at the outset. 1he
iltering regime will either block access to too much or too little Internet
content. Very oten, this decision is tied to whether to use a home-grown
system or whether to adopt a commercial sotware product, such as Smartlilter,
\ebSense, or an oering rom security proider lortinet, each o which are
products made in the United States and are belieed to be licensed to countries
that ilter the Internet. Bahrain, or instance, has opted or an under-broad`
solution or pornography, its ISPs appear to block access to a small and
essentially ixed number o black-listed` sites. Bahrain may seek to block
access to pornographic material online, while actually blocking only token
access to such material. 1he United Arab Lmirates, by contrast, seem to hae
made the opposite decision by attempting to block much more extensiely in
similar categories, thereby sweeping into its iltering basket a number o sites
that appear to hae innocuous content by any metric. And \emen was rebuked
by the United States-based \ebSense or allegedly using the company`s iltering
system to block access to material that was not pornographic in nature, contrary
to the company`s policies.
22

Most o the time, states make blocking determinations to coer a wide range o
\eb content, commonly grouped around a second-leel domain name or the IP
address o a \eb serice ,such as www.twitter.com or 66.102.15.100,, rather

22
ee Jillian C. \ork, !ebev.e ar. Yevev`. Corervvevt frov vrtber oftrare |aate., OpenNet
Initiatie, Aug. 12, 2009, #$%&&%'"&%',.3#+,CDDK,DT,!%.2%&2%H.)*2H4%>%&2H
+#J%*&>%&'H=-*'5%*H2#='!)*%H-$0)'%2.

than based on the precise URL o a gien \eb page ,such as
www.twitter.com,username,, or a subset o content ound on that page ,such as
a particular image or string o text,. Iran, or instance, has used such an
approach to block a cluster o weblogs that the state preers not to reach its
citizens. 1his approach means that the iltering process will oten not
distinguish between permissible and impermissible content so long as any
impermissible content is deemed nearby` rom a network standpoint.
Because o this wholesale acceptance or rejection o a particular speaker or site,
it is diicult to know exactly what speech was deemed unacceptable or citizens
to access. It`s een harder to ascertain why, exactly, the speech is blocked.
Bahrain, a country in which we only ound a handul o blocked sites at the
outset o our irst round o testing, blocked access to a discussion board at
www.bahrainonline.org. 1he message board likely contains a combination o
messages that would be tolerated independently as well as some that would
appear to meet the state`s criteria or iltering. Likewise, we ound minimal
blocking or internal political purposes in the United Arab Lmirates, but the
state did block a site that essentially acted as a catalog o criticism o the state.
Our tests cannot determine whether it was the material coering human rights
abuses or discussion o historical border disputes with Iran, but in as much as
the discussion o these topics is taking place within a broad dissention-based
site, the calculation we project onto the censor in the United Arab Lmirates
looks signiicantly dierent than that or a site with a dierent ratio o
oensie` to approed` content.
lor those states using commercial iltering sotware and update serices to
maintain a current list o blocked sites matching particular criteria, OpenNet
Initiatie has noted multiple instances where such sotware has mistaken sites
containing gay and lesbian content or pornography. lor instance, the site or
the Log Cabin Republicans o 1exas was blocked by the U.S.-based Smartlilter
as pornography, and thereore, the apparent basis or its blocking by the United
Arab Lmirates. ,OpenNet Initiatie research shows that gay and lesbian
content is itsel oten targeted or iltering, and een when it is not explicitly
targeted, states may not be oerly concerned with its unaailability.,
23

As content changes increasingly aster on the \eb and generalizations become
more diicult to make by URLs or domains,-thanks in part to the rise o
simpler, aster, and aggregated publishing tools, like those ound on weblog
sites and ia other social networking applications-accurate iltering is getting
trickier or iltering regimes to address unless they want to ban nearly
eerything. Mobile deices hae urther added to the complexity o the
problem rom the censor`s iewpoint.

23
OpenNet Initiatie, IN1LRNL1 lIL1LRING IN 1lL UNI1LD ARAB LMIRA1LS IN 2004-2005: A
COUN1R\ S1UD\, leb. 2005, #$%&&%'"&%',2'-01%2,-)%.

lor example, ree web hosting domains tend to group an enormous array o
changing content and thus prooke ery dierent responses rom state
goernments. In 2004, Saudi Arabia blocked eery page on reespace.irgin.net
and www.erols.com.
24
loweer, research indicated the www.erols.com sites
had been only minimally blocked in 2002, and the reespace.irgin.net sites had
been blocked in 2002, but were accessible in 2003 beore being re-blocked in
2004. In all three tests, Saudi Arabia imposed URL blocking on
www.geocities.com ,possibly through Smartlilter categorization,, but only
blocked 3 o more than a thousand sites tested in 2004. Vietnam blocked all
sites tested on the www.geocities.com and members.tripod.org domains. In
OpenNet Initiatie`s recent testing, it has ound that 1urkey and Syria hae
been blocking all blogs hosted on the ree Blogspot serice.
25

China`s response to the same problem proides an instructie contrast. \hen
China became worried about bloggers, they shut down the main blogging
domains or weeks in the summer o 2004. \hen the domains came back
online, the blogging systems contained ilters that would reject posts containing
particular keywords.
26
Len Microsot`s MSN Spaces blogging sotware
preented writers rom publishing terms like democracy` rom China. In
eect, China moed to a content-based iltering system, but determined that the
best place or such content ealuation was not the point o \eb page access,
but the point o publication, and it possessed the authority to orce these ilters
on the downstream application proider. 1his approach is similar to that taken
with Google in response to the accessibility o disaored content ia Google`s
caching unction. Google was blocked in China until a mechanism was
implemented to preent cache access.
2
1hese examples clearly demonstrate the
length to which regimes will go to presere good` access instead o simply
blocking an entire serice.
1hese examples also demonstrate the increasing reliance by states on just-in-
time` iltering, rather than iltering that occurs in the same, consistent way oer
time. \hile the paradigmatic case o Internet iltering was initially the state that
wished to block its citizens rom iewing any pornography online at any time

24
Saudi Arabia blocked eery page on !!!"%*#32"(#> except or the root page at
!!!"%*#32"(#> itsel, potentially indicating a desire to manage perceptions as to the extent
o the blocking.
25
All data rom OpenNet Initiatie testing can be ound in the country-by-country summaries
at !!!"#$%&&%'"&%',.
26
liltering by Domestic Blog Proiders in China, OpenNet Initiatie Bulletin 008, Jan. 14,
2005, !!!"#$%&&%'1&1'1)'1J%"&%',.-33%'1&2,DDT,.
2
1his mechanism turned out to be extremely rudimentary, as outlined in a preious OpenNet
Initiatie bulletin. ee Google Search & Cache liltering Behind China`s Great lirewall,
OpenNet Initiatie Bulletin 008, Sept. 3, 2004,
!!!"#$%&&%'1&1'1)'1J%"&%',.-33%'1&2,DDI,.

,or instance, Saudi Arabia,, the phenomenon o a state blocking particular
speech or types o speech at a sensitie moment has become commonplace.
lor instance, China blocked applications such as 1witter and \ou1ube at the
time o the 20
th
anniersary o the 1iananmen Square demonstrations in June
2009. A ew weeks later, Iran blocked similar applications, including lacebook,
at the time o demonstrations in the streets o 1ehran. 1hese blocks are oten
lited once the trouble has past. One means o tracking these changes in the
aailability o applications and websites is a project called lerdict.org, which
enables people rom around the world to submit reports on what they can and
cannot access in real-time.
28

Alternate approaches that demand a iner-grained means o iltering, such as the
use o automated keywords to identiy and expunge sensitie inormation on
the ly, or greater manual inolement in choosing indiidual \eb pages to be
iltered, are possible, so long as a state is willing to inest in them. China in
particular appears prepared to make such an inestment, one mirrored by
choices demonstrated through more traditional media. lor example, China
allows CNN to be broadcast within the country with a orm o time delay, so
the eed can be temporarily turned o as when, in one case, stories about the
death o Zhao Ziyang were broadcast.
29
Online policing o speech, een in
what appears to be a borderless world,` can be carried out through technical
controls at many layers.
Law, Surveillance & Soft Controls
Just as dozens o states use technical means to block citizens rom accessing
content on the Internet, most also employ legal and other sot controls.
Sureillance practices are most commonly coupled with outright technical
censorship. Many states that ilter use a combination o media,
telecommunications, national security, and Internet-speciic laws and regulatory
schemes to restrict the publication o and access to inormation on the Internet.
States oten require ISPs to obtain licenses beore proiding Internet access to
citizens. Some states-China and 1urkey, or instance, which hae each
enacted special regulations to this eect-apply pressure on cybercas and
ISPs to monitor Internet usage by their customers. \ith the exceptions o
Saudi Arabia and Qatar, no country seems to explicitly communicate to the
public about its process or blocking and unblocking content on the Internet.
Most countries, instead, hae a series o broad laws that coer content issues
online, both empowering states that need these laws to carry out iltering

28
ee !!!"5%*01('"#*+. 1he histories o reports o these just-in-time blocking patterns can
be iewed rom this website.
29
ee Lric Priest, Reactiov. to tbe vtervet c ociet, 2001 e..iov ov v.ive.. ;Decevber 11, 10:1:
12:1:), Spring 2005,
(4.%*"3)!"5)*J)*0"%0-,.3#+2,+%>2,'G),OV*1%2'S%)('1#&V)$%*C"$0=.

regimes, and putting citizens on general notice not to publish or to access
content online that iolates certain norms.
Oten these sot controls are exercised through social norms or through control
at the ar edges o the network. Sometimes the state requires non-goernmental
organizations and religious leaders to register beore using the Internet to
communicate about their work. In China and in parts o the ormer Soiet
Union, ery oten the most earsome enorcer o the state`s will is the old
woman on one`s block, who may or may not be on the state`s payroll. 1he
control might be exercised, as in Singapore, largely through amily dynamics.
1he call by the local police orce to the Malaysian blogger to come and talk
about his web publishing might hae as much o an eect on expression as any
law on the books or technical blocking system.
\hether through adanced inormation technology, legal mechanisms, or sot
controls, a growing number o states around the world are seeking to control
the global low o inormation. Ordinarily, this control takes the orm o
blocking, through technical means, state`s citizens rom accessing certain
inormation online. In other instances, the blocking stops the state`s citizens
rom publishing inormation online, in eect disallowing people outside the
state rom hearing the oices o the state`s citizens. As a result, most iltering
regimes cause a chilling eect on the use o inormation technologies as a
means o ree expression, whether or political, religious, or cultural purposes.
From How to Police Speech Online
to How to Limit Speech
Restrictions Online
It is commonplace to argue that states hae generally regulated the Internet
lightly, but it is increasingly untrue. 1he author o a chapter in an important
recent book wrote, goernments exercise relatiely little control oer the
Internet, een though it has a tremendous impact on society.`
30
1his statement
misleads readers into thinking that the Internet might somehow be a reer, more
open enironment than traditional spaces are. Such a statement might hae
been true in the United States iteen years ago. But as o today, it is inaccurate.
lrom a global perspectie, both the importance o digitally-mediated
communications and the extent o regulation o speech continue to grow oer
time.
1he policing o speech in a borderless world brings with it a series o problems
that merit public discussion. 1he types o controls that take place at a local

30
larold Kwalwasser, Internet Goernance, iv C\BLRPO\LR AND NA1IONAL SLCURI1\, 491
,lranklin D. Kramer et al., eds., Nat`l Deense Uni. Press 2009,.

leel, about bullying or hate speech or any other issue that may arise, can be
etted in a public meeting o a school district or in a state legislature. 1he
online speech issues that are consistently on ederal agendas in the United
States-intellectual property restrictions, child saety, network neutrality,
deamation, and so orth-tend to play out in public orums like the legislatures
or in the court system. O course, there are instances where the debate about
speech controls occurs behind closed doors at the local and national leels as
well. Corporate-leel iltering, which can aect millions o employees, is one
example where the issues are rarely etted in a meaningul way. But the place
where the debate is most consistently and conspicuously absent is on the
international stage. It ought to merit meaningul consideration in the Internet
goernance debate.
1he practice o state-mandated Internet iltering, and related regulations like
sureillance, is now a widely-known act, but the hard problems that stem rom
this practice are inrequently discussed as a matter o public policy outside o
human rights and academic circles and the occasional national-leel hearing.
1he blocking and sureillance o citizens` actiity on the Internet-by irtue o
the network`s architecture, an issue o international dimensions-calls or
discussion at a multilateral leel. Rather than retting oer the iner points o
the domain name system, time would be better spent in Internet goernance
discussions on issues like transparency in Internet iltering or broad issues o
interconnection o the global network. 1he Internet iltering problem oers
much more to be gained-through rank discussion, i not action-and
proides an exercise worthy o an extraordinary gathering o world leaders who
want to talk about the global Inormation Society.`
On one leel, Internet iltering is a priate matter between a state and its
citizens as to what inormation citizens may access online.
31
States that censor
the Internet assert the right to soereignty. lrom the state`s perspectie, the
public interest, as deined in one state, such as Saudi Arabia, is dierent rom
the public interest as deined by the state in Uzbekistan, China, or the United
Kingdom. States can and do exercise their soereignty through control o the
inormation enironment.
But een i one accepts the state soereignty argument, that iewpoint should
not end the conersation about Internet iltering. 1he state-based censorship

31
Some states make an eort to suggest that their citizens ,in Saudi Arabia and the United
Arab Lmirates speciically, are largely in support o the iltering regime, particularly when it
comes to blocking access to pornographic material. lor instance, the agency responsible or
both internet access and iltering in Saudi Arabia conducted a user study in 1999, and
reported that 45 o respondents thought too much` was blocked, 41 thought it
reasonable,` and 14 ound it not enough.` 1hese studies stand or the proposition, in
the context o our report, that some states that ilter seek to make the case that their iltering
regime enjoys popular support, not that such support necessarily exists.

and sureillance practices o any state aect the citizens and businesses o other
states in the context o the interconnected global communications network.
Increasingly, the censorship and sureillance practices o states reach past the
web-browsing habits o their own citizens. ligh-proile debates between
Canadian company Research in Motion, maker o BlackBerry smartphones, and
the United Arab Lmirates and Saudi Arabia about the ability o the state to
intercept BlackBerry communications make this point clear.
32
1he Internet
blocking that takes place in one state can also aect the network at a technical
leel in other jurisdictions, as Pakistan ound out when it brought down
\ou1ube globally or two hours in early 2009.
33

A global discussion about the relationship between these iltering and
sureillance practices and human rights is necessary and could be extremely
ruitul. Speciically, states might consider rules that relate to common
standards or transparency in Internet iltering and sureillance practices as they
relate to indiiduals and those corporations drawn into the process. On a
broader leel, the issue raised here is about interconnection between states and
the citizens o those states-and ultimately about what sort o an Internet we
want to be building and whether the global low o inormation is a sustainable
ision.
lor instance, we hae yet to join the ethical interests at play in iltering. States
ary greatly in terms o how explicitly the iltering regime is discussed and the
amount that citizens can come to know about it. No state OpenNet Initiatie
studied makes its block list generally aailable.
34
1he world leaders who gather

32
ee Adam Shreck, |. tac/err, Crac/aorv .ffect. 1i.itor. 1oo, ASSOCIA1LD PRLSS, August
3, 2010,
!!!"+##+3%"(#>,5#2'%0&%!2,)$,)*'1(3%,WB%XQ91Y8QB5WQR%SZ5MF5%-FB[!H
?+\:]ZK\@[ZY^D. ee at.o .vtbov, DiPaota c 1iriav atava, |. to v.eva tac/err,
errice Citivg ecvrit,, BLOOMBLRG, August 1, 2010, !!!".-21&%22!%%G"(#>,&%!2,CD8DH
DTHD8,-H)H%H'#H2-2$%&0H.3)(G.%**4H2%*J1(%2H(1'1&+H2%(-*1'4"5'>3.
33
ee Declan McCullagh, or Pa/i.tav Kvoc/ea Yov1vbe Offtive ;ava bor to va/e .vre it verer
baev. agaiv),` CNL1, lebruary 25, 2008, &%!2"(&%'"(#>,T:D8H8DETF/:HKTETI99HE"5'>3.
34
Saudi Arabia publishes its rationale and its blocking practices on an easily accessible website,
at !!!"12-"&%'"2),2)-01H1&'%*&%',(#&'%&%'H=13'*1&+,=13'*1&+"5'> ,1he Internet Serices
Unit oersees and implements the iltration o web pages in order to block those pages o
an oensie or harmul nature to the society, and which iolate the tenants o the Islamic
religion or societal norms. 1his serice is oered in ulillment o the directions o the
goernment o Saudi Arabia and under the direction o the Permanent Security Committee
chaired by the Ministry o the Interior.`,. In Saudi Arabia, citizens may suggest sites or
blocking or or unblocking, in either Arabic or Lnglish, ia a public website. Most sites
include a block-page, indicating to those seeking to access a website that they hae reached a
disallowed site. Most states hae enacted laws that support the iltering regime and proide
citizens with some context or why and how it is occurring, though rarely with any degree o
precision. As among the states we hae studied, China is one o the states that obscures the
nature and extent o its iltering regime to the greatest extent through a long-running series
o conlicting public statements about its practices in this respect.

periodically at United Nations-sponsored meetings and at the Internet
Goernance lorums could make the most o their leadership by seeking to
establish a set o best practices related to Internet iltering and the transparency
related to iltering regimes. 1hey might also ocus proitably on the diicult
problems acing those multinational companies that do business in regimes that
require iltering and sureillance o the network in ways that would not be
legally permissible in the company`s home jurisdiction, as the Global Network
Initiatie has in its priate capacity.
35
1he issue o speech regulation in a
borderless world is too important to leae only to those handul o companies,
right-minded though they may be, which are seeking to do the right thing in a
geopolitically complex regulatory enironment.
1he critical question in the next digital decade is not whether speech can be
policed in a borderless world, but whether we should set new limits on the
extent and manner in which it is policed today. \e should ask, too, whether it
is today easier, in act, than in the past to police speech in a digitally-mediated
world, and what the ramiications are or ciil liberties i so. 1he Internet is
becoming larger and more ractured each day. \e should not pretend the
Internet is a lightly regulated` medium as though the calendar on our kitchen
wall reads 1985` at the top. 1rends that support more speech rom more
people in more places around the globe-using mobile applications generally,
such as blogs, wikis, 1witter, SMS, and so orth-are countered by the
increasing sophistication and reach o Internet iltering and sureillance
practices. A richer understanding o the complexities at play in Internet iltering
and other speech restrictions would help deelop a oundation that does not yet
exist or building a sustainable, and truly global, network that will continue to
bring with it the innoation, jobs, serices, and other social beneits that it
promises.

35
ee !!!"+3#.)3&%'!#*G1&1'1)'1J%"#*+.

The Role of the Internet
Community in Combating
Hate Speech
By Christopher Wolf
*

In less than twenty years, the Internet has eoled to become an
unprecedented tool or inormation, communication, entertainment and
commerce. Much o the progress was made possible by the protections
o the lirst Amendment and the absence o legal constraints. But there
is a growing dark side to the Internet, which raises the question o how
to police harmul content. Online child predators are a well-known
blight, a requent ocus o headlines and teleision news. Cyber-bullying
also has receied signiicant attention recently, especially in the wake o
teen suicides.
Less reported but equally troubling is the act that the Internet has
become a technology embraced by racists, anti-Semites, homophobes
and bigots o all kinds to spread their messages o hate. 1he online
haters use all o the tools o the Internet, rom static websites, to
streaming audio and ideo, to social networking sites like lacebook.
No longer relegated to meeting in dark alleys and the basements o
abandoned buildings, or to mailing their propaganda in plain brown
wrappers, hate groups hae a platorm to reach millions around the
world. 1hey seek to ictimize minorities, to embolden and mobilize like-
minded haters, and to recruit ollowers. And in their wake, an online
culture has deeloped-aided by the mask o anonymity-in which
people who would neer consider themseles members o hate groups
employ racial, religious and other epithets as part o their ocabulary in
posting comments to news stories on mainstream sites and in other
aspects o online lie. In turn, the common appearance o such epithets
desensitizes readers, making hate speech and the denigration o
minorities normal.`

Christopher \ol is a partner in the law irm logan Loells LLP where he leads the Priacy
and Inormation Management practice. le ounded and chairs the Internet 1ask lorce o
the Anti-Deamation League, and is Past Chair o the International Network Against Cyber-
late ,INACl,.

One recent example o how haters are using the Internet occurred on the
lourth o July in 2010. As Americans were celebrating that eent, a new
Lent` was announced on lacebook, entitled Kill a Jew Day.`
1
1he
lacebook host` or the Lent wrote, \ou know the drill guys,` and he
urged ollowers to engage in iolence anywhere you see a Jew` between
July 4 and July 22. A Nazi swastika adorned the Lent page.
1he posting o that sickening Lent prompted a wae o anti-Semitic
rants on lacebook in support o the targeting o Jewish people. But it
also prompted a counter-eent on lacebook entitled One Million
Strong Against Kill a Jew Day` ,whose supporters actually numbered,
more modestly, in the thousands,.
2
And, pursuant to the lacebook
1erms o Serice, complaints about the Kill a Jew Day` eent to
lacebook administrators resulted in the company disabling the Lent
page.
1he outrage oer the lacebook Lent site was justiied, not just because
o the ile anti-Semitism underneath it or the gloriied display o a
swastika. People also objected to the site because they know that
Internet messages can and do inspire iolence. Online anti-Semitic hate
speech has been implicated in real-world acts o iolence, such as an
attack on Nobel Laureate and lolocaust surior Llie \iesel by a
lolocaust denier in 200,
3
and the 2009 murder o a guard at the
lolocaust Memorial Museum in \ashington, D.C., by a white
supremacist who maintained his own online hate site and who was
egged-on by ellow haters.
4
\ords hae consequences, and indeed
inspire acts o hate and iolence.
1his recent example o online hate proides another opportunity to
examine what society`s response to online hate speech should be. \hat

1
ee \aako Lappin, Kitt a ]er Page ov aceboo/ ar/. vror, 1lL JLRUSALLM POS1, July 5,
2010, 5''$U,,!!!"_$#2'"(#>,Y%!125`#*30,Y%!125P%!2,W*'1(3%")2$?61078TDF9I.
2
ee Amada Schwartz, .vtieviti.v r. aceboo/, JL\ISl JOURNAL, July 13, 2010,
5''$U,,!!!"_%!125_#-*&)3"(#>,(#>>-&1'4,)*'1(3%,)&'1H
2%>1'12>/J2/=)(%.##G/CD8DDE8:,.
3
Suzanne lerel, otocav.t vrriror, ôbet Peace Prie !ivver tie !ie.et .ttac/ea iv .. otet,
S.l. ClRONICLL, leb. 9, 200, 5''$U,,!!!"2=+)'%"(#>,(+1H
.1&,.3#+2,&!a(51G,0%')136.3#+107:Cb%&'*4/1078::T9.
4
Michael L. Ruane, Paul Duggan & Clarence \illiams, .t a Movevt of orror, . vr.t of
Deaat, 1iotevce, 1lL \ASlING1ON POS1, June 11, 2009,
5''$U,,!!!"!)251&+'#&$#2'"(#>,!$H
04&,(#&'%&',)*'1(3%,CDDK,DI,8D,WSCDDKDI8DD8EIT"5'>3.

is the best way to police sites like the Kill a Jew Day` Lent page on
lacebook, and the thousands o hate-illed ideos uploaded to \ou1ube,
and the white supremacist websites designed to recruit young people,
gloriying iolence against minorities
One isceral response to the prolieration o online hate is: 1here
ought to be a law.` Legal rules are the way a society decrees what is right
and what is wrong, and since hate speech is wrong, it seems logical that
the law would be employed to police it. A legal ban on hate speech and
the criminalization o its publication is indeed an alternatie in some
jurisdictions. But, o course, it is not an option in the United States where
the lirst Amendment gies broad latitude to irtually all speech, een the
most repugnant. ,Only direct threats against identiiable targets are
criminalized.,
Legislatures around the world hae heeded the call or laws
encompassing Internet hate speech. 1he hate speech protocol to the
Cybercrime 1reaty is a prime example o a heralded legal solution to the
problem.
5
It was designed to eliminate racist sites rom the Internet
through criminal penalties.
lrom Brazil to Canada, and rom South Arica to Great Britain, there are
legal restrictions on hate speech, online and oline. In much o Lurope,
denial o the lolocaust ,online or oline, is orbidden. In Germany,
een displaying the swastika is a crime. 1he enorcement o laws against
lolocaust deniers-gien the bitterly sad history o those countries-
seres as a message to all citizens ,especially impressionable children, that
it is literally unspeakable to deny the lolocaust gien the horrors o
genocide inlicted in those countries.
Still, there are many who beliee that prosecutions, such as that o
lolocaust denier Daid Iring in Austria,
6
promoted his isibility and
stirred up his benighted supporters, rather than quelling uture hate
speech and enlightening the public.
Moreoer, laws against hate speech hae not demonstrably reduced hate
speech or deterred haters. 1he hate speech protocol to the Cybercrime

5
1he Conention on Cybercrime No. 23, 2001, Lurop. 1.S. No. 185,
5''$U,,(#&J%&'1#&2"(#%"1&',M*%)'4,%&,M*%)'1%2,\'>3,8T9"5'>.
6
ee .v.tria ota. otocav.t Devier, BBC NL\S, No. 1, 2005,
5''$U,,&%!2"..("(#"-G,C,51,%-*#$%,FFFIIFI"2'>.

1reaty, or example, has not reduced online hate. 1he shield o Internet
anonymity and the iral nature o online hate make legal policing an
unrealistic challenge, except in cases where authorities want to set an
example.` And since the U.S., with its lirst Amendment is essentially a
sae-haen or irtually all web content, shutting down a website in
Lurope or Canada through legal channels is ar rom a guarantee that the
contents hae been censored or all time. 1he borderless nature o the
Internet means that, like chasing cockroaches, squashing one does not
sole the problem when there are many more waiting behind the walls-
or across the border.
Many see prosecution o Internet speech in one country as a utile
gesture when the speech can re-appear on the Internet almost
instantaneously, hosted by an Internet serice proider ,ISP, or social
networking site in the United States. Moreoer, in the social networking
era, the ability o people to upload ar outpaces the ability o the police to
track and pursue oending speech.
Like the prosecution in Austria o Daid Iring, the prosecutions in
Germany o notorious lolocaust deniers and hate site publishers Lrnst
Zundel
and lrederick 1oben

8
sent messages o deterrence to people that
make it their lie`s work to spread hate around the world that they may
well go to jail as well. And, again, such prosecutions expressed society`s
outrage at the messages. But all one need do is insert the names o those
criminals in a search engine, and you will ind websites o supporters
paying homage to them as martyrs and republishing their messages.
Len some ree speech adocates around the world applaud the use o
the law to censor speech when it is bate speech because o the pernicious
eects o hate speech on minorities and children, and because o its
potential to incite iolence. But many o those same people object to the
use o the law by repressie regimes like China to censor speech it deems
to be objectionable as hate directed towards the Chinese goernment. It
is not easy to draw the line between good and bad state use o censorship
because deining what is hate speech can be quite subjectie. Giing the
state the power to censor is problematic, especially gien the potential or
abuse.

ee Zvvaet Cet. ire Year. frov Cervav Covvt, \INNIPLG lRLL PRLSS, leb. 16, 200,
5''$U,,!!!"!1&&1$%+=*%%$*%22"(#>,512'#*1(,:C889IKF"5'>3.
8
ee Stee Kettmann, Cervav ate ar: ô Dev,ivg t, \IRLD, Dec. 15, 2000,
5''$U,,!!!"!1*%0"(#>,$#31'1(2,3)!,&%!2,CDDD,8C,FDIIKc1?aaD_M=93&B5.

1his is not to say that law has no role to play in ighting online hate
speech-ar rom it. But countries with speech codes intended to
protect minorities should make sure that the proper discretion is
employed to use those laws against Internet hate speech, lest the
enorcement be seen as ineectual and result in a diminished respect or
the law. And, again, the realities o the Internet are such that shutting
down a website through legal means in one country is ar rom a
guarantee that the website is shuttered or all time.
1hus, the law is but one, albeit limited, tool in the ight against online
hate.
Counter-speech-exposing hate speech or its deceitul and alse
content, setting the record straight, and promoting the alues o
tolerance and diersity, has an important role to play in the policing o
online hate speech. 1hat is the thrust o the lirst Amendment. 1o
paraphrase U.S. Supreme Court Justice Brandeis, sunlight is still the best
disinectant-it is always better to expose hate to the light o day than to
let it ester in the darkness. One answer to hate speech is more speech.
1he lacebook eent One Million Strong Against Kill a Jew Day,` een
i ar short o a million members, is a iid example o the power o
counter-speech as a ehicle or society to stand up against hate speech.
And, o course, education rom an early age on Internet ciility and
tolerance would go ar to stem the next generation o online haters.
An equally important and powerul tool against hate speech is the
oluntary cooperation o the Internet community, including Internet
Serice Proiders, social networking companies and others. \hen
lacebook enorced its 1erms o Serice ,which requires users not to
post content that: is hateul, threatening, or . incites iolence.`
9
, and
disabled the Kill Jew Day` eent site,
10
that was a powerul example o
an Internet company exercising its own lirst Amendment rights to
ensure that it remained an online serice with standards o decency. 1hat
oluntary act was quick and eectie. A legal action against lacebook or
hosting the site-impossible in the U.S. but iable elsewhere around the
world-would hae been expensie, time-consuming and no more

9
Statement o Rights and Responsibilities, lacebook,
5''$U,,!!!"=)(%.##G"(#>,'%*>2"$5$.
10
Lappin, .vra note 1.

eectie. 1he chilling eect o a legal action against lacebook may hae
resulted in undue restrictions by lacebook on uture user postings.
Voluntary enorcement by Internet companies o sel-established
standards against hate speech is eectie. I more Internet companies in
the U.S. block content that iolate their 1erms o Serice, it will at least
be more diicult or haters to gain access through respectable hosts. 1he
challenge, o course, is with social media sites where postings occur
constantly and rapidly. Social media companies normally wait or a user
complaint beore they inestigate hate speech posted on their serice, but
the prolieration o hate-illed postings outpaces the eectieness o such
a notice and take down` arrangement. New monitoring techniques to
identiy hate speech as it is posted may be in order.
In the era o search engines as the primary portals or Internet users,
11

cooperation rom the Googles o the world is an increasingly important
goal. 1he example o the Anti-Deamation League
12
and Google with
the site Jew \atch` is a good one.
13
1he high ranking o Jew \atch in
response to a search inquiry was not due to a conscious choice by
Google, but was solely a result o an automated system o ranking.
Google placed text on its site that explained the ranking, and gae users a
clear explanation o how search results are obtained, to reute the
impression that Jew \atch was a reliable source o inormation, and
linked to the ADL site or counter-speech.
14

In short, igilance and oluntary standards are more eectie than the
law in dealing with the increasing scourge o online hate speech. late
speech can be policed` in a borderless world, but not principally by the
traditional police o law enorcement. 1he Internet community must
continue to sere as a neighborhood watch` against hate speech online,

11
A study by Nielsen ound that 3 percent o respondents used search engines when looking
or inormation, as compared to 34 percent using portals and only 18 percent using social
media. ee Jon Gibs, ociat Meaia: 1be êt Creat Catera, for Covtevt Di.corer,., NILLSLN, Oct.
5, 2009, 5''$U,,.3#+"&1%32%&"(#>,&1%32%&!1*%,#&31&%/>#.13%,2#(1)3H>%01)H'5%H&%?'H
+*%)'H+)'%!)4H=#*H(#&'%&'H012(#J%*4,.
12
1he Anti-Deamation League was ounded to combat anti-Semitism. lor more inormation,
see 5''$U,,!!!")03"#*+,.
13
ee Coogte Re.ova. to ]er !atcb Covtrorer.,, \LBPRONL\S, April 15, 2004,
5''$U,,!!!"!%.$*#&%!2"(#>,'#$&%!2,CDDF,DF,89,+##+3%H*%2$#&02H'#H_%!H!)'(5H
(#&'*#J%*24.
14
Google, .v tavatiov of Ovr earcb Re.vtt., 5''$U,,!!!"+##+3%"(#>,%?$3)&)'1#&"5'>3.

saying something when it sees something,` and working with online
proiders to enorce community standards.


555

CHAPTER 10
WILL THE NET LIBERATE THE WORLD?
Can the Internet Liberate the World? 557
Lgeny Morozo
Internet Freedom: Beyond Circumvention 565
Lthan Zuckerman

556 CHAPTER 10: WILL THE NET LIBERATE THE WORLD?


Can the Internet
Liberate the World?
By Evgeny Morozov
*

It may be useul to start by laying out the basics. Anyone pondering the
question posed in the title o this essay most likely assumes that there exists
some powerul orces o oppression rom which the world could and should be
liberated. A list o such problems is as ininite as it is intuitie: lrom poerty to
racism and rom pollution to obesity, oppressie orces` seem to be all around
us.
\et, or some reason, these are rarely the kind o problems that one wants to
ight with the help o technology, let alone the Internet. It`s in soling political
rather than socio-economic problems that the Internet is presumed to hold the
greatest promise. Most speciically, it is its ability to undermine repressie
goernments that is widely discussed and admired, een more so as Internet
companies like Google ind themseles struggling with the likes o the Chinese
goernment.
1wo eatures o the Internet are oten praised in particular: 1, its ability to
quickly disseminate any kind o inormation-including the inormation that
authoritarian goernments may not like-and 2, its ability to allow like-minded
indiiduals to ind each other, to mobilize supporters and to collectiely pursue
uture goals-including democratization. 1he hype surrounding Iran`s 1witter
Reolution o 2009 was probably the strongest public maniestation o high
hopes or the transormatie potential o the Internet, only a rare pundit did not
predict the eentual collapse o the Iranian regime under the barrage o angry
tweets rom its citizens.
Still, such praise is not without merit. Len the hardest skeptics would grant
these two eatures to the Internet, to deny that it does enhance the citizens`
ability to inorm ,and get inormed, and to mobilize-what the Internet theorist
Clay Shirky calls ridiculously easy group orming`
1
-would be to deny the
obious. 1he skeptics would also hae no trouble acknowledging that both o
these eatures are constantly under threat, as goernments keep implementing
new systems o censorship and control.

Lgeny Morozo is the author o 1lL NL1 DLLUSION: 1lL DARK SIDL Ol IN1LRNL1
lRLLDOM ,Public Aairs, 2011,. le`s also a isiting scholar at Stanord Uniersity, a ellow
at the New America loundation and a contributing editor to loreign Policy magazine.
1
CLA\ SlIRK\, lLRL COMLS LVLR\BOD\: 1lL PO\LR Ol ORGANIZING \I1lOU1
ORGANIZA1IONS 54 ,Penguin Press 2008,, quoting social scientist Sbastien Paquet.

Nor does anyone really contest the act that the Internet has proed quite
resilient against such attacks, giing rise to numerous tools to circument
goernment censorship. lor many, the act that an institution as powerul as the
U.S. goernment has trouble reining in a luid and mostly irtual organization
like \ikiLeaks is a testament to the power o the Internet, een though the
morality o \ikiLeaks` actions ,in publishing leaked inormation about the U.S.
military occupation o Aghanistan, is still widely disputed.
loweer, conceding that the Internet helps to disseminate inormation and
mobilize campaigners around certain causes is not quite the same as conceding
that authoritarian regimes are doomed or that democracy is ineitable. 1here
may be good independent reasons to campaign or greater reedom o
expression on the Internet-but one shouldn`t presume that such reedoms
would necessarily translate into democratization.
lor the Net-and its two powerul eatures discussed aboe-to be able to
liberate the world` rom authoritarianism, one needs to make a ew urther
assumptions. lirst, one needs to assume that modern authoritarian regimes
derie their power primarily by suppressing the actiities that the Internet helps
to ampliy: i.e., dissemination o inormation and popular mobilization around
speciic causes. Second, one also needs to assume that the Internet won`t hae
any other political and social eects that may-i only indirectly-create new
modes o oppression,` entrenching authoritarianism as a result. In other
words, the Internet can only delier on its liberating promise as long as the
things it oers are also the things that the ight against authoritarianism
requires-and as long as it doesn`t produce any other regime-strengthening
eects that may inadertently undermine that ight.
On initial examination, the irst assumption seems to hold. 1here is no shortage
o suppression o both inormation and mobilization opportunities in modern
authoritarian states. 1heir rulers hae not lost the desire to guard their secrets or
regulate how their citizens participate in public lie. 1he act that some orms o
censorship persist een in democratic societies suggests that goernments do
not really aspire to lit all the digital gates and let inormation low reely. 1he
urge, then, is to ind ways to break through those gates-and the Internet seems
to excel at the job. I there is one thing that techies and hackers know how to
do well, it`s to build tools to pierce irewalls.
But suppose that such tools can be ound and can een achiee the kind o
scale where all Internet users in China or Iran hae access to them. \hat would
the eect be on their populations and their goernments I`d like to propose
that one`s answer to this question depends mostly on one`s iews about the
sources o legitimacy o modern authoritarianism.
1hose who beliee that such legitimacy is deried primarily through the
brainwashing o their citizens are justiiably ery excited about the Internet.

Moreoer, they are usually ery quick to predict the ineitable all o
authoritarian goernments. Ater all, their theoretical conception o
authoritarianism posits that once the inormation gates are open, brainwashing
loses much o its eectieness, people realize they hae been lied to all along-
and they rebel as a result.
At irst sight, the contemporary global situation may seem to indicate such
iews. Many modern authoritarian goernments-and here cases like Belarus,
China, Russia come to mind-enjoy strong leels o support rom ast swathes
o their populations. One may quibble about the ways by which the Kremlin has
solidiied its power in the last decade-many o those ways are ar rom
democratic-but irtually all opinion polls reeal that the Kremlin`s policies are
genuinely popular. Ditto China, where the goernment is one o the most
trusted institutions in the country, enjoying a leel o trust that the U.S.
Congress could only dream o.
Is it all because o brainwashing I the answer is yes,` then there are, indeed,
good reasons to be optimistic about the power o the Internet. 1he moment the
authoritarian goernments` monopoly oer inormation disappears, any
manipulations o truth that were possible in an age o inormation scarcity
would be impossible.
1his, I`d like to propose, may be a ery simplistic reading o the situation-and
a reading that is also extremely insensitie to historical and social orces. 1here
is much more to the legitimacy o modern authoritarian states than just their
skillul manipulation o inormation. Many authoritarian regimes-Belarus,
China and Russia are again excellent examples but one could also add
Azerbaijan, Kazakhstan, and Vietnam to the list-hae made genuine adances
in economic deelopment, many o them thanks to their embrace o technology
in general and the Internet in particular. lurthermore, in many o these
countries, the once extremely contentious political lie has stabilized as well,
allowing their populations to enjoy a rare period o peace and prosperity-een
it came at the hety price o haing their goernments tighten the ales on
reedom o expression or reedom o assembly.
It seems disingenuous to argue that modern Russians or Chinese do not
appreciate the act that today they purchase considerably more commodities-
including luxurious ones-than they could 20 years ago, or that they can trael
the world much more reely, or that they-at least online-can consume any
kind o entertainment they want, regardless o its origins. Placed in the historical
context and compared against other possible scenarios o where these two
countries may hae been had their rulers not embarked on a series o reorms,
such achieements look een more impressie.
Should it turn out that large segments o the populations o authoritarian states
are well aware o the kind o human rights iolations that are needed in order to

sustain the impressie rates o economic progress-or, een more shockingly,
that they are actually supportie o such iolations-a strategy o un-
brainwashing` simply would not work. I modern authoritarianism is as much a
product o a social contract as modern democracy is, then changing the
attitudes o those who hae long gien up the ight or reedom would take
much more than just exposing them to acts.
1he big question that \estern do-gooders should be asking themseles here is
not whether the Internet can liberate the world but whether the world actually
ravt. to be liberated. Aboe all, this is a question o whether capitalism
unburdened by democratic norms and ideals is sustainable in the long term-a
possibility that goes directly against the theory that the logic o capitalism
ineitably leads to democracy, a iew that was extremely popular in the early
1990s.
2

I capitalism can get by without democracy, the ability to spread subersie
inormation that might reeal the horrors o the regime looks considerably less
impressie, or the populations that the \est seeks to liberate are already well-
aware o what`s going on and many o them may hae simply chosen to look
the other way in expectation o a better lie or their children.
Granted, it`s not just acts that may help change their attitudes. One may use the
Net to distribute subersie poetry and iction that would reawaken ,or, in most
cases, create rom scratch,, the political consciousness o those liing under
authoritarianism. 1echnology would certainly be o great help here, both in
terms o helping to distribute such materials but also in terms o protecting
those who access them. But the ability o such materials to incite people to
democratic change is not just a unction o how many people read them, rather,
it`s a unction o how well-argued such materials are.
It seems that een i the \est succeeded in distributing 11, Dar/ve.. at ôov,
or rare êr !orta to eery single citizen o an authoritarian state, this might
not lead to a reolution, simply because those books oer a poor critique o the
actually existing modern-day authoritarianism, which has come to terms with
both \estern popular culture and globalization.

2
ee, e.g., Commission on Security & Cooperation in Lurope ,U.S. lelsinki Commission,,
Brieing on 1ritter .gaiv.t 1,ravt.: êr Meaia v .vtboritariav Regive., Oct. 22, 2009, aailable
at !""#$%%&'&()*+,%-./(0)&12345'(6&"-+.78+."(."9(&+:/');-(<=(">-?@
8+."(."9(&+:/B-/7CDE@9(*-+.B-/7F@G''5(B-/7F@8+."(."HI#(7JKL@8+."(."9(
&+:/HI#(7L@84G=7MENOOEDM@84HPQRS7TDEOCUUN, Nicholas Kristo, 1ear Dorv
1bi. C,berratt!, N.\. 1IMLS, June 1, 2009, araitabte at !""#$%%<<<).I"-2(')&+2%EFFT%FD%
NV%+#-.-+.%NVW:-'"+1)!"2?3B:7N, L. Gordon Croitz, Mr.. Ctivtov, 1ear Dorv tbi. C,berratt,
\ALL S1. JOURNAL, May 3, 2010, aailable at !""#$%%+.?-.()<'X)&+2%>:"-&?(%
YLNFFFNCECFUEOCVOFCDFVNFCUOUENTFEECTECOUMDC)!"2?, lRANCIS lUKU\AMA, 1lL LND
Ol lIS1OR\ AND 1lL LAS1 MAN ,1992,.

I support or modern authoritarianism is not the product o ignorance and
brainwashing but rather o a rational calculation that, under the present
conditions, authoritarianism is the best way to generate and presere economic
growth, the Internet`s ability to disseminate and mobilize, while ery impressie
in itsel, may not delier the kind o beneits that so many in the \est expect.
1o put it simply, citizens o authoritarian states may not be vvivforvea-so
getting them ivforvea is going to be o only limited alue. As such, the Net`s
ability to liberate the world is seerely constrained by the absence o a strong
intellectual ision or how a liberated Russia or China would look ,and work,
like.
Now, i the Soiet experience is anything to judge by, reolutions don`t need
the absolute majority o the population to be successul. In other words, it may
be possible that a small group o politically actie citizens could take adantage
o political openings at the right moment and push or signiicant reorm-or
the oerthrow o a goernment altogether. In situations like this, the Internet`s
ability to mobilize may indeed come ery handy.
Seeral caeats are in order here. lirst, obiously, this doesn`t mean that the
Internet can help create such political openings-those are usually created by
structural actors. As much as it is tempting to beliee that it was ax machines
and photocopiers that brought down communism in Lastern Luropean
countries, one would probably be better o studying their dismal economic
record in the late 1980s. So ar, it seems that the inormation reolution-which
many hae taken to mean the end o authoritarianism eerywhere-has, oerall,
had a positie impact on the rates o economic growth in modern authoritarian
regimes-and, to this extent, that reolution may actually hae .trevgtbevea these
regimes.
1he second caeat is that there is little certainty that the group with the best
ability to mobilize will also be the group with the most impressie democratic
agenda. Once again, the Soiet example-with a tiny group o Bolsheiks
gaining control o a country as massie as Russia-is quite instructie. Not all
reolutions are democratic in character, and more than one o them ended up
with the least democratic groups gaining power. Al-Qaeda is ar better at using
technology to mobilize the masses than the liberal oices o the Middle Last,
the Russian nationalists, likewise, are ar more creatie online than the
democratic and pro-\estern opposition.
1hird-and, perhaps, most important-what happens in between political
openings matters a great deal as well. It`s simply not true that, rom the
perspectie o an authoritarian state, all social mobilization is harmul. 1ake the
case o China. 1hanks to the extremely ibrant nationalist sector o the
country`s blogosphere, the Chinese goernment is oten pushed to adopt a
much more aggressie posture-towards 1aiwan, Japan, South Korea-than

they might otherwise hae done. Such hawkishness in their oreign policy may
or may not bolster their legitimacy, the short answer is that we hae to look at
the context. lor our purposes, it seems clear that we won`t be able to
understand the role o the Internet-een i one concedes that it is, indeed,
conducie to more mobilization and contestation-i we study that role outside
o the socio-political enironment in which it is embedded. 1he assumption that
greater opportunities or social mobilization equals greater odds that
democracy will preail in the long run` simply is not true.
1his last point highlights the problem with the second grand assumption that
we still need to examine-namely, that the Net won`t hae any other primary or
secondary eects on the quality and sustainability o authoritarianism. 1he
account that prioritizes the role o inormation dissemination and mobilization
usually rests on a ery simplistic, een reductionist, theory o authoritarianism.
It presumes the existence o an authoritarian chimera-the goernment-which
controls its citizens through a combination o sureillance and coercion.
Citizens, the theory goes, would take immediate adantage o the Internet and
use it to push against the goernment.
But why wouldn`t the goernment do the same to push against the citizens In
order to understand the oerall impact o the Net on the struggle or
liberation,` one must study how it may hae also acilitated goernment
monitoring and control o what their citizens do. Len the most optimistic
obserers o the Net would concede that social networking, un as it is, may not
necessarily be the best way to protect one`s data-both because no social
networking site is secure rom occasional data leaks and because secret police
around the world hae now, inadertently, obtained the ability to map the
connections between dierent actiists, see how they are related to oreign
unders, and so on.
lurthermore, there is a ibrant and rapidly-expanding global market in actiities
like ace recognition, which makes the identiication o those who participate in
anti-goernment protests much easier-oten this actually happens by
comparing party photos they themseles upload to social networking sites with
the photos taken at the protest rallies. Seasoned actiists may, o course, be
smart enough to steer away rom social networking sites, but this hardly applies
to the rest o the population.
But conceding that it`s not just anti-goernment actiists who hae been
empowered by the Net is only part o the story. 1he truth is that we simply
can`t easily classiy all social orces into pro-` and anti-goernment` simply
based on the location o their oices ,e.g., the secret police are in, the unions are
out,. In reality, modern authoritarian regimes derie their power rom a much
more dierse pool o resources than sheer brute orce or sureillance. 1o
understand what makes modern authoritarian regimes tick, one thus needs to

look at a whole range o other political, social, and cultural actors: religion,
history, nationalism, geography ,e.g., the relations between the ederal center and
the periphery,, rates o economic growth, corruption, goernment eiciency,
ear o a oreign inasion and so orth. Many o these actors hae successully
been co-opted by modern authoritarian rulers to justiy and prolong their rule.
It`s easy to imagine an authoritarian regime that would become stronger as a
result o ,a, an increase in religious sentiment among its population, ,b, the
promotion o a particular interpretation o recent history that would justiy the
current political regime as ineitable and an unambiguous improement oer its
predecessors, and ,c, impressie rates o economic growth, with little
corruption or goernment bureaucracy. Likewise, it`s easy to imagine how all o
these deelopments would be ampliied i ,d, the Internet ends up proiding
more access to more religious materials to more belieers ,e.g., through mobile
phones,, ,e, goernments ind a way to hire and compensate loyal bloggers or
touting a particular reading o history, and ,, goernments set up websites that
allow citizens to report on corrupt oicials, problems with existing
inrastructure, or goernment waste.
1hat last deelopment may seem like a good thing-until one realizes that an
authoritarian goernment with less goernment waste is not necessarily a
weaker authoritarian goernment. It may actually be more eectie and the
country may enjoy aster rates o economic growth-but that, alas, still does not
always translate into a more democratic goernment.
All o this is to say that the only way to understand how the Internet inluences
authoritarianism is to irst deine a theory o authoritarianism itsel-preerably,
a theory that goes beyond Manichean theories o the totalitarian state` ersus
the dissidents`-and then use it to closely inestigate how the Internet aects
each o its components.
As such, our ability to harest the potential o the Net to liberate the world`
depends not so much on our ability to understand the Net but on our ability to
understand the world itsel. It`s much easier to understand how the Internet
aects goernment eiciency than to understand how goernment eiciency
aects goernment legitimacy under conditions o capitalism-riendly
authoritarianism.
Political scientists, unortunately, don`t hae much to boast o on this ront:
their understanding o this completely new breed o authoritarianism is at best
rudimentary-and their understanding o how such a luid and complex
technology as the Internet can aect it is een worse. Gien the immense
poerty o our current conceptual apparatus, een i the Net does end up
liberating the world, most likely we won`t know it or quite some time.


Internet Freedom:
Beyond Circumvention
By Ethan Zuckerman
*

U.S. Secretary o State lillary Clinton`s January 2010 speech on Internet
lreedom signaled a strong interest rom the State Department in encouraging
the use o the Internet to promote political reorms in closed societies.
1
It
makes sense that the State Department would look to support existing projects
to circument Internet censorship. 1be êr Yor/ 1ive. reports that a group o
senators subsequently urged the Secretary to apply existing unding to support
the deelopment and expansion o censorship circumention programs.
2

My colleagues lal Roberts,

John Palrey and I hae studied the deelopment o
Internet circumention systems oer the past ie years, and released a study last
year that compared the strengths and weaknesses o dierent circumention
tools.
3
Some o my work at 1he Berkman Center or Internet & Society at
larard Uniersity is unded by a U.S. State Department grant that ocuses on
the continuing study and ealuation o these sorts o tools. As a result, I spend
a lot o time coordinating eorts between tool deelopers and people who need
access to circumention tools to publish sensitie content.
I strongly beliee that we need strong, anonymized and useable censorship
circumention tools. But I also beliee that we need lots more than censorship
circumention tools, and I ear that both unders and technologists may oer-
ocus on this one particular aspect o Internet reedom at the expense o other
aenues. I wonder whether we`re looking closely enough at the undamental
limitations o circumention as a strategy and asking ourseles what we`re
hoping Internet reedom will do or users in closed societies.

Senior Researcher at the Berkman Center or Internet and Society at larard Uniersity.
1hanks to lal Roberts, Janet laen and Rebecca MacKinnon or help editing and
improing this essay. 1hey`re responsible or the good parts. \ou can blame the rest on
me.
1
lillary Rodham Clinton, Secretary o State, Remarks on Internet lreedom at the Newseum
,Jan. 21, 2010,, !""#$%%<<<)'">"()*+,%'(&:(">:I%:2%EFNF%FN%NMUUNT)!"2.
2
Brad Stone, .ia |rgea for Crov. igbtivg vtervet Cev.or., N.\. 1IMLS, Jan. 20, 2010,
!""#$%%<<<).I"-2(')&+2%EFNF%FN%EN%"(&!.+?+*I%EN&(.'+:)!"2?3B:7N. lor more
inormation on 1or, see !""#$%%<<<)"+:#:+X(&")+:*. lor more inormation on Psiphon,
see !""#$%%#'-#!+.)&>. lor more inormation on lreegate, see !""#$%%<<<)/-"Z
-.&)5'%1:((*>"(.
3
lAL ROBLR1S, L1lAN ZUCKLRMAN & JOlN PALlRL\, 200 CIRCUMVLN1ION LANDSCAPL
RLPOR1: ML1lODS, USLS, AND 1OOLS ,March 2009,, !""#$%%/>'!)!>:,>:/)(/5%
[-"'":(>2%!>./?(%N%EOTCTMM%EFFOB8-:&52,(."-+.B\>./'&>#()#/13'(]5(.&(7E.

So here`s a proocation: ^( &>._" &-:&52,(." +5: <>I >:+5./ G."(:.("
&(.'+:'!-#)
I don`t mean that Internet censorship circumention systems don`t work. 1hey
do-our research tested seeral popular circumention tools in censored
nations and discoered that most can retriee blocked content rom behind the
Chinese irewall or a similar system.
4
1here are problems with priacy, data
leakage, the rendering o certain types o content, and particularly with usability
and perormance, but the systems we tested can indeed circument censorship.
\hat I mean is this: \e couldn`t aord to scale today`s existing circumention
tools to liberate` all o China`s Internet users een i they all wanted to be
liberated.
Circumention systems share a basic mode o operation-they act as proxies to
let users retriee blocked content. A user is blocked rom accessing a website
by her Internet Serice Proider ,ISP, or that ISP`s ISP. She may want to read a
page rom luman Rights \atch`s ,lR\, website, which is accessible at IP
address 0.32.6.212. But that IP address is on a national blacklist, and she`s
preented rom receiing any content rom it. So, she points her browser to a
proxy serer at another address-say, 123.45.6.89-and asks a program on
that serer to retriee a page rom the lR\ website. Assuming that
123.45.6.89 isn`t on the national blacklist, she should be able to receie the
lR\ page ia the proxy.
During the transaction, the proxy is acting like an Internet serice proider. Its
ability to proide reliable serice to its users is constrained by bandwidth-
bandwidth to access the destination site and to delier the content to the proxy
user. Bandwidth is costly in aggregate, and it costs real money to run a proxy
that`s heaily used.
Some systems hae tried to reduce these costs by asking olunteers to share
them-the irst release o Citizen Lab`s Psiphon used home computers hosted
by olunteers around the world as proxies, and then used their consumer
bandwidth to access the public Internet. Unortunately, in many countries,
consumer Internet connections are optimized to download content and are
much slower when they are uploading content. 1hese proxies could access the
luman Rights \atch website pretty quickly, but they took a ery long time to
delier the page to the user behind the irewall. As a result, Psiphon is no
longer primarily ocused on trying to make proxies hosted by olunteers work.
1or, on the other hand, is, but 1or nodes are requently hosted by uniersities
and companies that hae access to large pools o bandwidth. Still, aailable
bandwidth is a major constraint o the 1or system. 1he most usable
circumention systems today-irtual priate network ,VPN, tools like

4
ee, geveratt,, ROBLR1S, ZUCKLRMAN & PALlRL\, .vra note 3.

Relakks
5
or \i1opia
6
-charge users between >3 and >6 per month to deray
bandwidth costs.
Assume that systems like 1or, Psiphon and lreegate receie additional unding
rom the U.S. State Department. low much would it cost to proide proxy
Internet access or . well, China China reports 384 million Internet users,

meaning we`re talking about running an ISP capable o sering more than 25
times as many users as the largest U.S. ISP.
8
According to the China Internet
Network Inormation Center ,CNNIC,, China consumes 998,21 Mbps o
international Internet bandwidth.
9
It`s hard to get estimates or what ISPs pay
or bandwidth, though conentional wisdom suggests prices between >0.05 and
>0.10 per gigabyte. Using >0.05 as a cost per gigabyte, the cost to proide the
uncensored Internet to China would be >13,608,000 per month, or >163.3
million a year in pure bandwidth charges, not including the costs o proxy
serers, routers, system administrators and customer serice. laced with a bill
o that magnitude, the >45 million U.S. senators are asking Secretary Clinton to
spend quickly looks pretty paltry.
10

1here`s an additional complication-we`re not just talking about running an
ISP-we`re talking about running an ISP that`s ery likely to be abused by bad
actors. Spammers, raudsters and other Internet criminals use proxy serers to
conduct their actiities, both to protect their identities and to aoid systems on
ree webmail proiders, or instance, which preent users rom signing up or
dozens o accounts by limiting an IP address to a certain number o signups in a
limited time period. lor example, \ikipedia ound that many users used open
proxies to deace their system and now resere the right to block proxy users
rom editing pages.
11
Proxy operators hae a tough balancing act-or their
proxies to be useul, people need to be able to use them to access sites like
\ikipedia or \ou1ube, but i people use those proxies to abuse those sites, the
proxy will be blocked. As such, proxy operators can ind themseles at war
with their own users, trying to ban bad actors to keep the tool useul or the rest
o the users.

5
lor more inormation on Relakks, see !""#$%%<<<):(?>WW')&+2.
6
lor more inormation on \i1opia, see !""#$%%<<<)<-"+#->).(".
Chris Buckley, Cbiva vtervet Povtatiov it. 1 vittiov, RLU1LRS, Jan. 15, 2010,
!""#$%%<<<):(5"(:')&+2%>:"-&?(%-/`YHPRDFRFDYEFNFFNNU.
8
ee 1o 2 |.. P. b, vb.criber: Q 200, ISP Planet, !""#$%%<<<)-'#Z
#?>.(")&+2%:('(>:&!%:>.W-.*'%5'>)!"2?.
9
ee China Internet Network Ino. Ctr., vtervet vvaavevtat Data,
!""#$%%<<<)&..-&).(")&.%(.%-./(0%FP%-./(0)!"2 ,last isited July 29, 2010,.
10
Brad Stone, .ia |rgea for Crov. igbtivg vtervet Cev.or., N.\. 1IMLS, Jan. 20, 2010,
!""#$%%<<<).I"-2(')&+2%EFNF%FN%EN%"(&!.+?+*I%EN&(.'+:)!"2?.
11
ee Oev Proie., \IKIPLDIA, !""#$%%(.)<-W-#(/->)+:*%<-W-%^-W-#(/->$P#(.B#:+0-('.

I`m skeptical that the U.S. State Department can or wants to build or und a
ree ISP that can be used by millions o simultaneous users, many o whom may
be using it to commit click raud or send spam.
12
I know-because Iè talked
with many o them-that the people who und blocking-resistant Internet
proxies don`t think o what they`re doing in these terms. Instead, they assume
that proxies are used by users only in special circumstances, to access blocked
content.
lere`s the problem: A goernment like China is blocking a lot o content. As
Donnie Dong notes in a recent blog post, ie o the ten most popular websites
worldwide are blocked in China.
13
1hose sites include \ou1ube and lacebook,
sites that eat bandwidth through large downloads and long sessions. Perhaps it
would be realistic to act as an ISP to China i we were just proiding access to
luman Rights \atch-but it`s not realistic i we`re proiding access to
\ou1ube, too.
Proxy operators hae dealt with this question by putting constraints on the use
o their tools. Some proxy operators block access to \ou1ube because it`s such
a bandwidth hog. Others block access to pornography, both because it uses
bandwidth and to protect the sensibilities o their sponsors. Others constrain
who can use their tools, limiting access to people coming rom Iranian or
Chinese IP addresses, trying to reduce bandwidth use by American high school
kids whose schools hae blocked \ou1ube. In deciding who or what to block,
proxy operators are oering their personal answers to a complicated question:
!bat art. of tbe vtervet are re tr,ivg to oev v to eote iv cto.ea .ocietie.. As we`ll
address in a moment, that`s not such an easy question to answer.
Imagine or a moment that we could aord to proxy China, Iran, Myanmar and
others` international traic. \e igure out how to keep these proxies unblocked
and accessible ,it`s not easy-the operators o heaily used proxy systems are
engaged in a ast-moing cat and mouse game, and determine how to mitigate
the abuse challenges presented by open proxies. \e still hae problems.
Most Internet traic is domestic. In China, we estimate that, at minimum, 95
o total traic is within the country. Domestic censorship matters a great deal,
and perhaps a great deal more than censorship at national borders. As Rebecca

12
Matthew Broersma, Re.earcber. ,e Oev Pro, .ttac/., 1LCl\ORLD, No. 15, 200,
!""#$%%.(<')"(&!<+:?/)&+2%'(&5:-"I%NFDDM%:('(>:&!(:'Z(I(Z+#(.Z#:+0IZ>"">&W'.
13
Donnie Dong, Coogte`. .vgr,, acrifice ava tbe .cceteratea tittivg vtervet, BLA\GDOG, Jan. 13,
2010, !""#$%%(.*?-'!)[?><*/+*)&+2%EFNF%FN%*++*?('Z>.*:IZ'>&:-1-&(Z>./Z
>&&(?(:>"(/)!"2?.

MacKinnon documented in Cbiva`. Cev.or.bi 2.0,
14
Chinese companies censor
user-generated content in a complex, decentralized way. As a result, a good deal
o controersial material is neer published in the irst place, either because it`s
blocked rom publication or because authors decline to publish it or ear o
haing their blog account locked or cancelled. \e might assume that i Chinese
users had unettered access to Blogger, they`d publish there. Perhaps not-
people use the tools that are easiest to use and that their riends use. A
seasoned Chinese dissident might use Blogger, knowing she`s likely to be
censored-an aerage user, posting photos o his cat, would more likely use a
domestic platorm and not consider the possibility o censorship until he ound
himsel posting controersial content.
In promoting Internet reedom, we need to consider strategies to oercome
censorship inside closed societies. \e also need to address sot censorship`:
the co-opting o online public spaces by authoritarian regimes, which sponsor
pro-goernment bloggers, seed sympathetic message board threads, and pay or
sympathetic comments. Lgeny Morozo oers a thoroughly dark iew o
authoritarian use o social media in low Dictators \atch Us on the \eb.`
15

\e also need to address a growing menace to online speech-attacks on sites
that host controersial speech. \hen 1urkey blocks \ou1ube
16
to preent
1urkish citizens rom seeing ideos that deame Ataturk, they preent 20
million 1urkish Internet users rom seeing eerything on \ou1ube. \hen
someone-the Myanmar goernment, patriotic Burmese, mischieous
hackers-mount a distributed denial o serice attack on 1be rraraaa,,
1
an
online newspaper highly critical o the Myanmar goernment, this temporarily
preents eeryone eerywhere rom seeing it.
Circumention tools help 1urks who want to see \ou1ube get around a
goernment block, but they don`t help Americans, Chinese or Burmese see 1be
rraraaa, i the site has been taken down by a Distributed Denial o Serice
,DDoS,
18
or hacking attack. Publishers o controersial online content hae
begun to realize that they`re not just going to ace censorship by national

14
Rebecca MacKinnon, Cbiva`. Cev.or.bi 2.0: or Covavie. Cev.or togger., 14 lIRS1 MONDA\
,leb. 2009,, !""#$%%1-:'"2+./>I)+:*%!"[-.%&*-<:>#%[-.%+X'%-./(0)#!#%12%>:"-&?(%
,-(<%EMOV%EFVT.
15
Lgeny Morozo, or Dictator. !atcb |. ov tbe !eb, PROSPLC1, No. 18, 2009,
!""#$%%<<<)#:+'#(&"2>*>a-.()&+)5W%EFFT%NN%!+<Z/-&">"+:'Z<>"&!Z5'Z+.Z"!(Z<([.
16
Nico lines, Yov1vbe avvea iv 1vr/e, .fter 1iaeo v.vtt., 1lL 1IMLS, Mar. , 200,
!""#$%%<<<)"-2('+.?-.()&+)5W%"+?%.(<'%<+:?/%(5:+#(%>:"-&?(NCVMVCF)(&(.
1
Aung Zaw, 1be vrve.e Regive`. C,ber Offev.ire, 1lL IRRA\ADD\, Sept. 18, 2008,
!""#$%%<<<)-::><>//I)+:*%+#-.-+.B'"+:I)#!#3>:"B-/7NCEVF .
18
A DDoS attack uses multiple computer systems to target and attack a single system, or
website, thus preenting users rom accessing the targeted system.

iltering systems-they`re going to ace a ariety o technical and legal attacks
that seek to make their serers inaccessible.
1here`s quite a bit publishers can do to increase the resilience o their sites to
DDoS attack and to make their sites more diicult to ilter. 1o aoid blockage
in 1urkey, \ou1ube could increase the number o IP addresses that lead to the
web serer and use a technique called ast-lux DNS`
19
to gie the 1urkish
goernment more IP addresses to block. 1hey could maintain a mailing list to
alert users to unblocked IP addresses where they could access \ou1ube, or
create a custom application that disseminates unblocked IPs to \ou1ube users
who download the application. 1hese are all techniques employed by content
sites that are requently blocked in closed societies.
\ou1ube doesn`t utilize these anti-blocking measures or two reasons. One, it
has historically preerred to negotiate with nations who ilter the Internet to
make \ou1ube sites accessible again, rather than to work against these nations
by ighting iltering. ,1his may be changing, now that Google has decided to
disengage rom China due to censorship and hacking issues., Second, \ou1ube
doesn`t really hae an economic incentie to be unblocked in 1urkey. I
anything, being blocked in 1urkey, and perhaps een in China, may een be to
its economic adantage, since sering these countries is likely to be unproitable.
Sites that enable distribution o user-created content are supported by
adertising traic. Adertisers are generally more excited about reaching users
in the U.S. who hae credit cards, more disposable income and are inclined to
buy online than users in China or 1urkey. Some suspect that the introduction
o lite` ersions o serices like lacebook is designed to sere users in the
deeloping world at lower cost, since those users rarely create income or the
sites.
20
In economic terms, it may be hard to conince lacebook, \ou1ube and
others to continue proiding serices to closed societies, where they hae a
tough time selling ads. \e also may need to ask more o them to take steps to
ensure that they remain accessible and useul in censorious countries.
In short:
Internet circumention is diicult and expensie. It can make it easier
or people to send spam and steal identities.
Circumenting censorship through proxies gies people access to
international content, but doesn`t address domestic censorship, which
likely aects the majority o people`s Internet behaior.

!"
last-lux DNS preents the identiication o a host serer`s IP address.
20
Brad Stone & Miguel lelt, v Deretoivg Covvtrie., !eb Cror. !itbovt Profit, N.\. 1IMLS,
April 26, 2009, !""#$%%<<<).I"-2(')&+2%EFFT%FC%EO%"(&!.+?+*I%'">:"Z
5#'%EO*?+[>?)!"2?3B:7N.

Circumenting censorship doesn`t oer a deense against DDoS or
other attacks that target publishers.

1o igure out how to promote Internet reedom, we need to start addressing the
question: low do we think the Internet changes closed societies` In other
words, do we hae a theory o change`
21
behind our desire to ensure people in
Iran, Burma, China, etc., can access the Internet \hy do we beliee this is a
priority or the U.S. State Department or or public diplomacy as a whole
Much work on Internet censorship isn`t motiated by a theory o change-it`s
motiated by a deeply-held coniction-one that I share-that the ability to
share inormation is a basic human right. Article 19 o the Uniersal
Declaration o luman Rights states that Leryone has the right to reedom o
opinion and expression, this right includes reedom to hold opinions without
intererence and to seek, receie and impart inormation and ideas through any
media and regardless o rontiers.`
22
1he Internet is the most eicient system
weè eer built to allow people to seek, receie and impart inormation and
ideas, and thereore, we need to ensure eeryone has unettered Internet access.
1he problem with the Article 19 approach to censorship circumention is that it
doesn`t help us prioritize. It simply makes it imperatie that we sole what may
be an unsolable problem.
I we beliee that access to the Internet will change closed societies in a
particular way, we can prioritize access to those aspects o the Internet. Our
theory o change helps us igure out what we must proide access to. 1he our
theories I list below are rarely explicitly stated, but I beliee they underlie much
o the work behind censorship circumention.
H!( Y5##:(''(/ G.1+:2>"-+. H!(+:I: I we can proide certain
suppressed inormation to people in closed societies, they`ll rise up,
challenge their leaders and usher in a dierent goernment. \e might
choose to call this the lungary 56 theory`
23
-reports o struggles against
communist goernments around the world, reported into lungary ia
Radio lree Lurope, encouraged lungarians to rebel against their leaders.
,Unortunately, the U.S. didn`t support the reolutionaries militarily-as
many in lungary had expected-and the reolution was brutally quashed
by a Soiet inasion.,

21
Mark Schmitt, 1be 1beor, of Cbavge Privar,, 1lL AMLRICAN PROSPLC1, Dec. 21, 200,
!""#$%%<<<)#:+'#(&")+:*%&'%>:"-&?('3>:"-&?(7"!(B"!(+:IB+1B&!>.*(B#:-2>:I.
22
1he Uniersal Declaration o luman Rights, art. 19, G.A. Res. 21A,III,, U.N. GAOR, 3d
Sess., U.N. Doc. A,810 ,Dec. 10, 1948,,
!""#$%%<<<)5.)+:*%(.%/+&52(."'%5/!:%-./(0)'!"2?.
23
lor more inormation on the lungarian Reolution o 1965, see vvgariav Rerotvtiov of 1:,
\IKIPLDIA, !""#$%%(.)<-W-#(/->)+:*%<-W-%J5.*>:->.B9(,+?5"-+.B+1BNTUD.

Or we could term this the North Korea theory,` because a state as closed
as North Korea might be a place where unsuppressed inormation-about
the iscal success o South Korea, or instance-could prooke reolution.
Barbara Demick`s beautiul piece in the New \orker, 1he Good Cook,`
gies a sense o how little inormation most North Koreans hae about the
outside world and how dierent the world looks rom Seoul.
24

Nonetheless, een North Korea is less inormationally isolated than we
think-1be Dovg. tbo, a South Korean newspaper, reports an
inormation belt` along the North Korea,China border where calls on
smuggled mobile phones are possible between North and South Korea.
25

Other nations are ar more open-the Chinese tend to be extremely well
inormed about both domestic and international politics, both through
using circumention tools and because Chinese media reports a great deal
o domestic and international news.
It`s possible that access to inormation is a necessary, though not suicient,
condition or political reolution. It`s also possible that we oerestimate
the power and potency o suppressed inormation, especially as inormation
is so diicult to suppress in a connected age.
H!( H<-""(: 9(,+?5"-+. H!(+:I$ I citizens in closed societies can use the
powerul communications tools made possible by the Internet, they can
unite and oerthrow their oppressors. 1his is the theory that led the U.S.
State Department to urge 1witter to postpone a period o scheduled
downtime during the Iran election protests.
26
\hile it`s hard to make the
case that technologies o connection are going to bring down the Iranian
goernment,
2
good examples exist, like the role o the mobile phone in
helping to topple President Lstrada in the Philippines.
28

1here`s been a great deal o enthusiasm in the popular press or the 1witter
reolution theory, but careul analysis reeals some limitations. 1he
communications channels opened online tend to be compromised quickly,

24
Barbara Demick, 1be Cooa Coo/, 1lL NL\ \ORKLR, No. 2, 2009, at 58,
!""#$%%<<<).(<I+:W(:)&+2%:(#+:"-.*%EFFT%NN%FE%FTNNFE1>B1>&"B/(2-&W.
25
ôrtb Koreav. Directt, Covvect ritb ovtb Koreav. ria Cbive.e Cett Pbove., ASK A KORLAN!, Jan. 1,
2010, !""#$%%>'W>W+:(>.)[?+*'#+")&+2%EFNF%FN%(0&(??(."Z>:"-&?(Z+.Z/+.*Z-?[+Z
>[+5")!"2?.
26
Sue Pleming, |.. tate Deartvevt ea/. to 1ritter Orer rav, RLU1LRS, June 16, 2009,
!""#$%%<<<):(5"(:')&+2%>:"-&?(%-/`Y^LHFNNMOCEFFTFDND.
2
ee Cameron Abadi, rav, aceboo/, ava tbe ivit. of Ovtive .ctiri.v, lORLIGN POLIC\, leb. 12,
2010, !""#$%%<<<)1+:(-*.#+?-&I)&+2%>:"-&?('%EFNF%FE%NE%-:>.'B1>-?(/B1>&([++WB
:(,+?5"-+..
28
ee ]o.eb .traaa Covtrorer.ie., \IKIPLDIA,
!""#$%%(.)<-W-#(/->)+:*%<-W-%b+'(#!BR'":>/>c8+.":+,(:'-('.

used or disinormation and monitoring actiists. And when protests get
out o hand, goernments o closed societies don`t hesitate to pull the plug
on networks-China has blocked Internet access in Xinjiang or months,
and Lthiopia turned o SMS on mobile phone networks or years ater
they were used to organize street protests. And it`s worth noting that
prophesied twitter reolutions` in Moldoa and Iran both ailed in the
ace o authoritarian goernments.
H!( d5[?-& Y#!(:( H!(+:I: Communication tools may not lead to
reolution immediately, but they proide a new rhetorical space where a
new generation o leaders can think and speak reely. In the long run, this
ability to create a new public sphere, parallel to the one controlled by the
state, will empower a new generation o social actors, though perhaps not
or many years.
Marc Lynch made a pretty persuasie case or this theory in a talk last year
about online actiism in the Middle Last.
29
In the ormer Soiet Union,
samizdat ,sel-published, clandestine media, was probably more important
as a space or ree expression than it was as a channel or disseminating
suppressed inormation.
30
1he emergence o leaders, like Vacla lael,
whose authority was rooted in cultural expression as well as political power,
makes the case that simply speaking out is powerul. But the long timescale
o this theory makes it hard to test.
1he theory we accept shapes our policy decisions. I we beliee that
disseminating suppressed inormation is critical-either to the public at large or
to a small group o inluencers-we might ocus our eorts on spreading
content rom Voice o America or Radio lree Lurope. Indeed, this is how
many goernment orays into censorship circumention began-national news
serices began supporting circumention tools so their content, painstakingly
created in languages like Burmese or larsi, would be accessible in closed
societies. 1his is a ery eicient approach to anti-censorship-we can ignore
many o the problems associated with abusing proxies and ocus on prioritizing
news oer other less-important bandwidth-hogging uses, like the ideo o the
cat lushing the toilet. Unortunately, weè got a long track record that shows
that this orm o anti-censorship doesn`t magically open closed regimes, which
suggests that increasing our reliance on this strategy might be a poor idea.

29
Lthan Zuckerman, Marc ,vcb ../. |. to be Reati.tic .bovt Digitat .ctiri.v iv tbe Miaate a.t,
April 2, 2009, !""#$%%<<<)("!>.a5&W(:2>.)&+2%[?+*%EFFT%FC%EO%2>:&Z?I.&!Z
>'W'Z5'Z"+Z[(Z:(>?-'"-&Z>[+5"Z/-*-">?Z>&"-,-'2Z-.Z"!(Z2-//?(Z(>'".
30
ee geveratt,, Peter Steiner, vtroavctiov: Ov aviaat, 1aviaat, Magvitiaat, ava Otber travge
!ora., 29 POL1ICS 1ODA\ 613 ,2008,
!""#$%%#+("-&'"+/>I)/5W(X+5:.>?')+:*%&*-%:(#:-."%ET%C%DNM)#/1.

I we adopt the 1witter Reolution theory, we should ocus on systems that
allow or rapid communication within trusted networks. 1his might mean tools
like 1witter or lacebook, but it probably means tools like LieJournal and
\ahoo! Groups, which gain their utility through exclusiity, allowing small
groups to organize outside the gaze o the authorities. I we adopt the public
sphere approach, we want to open any technologies that allow public
communication and debate-blogs, 1witter, \ou1ube, and irtually anything
else that its under the banner o \eb 2.0. 1his, unortunately, presents
technical challenges that are proing extremely diicult to sole.
\hat does all this mean in terms o how the U.S. State Department should
allocate their money to promote Internet lreedom My goal was primarily to
outline the questions they should be considering, rather than oering speciic
prescriptions. But here are some possible implications o these questions:
I we beliee the U.S. goernment should be exporting Internet reedom`-
and there are good reasons to argue that a goernment, and particularly the US
goernment, shouldn`t take on this task-we need to continue supporting
circumention eorts, at least in the short term. But we need to disabuse
ourseles o the idea that we can sole` censorship through circumention.
\e should support circumention until we ind better technical and policy
solutions to censorship, not because we can tear down the Great lirewall by
spending more on proxies, etc.
Second, i we want more people using circumention tools, we need to ind
ways to make these systems iscally sustainable. Sustainable circumention is
becoming an attractie business or some companies.
31
It needs to be part o a
comprehensie Internet reedom strategy, and we need to deelop strategies
that are sustainable and proide low- to zero-cost access to users in closed
societies.
1hird, as we continue to und circumention, we need to address usage o these
tools to send spam, commit raud and steal personal data. \e might do this by
relying less on IP addresses as an extensie, undamental means o regulating
bad behaior, but we hae to ind a solution that protects networks against
abuse while maintaining the possibility o anonymity, a diicult balancing act.
Additionally, we need to shit our thinking rom helping users in closed
societies access blocked content to helping publishers reach all audiences. In
doing so, we may gain those publishers as a aluable new set o allies as well as
opening a new class o technical solutions.

31
Lara larrar, Ca.bivg iv ov vtervet Cev.or.bi, CNN, leb. 19, 2010,
!""#$%%<<<)&..)&+2%EFNF%HR8J%FE%NV%-."(:.(")&(.'+:'!-#)[5'-.(''%3!#"7Y[-..

lurthermore, i our goal is to allow people in closed societies to access an
online public sphere or to use online tools to organize protests, we need to
bring the administrators o these tools into the dialog. Secretary Clinton
suggests that we make ree speech part o the American brand identity-let`s
ind ways to challenge companies to build blocking resistance into their
platorms and to consider Internet reedom as a central part o their business
mission. \e need to address the act that making platorms unblockable has a
cost or content hosts and that their business models currently don`t reward
companies or proiding serices to blocked users.
1he U.S. goernment should treat Internet iltering-and more aggressie
hacking and DDoS attacks-as a barrier to trade. 1he U.S, should strongly
pressure goernments in open societies like Australia and lrance to resist the
temptation to restrict Internet access, as this behaior helps China and Iran
make the case that their censorship is in line with international norms. And we
need to ix U.S. treasury regulations that make it diicult and legally ambiguous
or companies like Microsot and projects like Sourcelorge
32
to operate in
closed societies. I we beliee in Internet lreedom, the irst step is rethinking
these policies so they don`t hurt ordinary Internet users.
linally, i attempts to export Internet reedom are to be met with something
other than cynicism or skepticism, the U.S. goernment needs to do a better job
o protecting ree speech domestically. 1he pressure exerted by indiidual
Senators and by the State Department on companies like Amazon and PayPal to
terminate serices to \ikiLeaks calls into question the U.S. goernment`s
commitment to online ree speech. I the U.S. wants countries like China to
consider a more ree and open Internet, control o the Internet in the U.S. must
also ollow the rule o law, and not all ictim to political expediency.
I we take seriously Secretary Clinton`s call, the danger is that we increase our
speed marching in the wrong direction. As we embrace the goal o Internet
lreedom, now is the time to ask what we`re hoping to accomplish and to shape
our strategy accordingly.

32
Sourcelorge is an open source code depository rom which sotware can be deeloped and
downloaded. lor more inormation, see !""#$%%'+5:&(1+:*().("%.

!"#$ &'#(&) *+,,)*-#+' ./#'0$ -+0)-")/ 12 -"+&0"- ,)34)/$ +' 5'-)/')-
,367 8"#,+$+8"97 8+,#*9 3'4 )*+'+:#*$ -+ *+'$#4)/ 6"3- -") ');-
4#0#-3, 4)*34) :#0"- ./#'0< =3$ -") 5'-)/')- .))' 0++4 >+/ +&/ *&,-&/)?
5$ -") 5'-)/')- 3- /#$@ >/+: -") 4/#A) -+ .&#,4 :+/) $)*&/)7 .&- ,)$$
B+8)'C $9$-):$ 3'4 4)A#*)$? 5$ -") 5'-)/')- /)3,,9 $+ B);*)8-#+'3,?C
=3$ #- >&'43:)'-3,,9 *"3'0)4 )*+'+:#*$? D"+E3'4 6"3- #4)3$E
6#,, 0+A)/' -") F)- #' 1G1G? H"+&,4 +',#') #'-)/:)4#3/#)$ ,#@) 3**)$$
8/+A#4)/$7 "+$-#'0 8/+A#4)/$7 $)3/*" )'0#')$ 3'4 $+*#3, ')-6+/@$ 4+
:+/) -+ B8+,#*)C -")#/ ')-6+/@$7 #'*/)3$) -/3'$83/)'*97 +/ +8)/3-)
B')&-/3,,9?C D"3- >&-&/) #$ -")/) >+/ 8/#A3*9 +',#')? I3' +',#') >/))
$8))*" .) /)0&,3-)4? I3' #- /)3,,9 &'$)3- -9/3'-$? !")$) JK -"+&0"-L
8/+A+@#'0 )$$39$ -3*@,) -")$) (&)$-#+'$ 3'4 :+/)< !"#$ .++@ #$
)$$)'-#3, /)34#'0 >+/ 3'9+') 03M#'0 -+63/4 -") 4#0#-3, >&-&/)<
THE BEST THINKING ABOUT
THE FUTURE OF DIGITAL POLICY
CONTRIBUTORS
Rob Atkinson
Stewart Baker
Ann Bartow
Yochai Benkler
Larry Downes
Josh Goldfoot
Eric Goldman
James Grimmelman
H. Brian Holland
David Johnson
Andrew Keen
Hon. Alex Kozinski
Mark MacCarthy
Geo Manne
Evgeny Morozov
Milton Mueller
John Palfrey
Frank Pasquale
Paul Szynol
Adam Thierer
Hal Varian
Christopher Wolf
Tim Wu
Michael Zimmer
Jonathan Zittrain
Ethan Zuckerman
TechFreedom
techfreedom.org
1899 L ST NW, 12th Floor
Washington, D.C. 20036
NextDigitalDecade.com

The Next Digital Decade

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Next Digital Decade

Uploaded by

Copyright:

Available Formats

THE NEXT DIGITAL DECADE

ESSAYS ON THE FUTURE OF THE INTERNET

ee 1be i.cvit Reort,

Indeed, beore most o us had een heard o

\illiam Powers, author o lamlet`s BlackBerry: A Practical Philosophy or Building a Good

Nearly all o those

ZI11RAIN, 1lL lU1URL Ol 1lL IN1LRNL1, .vra note 3, at 155.

Proessor, Columbia Law School, lellow, New America loundation

while another court ound a website

Barrett . Rosenthal, 146 P.3d 510, 52 ,Cal. 2006,.

Some legal scholars argue that goernment

Under the country o destination approach, the law

- In the midst o the panic oer peer-to-

which allows eBay`s buyers and sellers to rate

\lO CON1ROLS 1lL IN1LRNL1, .vra note 2.

ISPs are oten called carriers` to relect the parallel

DAVID S1ARK, 1lL SLNSL Ol DISSONANCL: ACCOUN1S Ol \OR1l IN LCONOMIC LIlL 1

1he court said the credit card companies did

In its current state, the law threatens to

ov, r. |virer.at Cit, tvaio., .vra note 2.

\ikipedia, Slashdot, and eBay all

ee ovvaev`. Coogte tor,, SLARClNLU1RALI1\.ORG ,Aug. 18, 2009,,

or may incorporate only part o a web page.

School o Inormation Studies, Uniersity o \isconsin-Milwaukee

Julie L. Cohen, . Rigbt to Reaa .vov,vov.t,: . Cto.er oo/ at Co,rigbt Mavagevevt` iv

series took place at the Uniersity o Caliornia, Berkeley campus.

ee 1C Rovvatabte erie., .vra note 5.

In other jurisdictions, regulation o intermediaries may soon be the

and lrederick 1oben

You might also like