You are on page 1of 3

Issue 232 • July/August 2009

IN THIS ISSUE

NEWS
3 President’s Column
6 Book Review
8 The “G” in your
“IS”— Redux
The Association for GIS Professionals A Call for Action
8 News from the SSI
– One voice created for
surveyors and spatial
scientists
9 Putting Health in Place
with GIS
FluTracker: 10 Welcome New urisa
Members
How a 6-Person Company Launched a Global Pandemic Information Portal in
Under 48 Hours
By Jeff Christensen, Chief Technology Officer, Rhiza Labs

In late April 2009, the international me- founder and president of the firm of interpretation, and to reuse data in
dia converged on one story: the swine Recombinomics, was diligently tracking other applications.
flu epidemic. Since epidemics are root- all cases reported by the media, local The second group of issues
ed in place, containing them requires governments, and others. He was not concerned scalability. Since Google
authorities to identify those places and affiliated with any official government does not allow a map to be maintained
act quickly. Using fast, flexible geospa- agency or nonprofit institution; he by multiple curators, Dr. Niman was
tial technology and a new workflow was simply a concerned and highly quickly becoming overwhelmed by the
that harnesses the power of crowds knowledgeable citizen. exploding epidemic. Additionally, the
but ensures data validity, Rhiza Labs While Dr. Niman’s map was technology was reaching its functional
partnered with a biomedical researcher popular, we recognized two groups of limits. The total number of cases had
and an army of volunteers to launch issues that hampered its usefulness in almost exceeded Google’s maximum
FluTracker, a swine flu data portal. Ap- responding to the spread of swine flu. number of records, and the clusters
proaching the problem not as cartogra- The first concerned the data’s structure, of pins were growing so large that
phers but instead as designers yielded a or lack thereof. Since Dr. Niman input they were crashing web browsers and
system that was acknowledged as being new cases by manually dropping becoming visually meaningless for
more current and comprehensive than “pins” on the locations of swine flu interpreting the data.
most official sources. Using these same reports which he identified visually, Unfortunately, official maps did
techniques, any organization can build each case’s geographic accuracy was not offer more usable systems or
geographic information systems that questionable. Since each case was higher-quality data. The CDC, WHO
leverage the omnipresence of crowds to organized as unstructured narrative and others were slow to release new
provide the most complete information paragraphs associated with a pin, the data, even though the epidemic was
possible. data couldn’t be easily reused in other rapidly progressing. The data they did
Given Rhiza Labs’ focus on systems, like GIS, spreadsheet and release was of very coarse granularity,
visualizing locational data, when the statistical analysis packages. Also, the with the CDC giving data only at the
flu threat emerged, we wanted to precision of the represented data was U.S. state level and the WHO only at
know where swine flu was happening. inconsistent. Sometimes one pin would the national level. Considering that
We found a map on Google represent a single case, and sometimes epidemics are rooted in place, such
MyMaps created by a biomedical it would represent more. This lack of broad geographies are practically
researcher located near our offices structure made it impossible to visually
in Pittsburgh. Dr. Henry Niman, aggregate data on the map for ease
continued on page 4
FluTracker 2009 continued from page 1 drew upon our experience to create
a speculative set of requirements and
useless for tracking localized outbreaks started building. We determined that
or generating actionable information. people want to be able to locate cases
Furthermore, official sources trapped the on a map, with specific, consistent
data as infographics – static, flat images information for each case. Given that
– so it could not be reused in other the data set would include both media
systems. reports and government sources, we
Comments from users of Dr. Niman’s also determined that it was imperative
map – both corporate and individual to include source references with
– showed that people wanted something every case so users could evaluate the
more dynamic and with more detail. As a veracity of each report. This would also
social venture, Rhiza Labs saw lending its establish the credibility of the dataset as data with minimal effort, attention or
experience to this emerging threat to be a whole. training. The form also allowed users
a key opportunity to pursue its mission. Based on the commentary from to input the incident data, the source
We approached Dr. Niman and offered users of the original MyMaps site, reference, and detailed information
our expertise pro bono to create a free concerned citizens around the world about the case. Additionally, users could
public resource that would provide more were tracking the spread of the disease contribute volunteered geographic
up-to-date, comprehensive, usable flu both locally and globally. This meant information to the map, but a number of
data, and require less effort to maintain. that many eyes were scanning media expert curators verified the data before
Rhiza’s process begins with defining outlets across the globe and contributing it entered the public site. This technique,
user communities’ needs before applying their findings. Tapping into this vibrant controlled crowd-sourcing, combines
technology to their problems. Since community was essential to keeping up the omnipresence of crowds with the
swine flu was spreading so rapidly, we with the flow of source information, so reliability of a smaller group of experts
we needed to include a clear and simple to create more complete, and more
mechanism for people to contribute accurate information than either could
case suggestions for verification by the gather independently.
curators. Creating the data is one thing; doing
To address these needs, we created something with it is another. No matter
an easy, validated data entry form. how well-crafted the visualizations we
Since most reports have a place name create can be, users will still want to
attached, the form enabled curators to make maps that have personal meaning
simply enter information and obtain to them and their communities. We
exact coordinates on a map. We addressed this issue in three ways.
designed this form so that curators First, we created a map that allows
could create structured, high-quality users to navigate the information easily.
The map allows users to zoom to their
geography of choice, and aggregates
geographically-contiguous data into
single points to ensure that the map
retains meaning. For example, at certain
levels of the map, all the cases in the
state of New York are collected into a
single ball showing the total number
of cases and coloring that represents
suspected, confirmed, and fatal cases.
This enables the map to show detail yet
remain accessible at a fine grain level.
Second, we enabled users to
make their own maps using whatever
symbology they like – heat maps
showing the density of flu cases,
choropleth maps showing concentrations
in geographic areas, and progressions
of either map over time. The diversity
of these user-created visualizations
continued on page 5

4 July/August 2009 • URISA News


FluTracker 2009 continued from page 4

shows a broader picture of the spread of


swine flu than a smaller array of expert-
created maps could. Ensuring that these
contributions were generally included in
the global distributed data set kept the
user community satisfied and engaged.
Finally, we knew it would be difficult
to create tools that satisfied both novice
and expert users alike, so we allowed
users to download raw data in different
formats for use in their systems of choice.
Instead of burying FluTracker’s data in
infographics, we broadcasted it as RSS
and KML feeds, as well as CSV files, so
that users could analyze the latest flu
data in their systems of choice. We also
allowed users to embed FluTracker on
their own webpages.
We’ve always believed at Rhiza
that when organizations share data,
users will come to it. FluTracker proved
us right. After five weeks, FluTracker
had 4,000,000 views, 5,000 flu reports
of 30,000 flu cases, visitors from 200
countries, and 11,000 submitted
comments. Many claimed that the data
was 4-7 days ahead of the CDC and
WHO. While FluTracker remained (and
remains) free for personal use, ABC,
Cisco, Hallburton, Walgreens, AT&T,
and Barclays all turned to FluTracker to
support their corporate preparedness
and continuity planning. URISA Board of Directors
Election Results
Swine flu was evolving quickly
and necessitated a quick response. Yet
official data for tracking swine flu was
too coarse, slow, and inaccessible. Rhiza As you know from recent email announcements, this year’s URISA President-Elect
Labs reacted to the fact that people want voting resulted in a tie. The exact same number of votes was cast for candidates Re-
more data in more formats with more becca Somers and Cy Smith. This is unique to URISA and our bylaws do not offer a
flexible tools, and built a better system way to handle it. While we will amend our bylaws to accommodate a tie should one
in just under 48 hours. Using these same occur again, the Board of Directors has decided to hold a run-off vote for President-
human-centered design principles, any Elect between the two candidates. You should have received an email with voting in-
organization, regardless of platform, can structions on June 23. If for some reason you cannot locate that email, please contact
implement a fast, flexible GIS in response URISA (wnelson@urisa.org) at your earliest opportunity. Voting will remain open until
to an emerging challenge. July 31.

For more information : http://flutracker. The election for Board Members did result in three Directors-Elect who will begin
rhizalabs.com/ their terms following the Anaheim conference. Congratulations to:
Carl Anderson, GISP - Fulton County, Atlanta, Georgia
David DiBiase, GISP - Penn State University, University Park, Pennsylvania
Twyla McDermott, GISP - City of Charlotte, North Carolina

Thank you to all of the candidates who participated in this election! And thank you,
URISA members, for participating in this process!

July/August 2009 • URISA News 5

You might also like