Professional Documents
Culture Documents
21 Deciding to Act
Using the Decision Logic diagram explained in this article gives project managers a tool to effectively
manage projects and report actions at project reviews with both customers and superiors.
by Walt Lipke
Open Forum
I f you look back at the managers you worked for in your career, I’m sure the names of
those who you considered quite good at their jobs come to mind quickly. Moreover,
I’m sure the names of those that were not so good also come to mind. Now that I am a
manager, I often try to remember what made the good managers in my career so good.
How can I learn from them? All managers want to be good managers, but how do they
ensure that they are doing their best and meeting their employees’ needs and expecta-
tions? This month’s issue of CrossTalk highlights management basics. To do their
best, I believe that managers need to conquer the basics first and then continually improve upon a
sound foundation of management principles.
We begin our issue with People Factors in Software Management: Lessons From Comparing Agile and
Plan-Driven Methods by Richard Turner and Barry Boehm. In this article, the authors address five
basic, people-related management areas: staffing, culture, values, communications, and expecta-
tions management. Whether developing software under agile, plan-driven, or hybrid methods, the
authors emphasize that managers improving upon these five key areas are more likely to succeed
at software project management.
Next, in Back to the Basics: Measurement and Metrics by Tim Perkins, Roald Peterson, and Larry
Smith, we are reminded of the importance of a measurement program as a basic, decision-mak-
ing tool for managers. The authors have included a Measurement and Metrics Checklist to assist
managers in developing, implementing, and reviewing their metrics programs.
Our Management Basics section continues with How to Talk About Work Performance: A Feedback
Primer by Esther Derby. One of the most uncomfortable tasks for a manager is to provide criti-
cism to an employee who is not performing well. Providing feedback – whether the news is good
or bad for the employee – is something every manager must do. This author presents good advice
and 11 feedback guidelines to follow to make feedback effective. Next, Johanna Rothman of
Rothman Consulting Group, Inc. writes of her 15 years of management experience in Successful
Software Management: 14 Lessons Learned. This author shares management lessons she has learned
while balancing business, employee, and work environment needs.
Wrapping up our theme section is Deciding to Act by Walt Lipke. Project managers focus much
of their time on monitoring their project’s performance in terms of cost, schedule, and quality.
This author presents an approach for project analysis and decision-making to assist a project man-
ager on knowing when to act if performance begins to weaken, and what action should be taken
to strengthen project performance.
Our Open Forum article this month is Requirements Engineering Maturity in the CMMI by Dennis
Linscomb. This author expresses his opinion on the deficiency of the Capability Maturity Model®
Integration (CMMI®) in defining requirements engineering (RE) maturity and shares his ideas on
how the CMMI model should be revamped in the RE area. If you would like to comment on this
author’s opinion, drop us a line or a Letter to the Editor at <stsc.customerservice@hill.af.mil>.
Another calendar year for CrossTalk is wrapping up. Thanks to all of our authors for their
fine contributions and to the many readers who continue to turn to our publication as a source for
defense software engineering best practices and lessons learned. CrossTalk’s 2003 Article Index
can be found on page 29. All of our articles and back issues can be found on our Web site at
<www.stsc.hill.af.mil/crosstalk>.
On behalf of the CrossTalk staff, I wish you a safe and very happy holiday season.
Tracy L. Stauder
Publisher
and user stories were not as crisp as designers sitting together can produce a High
they were before. [14] better design than each could produce
alone,” are valid. If not, you are more like- Typical
Agile
Plan-driven methods also need ly to get design by committee, with the X Methodology
(Skill, Understanding)
CRACK customer representatives and opposite effect. The plan-driven methods
benefit from full-time, on-site participa- do better with great people, but are gener-
Adapting
tion. Good planning artifacts, however, ally more able to plan the project and
enable them to settle for part-time architect the software so that less-capable Typical
Rigorous
CRACK representatives who provide fur- people can contribute with low risk. A sig- X Methodology
ther benefits by keeping active in customer nificant consideration here is the unavoid-
operations. The greatest customer chal- able statistic that 49.999 percent of the
lenge for plan-driven methods is to keep world’s software developers are below
project control from falling into the hands average (slightly more precisely, below Low
1
Lives Discretionary toward following plan-driven, SW-CMM-
5
Funds 0 35
compliant processes.
10
Essential
Funds
30
Comfort The new CMM IntegrationSM (CMMI®)
50
3 [22] upgrades the SW-CMM in more agile
90
Agi
le directions, with new process areas for inte-
10
P la grated teaming, risk management, and
70
n-D overall integrated systems and software
30 rive
n
engineering. A number of organizations
10 50
0 are welcoming this opportunity to add
30 30
more agility to their organizational culture.
0 Others that retain a more bureaucratic
10 interpretation of the SW-CMM are facing
Size Culture the challenge of change-averse change agents
(Number of personnel) (% Thriving on chaos vs. order) who have become quite comfortable in
Figure 2: Key Discriminators of Agile and Plan-Driven Home Grounds their bureaucratic culture.
of the project’s people to have Level 2 and ered when there are many degrees of freedom Values
3 skills, with no more than 10 percent of available for them to define and work Along with people come values – different
the people with Level 1B skills. The home problems. This is the classic craftsman values. One of the most significant and
ground for plan-driven methods can suc- environment, where each person is expect- underemphasized challenges in software
ceed with up to 30 percent to 40 percent ed and trusted to do whatever work is nec- engineering is to reconcile different users’,
Level 1B people, and as few as 15 percent essary for the success of the project. This customers’, developers’, and other suc-
to 20 percent Level 2 and 3 people. includes looking for common or unno- cess-critical stakeholders’ value proposi-
In fact, three of the five key discrimi- ticed tasks and completing them. tions about a proposed software system
nators in Figure 2 are people-related: per- In a plan-driven culture, the people feel into a mutually satisfactory win-win sys-
sonnel (as discussed above), size, and culture. comfortable and empowered when there tem definition and outcome. Unfor-
The size of the project is measured in the are clear policies and procedures that define tunately, software engineering is caught in
number of people. Agile methods succeed their role in the enterprise. This is more of a value-neutral time warp, where every
best on projects of 10 people or less, while a production-line environment where each requirement, use case, object, test case,
plan-driven methods work better on proj- person’s tasks are well defined. The expec- and defect is considered to be equally
ects of 100 people and up. In his landmark tation is that they will accomplish the tasks important.
XP book, Kent Beck says, to specification so that their work prod- Most process improvement initiatives
ucts will easily integrate into others’ work and debates, including the silver-bullet
Size clearly matters. You probably products with limited knowledge of what debate are inwardly focused on improving
couldn’t run an XP project with a others are actually doing. software productivity rather than outward-
hundred programmers. Not fifty. These cultures are reinforced as people ly focused on delivering higher value per
Nor twenty, probably. Ten is defi- tend to self-select for their preferred cul- unit cost to stakeholders. Again, agile
nitely doable. [20] ture, and as people within the culture are methods and their attention to prioritizing
promoted to higher management levels. requirements and responding to changes
Projects in the middle range of the key dis- Once a culture is well established, it is dif- in stakeholder value propositions are
criminator factors need a hybrid mix of ficult and time consuming to change. This pushing us in more high-payoff directions.
agile and plan-driven methods [1]. We will cultural inertia may be the most significant Other aspects of value-based software
next look more closely at culture. challenge to the integration of agile and engineering practices and payoffs are
plan-driven approaches. described in “Value-Based Software
Culture To date, agile cultural change has had a Engineering” [23]. These include the
The second area of people possibilities, bottom-up, revolutionary flavor. Failing DMR Consulting Group’s Benefits
and the third people-related key discrimi- projects with no hope of success have Realization Approach and Results Chains
nator between agile and plan-driven home been the usual pilots, supported by an it [24], stakeholder win-win requirements
grounds, is culture. In an agile culture, the can’t hurt attitude from management and a negotiation [25], business case analysis
people feel comfortable and are empow- no challenge is too hard adrenaline-charged [26], and the Kaplan-Norton Balanced
response from practitioners. Successes Scorecard technique [27].
SM CMM Integration is a service mark of Carnegie Mellon have been extraordinary in many cases and
University.
® CMM is registered in the U.S. Patent and Trademark have been used to defend migration to less Communications
Office. troubled projects. Even with closely knit in-house develop-
ment organizations, the “I can’t express We should note that this distinction discussed earlier, human success and fail-
exactly what I need, but I’ll know it when I see between agile-tacit and plan-driven-explicit is ure modes, information radiators and con-
it” syndrome limits people’s ability to com- not absolute. Agile methods’ source code vection currents, and the effects of dis-
municate an advance set of requirements and test cases certainly qualify as explicit tance on communication effectiveness.
for a software system. If software defini- documented knowledge, and even the
tion and development occurs across orga- most rigorous plan-driven method does Expectations Management
nizational boundaries, even more commu- not try to get along without some inter- Our primary conclusion in analyzing soft-
nications work is needed to define and personal communication to ensure a con- ware project critical success factors has
evolve a shared system vision and devel- sistent, shared understanding of docu- been that the differences between suc-
opment strategy. The increasingly rapid mentation intent and semantics. cessful and troubled software projects are
pace of change exacerbates the problem When agile methods employ documen- most often the difference between good
and raises the stakes of inadequate com- tation, they emphasize doing the minimum and bad expectations management. This
munication. essential amount. Unfortunately, most coincides with a major finding in a recent
Agile methods rely heavily on commu- plan-driven methods suffer from a tailoring- root-cause analysis of trouble factors in
nication through tacit, interpersonal knowledge down syndrome, which is sadly reinforced Department of Defense software proj-
for their success. They cultivate the devel- by most government procurement regula- ects [28].
opment and use of tacit knowledge, tions. These plan-driven methods are Most software people do not do well at
depending on the understanding and expe- developed by experts who want them to expectations management. They have a
rience of the people doing the work and provide users with guidance for most or all strong desire to please and to avoid con-
their willingness to share it. Knowledge is foreseeable situations. The experts, there- frontation, and have little confidence in
specifically gathered through team plan- fore, make them very comprehensive, but their ability to predict software project
ning and project reviews (an activity tailored down for less critical or less com- schedules and budgets, making them a
agilists refer to as retrospection). It is shared plex situations. The experts understand pushover for aggressive customers and
across the organization as experienced tailoring the methods and often provide managers trying to get more software for
people work on more tasks with different guidelines and examples for others to use. less time and money.
people. Unfortunately, less expert and less self- The most significant factor in success-
Agile methods generally exhibit more confident developers, customers, and ful agile or plan-driven teams is that they
frequent, person-to-person communica- managers tend to see the full-up set of have enough process mastery, preparation,
tion. As stated in the Agile Manifesto, plans, specifications, and standards as a and courage to be able to get their cus-
emphasis is given to individuals and interac- security blanket. At this point, a sort of tomers to agree to reduce functionality or
tions. Few of the agile communications Gresham’s Law (bad money drives out good increase schedule in return for accommo-
channels are one-way, showing a prefer- money) takes over, and the least-expert par- dating a new high-priority change. They
ence for collaboration. Stand-up meetings, ticipant can drive the project by requiring are aware that setting up unrealistic expec-
pair programming, and the planning game the full set of documents rather than an tations is not a win for the customers
are all examples of the agile communica- appropriate subset. While the nonexperts either, and are able to convince the cus-
tion style and its investments in developing rarely read the ever-growing stack of doc- tomers to scale back their expectations.
shared tacit knowledge. uments, they will maintain a false sense of Both agile short iterations and plan-driven
Relying completely on tacit knowledge security in the knowledge they have fol- productivity calibration are keys to suc-
is like performing without a safety net. lowed best practices to ensure project pre- cessfully managing software expectations.
While things go well, you avoid the extra dictability and control. Needless to say, the
baggage and setup effort, but there may be expert methodologists are then frustrated Conclusion
situations that will make you wish for that with how their tailorable methods are used Giving top-priority attention to such peo-
net. Assuming that everyone’s tacit knowl- – and usually verbally abused – by devel- ple-related factors as staffing, culture, val-
edge is consistent across a large team is opers and acquirers alike. Agilists have cer- ues, communications, and expectations
risky, and as people start rotating off the tainly highlighted this significant problem management is critical to successful soft-
team, the risk gets higher. in plan-driven methods. ware development and management.
At some point, a group’s ability to Except for the landmark people-ori- Beyond this top-level summary of key
function exclusively on tacit knowledge ented sources mentioned above, there are factors, there are many valuable sources
will run up against well-known scalability frustratingly few sources of guidance and of guidance on how to succeed with the
laws for group communication. For a team insight on what kinds of communications people-related aspects of your software
with N members, there are N(N-1)/2 dif- work best in what situations. Cockburn’s projects.
ferent interpersonal communication paths “Agile Software Development” [18] is a Besides the classic Weinberg, Ehn, and
to keep current. Even broadcast tech- particularly valuable recent source. It gets DeMarco-Lister books previously cited,
niques such as stand-up group meetings its priorities right by not discussing meth- there are some further references that can
and hierarchical team-of-teams techniques ods until the fourth chapter, and spending help you improve your people factors
run into serious scalability problems. the first hundred or so pages discussing whether you use agile, plan-driven, or
Plan-driven methods rely heavily on why we have problems communicating, hybrid development approaches. Good
explicit documented knowledge. With plan-driv- and what can be done about it. It nicely agilist treatments of people and their
en methods, communication tends to be characterizes software development as a ecosystems are provided in Jim
one-way. Communication is generally from cooperative game of invention and com- Highsmith’s “Agile Software Develop-
one entity to another rather than between munication, and provides numerous help- ment Ecosystems” [19] and Alistair
two entities. Process descriptions, progress ful communication concepts and tech- Cockburn’s “Agile Software Develop-
reports, and the like are nearly always com- niques. Some examples are his definition ment” [18]. Complementary plan-driven
municated as unidirectional flow. of the three skill levels based on Aikido approaches are provided in Watts
Humphrey’s “Managing Technical 14. van Duersen, A. “Customer Involve- gineering Notes Mar. 2003.
People” [29] and his Personal Software ment in Extreme Programming.” ACM 24. Thorp, J. The Information Paradox.
ProcessSM [30], as well as the People CMM Software Engineering Notes Nov. McGraw-Hill, 1998.
developed by Bill Curtis, Bill Hefley, and 2001: 70-73. 25. Boehm, B., P. Bose, E. Horowitz, and
Sally Miller [31]. 15. Britcher, R. N. The Limits of Software. M. J. Lee. Software Requirements as
As engineers, our selection of reading Reading, MA: Addison-Wesley, 1999. Negotiated Win Conditions. Proc. of
materials tends to gravitate toward pro- 16. Highsmith, J., and A. Cockburn. “Agile the First International Conference on
gramming, architecture, or processes for Software Development: The Business Requirements Engineering, Colorado
our next learning experience. We strongly of Innovation.” Computer Sept. 2001: Springs, CO. IEEE Computer Society
recommend you choose one of the books 120-122 Press, Apr. 1994.
above as a way to balance your technical 17. Constantine, L. “Methodological 26. Reifer, D. Making the Software
and people skills.◆ Agility.” Software Development June Business Case. Boston: Addison-
2001: 67-69. Wesley, 2002.
References 18. Cockburn, A. Agile Software Devel- 27. Kaplan, R., and D. Norton. The
1. Boehm, B., and R. Turner. Balancing opment. Boston: Addison-Wesley, Balanced Scorecard: Translating
Agility and Discipline: A Guide for the 2002. Strategy into Action. Boston: Harvard
Perplexed. Boston: Addison-Wesley, 19. Highsmith, J. Agile Software Devel- Business School Press, 1996.
2004. opment Ecosystems. Boston: 28. McGarry, J., and Charette, R. “Systemic
2. Beck, K., et al. “The Agile Manifesto.” Addison-Wesley, 2002. Analysis of Assessment Results from
The Agile Alliance, 2001 <www.agile 20. Beck, K. Extreme Programming DoD Software-Intensive System
alliance.com>. Explained. Boston: Addison-Wesley, Acquisitions.” Tri-Service Assessment
3. Weinberg, G. The Psychology of 1999: 157. Initiative Report, Office of the Under
Computer Programming. New York: 21. Paulk, M., et al. The Capability Secretary of Defense (Acquisition,
Van Nostrand-Reinhold, 1971. Maturity Model. Reading, MA: Technology, Logistics), 2003.
4. Ehn, P., Ed. Work-Oriented Design of Addison-Wesley, 1994. 29. Humphrey, W. Managing Technical
Computer Artifacts. Mahwah, NJ: 22. Ahern, D. M., A. Clouse, and R. People. Boston: Addison-Wesley, 1997.
Lawrence Earlbaum Associates, Mar. Turner. CMMI Distilled: A Practical 30. Humphrey, W. Introduction to the
1990. Introduction to Integrated Process Personal Software Process. Boston:
5. DeMarco, T., and T. Lister. People- Improvement. 2nd ed. Boston: Addison-Wesley, 1997.
ware: Productive Projects and Teams. Addison-Wesley, 2003. 31. Curtis, B., B. Hefley, and S. Miller. The
New York: Dorset House, 1999. 23. Boehm, B. “Value-Based Software People Capability Maturity Model.
Engineering.” ACM Software En- Boston: Addison-Wesley, 2001.
6. Curtis, B., H. Krasner, and N. Iscoe. “A
Field Study of the Software Design
Process for Large Systems.” Comm. About the Authors
ACM 31. 11 (Nov. 1988): 1268-1287.
7. Curtis, B. et al. People Capability Richard Turner, D.Sc., Barry Boehm, Ph.D.,
Maturity Model. Reading, MA: is a member of the is the TRW professor
Addison-Wesley, 2001. Engineering Manage- of software engineer-
8. Grant, E., and H. Sackman. “An ment and Systems En- ing and director of the
Exploratory Investigation of Pro- gineering Faculty at The Center for Software
grammer Performance Under On-Line George Washington Engineering at the
and Off-Line Conditions.” Report SP- University in Washington, D.C. University of Southern California. He
2581, System Development Corp., was previously in technical and man-
Sept. 1966. Currently, he is the assistant deputy
director for Software Engineering and agement positions at General
9. Boehm, B. Software Engineering
Economics. Upper Saddle River, NJ: Acquisition in the Software Intensive Dynamics, Rand Corp., TRW, and the
Prentice Hall, 1981. Systems Office of the Under Secretary Office of the Secretary of Defense as
10. Boehm, B., et al. Software Cost of Defense (Acquisition, Technology, the director of Defense Research and
Estimation With COCOMO II. Upper and Logistics). Turner is co-author of Engineering Software and Computer
Saddle River, NJ: Prentice Hall, 2000. the book “CMMI Distilled.” Technology Office. Boehm originat-
11. Dijkstra, E. Panel Discussion. Fourth ed the spiral model, the Constructive
International Conference on Software 1931 Jefferson Davis Highway Cost Model, and the stakeholder win-
Engineering, 1979. Suite 104 win approach to software manage-
12. Paulk, M., et al. The Capability Arlington, VA 22202 ment and requirements negotiation.
Maturity Model for Software: Phone: (703) 602-0581 ext. 124
Guidelines for Improving the Software E-mail: rich.turner.ctr@osd.mil University of Southern California
Process. Reading, MA: Addison- Center for Software Engineering
Wesley, 1994. Los Angeles, CA 900989-0781
13. Tucker, A. “On the Balance Between
Phone: (213) 740-8163
Theory and Practice.” IEEE Software
(213) 740-5703
Sept.-Oct. 2002.
Fax: (213) 740-4927
SM Personal Software Process is a service mark of Carnegie
E-mail: boehm@sunset.usc.edu
Mellon University.
Measurements and metrics are key tools to understanding the behaviors, successes, and failures of our programs and projects.
This article highlights the basic principles of measures and metrics and encourages the reader to improve his or her use of these
tools. The article is adapted from [1].
Can questions be answered by provid- ner to those who need them? Feb. 2003.
ing specific information? (Is it unam- Are metrics being used in the deci- 2. DeMarco, Tom. Controlling Software
biguous?) sion-making process? Projects. New York: Yourden Press,
Do the questions ask for all the infor- 1982.
mation needed to determine progress Metrics Program Evaluation 3. Perkins, Timothy K. “The Nine-Step
or completion of the goal? Are the metrics sufficient? Metrics Program.” CrossTalk, Feb.
Is each metric required for specific Are all metrics or measures required, 2001 <www.stsc.hill.af.mil/crosstalk/
decision-making activities? that is, non-superfluous? 2001/feb/perkins.asp>.
Is each metric derived from two or Are measurements allowing project
4. Bailey, Elizabeth, et al. Practical
more measurements (e.g., remaining work to continue without interfer- Software Measurement: A Foundation
budget vs. schedule)? ence? for Objective Project Management
Have you documented the analysis Does the analysis produce accurate
Ver. 4.0b1. Severna Park, MD: Prac-
methods used to calculate the met- results? tical Software and Systems Measure-
rics? Is the data collection interval appro-
ment, Mar. 2003 <www. psmsc.com/
Have you defined those measures priate?
PSMGuide.asp> under “Products.”
needed to provide the metrics? Is the metrics program as simple as it
Have you defined the collection can be while remaining adequate? 5. Park, Robert E., et al. Goal-Driven
process (i.e., what, how, who, when, Has the metrics program been modified
Software Measurement – A Guide-
how often, etc.)? to adequately accommodate any project book. Pittsburgh, PA: Software
or organizational goal changes?◆ Engineering Institute, Aug. 1996
Metrics Program Implementation <www.sei.cmu.edu/publications/
Does your implementation follow the References d o c u m e n t s / 9 6 . r e p o r t s / 9 6 . h b.
metrics program plan? 1. Software Technology Support Center. 002html>.
Is data collected the same way each Condensed Version (4.0) of Guide- 6. Augustine, Thomas, et al. “An
time it is collected? lines for Successful Acquisition and Effective Metrics Process Model.”
Are documented analysis methods Management of Software-Intensive CrossTalk June 1999 <www.stsc.
followed when calculating metrics? Systems. Hill Air Force Base, Utah: hill.af.mil/crosstalk/1999/jun/
Are metrics delivered in a timely man- Software Technology Support Center, augustine.asp>.
About the Authors
Tim Perkins has been Roald E. Peterson is Larry Smith is a sen-
involved in software a senior systems engi- ior software engineer
process improvement neer with Science Ap- and project manager
for the past 11 years, plications International for the Air Force’s
including leading the Corporation. He has Software Technology
effort to initiate soft- 22 years of electronic Support Center at Hill
ware process improvement at the systems development experience, Air Force Base, Utah. He provides
then five Air Force Air Logistics specializing in communications, software engineering, software
Centers. As the Software Engineering architecture, and software develop- process improvement, and project
Process Group leader at the Software ment. Peterson was an editor and management consulting for the U. S.
Engineering Division at Hill Air contributor for the “Guidelines for Air Force and other Department of
Force Base, Utah, he led the division the Successful Acquisition and Defense organizations as well as
in reaching Capability Maturity Management (GSAM) of Software commercial and nonprofit organiza-
Model® (CMM®) Level 3. The divi- Intensive Systems” and is the author tions. Smith is a faculty member at
sion has gone on to achieve CMM of the “Condensed GSAM the University of Phoenix. He is also
Level 5. Perkins is Acquisition Handbook.” He has a bachelor’s certified by the Project Management
Professional Development Program degree in physics and master’s Institute as a Project Management
Level 3 certified in Project degrees in computer resources man- Professional. Smith has a bachelor’s
Management and System Planning, agement and electrical engineering. degree in electrical engineering and a
Research, Development, and master’s degree in computer science.
Engineering. Science Applications
International Corporation Software Technology Support Center
Software Technology Support Center 920 W. Heritage Park Blvd. OO-ALC/MASE
OO-ALC/MASE Suite 210 6022 Fir Ave.
6022 Fir Ave. Layton, UT 84041 Bldg. 1238
Bldg. 1238 Phone: (801) 774-4705 Hill AFB, UT 84056
Hill AFB, UT 84056 Fax: (801) 728-0300 Phone: (801) 777-9712
Phone: (801) 775-5736 E-mail: roald.e.peterson@saic.com Fax: (801) 777-8069
Fax: (801) 777-8069 E-mail: larry.smith4@hill.af.mil
E-mail: tim.perkins@hill.af.mil
lunch yesterday? Like most people, some- ones are fine for small course corrections what you saw and heard. “In the 1 p.m.
times I do, and sometimes I do not. If and positive feedback. Schedule a sepa- meeting, you hit the conference table with
someone asks me about an event that rate private meeting to address situations your fist several times and your face
happened several weeks or months ago, I that involve a serious performance issue. became red.” Then you can discuss what
may or may not remember depending on responses are appropriate.
the significance the event held for me. Address Individual Issues Individually The same guideline holds for feed-
That is why it is important to give Some new managers try to avoid the awk- back about what is going well. Give spe-
feedback as close to the event as possible. wardness of providing feedback by mak- cific examples. Suppose your technical
People are not likely to remember past ing policy announcements in staff meet- lead did an exceptional job writing up a
events clearly, particularly when they do ings rather than address the individual site-visit report. Saying, “I really liked the
not have personal significance. directly. General policy announcements report you wrote,” is a nice compliment,
Marie, the employee whose manager, do not carry the same weight as an indi- but it does not tell your technical lead
Alex, decided he would wait three months vidual conversation. what you liked or what he or she did well.
to give her feedback, may not even When the general announcement con- Saying, “I particularly liked the way the
remember the incident by the time he cerns a specific behavior, the results can topics were prioritized. It made the report
brings it up. Marie’s behavior stood out to be even worse. One new supervisor made very easy to follow,” will let your technical
Alex, but since it is an ingrained habit, it a point in his weekly staff meeting of lead know that you value a well-organ-
might not stand out to Marie. reminding everyone not to use the speak- ized, easy-to-follow report. Without spe-
Worse, if Alex waits to tell Marie, how erphone to listen to voice mail. Only one cific information on what he or she did
many times will she repeat the gaffe in the well, your technical lead might have con-
intervening three months? Why rob cluded that you liked the fact the report
Marie of a chance to improve now rather
than later? If Marie’s actions affect the
“First and foremost was printed on pink paper.
Vague references, labels, and guessing
entire group by damaging the relationship feedback is information games leave employees feeling resentful
with the client, why accept the negative and confused. Employees are likely to
effects on the group for longer than nec- that we can use to ignore vague feedback. Specific observa-
essary? tions give people the information they
It might not be easy for Marie to understand the current need to know what result or behavior
change her habit, but it will be impossible they need to change.
if she is not aware of it. situation and make
Late feedback is a lost opportunity. Do Not Rely on Mind Reading
Give the feedback close to the event so choices.” Do not expect your employees to be mind
that the individual has a choice to change readers. One employee was surprised
and to improve group performance. when her manager informed her that she
person in the group had this habit and she was not meeting expectations. “What am
Provide Feedback on a Regular Basis felt publicly embarrassed when her super- I doing wrong?” she asked. “I expect
Regular one-on-one meetings are a great visor brought it up in the staff meeting. someone at your grade level to know what
place to provide feedback. Give informa- The other members of the team felt to do without me having to spell it out,”
tion for minor course corrections, and unfairly chastised. her manager replied.
discuss what is going well. If you do not If it is important enough to bring up, Another employee was likewise sur-
have a minor course correction, do not it is important enough to speak directly to prised when his manager told him the way
feel like you have to make one up. There the person involved. he had handled a customer meeting was
is no need to pounce on every little thing. unacceptable. “What did I do wrong?” he
If an employee breaks the build once in a Provide Specific Examples asked. “You know what you did,” the
year, or comes in late once in six months, General statements do not provide manager replied.
it is not a performance problem [3]. enough information for people to know Dropping hints does not work, nor
I have a rule of thumb, though, about what to improve. Labels, such as you are does making the employee guess what
noticing some positive aspect on a regular sloppy, set up an oppositional dynamic that you mean. If you want to see a change,
basis. Regular recognition for work well goes nowhere. Labels get in the way of say what it is.
done is one of the keys to being a great solving the problem. Rather than label the
manager [4]. Even if you do not have reg- behavior, describe it. Check for Agreement on the Data
ular meetings, find a way to notice and Whether you are describing results or Once you have given some specific exam-
appreciate the work people do every week. behavior, be as specific as you can. A tech ples of the result or behavior you have
People need to know what they are writer turned in a chapter that had dozens observed, obtain agreement on the data.
doing well – and should continue doing – of spelling and grammar errors. Telling If the feedback receiver does not agree
as much as they need to know when they the tech writer that her work is poor qual- with the data, it is going to be hard to
are missing the mark. ity is not specific enough. If you are con- move into problem solving.
cerned about accuracy, state what you The tech writer who turned in work
Give Serious Feedback in a have seen: “I noticed that in Chapter 1 with errors will probably acknowledge
Serious Setting there are numerous spelling and grammar that there were spelling errors in Chapter
Hints and off-hand comments in the hall- errors.” 1. She is not as likely to admit to being
way usually fail to get the message across. Say that one of your testers loses his sloppy. The tester who hit the table will
If you want an employee to act on your temper in a bug-fix meeting. Rather than probably admit that he hit the table, but
request, be intentional about it. One-on- tell him he has a temper problem, relate may disagree that he has an anger-man-
agement problem. week. He was concerned that the code burst or with tears – you will need some
Employees are more likely to accept was brittle and wanted her to do more strategies to bring the feedback conversa-
what you say if it is specific and observ- unit testing. When Charlie asked Shanna tion back on track.
able. to summarize their feedback conversa-
tion, Shanna replied, “You are disappoint- Check the Data
Request a Change ed in me. I need to be more careful.” When people are upset, they often do not
If you want a behavior to stop or contin- Charlie was able to correct Shanna’s hear clearly. Ask the feedback receiver to
ue, say that. If you want a change, say misperception. He was not disappointed repeat what he heard and what he saw.
that, too. Do not leave the employee in her – actually, he was quite pleased with When one manager asked a staff member
guessing what he or she needs to do to her work in general. He had noticed a pat- what he heard her say after a puzzling
correct the situation. tern and wanted her to take some addi- response, he described the way the man-
The tech writer’s manager might say tional steps so she and the entire group ager was leaning forward, the tone of her
something like, “I have marked spelling would be more successful. voice, her facial expression – but no
and proofing errors on the copy you gave For many people, hearing criticism is words. He was reacting primarily to what
me. In the future, I’d like you to run spell an emotionally charged situation. When he had seen. Once the manager repeated
check and proofread before you turn in people are in the throes of an emotional the words, and her staff member heard
your work.” response, they often do not hear clearly. them, they were able to move forward.
The table-hitting tester’s boss might
say, “I know you can’t control when your Check for Interpretation
face becomes red; hitting the table with
your fist is not acceptable.”
“The goal in giving If the words made it through intact but
the interaction is still tangled, check how
feedback is not to make the receiver interpreted your words. We all
Engage in Problem Solving make meaning of what we hear; in most
If you want a different result, but have sure everyone is cases, our interpretation is close enough
latitude on how that result is achieved, to allow productive communication.
move into problem-solving mode. The charming or performs Sometimes our interpretation is way off,
tech writer’s manager might say, “I was and then it helps to check.
surprised at the number of spelling and his or her work exactly Shanna interpreted Charlie’s feedback
proofing errors in the chapters you as criticism. Charlie had noticed a pattern
turned over to me. I expect copy to be the same way you would. and was offering information to help
clean when I receive it. What are three Shanna be more effective. When Charlie
ways you could make sure the copy is in It is to make sure that asked Shanna to summarize for under-
good shape?” standing, he was able to see that Shanna’s
The tech writer might reply, “Well, I the people who work interpretation was a little off.
could run spell check, proof it on paper,
or ask Jessica do a final check.” for you are productive.” Ask What Is Happening
When the employee arrives at the Sometimes a feedback response is still
solution, it is more likely to fit his style, puzzling even after you have checked the
and is more likely to stick. As a feedback giver, you do not have con- data and the interpretation. This may
trol over how someone else will respond indicate other forces are at work. Perhaps
Agree on How You Will Follow Up to your feedback. However, checking for your feedback triggered a memory or
Sometimes you will need to follow up on understanding can clear up some obvious association with a painful past event.
the changes you have requested. The tech misinterpretations. Perhaps the feedback receiver has a per-
writer and her manager might agree that fection rule, and any comment that indi-
he will skim for obvious spelling and Troubleshooting Feedback cates his work is not perfect is difficult.
grammar errors and if he finds them, he Conversations As a manager, it is not your job to psy-
will return the work to her to fix. When you are careful and intentional in choanalyze. It is your job to try to get the
giving feedback, you greatly increase the conversation back on track. I have found
Check for Understanding chance that you will be successful. You that when I can sincerely say, “I’m puz-
Saying the words is not enough. When will state what you have observed clearly. zled by your response. What’s happening
giving feedback is part of your job, you The receiver will agree with your observa- for you?” I’m often able to redirect the
also need to check to make sure the mes- tion, you will do some joint problem solv- conversation to the present circumstance.
sage you sent is the same one the employ- ing, and agree on a course of action. Sometimes it is tempting to think we
ee heard. Checking for understanding can In real life, even when you follow all know what is behind someone else’s
help correct some of the slippage that these guidelines, sometimes the feedback behavior. For example, one manager
normally occurs in conversation. receiver will have a response that is sur- reported after a year-end review that her
One manager I know wraps up feed- prising or puzzling. employee was angry and resistant during
back conversations by saying, “I’m going As we saw with Charlie and Shanna, the performance discussion.
to check for understanding now. I’d like you cannot control how the feedback “What did you see or hear that lead
you to summarize our conversation for receiver interprets what you say, how they you to that conclusion?” I asked.
me so I’m sure I have been clear.” react emotionally to that interpretation, “After I told him that our senior man-
Charlie told Shanna, a recent college and how they respond. ager wasn’t sure what he did, he asked for
graduate, that the code she had checked When the receiver responds in a puz- examples,” the manager said.
in had broken the build three times in a zling way – perhaps with an angry out- “That seems like a reasonable
organization, I would arrive at work in the personalities as the people already in our closed question: “Have you ever managed
morning, check my voice mail, and groups. Hiring people who are just like the a project where the team had trouble
respond to all message requests. That ones we have now does not always gain the meeting the schedule?” If the answer is
took until noon. Again after lunch, I best people for the job. no, you can decide if the project manager
would check my e-mail and voice mail and When you hire people your staff has enough experience to manage your
run around responding to those urgent thinks are great, you increase morale in team. If the answer is yes, ask the open-
requests. After a week of this, I realized I the group, and you increase your group’s ended behavior-description question,
was not performing any of the important capacity over time. I recommend you “What did you do? What actions did you
work such as planning for the group and develop a hiring strategy that identifies the take on that project to help the project
lab, reviewing critical development plans, technical and soft skills you are looking team meet the schedule?” The answers
or planning my hiring strategy. I also real- for, and that you choose a variety of tech- you hear will help you assess that candi-
ized that although people marked their e- niques for interviewing. date’s ability to work in your organization.
mails and voice mails high priority, they did I have found auditions [3, 4, 5] to be
not utilize the information I had given an essential technique for interviewing 6. Preserve Good Teams
them at the time I responded. technical staff. I normally create 30- to 45- Part of my hiring strategy is to hire people
I stopped responding immediately to minute auditions to see how a person who fit into my already-existing team, but
urgent requests and re-planned my days. works in a particular setting. Auditions sometimes you inherit teams or a project
While I still checked voice mail and e-mail, help candidates show what they can do. If has completed and a team is ready to
I tended to ask more questions about the you organize a congruent audition, you do move on. When a team is successful, I try
deadlines for requests. Prioritizing not trip people up on esoteric ideas or jar- to keep them together so they can contin-
requests helped me manage my manage- gon; you create a simplified situation that ue working well together. I may bring
ment time. I still had the problem of too more people into the team, one at a time,
many high priority projects coming into especially if the team has been highly pro-
my group, so I asked my manager these
questions:
“Hiring people who are ductive. But I do not scatter the produc-
tive team and hope they will form more
• If you could have one project first, just like the ones we productive teams. That just reduces their
which one would it be? productivity.
• What are the consequences if we have now does not Teams can overcome bad manage-
release any of these projects late? ment and bad processes, but they cannot
We talked and negotiated which projects always gain the best overcome a team un-jeller. A team un-
had to be completed when and why. jeller is the person who walks into the
When I understood the trade-offs people for the job.” lunchroom, and suddenly everyone else
between projects, I was able to manage leaves. Or, the un-jeller creates an argu-
the work coming into my group. ment out of every conversation. If you
the candidate could encounter at work. have a team un-jeller, find another place
4. Commit to Projects After Watching the candidate, or having the for that person to work, preferably for
candidate explain their answers/results is
Checking With Your Staff a powerful interview technique.
your competitor.
Business needs change. Sometimes your You can create auditions for any posi-
manager will grab you in the hall and say, 7. Avoid Micromanaging or
tion, including project managers, develop-
“Hey, can you do this project now, and ers, testers, writers, support staff, analysts, Inflicting Help
finish it in two months?” Or, a senior systems engineers, product managers, Many of us were software developers,
management planning committee will call program managers, and people managers. testers, analysts, or some other technical
you into its meeting and say, “We need Define the behaviors you require in a role before we became managers. When
this project now. Can you commit to it?” position, and then create an audition using we were technical contributors, we knew
It is very tempting to say yes. your products or open source products to how to perform the technical jobs.
However, saying yes is exactly the wrong see the person at work. Create auditions However, once you have been a manager
thing to do. You can say, “Let me check to that are 30 minutes long to start. If you for a while, you probably will not know
see if my previous estimate is still accu- are having trouble deciding between mul- precisely how to perform the employee’s
rate, and I’ll get back to you before 5 p.m. tiple candidates, define another audition job.
today.” that is one hour long and invite the candi- I once had a boss who liked to creep
If you say yes, you are training your dates back to see how they manage that into my office, stare over my shoulder,
senior management to ask you for audition. Auditions show you how the and say, “On line 16, shouldn’t that be a
answers when you do not know the person works at work – priceless informa- …” By the time he reached “16,” I had
answers. You have also committed your tion. jumped out of my chair, become flus-
staff to a project that may not be within I also recommend behavior-descrip- tered, and lost my concentration.
the scope you originally estimated. tion interview questions [5, 6] to under- Micromanagement neither gets the job
stand how a candidate has performed in done faster, nor does inflicting advice or
5. Hire the Best People for previous jobs. Behavior-description ques- help.
the Job tions are open-ended and ask the candi- On the other hand, you and the
Especially if you manage many projects, date to tell you the story of previous employee both need to know that the
your greatest leverage point is in hiring work. For example, to understand how a employee is progressing. I ask my staff to
appropriate staff. Too often, we hire peo- project manager deals with a project team decide when they have been stuck for too
ple who have similar technical skills and who has not yet met a schedule, ask this long (time-box the work). Some tasks
18 CROSSTALK The Journal of Defense Software Engineering December 2003
Successful Software Management: 14 Lessons Learned
require weeks of study, but most tasks progressing. I use one-on-ones weekly to learning new things. If you have a budget
require days or hours. If the employee meet with each person. We discuss the for formal training, that is great. Even if
spends more than the agreed-upon time employee’s progress on his or her tasks. you do not have a budget, plan weekly
on the task, their job is to ask for help. As Sometimes, tasks are amorphous and it is training time in the form of brown-bag
the manager, your job is to find them help, difficult to know when to stop or if the lunches, presentations from other groups
not necessarily inflict your help. employee needs help. I ask each employee in your organization, an internal user-
to show me visible progress on each task: group meeting of one of your tools, or
8.Treat People Individually drafts of plans, multiple designs, proto- presentations from people in your group
and With Respect type test results, anything that shows me about their successes or difficulties.
Buckingham and Coffman [7] claim that the employee is making progress and is I use the weekly group meeting as a
each employee’s relationship with his or not stuck. If the employee needs help time to deliver the training. When I man-
her manager is key to that employee’s suc- completing the task, we discuss what aged development groups, I organized
cess and long-term happiness in the kinds of help are appropriate. this internal training, including technical
organization. That means we need to treat I receive many benefits from these leads of other sub-projects to explain
people fairly, but uniquely, so that we weekly meetings. I learn what everyone is their architecture and application pro-
build and maintain the best possible rela- doing and can track it in my notebook. It gramming interface (API) to other
tionships with each employee. is easy to write up useful performance groups, testers to explain patterns of
Everyone has his or her own prefer- evaluations, including examples of suc- defects they found, different techniques
ences, especially in their communication cessful and not so successful actions the for peer review, or discussion of a partic-
patterns, and how they organize their employee has taken over the year. And, ularly interesting article in one of the
thoughts about work. Some people prefer because we meet weekly, I can give feed- technical magazines someone had read.
e-mail communications; some prefer in-
person discussions. Some people want to 11. Fire People Who Cannot
understand all the reasons behind your
requests, and others will take the request
“Buckingham and Perform the Work
Even when you meet regularly with your
at face value. Some people need to gather Coffman [7] claim that staff and encourage them to acquire help
data to make decisions; others will devel- when they need it, some people in your
op a model about the situation and make each employee’s group may not be able to perform at the
a decision based on that model.
It does not matter if people work top-
relationship with his or level you require. First, make sure you
have been specific and have given feed-
down or bottom-up, or if they want to her manager is key to back to the employee with examples of
talk in person or by e-mail. What matters inadequate behavior. If the employee
is that you, within reason, accommodate
everyone’s uniqueness.
that employee’s success understands the lack of performance, you
can choose whether to coach the person
I once managed two very talented
developers who shared a large office.
and long-term happiness or perform a get-well plan, or in radical
circumstances, escort the employee out
Begrudgingly, they allowed me to have 20-
minute one-on-ones with each of them
in the organization.” the door.
Retaining non-productive employees
every two weeks. In between, if I wanted has direct and indirect costs. The direct
to talk to either of them, I had to e-mail back then, not when we make time. I also costs are easier to define: You are paying a
them first – dropping in was not allowed. reduce the number of staff interruptions salary and benefits and not receiving the
I treated them differently than the other because everyone knows they can ask me expected work. The indirect costs are
people in my group, but fairly, considering non-urgent questions during the one-on- much subtler and more damaging.
their preferences. one. I can perform weekly career develop- When you continue employing an
They frequently worked on the same ment and learn if my staff has personal inadequate employee, the morale of the
software. They never spoke to each other issues affecting their ability to do their entire workgroup declines. If morale
aloud, they only communicated via e-mail jobs. declines enough, your best people will
even though they shared an office. If I am managing more than eight leave. Not only do you have someone in
Because they were so successful at their people, I meet biweekly with more senior your group who is not successful, that
work together, and even mentored others staff because they need less direct super- person has driven away the people who
in the organization by e-mail, their com- vision. are the most successful.
munications preferences were a bit odd Some of you are probably thinking In addition to low morale, you and
but acceptable. If I had tried to change you do not have time to meet with every- your group accomplish less than you
them to meet my needs and work with one once a week. However, if you do not expected. You are not just accomplishing
them the same way I worked with the set up specific times to meet with every- less because of the one employee who
other people, none of us would have been one, you tend to either not know what cannot work at the level you require; that
happy. people are doing, or you are interrupted person probably has to hand off work to
frequently by your staff with questions. others in the group, and those other peo-
9. Meet Weekly With Each ple will be delayed by the inadequate
Person 10. Plan Training Time Each work.
Even if you have hired stars, you still need Week I once inherited a group where the
to know each person’s progress on their Technical work is constantly changing; previous management had spared an
tasks, and how the project as a whole is most of the technical people I know enjoy employee from layoffs because he was
having personal problems. Those person- they are paid fairly, then more money is References
al problems affected his work – he did not not reward enough. Recognition of good 1. Magretta, Joan. What Management Is:
always come to work, he was late on every work and the opportunity to perform How It Works and Why It Is
deliverable, and he was unable to perform meaningful work [9] is much more impor- Everyone’s Business. New York: The
most of his work. In my one-on-ones tant. Lack of money can be a demotivator, Free Press, 2002.
with the employee, I gave him examples but only money is not sufficient when rec- 2. Covey, Stephen R. The Seven Habits
of his work and asked if he was able to ognizing good work.
of Highly Effective People. New
continue to work. He said yes. (If he had Kohn says, “[Rewards] motivate peo-
ple to get rewards.” If your organization York: Simon & Schuster, 1989.
said no, we would have put him on short-
term or long-term disability.) We chose to has trained employees to expect money as 3. DeMarco, Tom. Slack. New York:
perform a get-well plan, which the a reward, this appreciation technique may Broadway Books, 2001.
employee stopped after a week. After the seem small. Try it anyway. 4. Weinberg, Gerald M. “Congruent
employee left, the morale in the group When I use appreciation as a recogni- Interviewing by Audition.” Amplifying
jumped dramatically and we were able to tion technique I say, “I appreciate you, Your Effectiveness: Collected Essays.
accomplish more work. Jim, for your work on the blatz module New York: Dorset House, 2000.
and API definition. Your work made it 5. Rothman, Johanna. Hiring Technical
12. Emphasize Results, Not possible for Joe to write great tests and for People. New York: Dorset House,
Time me to predict the project’s progress.” 2003.
I have worked for senior managers who Appreciation between peers could mean 6. Janz, Tom, et al. Behavior Description
rewarded individuals based on their work even more than money from you. When Interviewing. Englewood Cliffs, NJ:
hours, i.e., those who started early and you appreciate a person for good work Prentice Hall, 1986.
stayed late. Unfortunately, these managers and you explain what the work meant to 7. Buckingham, Marcus, and Curt
had no ability to understand the results you, you are motivating the person to con-
Coffman. First, Break All the Rules:
the long-working employees imposed on tinue performing similar work.
In addition, consider time off, group What the World’s Greatest Managers
the rest of the organization: buggy code, Do Differently. New York: Simon &
inadequate designs, and tests that did not activities, movie tickets, or funny awards
such as best recursion of the week as recogni- Schuster, 1999.
find obvious problems. When people 8. DeMarco, Tom, and Tim Lister.
work long hours, their productivity tion techniques.
The most important part of a reward Peopleware: Productive Projects and
decreases, not increases [8]. In “Slack” [3], Teams. 2nd ed. New York: Dorset
is to make sure it is congruent with each
Tom DeMarco says, “Extended overtime
person’s performance. Your staff knows House, 1999.
is a productivity-reduction technique.”
who is performing well and who is coast- 9. Kohn, Alfie. Punished by Rewards.
The longer people stay at work, the less
ing. If you recognize and reward evenly, New York: Houghton-Mifflin, 1993.
work they do. Instead, they perform the
you are not differentiating between out-
life activities they are not performing out- About the Author
standing performance and adequate per-
side of work.
formance. Make sure you reward a per- Johanna Rothman
Make it possible for people to only
son’s entire contribution (the entire work
work 40 hours a week. The less overtime consults on managing
product, including how good the work
people put in, the better their work will be. high technology prod-
product is, the timeliness of the deliver-
If people tell you they are working able, the person’s ability to work with oth- uct development, which
long hours because they cannot accom- ers, and whatever else is important to you), helps managers, teams,
plish anything in their regular work weeks, not just the size or quality of the work. and organizations be-
ask them where they spend their time. come more effective. Rothman uses
Look for patterns such as multi-tasking, Summary pragmatic techniques for managing
or meetings that do not have any produc- Managers exist to help people do their
tive output. Use your management power people, projects, and risks to create suc-
best work to serve the business of the cessful teams and projects. A frequent
to discover and remove the obstacles pre- organization. Technical people can make
venting people from working a 40-hour speaker and author on managing high
great managers as long as they understand technology product development, she
week. people and want to succeed at working
with them. Many successful technical has written numerous articles and is
13. Admit Your Mistakes managers took the time to learn about now a columnist for Software
Sometimes, those obstacles to people management, putting as much effort (if Development, Computerworld.com,
completing their work successfully in 40 not more) than the effort they took to and StickyMinds.com. Rothman served
hours arise from your management mis- learn the necessary technical background as the program chair for the Software
takes. It is difficult, and sometimes embar- for the technical jobs. Managers do not Management conference and is the
rassing to have to admit you have made a have to be perfect; they have to be good author of “Hiring Technical People.”
mistake. In my experience, when I admit- enough to create a working environment
ted mistakes to my staff, they have for their employees to deliver great Rothman Consulting Group, Inc.
respected me more for it. work.◆ 38 Bonad Road
Arlington, MA 02476
14. Recognize and Reward Acknowledgements Phone: (781) 641-4046
Good Work I thank Dwayne Phillips and the Fax: (781) 641-2764
Money is not an adequate reward for CrossTalk reviewers for their input on
E-mail: jr@jrothman.com
many technical people. If people think this article.
tional time and funding. Including the aforementioned indicators is certainly not too difficult for the first
The outcome of this negative spiral is of project performance, the manager two items. When the project is perform-
that the company and the project manag- needs information for the following ing well, the manager would be wise to
er gain poor reputations. Additionally, if areas: not make any changes. In addition, when
the product is extremely important and its • Project Performance. Do the indica- the project has poor performance, but
sunk cost is significant with respect to the tors show poor project performance? has insufficient data, it is prudent to
amount needed for completion, the agi- • Sufficiency of Data. Is enough data investigate for potential causes and simply
tated customer will likely agree to the available to make a good decision? monitor the indicator(s) for improve-
added cost and delivery date extension. • Possible Strategy. Can a strategy be ment.
Under these circumstances, the company created to recover the project? The Adjust/Realign and Negotiate
cannot expect repeat business with this • Sufficient Time. Is there enough actions are not so simply connected to the
customer. time remaining to use the strategy? analysis results. The project manager
Another common earned value By doing the analysis, and then should negotiate additional cost and/or
approach is to manage using the cost vari- answering these questions, a project man- schedule, or reduction of requirements,
ance (CV) percentage, i.e., CV divided by ager can be confident the decision and only when a recovery strategy is not pos-
the planned cost (BAC). With this action taken will have a much higher sible or there is insufficient time for the
method, the project manager takes cor- probability of success. Before moving on, recovery to be effective. Adjustment, i.e.,
rective action upon breaching an arbitrary a few words are needed concerning raising or lowering overtime or number of
limit, e.g., plus or minus 10 percent. It is Sufficiency of Data. This information is project personnel, requires several inputs.
common practice to ignore the schedule critical in controlling management’s ten- It is the proper action when performance
information and manage the project by dency to overreact. It is common knowl- is poor, there is enough data to make an
cost variance alone. Generally, the results edge we should not react to insufficient informed decision, a recovery strategy is
from the CV management method are as data. However, sometimes the pressure to possible, and there is sufficient time to
poor as described for the EVM indexes. do something is overwhelming, and we execute it.
Certainly, there are successful projects act foolishly. Also, once a recovery strate- Careful realignment of personnel can
that have been managed using earned gy is implemented, we need to allow it yield increased efficiencies. However, the
value indicators; we are not implying time. It is not effective to amend and forecast effects of realignment cannot be
EVM has no merit. Using earned value as change strategies constantly; in fact, it is quantified easily. It is recommended that
a project management method greatly wasteful. this management action be used sparing-
increases the opportunity for success, but Supposing the questions can be ly. Realignment can be an effective strate-
improvement is needed. Project perform- answered, and a viable project recovery gy when the values of CPI-1 and SPI-1 are
ance data is readily available, but rarely is strategy can be prepared, what actions are less than their associated cost and sched-
it used advantageously. This is the state of possible? There are four basic actions: ule ratio, but worse than their planned
today’s management practice. • No Action Required when performance value (1.0). Figure 1, Decision Logic, illus-
is good. trates coupling the decision data to the
Analysis and Decision • Investigate when there is insufficient management actions. The graphical dia-
Is there an alternative? Yes, there is. Simply data. gram uses the logic symbols and, or, and
reacting to poor performance indicators • Adjust/Realign overtime or personnel. not 4. Once the inputs for Poor Performance,
(CPI, SPI, or CV) is not good practice. • Negotiate cost, schedule, or require- Sufficiency of Data, Possible Strategy, and
There are other considerations needed to ments. Sufficient Time are known, the logic dia-
make the management decision. Connecting the analysis to the action gram can be used to identify the recom-
mended management action.
Figure 1: Decision Logic When the cumulative value of either
CPI-1 or SPI-1 is greater than its respective
Poor CPI- 1(cum) > CR Yes
ratio, the project is performing poorly.
Performance + • Similarly, when there are more than seven
SPI- 1(cum) > SR Yes • Adjust
(Personnel, periods of performance data, there is suf-
Overtime) ficient basis for taking action5. A possible
Yes
Sufficiency m>7 strategy is one in which the forecast val-
No
of Data (Count m is from last recovery ues of CPI-1 and SPI-1 at project comple-
or re-plan, which impacted WBS tion are less than the cost and schedule
or Task values.)
• Investigate ratios, respectively.
Yes
-1
CPIs < CR Developing a possible recovery strate-
Possible No
gy is a trade-off; improving one index
Strategy Yes
SPIs < SR
-1
negatively impacts the other [1]. For
No + • Negotiate example, if the problem is poor cost per-
(Cost, Schedule,
Sufficient T_PI< 1.0
Yes
Requirements) formance, then the strategy, which causes
Time No its improvement, will detract from sched-
ule performance, and vice versa. It is also
[1- BCWP%] to be noted that the project will experi-
T_PI = Investigate
{ _PI s- 1 - _ PI- 1 (BCWP%)] No Action
• Required
ence an added expense to cost and sched-
(The equation and comparison Adjust ule to implement the change.
applies to the worse of CPI-1 Once the strategy has been deter-
Negotiate
and SPI-1.) mined, the To Complete Index (T_PI) is
used to evaluate whether or not there is made. We have a yes for Poor Notes
sufficient time for the recovery strategy to Performance; CPI-1 exceeds CR. 1. The definitions of the cost and sched-
be successful6. When T_PI is less than Sufficiency of Data is yes; the value of m ule performance indexes (CPI and
1.0, we are assured the strategy is viable. (11) is greater than seven. Yeses are evident SPI, respectively), and cost variance
In other words, the project will not have for the Possible Strategy; both CPIs-1 and (CV) are:
to perform better than planned to achieve SPIs-1 are less than their respective ratios. CPI = BCWP/ACWP
the customer commitments. In addition, Sufficient Time is yes; the SPI = BCWP/BCWS
When the recommended action is computed value for TCPI is less than 1.0. CV = BCWP — ACWP
either Adjust or Negotiate, management From the evaluation of the logical where,
must then determine, how much? For comparisons, the Decision Logic diagram ACWP = Actual Cost for Work
Adjust, the project manager computes is then used to identify the recommended Performed
how many people to add or subtract from management action. Investigate is not an BCWP = Budgeted Cost for
the project, or how much increase or appropriate management action because Work Performed
decrease in overtime is needed to accom- we have 11 months of data. We have also (earned value)
plish the recovery. For Negotiate, the determined the recovery strategy is possi- BCWS = Budgeted Cost for
manager determines the amount of over- ble and there is sufficient time to execute Work Scheduled
run in cost and schedule. Knowing these it. Therefore, Negotiate is not the action (project performance
values, he can then identify the require- to use. Adjust is the action the logic leads baseline)
ments, which can be completed within us to. Of course, with Adjust selected, No For more in-depth explanation of
the remaining time and funding, or the Action Required cannot be the recom- earned value and its indicators, see ref-
increases to schedule and cost needed to mended action. erence [2].
complete all of the requirements. Thus, For the Adjust action, the manager 2. The definitions of the cost and sched-
the project manager has the data with will perform calculations to determine ule ratios are as follows:
which a contract change may be negotiat- either a revised overtime or staffing level. Cost Ratio = (BAC + MR)/BAC
ed. If all that is needed is a change in over- Schedule Ratio = (POP+SR)/POP
The calculation methods needed for time, the success of the project recovery where,
Adjust, Negotiate, and Possible Strategy is more certain. Within reason, modifying BAC and MR are the EVM terms,
are beyond the scope of this paper. The the overtime level has much fewer reper- Budget at Completion and
reader may obtain the methods from [1]. cussions than does changing staffing. Management Reserve (cost
Lastly, when Adjust, Investigate, and reserve), respectively. POP is the
Negotiate are simultaneously inappropri- Summary period of performance and SR is
ate, the project requires no management EVM provides incredible management the schedule reserve, measured in
action, i.e., No Action Required. The information. However, it does not pro- units of time.
logic for this outcome is depicted in the vide a good connection between the indi- 3. Although SPI, as defined by EVM,
lower right corner of Figure 1. cator values and the possible manage- may be used, it is recommended to use
ment actions. In today’s project manage- the cumulative value of SPI(t). The
Example ment climate, action is more likely to be time definition of the schedule per-
To illustrate the use of the Decision taken because the project manager per- formance index is:
Logic diagram in Figure 1, I will use ceives it to be the correct thing to do in SPI(t) = ES/AT
hypothetical data. Let us suppose for this the eyes of the customer and his superi- where,
example the cost ratio (CR) equals 1.2, ors. AT is the actual period of time
and the schedule ratio (SR) is 1.3. The The Decision Logic diagram provides
from project start to present, and
reciprocals of the performance index val- the project manager with another tool.
ES is the resultant time associated
ues are 1.250 for CPI-1 and 1.125 for SPI- Using this tool, the method for deciding
1 with BCWS, when evaluated at the
, respectively. The project is 40 percent to act on a poorly performing project has
cost equivalent to the earned value
complete (BCWP/BAC = 0.4) with 11 been significantly refined. Furthermore,
(BCWP).
months of data. the action recommended is the one that
4. Reference Figure 1 for this discussion
If the project continues its present will most benefit the project. The project
performance (CPI-1 exceeds CR), it can- of the logic symbols. The and symbol
manager now has a tool he or she can use
not be completed within cost. However, is identified by the heavy dot. The
effectively for managing his or her proj-
the schedule performance provides some ect, and for reporting his or her actions at operation of and is all of the inputs
hope. Although schedule performance is the project reviews with both the cus- (lines from the left) must be yes for the
not as good as planned, the project is tomer and superiors. Using the decision output to be yes. The or symbol has the
expected to complete before the cus- diagram, the manager has supporting + sign. For the or operation, the out-
tomer’s delivery date (1.125 < 1.3). rationale for his or her actions.◆ put is yes if any of the inputs are yes.
Therefore, a possible strategy is comput- The not symbol is the triangle with a
ed that elongates the schedule and References circle at its point. Its operation is to
improves cost efficiency. The possible 1. Lipke, W. “Project Recovery … It Can change the input (line from the right)
strategy is determined to be SPIs-1 and Be Done.” CrossTalk Jan. 2002: from yes to an output of no, and vice
CPIs-1 equal to 1.256 and 1.140, respec- 26-29. versa.
tively. Using the CPIs-1 strategy value 2. Fleming, Q. Cost/Schedule Control 5. The criteria for data sufficiency is that
(1.140), TCPI is computed to be 0.9375. Systems Criteria, The Management we must have, at minimum, 50 percent
With all of the numerical information Guide to C/SCSC. Chicago: Probus, confidence of knowing the true values
known, the logical comparisons can be 1988. of the performance indexes, CPIt and
WEB SITES
Product Development and Management Integrated Software Industry Benchmarking
Association Association
www.pdma.org www.isiba.com
The Product Development and Management Association’s The Integrated Software Industry Benchmarking Association
(PDMA) mission is to improve the effectiveness of people (ISIBA) is a free association of software companies. ISIBA con-
engaged in developing and managing new products – both new ducts benchmarking studies to compare operating performance
manufactured goods and new services. This mission includes and identify practices that improve the overall operations of its
facilitating the generation of new information, helping convert members. Consortium studies are offered to the membership as
this information into knowledge that is in a usable format, and a whole with costs divided. Single-company sponsored studies
making this new knowledge broadly available to those who addressing the interest of one member company can be offered
might benefit from it. A basic tenet of the PDMA is that to other selected members for no fee. Interest group roundtables
enhanced product innovation represents a desirable and neces- are organized throughout the year.
sary economic goal for firms that wish to achieve and retain a
profitable competitive advantage in the long term. Project Management Institute
www.pmi.org
What You Need to Know About Management Established in 1969, the Project Management Institute (PMI) is
www.management.about.com a not-for-profit, project-management professional association
The About Web sites are a network with each site run by a pro- with over 100,000 members in 125 countries. PMI members are
fessional Guide who is carefully screened and trained by About. in many different industry areas, including aerospace, automo-
Guides build a comprehensive environment around each of their tive, business management, construction, engineering, financial
specific topics, including the best new content, relevant links, services, information technology, pharmaceuticals, and telecom-
how-to's, forums, and answers to just about any question. What munications. PMI publishes “A Guide to the Project Manage-
You Need to Know About Management includes general, peo- ment Body of Knowledge,” and its Project Management
ple, and project management; leadership; communications busi- Professional certification is the world’s most recognized profes-
ness ethics; conflict resolution; and more. The Guide for this site sional credential for individuals associated with project manage-
is F. John Reh, who has a 30-year management career. ment. In 1999, PMI became the first organization in the world
to have its Certification Program attain International
New Grange Center for Project Organization for Standardization 9001 recognition.
Management
www.newgrange.org Software Program Managers Network
The New Grange Center for Project Management is a nonprof- www.spmn.com
it, all-volunteer organization dedicated to the principle of build- The Software Program Managers Network (SPMN) is sponsored
ing a community of practice among project managers. Its goal is by the deputy under secretary of defense for Science and
to get to the heart of project management by defining what real- Technology, Software Intensive Systems Directorate. It seeks out
ly works and why. The backbone of the organization is its five- proven industry and government software best practices and
minute e-mail list where members take five minutes to address conveys them to managers of large-scale Department of Defense
what they learned from their latest problem, which over time software-intensive acquisition programs. The SPMN provides
develops into a database. Topics include how to develop a proj- consulting, on-site program assessments, project risk assess-
ect communication plan, how to define the best reward struc- ments, software tools, guidebooks, and specialized hands-on
ture, and the right way to conduct a project post-mortem review. training.
With respect to terminology, it should you are supposed to be managing maturity levels.
be noted that CMMI treats the standard these requirements. How can you The following is my proposal for the
RE components (management, elicitation, manage them at ML2 if you do not SEI CMMI Project Team:
analysis, specification, and validation) dif- have an institutionalized way of elicit- 1. Review the entire RE discipline (and
ferently from that usually found in RE lit- ing requirements until ML3? The ML2 not just the requirements-related goals
erature. For example, REQM is defined as REQM Specific Practice (SP) 1.1 and practices currently in the CMMI)
a separate PA, but requirements elicita- “Obtain an Understanding of with the goal of determining how RE
tion, analysis, specification, and validation Requirements” does not contain should be presented in the CMMI.
are all lumped into one RD PA. I have not enough detail about the scope, source, The review should include holding
found any SEI documentation describing and specificity of requirements to CMMI workshops to get consensus
the rationale of their taxonomy, as does form a solid basis for managing those from a broad spectrum of RE practi-
the SWEBOK [5]. Part of the answer may requirements at ML2. Requirements- tioners about what they consider to be
lie in the fact that the RD PA in the CMMI related problems are closely tied to basic versus advanced requirements
was split out of the Software Product project failure9. Why wait until ML3 to practices.
Engineering PA in the CMM. This differ- institutionalize practices to ensure that 2. Work closely with the IEEE to ensure
ence in terminology is more than academ- you have complete and accurate that their standards and work prod-
ic. By placing REQM and RD not only in requirements? ucts, e.g., SWEBOK and the CMMI
separate PAs but also in separate MLs, 2. Requirements analysis and validation stay in sync with respect to terminolo-
there is an artificial dichotomy created are defined under the ML3 RD PA gy and processes.
between the components of RE. As I (under SG 3). However, you need to 3. Revise the CMMI model to reflect
shall discuss later, REQM cannot be done do a certain amount of analysis and consensus from the above steps.
in a vacuum. validation of requirements at ML2 in I think consensus from this effort will
At this point, you may object that I am order to get them in a mature enough support the following concepts:
mixing apples and oranges. Requirements state to manage them. 1. RE maturity should be represented at
management, elicitation, analysis, specifi- 3. Bidirectional requirements traceability more than one ML. It is just not prac-
cation, and validation are categories or a is required under the ML2 REQM PA. tical to assume that an organization
taxonomy of RE activities, one may argue, While a certain amount of require- can and should implement everything
whereas the CMMI is concerned with ments traceability is necessary at ML2, related to RE at one level.
describing process areas relating to RE. should an organization concentrate 2. A RE-related PA should, at minimum,
However, these categories may also be on this full-blown bidirectional trace- exist at ML2 and ML3. Perhaps a case
viewed as activities in the RE process. ability before institutionalizing can be made for some advanced RE
According to Linda Macaulay, requirements elicitation and analysis activity at ML4 and ML5. However,
(at ML3)? I think not. It is interesting until that case is made, I believe the
In general terms, the RE process to note that Rational Software puts CMM and CMMI are correct in plac-
can be thought of as a series of traceability at Level 4 in their Five ing RE activity at ML2 and ML3.
activities consisting of articulating Levels of Requirements Management 3. The dichotomy between requirements
the initial concept, problem analy- Maturity [7]. management and other RE activities
sis, feasibility and choice of The CMMI recognizes that there is RE should be minimized.
options, analysis and modelling activity present even in ML1 organiza- Based on my IT experience, my rec-
[sic], and requirements documenta- tions10. Also, the CMMI acknowledges the ommendation (though I am willing to
tion. [6] interrelationship of RE activities in the change it based on consensus from the
Introductory Notes to the REQM PA: above proposal) is that the CMMI Staged
Requirements life cycles have been Representation should be changed to
defined as consisting of three to five phas- … if the Requirements Develop- something like a Basic RE PA at ML2 and
es with the above RE categories, or equiv- ment process area is implemented, an Advanced RE PA at ML3. The concept
alent terms, as phase names8. Although its processes will generate product of basic and advanced is not foreign to
the CMMI does not require you to choose and product-component require- the CMMI. For example, there are basic
any specific RE life cycle, it should use ments that will also be managed by and advanced project and process man-
standard RE terminology in describing the requirements management agement PAs [9].
PAs, goals, and practices relating to RE. processes. When the Requirements The following are my recommenda-
With respect to the typical order of Management, Requirements De- tions for some of the goals and practices
RE activity, I believe there is room for velopment, and Technical Solution at Basic RE PA (for ML2):
improvement in the CMMI. While the process areas are all implemented, 1. Elicit/gather requirements. You do
CMMI does not dictate any specific RE their associated processes may be not have to have a trained staff of
life cycle, it does have something to say closely tied and be performed con- facilitators and many different ways of
about the order of implementation and currently. [8] eliciting or gathering requirements at
institutionalization of RE by its placement ML2. You just need at least one
of a certain RE activity under a specific Therefore, the issue is not that the repeatable method of obtaining proj-
ML. I contend that this order is not always CMMI is opposed in principle to a normal ect requirements. Why wait until ML3
logical. Consider the following examples: progression and maturity of RE activity. to institutionalize one method?
1. Requirements elicitation is supposed The issue is whether the CMMI defines it 2. Analyze requirements. To ensure they
to be institutionalized in the ML3 RD the best way, i.e., using terms and maturi- meet the characteristics of good
PA under Specific Goal (SG) 1. ty criteria that the industry can agree requirements, e.g., complete, clear,
However, under the ML2 REQM PA, upon, and puts RE at the appropriate consistent, verifiable, traceable, feasi-
ble, and design independent. These effect at ML2, but they do not have to be industry best practices that certain
characteristics are currently defined as documented and can be informal. While I ML3 RE practices (e.g., elicitation and
examples in the ML2 REQM PA agree with the CMM improvement strate- analysis) must be done as part of their
under SP 1.1. However, why use the gy, it should not be interpreted in such a life cycle. Therefore, they continue to
ambiguous title “Obtain an way as to exclude activities required to do them because they make sense and
Understanding of Requirements” make work products mature enough to be are required to deliver quality work
when many ML 1 and 2 organizations managed at ML2. products.
know what you mean by requirements In other words, you cannot manage in In conclusion, I believe that RE matu-
elicitation and analysis? a vacuum. A certain level of formalization rity makes sense as a concept and reflects
3. Document requirements. This is must be in place for some engineering reality in IT organizations seeking opera-
already in the ML2 REQM PA as a practices in order for the management tional excellence, whether or not they call
typical work product (an agreed-to set process areas to work properly. Consider it basic versus advanced RE. The attempt
of requirements) under SP 1.1. the ML2 Project Planning PA. You need of the CMMI to define this RE maturity is
4. Get approval of requirements from to perform a certain amount of technical admirable but deficient. However, this
appropriate stakeholders. This is (engineering?) activities for SP 1.4 (per- deficiency does not mean that we abandon
already in the ML2 REQM PA as SP haps using some sophisticated tools) in the model. The CMMI is being widely
1.2 (Obtain Commitment to order to get sound estimates of effort and used, and I have personally witnessed the
Requirements). cost so that you can put together the proj- success of CMM at several companies. I
5. Manage requirements changes. This is ect plan in order to manage it. In like man- want the model to continue its success.
already in the ML2 REQM PA as SP ner, the ML2 REQM PA assumes a cer- However, for it to be durable for many
1.3. tain amount of RE formalization and years to come, I believe it needs an over-
6. Develop and maintain requirements institutionalization in order to ensure that haul in the RE area.◆
traceability to the extent that you can requirements are mature enough to be
demonstrate that all requirements have managed12. References
been implemented. See comment Also, it should be noted that ML3 has 1. Abran, Alain, and James W. Moore.
below on traceability at ML3. never been composed of pure engineering Guide to the Software Engineering
The following are my recommenda- PAs. For example, management activities Body of Knowledge – SWEBOK.
tions for some of the goals and practices permeate the Integrated Software Eds. Pierre Bourque and Robert
at Advanced RE PA (for ML3): Management and Intergroup Coordina- Dupuis. New York: Institute of
1. Develop different techniques of elicit- tion PAs in the CMM and several PAs in Electrical and Electronics Engineers,
ing requirements, define criteria about the CMMI, such as Integrated Project May 2001: 9 <www.swebok.org>.
when to use each based on project Management, Risk Management, Inte- 2. Zave, Pamela. “Classification of
profiles, and institutionalize these grated Teaming, Integrated Supplier Research Efforts in Requirements
techniques with formal training and Management, and Decision Analysis and Engineering.” ACM Computing
mentoring. Resolution. That is the way it should be. Surveys 29.4 (Dec. 1997): 315-321.
2. Provide a staff (more than one – even Each ML should be composed of the cor- 3. Davis, Alan M. “Requirements
if part-time) of trained requirements rect mixture of technical and management Bibliography.” <http://web.uccs.edu/
facilitators. activities so that management can be adavis/UCCS/reqbib.htm>.
3. Develop and maintain a full-blown, effective for that ML. 4. Abran 15ff.
bidirectional requirements traceability You may be asking, “If this RE matu- 5. Abran 23f.
matrix showing that each requirement rity discrepancy is that obvious in the 6. Macaulay, Linda. Requirements for
is satisfied in design, development, CMM/CMMI, why has it not been a prob- Requirements Engineering Tech-
test, and implemented work products. lem for organizations that have attained niques. Proc. of the Second
I have never seen a ML2 organization ML2 or ML3?” My answer is twofold: International Conference on
do a good job at this type of full- 1. Some ML1 organizations fund their Requirements Engineering. York,
blown traceability matrix. Yet it is process improvement effort with the United Kingdom, 1995. New York:
required in SP 1.4 of the ML2 REQM goal of achieving ML3. In other IEEE Computer Society Press, Apr.
PA. words, they are not first assessed at 1996: 158.
4. Include all current ML3 RD goals and ML2 and then work toward ML3. 7. Heumann, Jim. “The Five Levels of
practices that involve showing interre- Why? Because two separate efforts are Requirements Management Maturity.”
lationship of requirements, require- more expensive than one. Also, they The Rational Edge Feb. 2003 <www.
ments decomposition, assumed system may be under management pressure to therationaledge.com/content/feb_03/
requirements, and requirements achieve ML3 by a certain date, and f_managementMaturity_jh.jsp>.
change metrics. In other words, every- there is not enough time to do this in 8. CMMI Product Team. CMMI Ver. 1.1.
thing beyond the ML2 basics defined two independent efforts. Whatever the Pittsburgh, PA: Software Engineering
above. reason, by including both ML2 and Institute, Mar. 2002: 82.
Probably, some people may not want ML3 in one process improvement 9. CMMI Product Team Chapter 5.
to tamper with REQM at ML2. They effort, all of RE goals and practices
believe this PA simply follows the overall are covered. Therefore, it never Notes
CMM process improvement road map to becomes an issue about how RE is 1. See [7] for Rational Software’s Five
get management infrastructure in place at split out between ML2 and ML3. Levels of Requirements Management
ML2 in order to support the engineering 2. For those ML1 companies who work Maturity. Some articles describe a
processes at higher levels11. They make the toward ML2 as their goal, they just Requirements Maturity Index (RMI),
point that engineering processes are in know from past experiences and but this has to do with the readiness of
Acquisition Deployment: Moving Technology Into the Operational Air Force Lt. Col. S. B. Dufaud (Ret.), 5 31
Dr. L. R. Carter
Lessons Learned From Another Failed Software Contract Dr. R. W. Jensen 9 25
Configuration Management But I Only Changed One Line of Code T. R. Leishman, Dr. D. A. Cook 1 20
COTS Decision Point: Will Using a COTS Component Help or Hinder T. J. Budden 11 18
Your DO-178B Certification Effort?
Improving Processes for Commercial Off-the-Shelf-Based Systems Dr. B. Tyson, C. Albert, 5 17
L. Brownsword
Information Security Defining a Process for Simulation Software Vulnerability Dr. J. A. Hamilton Jr., 11 22
Assessments Col. K. J. Greaney, G. Evans
Securing Your Organization's Information Assets Dr. B. Brykczynski, B. Small 5 12
Steganography 2nd Lt. J. Caldwell 6 25
Information Sharing Data Warehouse: Your Gateway to the Information Age K. L. Smith 10 29
Effective Collaboration: People Augmented by Technolog y R. L. Conn 10 12
An Information Architecture Strategy J. Wunder, Dr. W. Tracz 10 4
Information Assurance Post 9-11: Enabling Homeland Security D. W. Carey 5 8
Improving Information Management Software System Deployment Dr. J. A. Forbes, Maj. K. Bodiford, 6 17
Practices Dr. E. R. Baker
Prospecting for Knowledge J. Kelley 4 24
Serialized Maintenance Data Collection Using DRILS Capt. G. Lindsey, K. Berk 10 16
Warfighter's Access to Geospatial Intelligence P. Winter 10 8
Miscellaneous Airport Simulations Using Distributed Computational Resources W. J. McDermott, Dr. D. A. Maluf, 6 7
Y. Gawdiak, P.B. Tran
Introducing Global Software Competitiveness D. O'Neill 10 29
Pilot Testing Innovative Auto ID Technologies J. E. Bagley 6 21
SAASM and Direct P (Y) Signal Acquisition S. Callaghan, H. Fruehauf 6 12
Trafficability Analysis Engine Dr. K. R. Slocum, Lt. Col. J. R. Surdu, 6 28
2nd Lt. J. Sullivan, 2nd Lt. M. Rudak,
2nd Lt. N. Colvin, Cadet C. Gates
Upgrading Global Air Traffic Management E. Starrett 6 4
Network-Centric Architecture Designing Highly Available Web-Based Software Systems M. Acton 8 4
Enterprise Engineering: U.S. Air Force Combat Support Integration E. Z. Maass 8 16
A Fire Control Architecture for Future Combat Systems Dr. M. Morrison, Dr. J. Sherrill, 8 9
R. O'Guin, D. A. Butler
Technical Reference Model for Network-Centric Operations B. C. Logan 8 21
Requirements An Enterprise Modeling Framework for Complex Software Systems Dr. P. Donzelli 2 23
Requirements Engineering Maturity in the CMMI D. Linscomb 12 25
Risk Management Risk Management Applied to the Reengineering of a Weapon C. Y. Laporte, G. Boucher 1 24
System
Software Development Application of Lightweight Formal Methods in Requirements V. George, Dr. R. Vaughn 1 30
Engineering
Clarify the Mission: A Necessary Addition to the Joint Technical I. Ögren 3 25
Architecture
International Standardization in Software and Systems Engineering F. Coallier 2 18
A Pair Programming Experience Dr. R. W. Jensen 3 22
Software Architecture as a Combination of Patterns K. Petersson, T. Persson, 10 25
Dr. B. I. Sanden
Wireless Data Entry Device for Forward Observers P. Manz, Lt. Col. J. R. Surdu, 7 31
2nd Lt. A. M. Adas, 2nd Lt. Z. R. Miller,
2nd Lt. A. J. Peplinski,
2nd Lt. E. J. Watson
Software Inspections Determining Return on Investment Using Software Inspections D. O'Neill 3 16
High Quality, Low Cost Software Inspections L. A. Poulin 1 29
Systems Engineering Developing a Stable Architecture to Interface Aircraft to D. W. Christenson, L. Silver 11 26
Commercial PC's
The CrossTalk staff would like to wish you and yours the very best
this holiday season and the happiest of New Years.
30 CROSSTALK The Journal of Defense Software Engineering December 2003
BACKTALK
Misbehaving Toys
I t’s toy time folks, and as the song goes,
“You’d better watch out!” The bleeding
thumbs and tons of assembly required so popu-
Anyway, Mike was stashed in the master
closet, secreted away for Christmas morn-
ing.
What manufactures don’t do, the enter-
prising engineer can fix up at home. My
friend, let’s call him Bruce, felt such a call.
lar on past Christmas eves have been updat- Fates crossed when Grandma came to His mission was to fix the professional mal-
ed, so here’s fair warning. visit and at one point excused herself to the practice of the Big Wheel people. His son
You might say I have issues regarding bedroom to change. At some point of needed one of those plastic three-wheel
my kids’ toys, and it all started in the middle undress she heard a strangely familiar man’s bikes, so Bruce scoured the toy stores but all
of the night long after one Christmas revel- voice in the closet greeting her with, “I’ve he found were bikes with rotating wheels, no
ry. I awoke hearing voices in the dead of got my eye on you.” gadgets, lights, or sound.
night, a conversation going on in my living We understand Grandma did not require So Bruce went to work. Long nights and
room. Naturally, I panicked and was soon medical attention – even after hurdling a endless requirements changes caused the
wide-awake. My intruder turned out to be queen-sized bed – but her opinion of Billy usual challenges, and then there was the
Cookie Monster, and he was being rudely Crystal did go down a notch. I suspect Mike escalating budget. But after the holidays,
interrupted by Elmo. has already made a pass through the Bruce reported a successful on-time
Cookie would say, “Cookie Monster Goodwill cycle. Christmas delivery. A dream bike it was.
here, Cookie see you.” Somewhere in there, Of course, many of the toys under the This is what tricycles were meant to be.
Elmo’s squeaky voice would cut him off tree these days not only talk, but they also LCD readouts, turn signals (with multiple
with several rounds of “Let’s play.” The listen and respond to our commands – lights), speedometer, public address system,
fight would go on for a while, then stop, and rather like real pets. I’m very impressed with siren, and sound synthesizer – fortunately
then start up again hours later. I would walk two such creatures my girls have on their for the neighbors, lasers had yet to come
into the living room and from the direction shelves; they respond exactly like our down in price.
of the toy basket hear a gravelly, “Hey, Schnauzer – which is to ignore all of my Bruce went on to produce another ver-
scram,” followed by Oscar’s scariest laugh. commands. Oh, I have managed to get them sion of the bike, taking advantage of tech-
While I admit this feature could actually to beep and jump around a bit and flash nology advances and of course, lessons
learned. He offered his innovation to the
prove useful for clearing the stoop, we could their LED eyes, which shouldn’t surprise
manufacturer. They responded politely say-
never count on it, except to speak up when me, since the Schnauzer raises a ruckus and
ing it was too complicated. Funny thing
it felt like it. I last heard from the gang as I jumps like crazy without any commands
though, I came home recently and found a
passed a sack in the garage bound for the from me.
new tricycle in the garage. As I picked it up
local Goodwill, destined for another soon- I could go on. There was the dollhouse to move it, a stern woman’s voice said, “This
to-be-sleepless house. that came with several realistic sounds for is mommy. Don’t go near the street!” This
It’s funny how these talking toys seem to your typical household, including a crying was followed by the sound of a neighing
go off at just the right or wrong moments. baby that, yes, went off in the middle of the horse. The tricycle was well into a cute little
Consider this recent report: an acquaintance night. Then there was that cool toy bank that jingle about butterflies by the time I closed
of my wife bought a talking one-eyed Mike did all sorts of things, even more than was the door.
of “Monsters Inc.” fame. You no doubt advertised on the box. I came downstairs one
remember the character from the movie; morning to find my young daughter asleep — Tony Henderson
he’s essentially a green ball with one large on the couch. I found the bank, buried in Software Technology Support Center
eye and sounds a lot like Billy Crystal. blankets, beeping non-stop in her room.
MONTHLY COLUMNS:
ISSUE COLUMN TITLE AUTHOR
Issue 1: January Publisher: Best Training Includes Going Back to Basics Lt. Col. G. A. Palmer
Back to Basics BackTalk: Pandora and the Magic Vase R. Jensen
Issue 2: February Publisher: We've Come a Long Way From Machine Code to Current Programming Languages H. B. Allgood
Programming Languages BackTalk: The First Book of EPP G. Petersen
Issue 3: March Publisher: We Need the Right Tools for Quality Software E. Starrett
The Case of Quality Software BackTalk: Did I Say "Koala Tea?" D. A. Cook
Issue 4: April Publisher: Engineers at Their Best T. L. Stauder
The People Variable BackTalk: A More Perfect Union G. Petersen
Issue 5: May Publisher: Top 5 Contest Nominations Reveal Trends in COTS, E-Commerce, and Web Services Lt. Col. G. A. Palmer
Strategies and Technologies BackTalk: Everybody Knows It's True D. A. Cook
Issue 6: June Publisher: The Knowledge Flows Both Ways H. B. Allgood
Commercial and Military Applications Meet BackTalk: Shock and Awe G. Petersen
Issue 7: July Publisher: Top 5 Winners' Technologies Aim to Support the Warfighter: Several Used in Operation Iraqi Freedom J. Jarzombek
Top 5 BackTalk: Suggestion for "Bottom 5" Projects D. A. Cook
Issue 8: August Publisher: Network-Centric Warfare Brings Increased Combat Power Lt. Col. G. A. Palmer
Network-Centric Architecture BackTalk: Softbucks G. Petersen
Issue 9: September Publisher: Managing Defects Together T. L. Stauder
Defect Management BackTalk: Defect Mismanagement D. A. Cook
Issue 10: October Publisher: Developers Meet a Variety of Complex Information and Data Sharing Needs H. B. Allgood
Information Sharing BackTalk: CrossTalk Terminology Invitational G. Petersen
Issue 11: November Publisher: Real-Time Software Development Requires Rigid Constraints E. Starrett
Development of Real-Time Software BackTalk: Real Time - Military Style D. Ludwig
Issue 12: December Publisher: Management Basics: A Necessary Foundation T. L. Stauder
Management Basics BackTalk: Misbehaving Toys T. Henderson
Published by the
Software Technology
Support Center (STSC)