Volume 32, Number 1, April 2013 Decision Analysis Today
Decision Analysis Today
Vol. 32, No. 1, April 2013
The newsle er of the INFORMS Decision Analysis Society
Inside: President’s Letter------1 Call for Session Chairs------11 Research------25 Letter from the Editors------2 Professional News------11 Ask DAS------29 Upcoming Conferences------3 Decision Analysis Journal------14 Editorial Team------32 Professional Postings------4 DA Around the World------17 DAS Officers------33 Call for Submissions and DA Practice------20 Nominations------4 Call for Chapters------9 Society of Decision Professionals---23
_
President’s Letter Jeffrey Keisler
A lot people are putting a lot of effort into building DAS. Here are some updates, in no particular order: Florian Federspiel, Chris Karvetski, Ahren Lacy and Jeff Stonebraker are exploring ways to help DAS better serve its student members, starting with some activities at the fall INFORMS conference. Speaking of the fall conference, cluster co-chairs Jag Chattwal, Victor Jose and Alec Morton are working hard and tell me another strong lineup is coming together. In addition to formal submissions, you should let them know if you have any other ideas or interest regarding DAS social interactions at the conference. Speaking of conferences, Jason Merrick has put together a working group (Robin Dillon-Merrill, Philippe Delquié, Yael Grushka-Cockayne, Victor Jose, and Casey Lichtendahl) and a steering committee (including current and former editors of the DA outlets within INFORMS: Robin Keller, Kevin McCardle, Rakesh Sarin, and Jim Smith) setting up the first standalone DAS research conference, which will tentatively be around June 2014 and will be of modest size. This is an exciting and large development for our field, and details will be coming. Nadia Papamichal and Johannes Siebert are undertaking an initiative to build bridges between DAS and other decision making organizations, including the MCDM and EURO, and we will be discussing ways to do that over the summer. Vicki Bier is leading an effort to build our capabilities for getting publicity for our work, our field, and our society.
Page 1 Volume 32, Number 1, April 2013 Decision Analysis Today
Our web and social media presence, led by Jay Simon, is seeing healthy activity in LinkedIn, Facebook and our website. We will have more work to do in the fairly near future, when INFORMS launches its Golightly system. Eric Bickel reports that this system will give us a great deal of options to do many cool things fairly easily – with the help of DAS members as yet to be named – but INFORMS is taking a little longer than originally planned. Rakesh Sarin has taken charge of the DA Journal and is looking beyond the usual suspects to reach new potential contributors. Our relationships with the Society for Decision Professionals and the DAAG conference are strengthening, in large part thanks to Eric Bickel who chaired the DAAG conference last week in Austin. SDP leadership (e.g., SDP President Frank Koch, SDP President-Elect Jim Felli and SDP past-President Carl Spetzler, as well as Larry Neal who has been serving both SDP and DAS) is likewise interested in strengthening our collaborative activities. Patrick Noonan has started exploring some ideas for developing DA educational materials, and along these lines, Chris Spetzler from the Decision Education Foundation has an interest in hearing from DAS members. DAS award committees are getting to work, and you will be seeing more from Robin Dillon-Merrill (publication award), Jim Dyer (Ramsey medal), Frank Koch (practice award), Lea Deleris & Jun Zhuang (student paper award) and Jason Merrick (DA Journal award). VP/President-Elect Eric Bickel will be forming a nominations committee for open DAS positions (election information will be in the summer issue). The DAS council just had a long strategy phone call, and I hope to keep you in the loop as things go ahead. One topic we may start to think about in the future is ways of bringing in revenue beyond our conference activities, as there are probably some worthwhile things we could do if we had funds to do them. As usual, Secretary/Treasurer Yael Grushka-Cockayne is there behind anything that is going smoothly in DAS. Also as usual, Heather Rosoff, Jun Zhuang and Elizabeth Newell are making DA Today better and better. This long list of people and the ways they are contributing to DAS fills my heart with joy and bodes well for the future. I hope I haven’t left anyone off, but I surely have. Thanks to all of you listed above, and thanks to everyone who is making contributions of all kinds. If you are interested in getting involved with any of these initiatives, please contact any of the people mentioned above. If you want to get involved in other ways, please get in touch with me.
Letter from the Editors Heather Rosoff, Jun Zhuang, and Elizabeth Newell
Hello loyal followers!
As Jeff mentioned, many exciting things are happening within the DA community! In this issue, we highlight some of the amazing contributions being made by your peers. In DA Around the World, Alec Morton and Matthias Seifert provide us with a nice overview of the DAS Community’s reach around the globe. In DA Practice, Bill Klimack brings us an interesting contribution from Phil Beccue who writes about the benefits and insights gained from employing decision analytic techniques in the pharmaceutical sector. In SDP, Frank Koch writes about the state of the society and its tremendous growth and
Page 2 Volume 32, Number 1, April 2013 Decision Analysis Today
development. In Research, Debarun Bhattacharjya brings us an article by Lav Varshney about the use of DA techniques to enhance computational creativity. Finally in Ask DAS, John Coles and Florian Federspiel bring us ―a quick guide to effective and successful poster presentations.‖
It is also that time of year to start thinking about whether you would like to be a Session Chair at next year’s INFORMS conference and if you have nominees for the Ramsey Medal, DAS Publication, Practice and/or Student Awards. More details on the sessions and awards for the next INFORMS conference can be found starting on page 4.
We wish everyone a very productive and progressive spring/summer!!
Until next time,
Heather, Jun, and Elizabeth
Upcoming Conferences
June 6, 2013 – June 7, 2013 July 22, 2013 – July 25, 2013 INFORMS Revenue Management and Pricing MIP2013, University of Wisconsin-Madison Conference, Atlanta, GA Madison, WI http://www.informs.org/Community/revenue-mgt August 14, 2013 – August 16, 2013
June 16, 2013 – June 19, 2013 MOPTA 2013 INFORMS Transportation Science and Logistics Lehigh University Society Annual Workshop, Asilomar Conference Bethlehem, PA Grounds, California http://coral.ie.lehigh.edu/~mopta/ www.informs.org/Community/TSL/TSL-Workshop August 22, 2013 – August 25, 2013 June 23, 2013 – June 26, 2013 Operations Research for Surgical Services INFORMS Health Care 2013 Hampton Inn Chicago, IL Iowa City, IA http://meetings.informs.org/healthcare2013/ http://www.franklindexter.net/PDF%20Files/Surgical ServicesCourse.pdf June 23, 2013 – June 26, 2013 First conference on Validating Models of Adversary October 6, 2013 – October 9, 2013 Behavior, Buffalo/Niagara Falls, NY INFORMS Annual Meeting 2013 Minneapolis http://www.eng.buffalo.edu/~jzhuang/Conference13/ Minneapolis Convention Center and Hilton Minneapolis, Minneapolis, MN July1, 2013 – July 4, 2013 http://meetings2.informs.org/minneapolis2013/ EURO – INFORMS Joint International Conference June 16, 2014 – June 18, 2014 Rome, Italy Advances in Decision Analysis Conference http://euro2013.org/ Georgetown University Washington, D.C. For additional information: Jason Merrick ([email protected]) or Yael Grushka-Cockayne ([email protected])
Page 3 Volume 32, Number 1, April 2013 Decision Analysis Today
Professional Postings
JOB PLACEMENT SERVICES: Visit http://jps.informs.org/ for the INFORMS Job Placement Service (JPS). For many years INFORMS (and its founding societies) has offered a Job Placement Service to connect employers searching for O.R. professionals and qualified O.R. professionals looking for employment.
The Job Placement Service database offers: • Online access to job listings and applicant files • Expanded information about jobs and applicants • Updates of the database • Improved database search capabilities • Online data entry for applicants and employers • Extended availability of the database
INFORMS MEMBERSHIP: We are writing to relay some important and exciting news regarding INFORMS membership.
During 2013, INFORMS is offering a FREE membership in any society for new regular members! Anyone that joins INFORMS during 2013 can choose to join the Decision Analysis Society (DAS) for free. This membership would accrue all DAS benefits, including a free subscription to the Decision Analysis Journal. Note that this offer does not apply to new student members, who already receive a free society membership (without the DA Journal).
Here is a link you can forward to colleagues that you believe would benefit from this offer: http://www.informs.org/Membership/Join-INFORMS-and-or-INFORMS-Communities
Regards,
Eric Bickel (President-Elect), Vicki Bier (Past-President) and Jeff Keisler (President) Call for Submissions and Nominations
2013 Frank P Ramsey Medal Application Process Submission deadline: August 1, 2013
The Frank P. Ramsey Medal is the highest award of the DAS. It was created to recognize distinguished contributions to the field of decision analysis. The medal is named in honor of Frank Plumpton Ramsey, a Cambridge University mathematician who was one of the pioneers of decision theory in the 20th century. His 1926 essay "Truth and Probability" (published posthumously in 1931) anticipated many of the developments in mathematical decision theory later made by John von Neumann and Oskar Morgenstern, Leonard J. Savage, and others. The Ramsey Medalists are recognized for having made substantial further
Page 4 Volume 32, Number 1, April 2013 Decision Analysis Today
contributions to that theory and its application to important classes of real decision problems. The Medal is accompanied by a $1,000 honorarium.
Distinguished contributions can be internal, such as theoretical or procedural advances in decision analysis, or external, such as developing or spreading decision analysis in new fields. Thus the specific criteria for evaluating potential Ramsey Medal recipients are a candidate's
Theoretical, methodological, and procedural contributions to decision analysis Applications of decision analysis (including for new uses and into new fields) Other contributions promoting decision analysis (e.g. educational and public relations) Exceptional contributions to the Decision Analysis Society (e.g. service to society or journal)
A potential recipient need not meet all of the criteria, but contributions to each criterion are relevant.
For this award, decision analysis is defined as a prescriptive approach to provide insight for decision making based on axioms that are logically consistent with the axioms of von Neumann and Morgenstern and of Savage. Key constructs of decision analysis are utility to quantify values and probability to quantify the state of one's knowledge. There are overlapping aspects of decision analysis with other fields such as behavioral decision research, probabilistic risk analysis, and engineering and economic analyses. Behavioral decision research that addresses how people make decisions that has direct implications for how the practice of decision analysis might be improved is a contribution to decision analysis. Models of uncertain scenarios and possible consequences from risk analysis or engineering and economic modeling that are useful for decision analysis are contributions.
Please send nominations for the 2013 award to the Chair of the Frank P. Ramsey Medal awards committee by e-mail. The deadline for nominations is August 1, 2013.
2013 Committee Chair Professor Jim Dyer Energy Management and Innovation Center Information, Risk, and Operations Mgmt. [email protected]
DAS Publication Award Announcement Submission deadline: May 31, 2013
Please send Robin Dillon-Merrill (([email protected]) your nominations by Friday, 31 May 2013. Self nominations are always welcome, but if you read a paper published in 2011 that you really enjoyed or found particularly useful in your decision analysis practice or research, please let us know.
The Decision Analysis Publication Award is given annually to the best decision analysis journal article or book published in the second preceding calendar year. For consideration for this year's award, a work should have been published during CALENDAR YEAR 2011. The Decision Analysis Society of INFORMS award is accompanied by a plaque and a $750 honorarium.
The intent of the award is to recognize the best publication in ―decision analysis, broadly defined.‖ This
Page 5 Volume 32, Number 1, April 2013 Decision Analysis Today
includes, but is not limited to, theoretical work on decision analysis methodology (including behavioral decision making and non-expected utility theory), descriptions of applications, and experimental studies.
Please send the author's name(s) and the full journal citation or book title. Nominators should ensure that the Publication Award Committee has a copy of the publication, preferably in electronic form.
SELF-NOMINATIONS ARE ACCEPTABLE AND ARE RECOMMENDED. Historically, most nominations for this award have been self-nominations, so don't rely on your admiring colleagues to nominate your work. However, others who wish to write in support of a publication (in a substantive way regarding impact of the work) are encouraged to do so. Testimonials by those who have benefited from a work will be very helpful to our decision process.
Nominated publications will be judged for significance, relevance, originality, and readability. The award will be presented at the INFORMS Annual Meeting in Minneapolis on October 6-9, 2013. This award is sponsored by the Decision Analysis Society of INFORMS. Membership in the Decision Analysis Society is not a condition for being a nominator or a nominee, so please feel free to forward this announcement to other colleagues.
Names of past winners of the Decision Analysis Publication Award are posted on the DAS Awards web page.
Please send nominations no later than 31 May 2013 to:
Robin Dillon-Merrill McDonough School of Business Georgetown University ([email protected])
Announcing the 2013 Decision Analysis Society Practice Award Competition Submission deadline: May 31, 2013
The DAS Practice Award is given annually to the best decision analysis application, as judged by a panel of Society members. The Decision Analysis Society invites member submissions for the 2013 Decision Analysis Practice Award offered by the INFORMS Decision Analysis Society. The submissions should use the attached format and be submitted by 31 May 2013. The summary should be about two pages long. Once you submit your work, (1) The Practice Award committee will evaluate all submissions in the summer and name three finalists. (2) At the 2013 INFORMS annual meeting on October 6-9, 2013, Minneapolis Convention Center, Minneapolis, MN, there will be a special DAS sponsored Practice Award session in which the three finalists present their results to the award committee and interested attendees at the conference. (3) The Practice Award committee will then determine the winner of the practice prize using six criteria: importance of the problem, impact on the client's decision making, benefits to the client organization, use of decision analysis tools, quality of the analysis, and originality.
Page 6 Volume 32, Number 1, April 2013 Decision Analysis Today
The prize is $750, a plaque from the Decision Analysis Society and publicity (consistent with your proprietary restrictions). Submissions based on previously published work are also eligible for the prize. DAS invites you to submit your best decision analysis work to the competition. Submitters are responsible for obtaining any company or agency clearances necessary to allow presentation of the work. Aspects of the application or results may be disguised as necessary to preserve confidentiality, provided that the material that is disclosed is sufficient to allow evaluation of the work. We encourage co-authored submissions where appropriate. For more information on the prize and a list of past winners, see http://www.informs.org/Recognize-Excellence/Community-Prizes-and-Awards/Decision-Analysis- Society/DAS-Practice-Award For more information on the conference at which the practice competition will be conducted, see http://meetings2.informs.org/minneapolis2013/ Please send submissions and any questions to Frank Koch, the Chair of the 2013 Decision Analysis Practice Award Committee, at [email protected].
2013 INFORMS Decision Analysis Society (DAS) Student Paper Award Submission deadline: June 01, 2013
The Student Paper Award is given annually to the best decision analysis paper by a student author, as judged by a panel of the Decision Analysis Society of INFORMS. Students who did not complete their Ph.D. prior to May 1, 2012 are eligible for this year's competition.
The award is accompanied by a plaque and a $500 honorarium. The award will be presented and the winner will also be invited to present his or her paper at the DAS Awards Session at the INFORMS Annual Meeting to be held in Minneapolis, Minnesota, October 6-9, 2013.
All students doing work in or related to decision analysis (e.g., decision methodologies, experimental studies, and applications) are encouraged to submit a paper. The majority of work, including writing, must be that of the student, though faculty members or other mentors can be co-authors if appropriate. The paper should be 30 pages or less (double spaced and 12 point font) and, in the standard format of Management Science or Operations Research.
If you are a faculty member who is supervising students, please inform them of this opportunity. If you are a student reading this, please encourage your classmates (and yourself) to submit a paper and to join the Decision Analysis Society (http://www.informs.org/Community/DAS). While we encourage all applicants to join DAS, it is not necessary for students to be members in order to be eligible for the competition.
To be considered for this year's competition, please email both committee co-chairs, at the address given below, by the deadline, June 01, 2013, with your final submission of:
Page 7 Volume 32, Number 1, April 2013 Decision Analysis Today
(i) An electronic version of your paper in PDF format, and (ii) A letter in PDF format from one faculty co-author (if any) articulating your role in writing this paper
Let us know if you have any questions.
Sincerely,
2013 INFORMS Decision Analysis Society (DAS) Student Paper Award Co-Chairs
Léa Deleris Research Staff Member and Manager IBM Technology Center, Dublin, Ireland Email: [email protected]
Jun Zhuang Assistant Professor in Industrial and Systems Engineering University at Buffalo, State University of New York Email: [email protected]
Call for Participation and Poster Submission: First Conference on Validating Models of Adversary Behavior
Date: June 23-26, 2013 Location: Buffalo/Niagara Falls, NY, USA Web site and registration: http://www.eng.buffalo.edu/~jzhuang/Conference13/ Hotel Cutoff Deadline: May 23, 2013 Early-bird Registration Deadline: May 23, 2013
Hundreds of billions of dollars have been spent on homeland security since September 11, 2001, and numerous models have been developed to study the strategic interactions be tween defenders and adversaries (e.g., attackers or terrorists). Unfortunately, few if any models have yet been validated using empirical data, limiting the application of those models in practice.
Supported by the U. S. Department of Homeland Security (DHS) through the National Consortium for the Study of Terrorism and Responses to Terrorism (START) and the National Center for Risk and Economic Analysis of Terrorism Events (CREATE), Drs. Jun Zhuang and Vicki Bier are organizing a conference on validating models of adversary behavior in Buffalo/Niagara Falls, NY, in June 23-26, 2013. This conference is intended to bridge theoretical and empirical research on adversarial modeling, and facilitate transitioning of the best existing models of adversary behavior into practice by assessing and demonstrating their validity and applicability to real-world problems. A secondary goal of the conference is to encourage synergy and communication between risk analysts, statisticians, economists, and other social scientists engaged in terrorism modeling and research.
Our confirmed speakers include:
Page 8 Volume 32, Number 1, April 2013 Decision Analysis Today
*Morral Andrew (RAND Corporation) *Daniel Arce (University of Texas-Dallas) *David Banks (Duke University) *Vicki Bier (University of Wisconsin-Madison) *Seth Guikema (John Hopkins University) *Kjell Hausken (University of Stavanger, Norway) *Richard John (University of Southern California/CREATE) *Victor Jose (Georgetown University) *Jason Merrick (Virginia Commonwealth University) *Hamid Mohtadi (University of Wisconsin-Milwaukee) *Nageswara Rao (Oak Ridge National Laboratory) *Laura Razzolini (Virginia Commonwealth University) *Bob Ross (Department of Homeland Security, retired) *John Sawyer (University of Maryland/START). *Steve Streetman (Data Architecture Solutions, Inc) *Milind Tambe (University of Southern California/CREATE) *Chen Wang (University of Wisconsin-Madison) *Jun Zhuang (University at Buffalo)
The poster presentation and competition will be held on June 24 during dinner buffet reception. Student poster presentation winner will be awarded a certificate, as well as a $500 gift card from a brand at his/her choice. Let us know if you have any questions.
Sincerely,
Conference chairs:
Jun Zhuang Department of Industrial and Systems Engineering, University at Buffalo, SUNY
Vicki Bier Department of Industrial and Systems Engineering, University of Wisconsin-Madison
Call for Chapters
Call for Chapters: Springer book on "Game Theoretic Analysis of Congestion, Safety and Security"
We are editing a book for Springer with a tentative title ―Game Theoretic Analysis of Congestion, Safety and Security.‖ Many researchers working on congestion have not extensively considered safety/security, and vice versa. However, significant interactions exist between the two research areas, which motivated this book. We envision the book to establish a new and enhanced current state of affairs within this topic, illustrate linkages between research approaches, and lay the foundation for subsequent research. Congestion (excessive crowding) is defined broadly to include all kinds of flows; e.g., road/sea/air traffic,
Page 9 Volume 32, Number 1, April 2013 Decision Analysis Today
people, data, information, water, electricity, and organisms. We consider a system where congestion occurs, or a system which may be in parallel, series, interlinked, or interdependent, with flows one way or both ways. Congestion models exist in abundance. The book makes ground by introducing game theory and safety/security. For the analysis to be game theoretic, at least two players must be present. For example, in Wang and Zhuang (2011) one approver and a population of normal and adversary travelers are considered. Similarly, in Bier and Hausken (2013), one defender and one attacker are considered, in addition to drivers who choose the more time-efficient of two arcs of different lengths. Multiple players can be adversaries with different concerns regarding system reliability; e.g., one or several terrorists, a government, various local or regional government agencies, companies, or others with stakes for or against system reliability. Governments, companies and authorities may have tools to handle congestion, as well as ensure safety/security against various threats. The players may have a variety of individual concerns which may or may not be consistent with system safety or security. Much of the congestion literature is not game-theoretic, and does not extensively consider safety or security. Also, most game- theoretic analysis do not account for congestion. If you would like to contribute a chapter to this book, please let us know as soon as possible to the emails below. The chapter should be 12pt, double spaced, 30 pages or less. You are also welcome to forward this invitation to your colleagues who might be interested in this, or you may suggest potential authors to us. All the book chapters will be double-blinded peer- reviewed according to conventional standards.
The time schedule is as follows:
June 15, 2013: Title, authors, and 200 words abstract due June 30, 2013 or earlier: Decision on abstracts December 1, 2013: Full paper due February 15, 2014 or earlier: Distribution of referee reports April 30, 2014: Revised paper due (if needed) May 31, 2014: Delivery of manuscript to Springer
Please feel free to let us know if you have any questions or comments.
Kjell Hausken Faculty of Sciences, University of Stavanger [email protected]
Jun Zhuang Department of Industrial and Systems Engineering, University at Buffalo, SUNY [email protected]
References - Bier, V. and Hausken, K. (2013). Defending and Attacking Networks Subject to Traffic Congestion. Reliability Engineering & System Safety 112, 214-224. - Wang, X. and Zhuang, J. (2011). Balancing Congestion and Security in the Presence of Strategic Applicants with Private Information," European Journal of Operational Research 212, 1, 100-111.
Page 10 Volume 32, Number 1, April 2013 Decision Analysis Today
Call for Session Chairs
We have received a number of volunteers to chair sessions for the Decision Analysis cluster at this year's INFORMS meeting (http://meetings.informs.org/minneapolis2013/index.html), which is to be held in Minneapolis from 6th to 9th October 2013, and it looks like we will have a great meeting and a great cluster! Thanks to those of you who have volunteered!
For those of you who have not yet volunteered, we would like to invite and encourage you to consider chairing a session. Sessions comprise three or four papers linked by a common theme, and session chairs are responsible for finding speakers, persuading them to submit titles and abstracts in a timely fashion, and chairing the session on the day. Being a session chair is minimal work and is a great way to contribute to the Society and to INFORMS, to meet people, and to ensure that the meeting provides a forum for discussion on topics interesting to you.
Please consider organizing a session and contact Jag Chhatwal, Victor Jose or Alec Morton to reserve a slot. We would like to have volunteers for session chairs in place by the end of March and all speakers arranged by the end of April to meet this INFORMS deadline of May 15th for final abstract submission.
We look forward to hearing from you! Thank you for your continued involvement and support in the Decision Analysis Society. Best wishes,
Jag Chhatwal - chhatwal at pitt.edu Victor Jose - vrj2 at georgetown.edu Alec Morton - a.morton at lse.ac.uk
Professional News
As most readers know, when a magazine or newspaper writes an article on a paper published in the professional literature, the writer typically will have a telephone call with the author of the professional article. Afterwards, the author has no control over the content and first sees it when it appears in print. We ask Ralph what he thought about the Forbes article written by Susan Adams and he said "Overall, it is a positive article and written well for the audience. However, a few things attributed to what I said or think are a bit off, like the first sentence, but I guess the author must draw readers in."
4 Steps to Successful Brainstorming By: Susan Adams, Forbes Available online at: http://www.forbes.com/sites/susanadams/2013/03/05/4-steps-to-successful-brainstorming/
Almost everybody does brainstorming wrong, Ralph Keeney says, and turns it into an enormous waste of time. He wants to tell you how to do it right.
Page 11 Volume 32, Number 1, April 2013 Decision Analysis Today
An emeritus professor at Duke’s Fuqua School of Business and a consultant to such diverse organizations as the Department of Energy and, just last month, a German power company with $40 billion in revenues, Keeney has devoted his career to a discipline called ―decision science,‖ helping companies and government agencies bring focus and rigor to their decision-making process so that they can waste less time spinning their wheels and instead get clear on their objectives before they try to meet their goals. Thirteen years ago he penned a book, still in print, called Value-Focused Thinking: A Path to Creative Decision-Making, which says that most corporate executives put the cart before the horse. Instead of parsing the objectives they hope to achieve, they direct their energy at coming up with solutions to broadly-stated problems.
―When most people do brainstorming, they run all over the place and think outside the box,‖ he says. ―I think they should think inside the box—the right-sided box.‖
His latest paper, published in the December issue of a journal called Decision Analysis, spells out what he believes is the right approach. In Germany for instance, the company he counseled is trying to cope with a vastly changed energy landscape, where nuclear power will be banned as of 2022, coal emissions restricted, and by 2020 at least 20% of the company’s energy must come from non-carbon-emitting sources. ―The company has to change phenomenally in order to exist 10 years out,‖ he says.
Instead of packing executives into a conference room and brainstorming solutions, Keeney met for one hour each with 19 top people, including the CEO. He pressed them on what they thought the company’s objectives were. Then he compiled a list of 450 things the executives wanted to achieve. He took the hundreds of objectives and boiled them down to 40 major goals, with 200 subsets. Why? Because, as his paper says, before you brainstorm, it’s essential to go through the process of analyzing and focusing on objectives. Here are Keeney’s four steps to effective brainstorming:
1. Lay out the problem you want to solve. This may be easier said than done. Keeney describes a doctoral student who is at sea while trying to come up with a dissertation topic and advisor. The student grasps for ideas with only the vaguest idea of a goal, stated as negatives rather than positives. ―I don’t think I could do it,‖ ―it is not interesting to me,‖ ―it seems too hard,‖ and ―it would be too time consuming.‖ Then finally someone suggests an idea that doesn’t have any of those negatives. The doctoral student grabs the topic. But Keeney says this is a poor way to make a major decision. Instead the student should keep pushing until they come up with at least five more alternatives, and then, considering all those, ―identify your objectives for your dissertation, evaluate the alternatives and select the best.‖ It will be well worth the effort.
2. Identify the objectives of a possible solution. This is what Keeney did for the German energy company and what he’s done for several government agencies, including the Department of Homeland Security and the energy department. It’s not easy and it takes time but if you can approach your goals critically and hone in on what you want to achieve, your brainstorming session will be much more effective.
Keeney offers a great example of this process. David Kelley, the founder of renowned design firm IDEO, wanted to design a product that would enable cyclists to transport and drink coffee while they were riding. A couple of ways to describe what he wanted to design: ―spill-proof coffee cup lids,‖ or ―bicycle cup holders.‖ But a much better description is the following objective: ―helping bike commuters to drink coffee without spilling it or burning their tongues.‖ Keeney likes this statement because it clearly lays out IDEO’s objectives, to help bike commuters 1) drink coffee, 2) avoid spills, 3) not burn their tongues. He even contributes a few objectives of his own: avoid distractions while biking, don’t contribute to accidents,
Page 12 Volume 32, Number 1, April 2013 Decision Analysis Today
keep the coffee hot and minimize costs. Going into that much detail before brainstorming about ways to design the cup holder makes IDEO much more likely to succeed.
3. Try to generate solutions individually. Before heading into a group brainstorming session, organizations should insist that staffers first try to come up with their own solutions. One problem with group brainstorming is that when we hear someone else’s solution to a problem, we tend to see it as what Keeney calls an ―anchor.‖ In other words, we get stuck on that objective and potential solution to the exclusion of other goals. For instance, when Keeney was consulting with a cell phone maker years ago, the company had numerous objectives. It wanted to produce a lightweight phone that also had GPS capabilities (Keeney did this consulting gig some time ago, but he insists the example remains illustrative). When company executives got together to brainstorm ideas about how to build a better phone, one person brought up the issue of weight. Suddenly everyone became fixated on that idea and forgot about their other objectives. Coming into a meeting with potential solutions reduces the risk that participants will get bogged down on one objective.
4. Once you have gotten clear on your problems, your objectives and your personal solutions to the problems, work as a group. Though he acknowledges that it’s a challenge not to ―anchor‖ on one solution in a brainstorming session, Keeney believes that if participants have done their homework, clarifying the problem, identifying objectives, and individually trying to come up with solutions, a brainstorming session can be extremely productive.
At the end of the paper, he describes a 2008 workshop he held to try to come up with ways to improve evacuations in large buildings in case of a terrorist attack, based on a recommendation from the National Institute of Standards and Technology. Keeney brainstormed for two-and-a-half days with 30 people with expertise in everything from firefighting and building codes to handicapped people and human behavior. The result, after going through Keeney’s four-step process: a list of 300 new alternative ways to speed evacuation. Then the participants evaluated the many ideas, which included using cell phone alarms to guide people to exits and building linked sky bridges on every fifth floor. The hope, of course, is that these solutions will never be tested. But Keeney’s brainstorming method helped the group find effective suggestions.
Business Strategies for Electrical Infrastructure Engineering: Capital Project Implementation Reginald Wilson (Redawil Engineering Company, USA) and Hisham Younis (Wayne State University, USA) Released: January 2013
With the principles of business strategies in mind, the analysis of cost containment plans, project risk evaluation, and the wide-range of quality planning techniques is essential for the integration of renewable generation and capital-intense endeavors in the current electrical infrastructure.
Page 13 Volume 32, Number 1, April 2013 Decision Analysis Today
Business Strategies for Electrical Infrastructure Engineering: Capital Project Implementation brings together research on informed-decision making within the strategic planning sphere of system integration. By highlighting social responsibility and environmental issues, this book is essential for technologically- literate executives, engineers, application analysts and many more interested in high-impact process evaluation.
Topics Covered:
• Capital Planning • Informed Decision-Analysis • Capital Project Management and Progress • Operations Planning Philosophy • Process Evaluation • Environmental Effect • Quality Improvement Techniques Scenario • Fiscal Management and Cost Avoidance Analysis Strategies • System integration and challenges
Decision Analysis Journal The Decision Analysis June 2013 issue... http://da.journal.informs.org/content/10/2.toc (link will activate upon publication) Group Decisions, Preference Elicitation, Experienced Utility, Survival Probabilities, and Portfolio Value of Information: From the Editors Rakesh K. Sarin and L. Robin Keller This “From the Editors” column is co-authored by Editor-in-Chief Rakesh K. Sarin and former Editor-in- Chief L. Robin Keller, since the papers were all submitted before this year and were thus accepted by Keller. Foundations for Group Decision Analysis Ralph L. Keeney
http://da.journal.informs.org/content/early/2013/03/13/deca.2013.0265 The first article develops a general group decision analysis model. In this paper the author extends the logical foundations of decision analysis (Pratt, Raiffa, and Schlaifer, 1964) developed for individual decisions to group decisions. The focus of the paper is to provide prescriptive guidance in settings where a committee or a group is collectively responsible for the decision. They key feature that separates this paper from a voluminous literature in group decisions is that individuals in the group are permitted to have both different objectives and different events and therefore different frames for the same decision problem. Using a set of decision analysis assumptions for the group decision, it is shown that the group expected utility is the weighted sum of individual expected utilities. The model presented in the paper is consistent with the “additive collective choice rule” used by Dyer and Miles (1976) for selecting the trajectory pair for NASA’s Mariner Jupiter/Saturn 1977 project.
Page 14 Volume 32, Number 1, April 2013 Decision Analysis Today
Group decisions are made every day in a myriad of contexts in both private and public organizations. The result of this paper will encourage a prescriptive decision analysis approach to group decisions resulting in more thoughtful decisions. Toward an Improved Methodology to Construct and Reconcile Decision Analytic Preference Judgments Richard M. Anderson and Robert Clemen The second article in this issue provides a process for eliciting preferences that reduces or eliminates biases. A vast amount of behavioral research has demonstrated that peoples’ tradeoffs/preferences are highly volatile and depend on the framing of the decision problem and on the elicitation method. This is highly problematic for decision analysis which assumes a reasonably stable and consistent set of preferences. If revealed preferences depend on the questions asked, then which questions and therefore which preferences are the right ones to use in the prescriptive analysis? The authors recognize that people are learning about their preferences during elicitation and therefore the decision analytic technique should facilitate this process. They propose and demonstrate a three-step process of training, practice, and application to elicit decision weights and find that the resulting weights seem to be bias free.
Because of observed biases in preference assessments, researchers have proposed equal weighting or other simple rules. These simplifications, though appropriate in some contexts, produce clearly sub- optimal results. This paper gives the decision analysis community the hope that, for important decisions, one can use the prescriptive approach and elicit inputs required by the model in a reliable way.
Determinants of Experienced Utility: Laws and Implications Manel Baucells and Rakesh K. Sarin The key premise of the third paper is that our choices today influence our satisfaction in the future. It is, however, far from clear what the choices are that will improve life satisfaction or well-being (experienced utility). This paper advocates six laws that determine experienced utility. These laws capture habit formation and satiation. The implications of these laws such as wanting vs. liking, crescendo, recharge periods, variety seeking and craving are explored. These laws explain the puzzle that more money, beyond the amount needed for basic needs, does not improve well-being. Further, the sixth law, projection bias, gives the paradoxical result that higher income may lead to lower well-being because of incorrect predictions.
Estimating Second Order Probability Beliefs from Subjective Survival Data Péter Hudomiet and Robert J. Willis The fourth paper uses an econometric model to estimate personal longevity. The estimation of personal longevity or the probability of living for the next 10 to 20 years is important in financial as well as personal life planning. The current method used by the Health and Retirement Study asks: “What is the percent chance that you will live to be [TARGET AGE] or more?” This method produces a biased response with a large number of responses heaped on values of “0”, “50”, or “100”. The authors’ model provides a better estimate of probability of survival. Based on a sample of about 13,000 people, the authors show that people place too much weight on parents’ age at death in forming expectations about their own longevity but underweight factors such as health behavior.
Page 15 Volume 32, Number 1, April 2013 Decision Analysis Today
Components of Portfolio Value of Information Kun Zan and J. Eric Bickel Our final paper deals with portfolio decision analysis. The results of this paper are useful in shaping corporate strategy. The central problem in portfolio decision analysis is the selection of an optimal portfolio of projects under a budget constraint. The decision maker could maximize expected net present value (NPV) of the portfolio subject to the budget constraint. The decision maker could also obtain additional information about the NPV of each project. This information is used to update the probability distribution of project NPV. With the aid of the information, hopefully, the decision maker obtains a higher expected net present value of the portfolio. The portfolio value of information is simply the incremental expected net present value of the portfolio with information over that without information. The key contribution of this paper is to dissect the portfolio value of information into two parts: value of information that comes from better prioritization of projects and that comes from ability to exclude projects that do not meet a performance threshold.
Decision Analysis is included in the Social Sciences Citation Index: (Decision Analysis has an impact factor of 2.143 in the management category, ranking it in the top 25%, as 38th out of 166 journals.) http://thomsonreuters.com/products_services/science/science_products/a- z/social_sciences_citation_index/ Email Alerts is a free service providing table of contents, searching, article citation, and other email based alerts for Decision Analysis: http://da.journal.informs.org/cgi/alerts/etoc Decision Analysis archive available through Highwire Press: http://da.journal.informs.org/ Decision Analysis subscription information and rates: http://www.informs.org/Find-Research- Publications/Journals/Subscribe
See Decision Analysis papers which are most often cited and most read: (Rankings are based on citations to online articles from HighWire Press-hosted articles only, not all citations from any works published anywhere.): http://decision.highwire.org/reports/most-cited Hear INFORMS podcasts on decision analysis: http://www.informs.org/Pubs/DA/NEWS/INFORMS- Podcasts-on-Decision-Analysis
Attention INFORMS Decision Analysis Society Members! By special arrangement with the Decision Analysis Society Council, dues-paying regular members of the DAS receive a subscription to the journal as part of their membership dues. The DAS is a subdivision of INFORMS. For information on DAS: http://decision-analysis.society.informs.org/.
Decision Analysis is a quarterly journal dedicated to advancing the theory, application, and teaching of all aspects of decision analysis. The primary focus of the journal is to develop and study operational decision-making methods, drawing on all aspects of decision theory and decision analysis, with the ultimate objective of providing practical guidance for decision makers. As such, the journal aims to bridge the theory and practice of decision analysis, facilitating communication and the exchange of knowledge among decision analysts in academia,
Page 16 Volume 32, Number 1, April 2013 Decision Analysis Today
business, industry, and government. Decision Analysis is published in March, June, September, and December by the Institute for Operations Research and the Management Sciences (INFORMS) at 7240 Parkway Drive, Suite 300, Hanover, Maryland 21076. Please visit our website at http://www.informs.org/Pubs/DA.
DA Around the World
Column Editors: Alec Morton and Matthias Seifert
How International is the DAS Community? In previous newsletter issues we have tried to showcase some of the local Decision Analysis communities around the world. By doing so, we wanted to provide snapshots of the different types of activities and networks that can be found outside the INFORMS domain. The purpose of this current edition, however, is to take a look at what happens inside DAS. In particular, we will shed more light on the level of internationalization that characterizes different aspects of the Society. We hope that this article will help stimulate a discussion on how we can potentially improve in order to become more accessible to DA enthusiasts around the globe. Decision Analysis Authorship One obvious starting point for analyzing diversity is to examine the profile of contributors and editors associated with the Decision Analysis (DA) journal. In particular, Figure 1 depicts the proportion of authors in DA who were based at institutions outside the United States for each year since the inception of the journal in 2004.. The bar chart reveals that scholars and practitioners from around the world frequently contribute to the journal’s content and account on average for about 30% of DA’s authors in each issue. The peak year so far was 2007, where 18 out of 34 contributors were international, while in the most recent year 2012, 20 out of 49 authors were based at non-US institutions at the time of publication.
Page 17 Volume 32, Number 1, April 2013 Decision Analysis Today
Figure 1 Overview of non-US DA contributors
60%
50%
40%
30%
20%
10%
0% 2004 2005 2006 2007 2008 2009 2010 2011 2012
% Non-US Contributors
Decision Analysis Editorial Board In addition to authorship, diversity can also be examined by considering the current composition of the editorial board at DA. In more general terms, Decision Analysis is run by one Editor-in-Chief, 11 Associate Editors and 55 members of the Editorial Board. Figure 2 illustrates that 72% of the editorial team members are based at US institutions, whereas the remaining 28% of members work in 9 different countries. The largest proportion of members is thereby located in the United Kingdom (4), followed by Spain (4), France (3) and Finland (3). Figure 2 Diversity of the DA Editorial Board
China Finland France France/Singapore Germany Netherlands Spain UK USA
Page 18 Volume 32, Number 1, April 2013 Decision Analysis Today
INFORMS Annual Meeting in Phoenix 2012 Since not every member of DAS publishes research findings in Decision Analysis, an alternative approach to studying diversity in our Society relates to participation at the INFORMS Annual Meeting. The pie chart below (Figure 3) shows the proportion of international contributors to the sessions that had been sponsored by the Decision Analysis Society in Phoenix 2012 (including presenters and co-authors). More specifically, 61 of the 288 scholars and practitioners (22%) who had papers scheduled for presentation were based outside the USA. Among the countries that were most frequently represented are the United Kingdom (5%), Canada (4%), Spain (4%) and France (2%).1 Moreover, it can be seen that 17 countries were contributing to the presentation line-up in total. Figure 3 Non-US contributors to the Decision Analysis Track at the INFORMS Annual Meeting in Phoenix 2012
Brazil China Finland France/Singapor Venezuela Canada France e India Ireland Germany Italy Netherlands Norway Poland Spain
UK
USA
DAS Membership profile Finally, one could also look at the general profiles of members in the Decision Analysis Society. A quick search in the directory on the INFORMS webpage reveals that at the moment, regular, fee-paying members (i.e. excluding students) come from 35 different countries. Furthermore, it can be seen that the largest proportion of DA Society members outside the US come from Canada (19), followed by Brazil (10), the United Kingdom (9), France (6), Germany (6) and Turkey (6). The bar chart in Figure 4 summarizes these findings.
1 These figures only account for the total number of contributors regardless of the number of papers. Hence, the findings may be somewhat distorted if, for instance, one paper had been co-authored by 10 scholars.
Page 19 Volume 32, Number 1, April 2013 Decision Analysis Today
Figure 4 DA Society Membership (excluding US)
20 18 16 14 12 10 8 6 4 2
0
Italy
India
Chile Spain
Japan
Israel
China
Brazil
Korea
Russia
Greece
France
Turkey
Ireland
Croatia
Canada
Zambia
Taiwan
Finland
Sweden
Norway
Portugal
Malaysia
Romania
Australia
Germany
Indonesia
Argentina
Philipines
Singapore
Switzerland
Netherlands Great Britain Great New Zealand New So how international are we? In sum, these different figures illustrate that non-US institutions certainly play an important role in the Decision Analysis Society as they are substantially involved at all ends of the Society’s activities. It appears though that there is still room for improvement. In particular, given the continuously growing number of top research articles produced in Asia, it seems that countries such as China, India, Korea, Singapore and Japan are still relatively under-represented. The International INFORMS Meeting in Beijing in Summer 2012, however, where DAS sponsored a total of 13 papers, showed that a thriving DA scene does certainly exist. In the upcoming editions of the DAS Newsletter, we will therefore endeavor to raise more awareness of the Asian DA communities and introduce the different types of activities and initiatives that take place. Taking this into account, we welcome potential volunteers among our current members, who would like to introduce their local DA domain. As usual, please feel free to get in touch with us any time ([email protected]).
DA Practice
Column Editor: Bill Klimack
As I write this, we are preparing for the Analytics Conference in San Antonio. I anticipate a great DA track, organized by Jack Kloeber, Kromite, Dave Leonhardi, Boeing, and Drew Pulvermacher, ShopBop.com. The soft skills track is inspired by the interpersonal requirements of DA and is chaired by Freeman Marvin, Innovative Decisions, and Paul Wicker, Decision Strategies. The DAAG meeting has been scheduled for later in the week in Austin. So April is a busy month for DA people. The INFORMS topical conference in 2013 will be on healthcare and held in June in Chicago. The INFORMS international conference is a joint effort with EURO in Rome in July. Both of these will
Page 20 Volume 32, Number 1, April 2013 Decision Analysis Today
include strong DA components. In 2014 the topical conference will be on Big Data and held in Silicon Valley. All are great opportunities to see what is being done in the field and to grow professionally. See https://www.informs.org/Attend-a-Conference for more information. The San Antonio Analytics Conference will also host the first offering of the Certified Analytics Professional (CAP) examination. The development of the CAP program has been a major INFORMS strategic initiative. The exam will be offered at four other locations in 2013, including the INFORMS conferences. As of mid-March, there were 121 people in various stages of application to take the exam. Information on CAP is available at https://www.informs.org/Build-Your-Career/Analytics-Certification. This month’s guest columnist is Phil Beccue. He shares with us valuable insights gained from his experience in the pharmaceutical sector. He has practiced DA both as an internal expert and as an external consultant in a number of firms. A long time member of the Decision Analysis Society, he is a recipient of the DAS Practice Award. Please send your comments, suggestions, and, especially, offers to be a guest columnist to me at [email protected]. You can help improve the practice of decision analysis!
“Create Value by Following the DA Process” Phil Beccue, Principal White Deer Partners
―This meeting is over!‖ stated the senior vice-president of a large biopharmaceutical, and after only 10 minutes of a planned 30-minute presentation, the shocked and puzzled team members filed out of the paneled conference room. I’ve been part of an organization where this event has occurred more than once. What was the problem? The senior leader sent the project team back to the drawing board because he had no confidence in the analysis for a top R&D program, a single alternative was presented, potential futures were assumed away, and the presenter was ―selling‖ the strategic plan with limited facts to back his assumptions. In my past roles as an internal provider of decision support, and now as an external consultant for biopharmaceutical companies, I have seen numerous challenges to driving high quality decisions, and have found the philosophy, tools, and processes of Decision Analysis (DA) are well-suited to provide tremendous improvement to decision-making, as well as increased value to the organization’s long-term profitability. The value of the DA framework takes on numerous forms. Good framing up front, coupled with a decision mindset (e.g., clearly distinguishing between factors that can be controlled from factors not in our control), opens the team’s thinking to new alternatives and a greater awareness of potential futures. Well-planned and facilitated assessment sessions provide consensus around probabilities of technical success for clinical experiments. When there are many and interrelated parameters, tornado sensitivity analyses helps to understand the top drivers of value. Focus on the DA Framework Depending on the particular business need, DA provides many avenues for adding value. In this article, I’d like to focus on one element: the application of a systematic decision process based on a DA framework. There are many good business processes that are widely used and very helpful. Business
Page 21 Volume 32, Number 1, April 2013 Decision Analysis Today
processes are not unique to DA, but coupling a business process with practical DA principles and tools has differentiated DA from the status quo for many of my colleagues. There are many flavors of the DA process. The one I have found useful contains four steps: Problem Structuring, Data Collection, Evaluation, and Communication.
Decision Problem Data Evaluation Communication Problem Structuring Collection Action
Following a systematic DA process helps avoid many common problems in the organizations I have worked with, such as meetings that waste time and follow tangent paths of thinking, decisions that continue to be revisited, and lack of confidence in conclusions because uncertainties are ignored, to name a few. Skipping steps I will highlight three common failures of not following each of the four steps in the DA process. The first failure example is the team that jumps immediately to the last step, Communication. In this example, the team starts to assemble all the information for a particular plan, without thinking about alternate plans, or collecting information to back up their conclusions. The presentation ends up as qualitative gibberish, often failing to hide the true lack of preparation and thoughtfulness of the team. The second failure example is also common: skipping the Structuring step and jumping right into Data Collection. Many teams I have worked with grab existing spreadsheet models with layers upon layers of detail, and attempt to fit them into the current decision at hand. This causes the usual biases of anchoring on what is easily accessible rather than what is most important, and makes it too easy to focus efforts on information that is not relevant to comparing alternatives. The motivation is understandable. It comes from a desire to save time by building on past models, but the results are just the opposite … even more time spent and worse, limiting the alternatives under consideration and the number of uncertainties that should be addressed. The third failure example comes from a team that completed the first and last steps, but skipped over both the Data Collection and Evaluation steps. This team did a commendable job brainstorming six strategic options and a range of uncertain factors, but the urgency of the final presentation prevented them from building a complete, fact-based analysis to support the conclusions. Instead, the preferred plan from a key function was presented, supporting their wishful thinking unencumbered by technical and market facts.
Lessons learned The many failures that I hear about (and sometimes experience) have taught me how important it is to take a strong leadership role in taking teams through a systematic four-step process. I have learned to: 1. Allow plenty of time for the Problem Structuring step. Many are inclined to skip this first step, and jump right into Data Collection and Evaluation. I usually schedule two to four meetings just for this part alone, as it is critical to adequately create good alternatives, ensure we have the best metrics in place to compare them, and include all relevant information.
Page 22 Volume 32, Number 1, April 2013 Decision Analysis Today
2. Create a Decision Statement and use it at every meeting. The Decision Statement is a concise, one-sentence summary of the scope of the decision context. In order to make sure the team stays on track, I review this at every opportunity to motivate and confirm our purpose. 3. Remind teams of progress against the four-step process. At each meeting, I review the process, and tell them where we have come from, what our purpose is now, and what remains. This constant reminder helps to keep our focus. If I receive pushback from any members, I recollect some of the failure examples mentioned earlier.
At the end of each DA project, I ask teams to reflect on the value of the time spent and approach taken. One common answer is that the team spent extra time thinking through multiple strategies and many future scenarios, but they also quickly counter that without a systematic DA process, they were spending months (and sometimes years) revisiting plans and failing to arrive at a consensus that was comprehensive and fact-based. By strictly following a rigorous process based on foundational DA principles, cross functional product development teams are more likely to avoid abrupt closure of final presentations, and more likely to find the following response from the General Manager of a key business unit after sponsoring a DA project: ―We were looking at a product opportunity that was complex….with different formulations, different indications, all with different risks, different timelines, and different types of competitive advantages in the marketplace. The shift to a solid, analytical decision-making process allowed us to model various scenarios, and ultimately make the right call on a complex product opportunity that gave the most value to the company.‖
Society of Decision Professional State of the Society 2013 Frank Koch, President, Society of Decision Professionals Koch Decision Consulting, Eugene, OR
I am writing this as I fly home to Eugene from Austin, TX. This has been a great week! It started in San Antonio at the INFORMS Conference on Business Analytics and Operations Research and ended in Austin at the Decision Analysis Affinity Group’s annual meeting of Decision Analysis practitioners. As I enter the final quarter of my presidency, I would like to reflect on the progress SDP has made. The Society of Decision Professionals is a young society with about 275 members. I belong to other professional organizations that are in their second century with a rich tradition and a large membership; SDP is barely a toddler by comparison. And yet we have accomplished a lot in the past two years. SDP is focused on the decision professional, specifically the practitioner who helps decision-makers in their day- to-day work of making good, strategic choices and creating value for their respective organizations. Let me outline some of the accomplishments this year. SDP has launched a decision-makers ―landing page‖ – a website focused on our customers, the decision-makers who use our analyses. This website is
Page 23 Volume 32, Number 1, April 2013 Decision Analysis Today
designed to provide resources and ideas to decision-makers seeking to improve their organizations’ decision processes and capture more value from the opportunities they face. The website, www.decisionquality.org , is great starting point for decision-makers who want to understand what decision analytics can do for them. SDP has also produced 9 webinars in the past 9 months. These webinars are broadcast at no cost to the decision analysis community. Speakers have included Jay Russo, Vince Barabba, Greg Parnell & Jim Matheson and have featured topics from decision analysis & company culture, to portfolio management and value of information. SDP has also launched a ―Fundamental of Decision Analysis‖ webinar series designed for the new and young (or old and forgetful) members of our profession. These monthly webinars are open to all and are publicized on the SDP website (http://www.decisionprofessionals.com/) as well as the Linked-In SDP Open Network. We are also launching a new knowledge sharing area on our website which will feature articles, presentations, and simple tools that SDP members may download. SDP was a co-sponsor of the 2013 Decision Analysis Affinity Group (DAAG) annual meeting in Austin, along with the University of Texas at Austin, Chevron and Shell. The meeting was scheduled to directly follow the INFORMS Analytics conference in San Antonio to allow interested practitioners to extend their stay in Texas and participate in both meetings. DAAG included about 95 attendees, two days of sessions, 30 speakers and two evening networking events (drinking and talking). We had about 15 people attend both DAAG and the Analytics meeting (thanks to Chevron for providing transportation up I-35 from the Grand Hyatt to Austin on the evening of April 9th). Next year our plan is to hold DAAG in Boston adjacent to the 2014 Analytics conference, again to encourage our members to attend both meetings and to attract the Analytics attendees to spend a couple more days focusing on best practices in decision analysis with case studies, applied workshops and of course more networking. One of the co-chairs for the 2014 DAAG meeting is Jeff Keisler, DAS president; Eric Bickel, DAS President-elect was the 2013 DAAG Chairman. Our emphasis is to be complementary to the INFORMS/DAS sessions, providing more opportunities to delve into case studies, and get better understanding of the decision-makers’ focus. In May, we will be holding our annual elections. Running for Vice President/ President-elect are Bill Klimack, Chevron & INFORMS VP-Meetings, and Marilyn Metcalf, GlaxoSmithKline. Running for Secretary are Jay Anderson, Lilly, and Rob Kleinbaum, RAK & Co. They will join Jim Felli, Lilly, who becomes President on July 1 and Pat Leach, DSI, who continues into his 2nd term as Treasurer. We also have five candidates running for the SDP Board: Terry Bresnick, Innovative Decisions Inc., Chris Dalton, Syncopation, Raymond Fonk, Shell, C. Kwon Kim, Kromite, and Tim Neiman, Decision Applications. If you have not yet joined the Society of Decision Professionals, I welcome you to join now and vote in our upcoming elections. I have enjoyed my two years as an SDP officer and look forward to networking and meeting with you at future SDP & DAS events.
Page 24 Volume 32, Number 1, April 2013 Decision Analysis Today
Research
Column Editor: Debarun Bhattacharjya
My colleague Lav Varshney (IBM T.J. Watson Research Center) has written a fascinating article for the research column about how decision analysis techniques may be applied to computational creativity. Computational creativity is an interdisciplinary domain that explores how machines might be able to support and enhance human creative endeavors. In the article, he explains in particular how the notion of Bayesian surprise (a metric based on Bayes' rule) may be used to potentially generate artifacts that a user may perceive to be novel.
Surprise in Computational Creativity and Machine Science By: Lav R. Varshney (IBM T. J. Watson Research Center)
Introduction When we think about human intelligence, we think about the kinds of abilities that people have, such as memory, deductive reasoning, association, perception, abductive reasoning, inductive reasoning, and problem solving. With technological advancement over the past century, computing technologies have progressed to the stage where they too have many of these abilities. The pinnacle of human intelligence is often said to be creativity, ensconced in such activities as music composition, scientific research, or culinary recipe design. One might wonder, then, can a computer be creative? Whether considering literary manuscripts, musical compositions, culinary recipes, or scientific results, this article argues that it is indeed possible for computers to create novel, high-quality artifacts. As one typical example, consider a culinary computational creativity system that takes repositories of existing recipes, data on the chemistry of food, and data on human hedonic perception of flavor to create new recipes that have never been cooked before, but that are flavorful (Varshney et al., 2013). As another example, consider a machine science system that takes the scientific literature in genomics, generates hypotheses, and tests them automatically to create new scientific knowledge (King et al., 2009). Some classical examples of computational creativity include AARON, which creates original artistic images that have been exhibited in galleries around the world (Cohen, 1995), and BRUTUS, which tells stories (Bringsjord and Ferrucci, 2000). Several new applications, theories, and trends are now emerging in the field of computational creativity (Cardoso et al., 2009). Computational creativity algorithms proceed by first taking existing artifacts from the domain of interest and intelligently performing a variety of transformations and modifications to generate new ideas; the design space has combinatorial complexity. Next, these generated possibilities are assessed to predict if people would find them compelling as creative artifacts and the best are chosen. A similar approach applies for machine science. A critical aspect of any such algorithm is determining a meaningful characterization of what constitutes a good artifact. The field of decision analysis has developed a wide variety of methods and tools to formally identify, represent, and assess important aspects of decision-making: the central guiding principle is to maximize
Page 25 Volume 32, Number 1, April 2013 Decision Analysis Today
expected utility. The key point to note in the assessment phase of computational creativity algorithms is that it is important to assess novelty of alternatives along with other attributes of value pertaining to quality. Note that each domain—whether literature or culinary art—has its own specific metrics for quality. However, in all cases of exploration and discovery independent of domain, people like to be surprised. This article discusses how to create surprising artifacts through computational creativity or discover surprising facts through machine science. Bayesian Surprise How can we compute whether something is likely to be perceived as surprising? Itti and Baldi have defined an information-theoretic measure of surprise termed Bayesian surprise, and have experimentally demonstrated its connection to attraction of human attention across different spatiotemporal scales, modalities, and levels of abstraction (Itti and Baldi, 2009; Baldi and Itti, 2010). The surprise of each location on a feature map is computed by comparing beliefs about what is likely to be in that location before and after seeing the information. An artifact that is surprising is novel, has a wow factor, and changes the observer’s world view. This can be quantified by considering a prior probability distribution of existing artifacts and the change in that distribution after the new artifact is observed, i.e. the posterior probability distribution. The difference between these distributions reflects how much the observer’s world view has changed. It is important to note that surprise and saliency depend heavily on the observer’s existing world view, and thus the same artifact may be novel to one observer and not novel to another. That is why Bayesian surprise is measured as a change in the observer’s specific prior probability distribution of known artifacts.
Mathematically, the cognitively-inspired Bayesian surprise measure is defined as follows. Let be the set of artifacts known to the observer, with each artifact in this repository being . Furthermore, a new artifact that is observed is denoted . The probability of an existing artifact is denoted , the conditional probability of the new artifact given the existing artifacts is , and via Bayes’ theorem the conditional probability of the existing artifacts given the new artifact is . The Bayesian surprise is defined as the following relative entropy (Kullback-Leibler divergence):
Surprise and Information One might wonder if Bayesian surprise, , has anything to do with measures of information such as Shannon’s (1948). In fact, if there is a definable distribution on new artifacts , the expected value of Bayesian surprise is the Shannon mutual information.
,
which by definition is the Shannon mutual information . The fact that the average of the Bayesian surprise equals the mutual information points to the notion that surprise is essentially the derivative of information. The precise relationship can be formalized as follows. For a fixed reference distribution , the weak derivative of mutual information is:
Page 26 Volume 32, Number 1, April 2013 Decision Analysis Today
Indeed, even the Shannon capacity of communication over a stochastic kernel can be expressed in terms of the Bayesian surprise:
, therefore all communicated signals should be equally surprising when trying to maximize information rate of communication. Capturing Attention In information overload regimes, it is necessary for messages to not only provide information but also to attract attention in the first place. In many communication settings, the flood of messages is not only immense but also monotonously similar. Some have argued that ―it would be far more effective to send one very unusual message than a thousand typical ones‖ (Davenport and Beck, 2001). Indeed, novel and surprising stimuli spontaneously attract attention (Kahneman, 1973), a fact well-known in marketing (Schoormans and Robben, 1997). There is a basic tradeoff between messages being informative and being surprising (Varshney, 2013). But what does this have to do with either computational creativity or machine science (Evans and Rzhetsky, 2010)? It is our contention that the wow factor of newly creative things or newly discovered facts is especially important in regimes with an excess of creative artifacts or growing scientific literature. Hence, surprise is not only important for ensuring novelty but also for capturing people’s attention. Designing for surprise is of utmost importance. For machine science in particular, the following analogy to the three layers of communication put forth by Warren Weaver (1949) seems rather apt:
Level Communication Machine Science A (The technical problem) How accurately can the symbols How accurately does gathered of communication be data represent the state of nature? transmitted? B (The semantic problem) How precisely do the transmitted How precisely does the measured symbols convey the desired data provide explanation into the meaning? nature of the world? C (The effectiveness problem) How effectively does the How surprising are the insights received meaning affect conduct that are learned? in the desired way?
A key element of machine science is not just producing accurate and explanatory data, but insights that are surprising as compared to current scientific understanding. Conclusion Computational creativity and machine science are domains of study that are not within the traditional boundaries of decision analysis, yet its tools and techniques may have a role to play in formalizing typical problems that arise. Once formalized, optimizing artifacts/insights for quality and surprise can lead to new
Page 27 Volume 32, Number 1, April 2013 Decision Analysis Today
products and ideas that the world has never seen before, thereby helping people discover the previously unexplored. This paper has described an assessment approach for evaluating new artifacts based on Bayesian surprise. Computationally, this method may result in a big data problem as there may be quintillions of design possibilities – but upon learning and fixing the reference beliefs from existing artifact repositories, computation of surprise is eminently parallelizable. For additional information contact: Lav R. Varshney ([email protected])
References P. Baldi and L. Itti, ―Of bits and wows: A Bayesian theory of surprise with applications to attention,‖ Neural Netw., vol. 23, no. 5, pp. 649–666, Jun. 2010. S. Bringsjord and D. A. Ferrucci, Artificial Intelligence and Literary Creativity: Inside the Mind of BRUTUS, a Storytelling Machine, Mahwah, NJ: Lawrence Erlbaum Associates, 2000. A. Cardoso, T. Veale, and G. A. Wiggins, ―Converging on the divergent: The history (and future) of the international joint workshops in computational creativity,‖ A. I. Mag., vol. 30, no. 3, pp. 15–22, Fall 2009. H. Cohen, ―The further exploits of AARON painter,‖ in Constructions of the Mind: Artificial Intelligence and the Humanities, eds. S. Franchi and G. Guzeldere. Special edition of Stanford Humanities Review, vol. 4, no. 2, pp. 141–160, 1995. T. H. Davenport and J. C. Beck, The Attention Economy: Understanding the New Currency of Business. Boston: Harvard Business School Press, 2001. J. Evans and A. Rzhetsky, ―Machine science,‖ Science, vol. 329, no. 5990, pp. 399–400, Jul. 2010. L. Itti and P. Baldi, ―Bayesian surprise attracts human attention,‖ Vis. Res., vol. 49, no. 10, pp. 1295–1306, Jun. 2009. D. Kahneman, Attention and Effort. Englewood Cliffs, NJ: Prentice-Hall, 1973. R. D. King, J. Rowland, S. G. Oliver, M. Young, W. Aubrey, E. Byrne, M. Liakata, M. Markham, P. Pir, L. N. Soldatova, A. Sparkes, K. E. Whelan, and A. Clare, ―The automation of science,‖ Science, vol. 324, no. 5923, pp. 85–89, Apr. 2009. J. P. L. Schoormans and H. S. J. Robben, ―The effect of new package design on product attention, categorization and evaluation,‖ J. Econ. Psychol., vol. 18, no. 2-3, pp. 271–287, Apr. 1997. C. E. Shannon, ―A mathematical theory of communication,‖ Bell Syst. Tech. J., vol. 27, pp. 379–423 and 623–656, Jul. and Oct. 1948. L. R. Varshney, ―To surprise and inform,‖ working paper, 2013. L. R. Varshney, F. Pinel, K. R. Varshney, A. Schörgendorfer, and Y.-M. Chee, ―Cognition as a part of computational creativity,‖ working paper, 2013
Page 28 Volume 32, Number 1, April 2013 Decision Analysis Today
Ask DAS Column Editors: John Coles and Florian Federspiel
A quick guide to effective and successful poster presentations Poster presentations are something relatively unique in the academic world, thus it is important to give context to newcomers, as well as provide a reminder for those of us who have presented posters before and would like to continue to improve. To get some perspective, we asked the poster award winners and judges from the 2012 Society for Risk Analysis Annual Meeting to share some observations, thoughts, and insight about making and presenting posters. These contributors provided some great insights which we wanted to share with you.
What goal are you trying to achieve when presenting a poster at a conference? The general purpose of a poster presentation is to provide a quick snapshot of your research, and to draw in an audience, while at the same time offering enough material to delve deeper into the topic as appropriate. One of the insightful responses people gave on this topic dealt with the amount of material that should be covered in a poster. Researchers can easily get caught up in details and overload a presentation. Writing focused abstracts may be a valuable exercise to practice and help to create effective posters. A well-written abstract (200-300 words) should be concise, quick to grasp, and focus only on the most important points of your research. This exercise can help to focus your poster content and provide a clear outline for your presentation. Below we list a few of the goals that our contributors had for their posters:
- To present preliminary research findings and get early feedback - To find potential collaborators for future work - To get people interested in work that you will eventually publish - To start a conversation about the topic, including the sharing of new and emerging ideas
To achieve these goals, our contributors emphasized the importance of first capturing the attention of the passersby, by any means possible. The most likely presentation scenario is that you will not get to talk to most of the people that look at your poster, so it is very important to design your poster with this in mind. Assuming this, it is important to be able to capture your audience purely with visual aids and large text. As a test, you can have a friend or colleague come look at your poster and see what things they notice in the first 5 to 10 seconds. If what they point out does not represent what you were hoping to communicate, rework your poster until the emphasis is correct. Another possible situation is that you will not be around when someone does come by your poster. For this reason it is very important that your poster be self-guiding. By ―self-guiding,‖ we mean that a person walking up to your poster should be able to follow the reasoning and graphics without you walking them through it. You could have a friend test this as well using an approach similar to the one proposed above. When you do get to talk to someone looking at your poster, then you have about 30 to 60 seconds (the famous elevator pitch) to capture your audience for a full presentation. A full presentation should give a walk-through of your research in about 5 minutes. If you are uncertain what things should be included in a full walk-through of your presentation, refer back to the abstract you used to help focus the poster.
Page 29 Volume 32, Number 1, April 2013 Decision Analysis Today
Have you found poster sessions more or less useful than conference presentations for meeting new people with similar research interests? Many of our contributors noted that, in their experience, poster sessions were not actually the best time to share your work with many people with similar research interests. Oral sessions are probably a better venue for sharing your work (in-depth) with people whose research is most closely related to your field. However, poster sessions do provide a good opportunity to have one-on-one conversations about your work with people who might be interested. As one contributor noted, the interaction during a poster session ―can be significantly enhanced with a little advance work, by contacting individuals you'd like to visit with who are going to be at the conference and inviting them to stop by your poster.‖ Have you found poster sessions more or less useful than conference presentations for learning to effectively communicate your research to a broad audience?
This was noted as one of the main strengths of poster sessions. Poster sessions are a great opportunity to try out an elevator speech or present a new, inter-disciplinary concept that you are developing. As one contributor said: ―There are some posters whose area of research I really didn’t know anything about, but I happened to be attracted to the look of their poster, and I learned a little bit about their research through the poster. Because posters are more generally going to be communicating to a broad audience, I think it’s important to make the posters appealing to people who might not be in your field.‖
How do you choose how to balance words, diagrams, pictures, and equations? We got many great responses to this question, so we decided to list some of the key advice points from our contributors in bullets. Developing a Theme and Focus - The poster should tell a story, hence include an introduction or motivation, the main findings or theoretical contribution, and some closing thoughts - Make your desired poster outline, then cut the material in half, increase the font size, and find some appropriate and attractive visuals - The best posters limit the amount of detail presented and instead opt to also achieve some degree of visual attraction Text and Equations - Words could be in sentence and paragraph form, but most generally prefer shorter bullet points - Displaying key equations and/or findings prominently is important, but are unlikely to capture an audience - There should be enough information on the poster that if you step away for a moment, people will still be able to quickly grasp and follow your contribution - Avoid equations as much as possible. Even if the work is essentially theoretical, it is often better to describe the work with words and show the results of the work through simple numerical examples Pictures and Diagrams - Colorful diagrams and pictures are wonderful and you usually need an explanation to accompany the diagrams and pictures to fully tell the story of your research - There should be enough relevant graphics and pictures to catch a person’s eye - Pictures can help to provide visual context to the research - Use diagrams to succinctly and clearly display information; these can help prompt questions from attendees, and also allow you to describe your research in more detail without using a lot of space Other Useful Principles for Poster Design
Page 30 Volume 32, Number 1, April 2013 Decision Analysis Today
- White space is acceptable on a poster; do not feel the need to cover up every little part of the poster with text or pictures - It takes time as well as some trial and error to determine the best placement for the various elements; ideally all elements should follow a natural sequence leading a viewer from beginning to end - By utilizing basic design principles such as complementary colors, or even simply adding some color to a design, you can make your poster stand out
What particular features or aspects of a poster make it competitive for judged competitions?
While there are many other reasons to present a poster, when participating in a poster competition it is also nice to win! It is true that even if everyone improved their posters after reading this column, we would not have an increase in the number of winners. However, perhaps whoever reads this column and takes it to heart might enjoy a competitive edge! We thus conclude with the responses of our contributors with regards to poster competitions. Two main themes emerged: First: It is important for you to be knowledgeable about your topic and have something substantive to say and show. As one contributor said, ―regardless of design, if a poster lacks substantive content, it is a waste of time. Know your research inside and out. Do not just spout off a few figures but be able to have a conversation about the research and results.‖ When a poster is poorly thought-through or poorly presented, it is not hard for your audience (and judges) to figure it out. Make sure that you are ready and available to answer questions that your audience might have. Second: Creativity and simplicity are critical to garnering votes and getting attention. To win a competition you want to catch people’s eye. Color and simplicity can help to draw your audience in; creativity helps them to remember. One the contributors stated that a winning poster they saw had ―a large image that demanded a second look, a clear and concise topic stated in large type, a[nd] relatively limited detail.‖ As another contributor put it, ―’Less is more’ is a good motto to follow when creating posters.‖ The ―Ask DAS‖ column is intended to target the interests, needs, and questions of members of the Decision Analysis Society. Special thanks go out to Lisa Kenney, Cameron MacKenzie, Sarah Timberlake, and Craig Trumbo for providing valuable insights and advice. If you have ideas or questions that you would like us to deal with in future Ask DAS columns, please just send us an email ([email protected], [email protected]).
Page 31 Volume 32, Number 1, April 2013 Decision Analysis Today
Editorial Team
Research: DA Around the World: Dr. Debarun Bhattacharjya Dr. Matthias Seifert IBM T. J. Watson Research IE Business School Center [email protected] [email protected]
Ask DAS: Editor Assistant: Mr. John Coles Ms. Elizabeth Newell SUNY University at Buffalo SUNY University at Buffalo [email protected] [email protected]
Ask DAS: Mr. Florian Federspiel IE Business School Co-Editor: (Instituto de Empresa) Dr. Heather Rosoff ffederspiel.phd2014@studen University of Southern t.ie.edu California, Sol Price School of Public Policy and CREATE [email protected] DA Practice: Dr. Bill Klimack Chevron Corporation Co-Editor: [email protected] Dr. Jun Zhuang SUNY University at Buffalo [email protected]
DA Around the World: Dr. Alec Morton London School of Economics [email protected]
Page 32 Volume 32, Number 1, April 2013 Decision Analysis Today
DAS Officers DAS Council
President: Phil Beccue Jeffrey Keisler Independent consultant
College of Management [email protected] University of Massachusetts Boston [email protected] Seth Guikema Whiting School of Engineering VP/President-Elect: Johns Hopkins University Eric Bickel [email protected] College of Engineering University of Texas at Austin [email protected] Victor Jose McDonough School of Business Georgetown University Past President: [email protected] Vicki M. Bier Industrial and Systems Engineering University of Wisconsin Gilberto Montibeller [email protected] London School of Economics and Political Science [email protected] Secretary-Treasurer: Yael Grushka-Cockayne Darden School of Business Larry Neal University of Virginia Chevron Corporate [email protected] [email protected]
Social Media Officer/Webmaster: Canan Ulu Jay Simon McCombs School of Business Defense Resources Management Institute University of Texas at Austin Naval Postgraduate School [email protected] [email protected]
Page 33