To the Editor:

What a great interview with INFORMS President Brenda Dietrich! ["Dietrich Gets Down to Earth, " OR/MS Today, December 2006] What a great practitioner! Finally someone who knows about dealing with people! Finally a realistic approach to some long-term strategies! Finally, someone who knows that marketing needs to be done by a full-time professional who knows how people work in the real world! She hit the nail on the head about the abortive Science of Better campaign.

She is right about the research. The field in which I work does not fund research beyond basic inquiry unless a product developer is willing to accept the effort as a positive ROI. Too often, what passes for "INFORMS research" is just academic intellectual masturbation.

And she is right about other fields taking over the INFORMS arena. I don't think statistics or computer science per se will take over the INFORMS arena. However, I do think data mining, computer-based decision support — especially modeling and simulation (M&S) — pose bigger threats. I myself started the migration to decision support and M&S several years ago because that's where the interesting new work is being done, and INFORMS was simply not open to my ideas.

So here's a freebie for INFORMS: Dietrich is also right that INFORMS is "clueless" regarding human behavior, but they need to get hot solving this problem because it is the next big wave. DOD, Homeland Security, other government venues, as well as and various commercial venues, are going to the gaming industry for their human behavior modeling capabilities. Artificial intelligence failed to deliver, and the present state of human behavior modeling, even in the gaming world, is rudimentary at best. The gaming industry's approach is mostly trial and error, but eventually by dint of effort, they will codify on a decent, productive model. INFORMS has a twofold advantage in this contest. INFORMS knows how to work with other disciplines. INFORMS knows how to avoid excessive trial and error in constructing a practical model.

I read the descriptions of the 38 decision analysis tools in the software survey ["Improving Hard Decisions," OR/MS Today, December 2006] and none included any specific mention of psychology or modeling human behavior.

Unlike WWII where the O.R. community stepped up, the performance of O.R. in helping to solve basic problems in Iraq has been less than stellar. Human behavior-based M&S of sectarian violence and terrorism is a wide-open field. Let's not yield the next big wave because it's a hard problem. As the cover says, aren't we all about solving puzzling problems?

Ralph Nebiker
San Diego, Calif.

To the Editor:

I was pleased to see the article by John Kettelle ["When Three's Not a Crowd," OR/MS Today, October 2006] and thought that your readers might like to know that there are also other approaches to employing a computerized third party for group decision making and negotiations. I agree with much of the article, especially the huge benefit that a computerized third party can potentially bring to complex negotiations, so aptly illustrated by the examples provided. However, I have come to some different conclusions about how to achieve those benefits in practice.

The article centers on the so-called "equi-max" solution, which is essentially the same as "maximize the minimum gain" as first described by Thiessen and Loucks (1992). Raiffa (1996) lists this algorithm among several other methods for achieving fair optimal solutions. His analysis makes clear that none of these methods can be categorically declared best. However, Raiffa's preferred method is similar in some respects to algorithms implemented in the Smartsettle eNegotiation system (www.smartsettle.com), which is the result of my own research, which began at Cornell University (Thiessen, 1993).

The process described in the article's hypothetical pizza problem attempts to achieve both fairness and efficiency in one fell swoop. Recognizing that this is not realistic in practice, the remainder of the article describes a different process that depends on negotiators revealing their minimum requirements for a deal (their best alternative to no action (BATNA)) and other aspects of their preferences. Real negotiators are reluctant to reveal this information when it is possible for another party to take advantage of it. Even if parties try to be transparent; non-linear preferences, interdependencies, changing circumstances and other factors continue to be barriers to accurate preference representation, especially when parties have no idea what the final solution might be.

Smartsettle addresses these problems by incorporating multivariate blind bidding along with optimization using a flexible and comprehensive interface that can represent even the most complex problems. Essentially, the entire process is collaborative, but no potential solution is revealed or imposed on the parties until they both accept it. There are no "surprises" prior to parties reaching at least a tentative outcome, and thus no need for a "digesting" phase, during which things might go bad.

Any attempt by a party to game the system by misrepresenting preferences simply generates unacceptable suggestions. Smartsettle also does not need to inhibit the process by letting parties know whether or not there is "room for a deal" before they actually have a deal. In impasse situations that must resolve, parties can agree to instruct an arbitrator to finalize the negotiation after a deadline by simply picking one of the two most acceptable packages (similar to "baseball arbitration" or "last best offer").

Using the pizza problem as an example [Tom and Dick buy a pizza that is half mushrooms and half pepperoni. Tom likes mushrooms 1.5 times more than pepperoni. Dick likes pepperoni 9 times more than mushrooms.], Smartsettle could be used by Tom and Dick to generate the same solution as identified in the article (i.e., in one fell swoop), but it would require a degree of cooperation and trust that are rarely found in real-life negotiations. In practice, parties using the Smartsettle system do not need to accept conditions of fairness prior to negotiation. An unbiased computer guiding the process ensures that the parties themselves determine the outcome and are not compelled to accept anything they wouldn't consider fair.

For example, suppose that Tom and Dick first negotiate a tentative solution that gives them each 50 percent of each side of the pizza. In that case, an efficient improvement can be generated that gives all the mushrooms to Tom but only 17 percent of the pepperoni to him. (With a tentative solution in hand, parties have already agreed on how they will divide the benefits and only one more package needs to be generated to reach an efficient solution. It is not necessary to derive the entire set of deals that make up the Efficiency Frontier.)

On the other hand, if parties begin with the "normal" resolution, where Tom gets all the mushrooms and Dick gets all the pepperoni, that would also be the final solution, since it is already Pareto efficient. (Introducing a side payment doesn't change things either, unless the parties have different values of the entire pizza, i.e., one is a lot hungrier or richer.)

Smartsettle is not the only emerging implementation of this exciting technology, but it is a system that is working for any number of parties and ready to be challenged by serious problems.

Ernest M. Thiessen
Abbotsford, British Columbia

1. Raiffa, Howard, 1996, "Lectures on Negotiation Analysis, " program on negotiation at Harvard Law School.
2. Thiessen, E.M., 1993, "ICANS: An Interactive Computer-Assisted Multi-party Negotiation Support System," Ph.D. dissertation, School of Civil & Environmental Engineering, Cornell University, Ithaca, N.Y.
3. Thiessen, E.M., and D.P. Loucks, 1992, "Computer-Assisted Negotiation of Multi-objective Water Resources Conflicts," Water Resources Bulletin, American Water Resources Association, Vol. 28, No 1, pps. 163-177.

Table of Contents

OR/MS Today Home Page


OR/MS Today copyright © 2007 by the Institute for Operations Research and the Management Sciences. All rights reserved.


Lionheart Publishing, Inc.
506 Roswell Rd., Suite 220, Marietta, GA 30060 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com


Web Site © Copyright 2007 by Lionheart Publishing, Inc. All rights reserved.