Thursday, November 17, 2005


ROADEF Celebrates EURO30

Tomorrow, in Paris, the French operational research society will celebrate the thirtieth anniversary of the founding of EURO, the Association of European OR Societies. This year was also the thirtieth anniversary of my degree in "Administrative Science, Mathematics, and Statistics" at Yale. So, I'd like to pull together some thoughts on the subject that I've accumulated over the years on what OR is, was, and might be.

My initiation to this field came at Yale, so let's start with what Yale has to say:

The discipline of operations research is concerned with the organization and management of productive activities. Operations research includes mathematical analysis of techniques of optimization, efficient organization of production, calculation of operating characteristics of complex systems, and application of these and other techniques to operating problems throughout business and government. The subject, also known as management sciences, is closely related to computer science, economics, statistics, engineering, and pure and applied mathematics.

Operations research developed out of an awareness that effective decision making in organizations requires more than intuition. To practice operations research, one must identify the objectives of the operation under study, describe alternative actions, define measures of effectiveness for them, create a model of the system under study, and select the action that best meets the stated objectives.

Operations research has four major subfields. Mathematical programming concerns the optimal operation of systems with many variables that are linked by simple relationships. Stochastic processes describes the evolution over time of systems whose "laws of motion" are affected by chance. Game theory describes models of cooperation and competition between members of an organization or participants in a market. Production and inventory control is a family of models that applies to manufacturing and service systems. Operations research can prepare the mathematically inclined student for a career in the management of technology or in administration, for graduate study in the mathematical sciences, or for graduate study in management. Yale College offers no major in this subject.

A bit long and ponderous for a subject hardly taught any more. I'm not sure I totally agree with the productive activities, either: bombing u-boats (one of the first, defining O.R. projects) was productive in what sense?

No major in this subject. Sniff. How sad. An article in the Yale Daily News chronicles what happened. Once the SOM (School of Organization and Management) got going and decided to start selling MBA instead of MPPM degrees(am I right about the timing of events?) and drop "Organization and" from their name, they kicked OR out to the Faculty of Arts and Sciences in 1989. By 1992 they stopped accepting PhD applications, and attached OR to Applied Mathematics until all the majors graduated. The Only OR prof for the past thirteen years has been Eric Denardo (my faculty advisor in 1975, incidentally.)

I can easily list some top OR people who have taught at Yale. About a quarter of the recipients of the Lanchester award for O.R. publication have taught at Yale (although not necessarily when producing their award-winning writing). Why, even on Yale's MBA research page today, they show "This model formulated by Professor Edward Kaplan, winner of the two top honors in the operations research field, helps policy makers better allocate resources in the fight against the spread of HIV infections." Ward Whitt, Harvey Wagner, George Fishman, Martin Shubik, Edward Kaplan. Even one of the inventors of linear programming, T.J. Koopmans, was Alfred Cowles Professor of Economics at Yale from 1955 to 1980 (his linear programming began in the 40's, as did those of Kantorovich and Dantzig).

Oh well. What does Cal (Berkeley), where I earned an M.S. in Operations Research, have to say about OR today?
Operations research emphasizes the basic understanding of the functioning of complex systems of technology and management through mathematical models for the purpose of predicting system behavior or optimizing system performance under economic and technological constraints.

That is shorter and to the point, although "emphasizes" is not quite the same as "is". It hasn't changed much in thirty years: same emphasis on mathematical models, although in my day we didn't touch a computer. One of the definers of operations research, West Churchman, taught across the street in the business school, but he had no contact with our department. OR at Cal was (and is) about mathematics, not about figuring out how to do better (or best).

Next stop, the professional society (12,000 members), INFORMS, which says succinctly : "Operations Research (OR) and the Management Sciences (MS) are the professional disciplines that deal with the application of information technology for informed decision-making." (emphasis mine). Their slogan: "The science of better." Maybe. How exclusive a claim to this territory can they defend? Isn't accounting information technology for informed decision-making? Nutritional content labelling of foods?

From my point of view, to capture both the generality and the specificity, and to acknowledge the heritage from economics, it should have been called "performance research": the investigation of how to control the performance of systems, and particularly how to improve human judgement and choice of action.

In "Game Theory and Operations Research: Some Musings 50 Years Later", Martin Shubik included some helpful annecdotes. One was
...George Feeney’s experience at Stanford Research involved explaining what operations research was to one of their vice-presidents who reacted immediately. “I see”, he said, “operations research involves utilizing big minds to work on small problems.”

Jay Forrester made a similar assessment in his talk on the "Beginnings of System Dynamics" :
As I looked at the field of operations research [in 1956!], it seemed interesting; it no doubt was useful; but it was not working with issues that made the difference between corporate success and failure. Operations research did not have that practical importance that I had always worked toward.

Ackoff acknowedged that corporate O.R. was vanishing in his classic "The Future of Operational Research is Past", (1979).

More Shubik : "The great power of the military operations research at that time (as is shown in Morse and Kimball, 1970, Kirby, 2000, Blackett 1962 and others) was that the operations researchers were closely concerned with formulating the problems and evaluating how important they were. Similar observations combined with questions concerning the obtaining of data within an organization have been raised in the discussion of the CONDOR report (Committee on the Next Decade in Operations Research)."

Indeed: if they can keep you from asking the right question, they don't have to worry about the answers. Figuring out what the right performance measures are.

Perhaps this is because, as Shubik notes, during institutionalisation in America O.R. was appropriated by some excellent mathematicians:
At Princeton there was some direct talk about operations research per se and only a few of us were aware of the newly formed Operations Research Society. But in a few years around Fine Hall (and elsewhere) much of the mathematics relevant to its development was being developed. Among the visitors, students and faculty with whom I interacted were Bellman, Feller, Gomory, Karlin, Kemeny, Kuhn, McCarthy, Mills, Minsky, Nash, Scarf, Shapley, Morgenstern, Tucker and Tukey and von Neumann who was at the institute. Dynamic programming, linear programming, convex programming, integer programming, inventory theory, game theory, artificial intelligence and applied probability were all being developed.

He continues a bit farther on:

As my main purpose is to cover the evolution of the relationship between game theory and operations research rather than to review all of operations research I limit my broader comments on operations research with a few Panglossian remarks. It is my belief that operations research has been so successful that it may have put itself out of business, at least in it’s easy to recognize sense. It has succeeded to the extent that it is taught in a more or less routine and watered down manner in every business school. Linear programming, queuing studies and elementary competitive models go with the turf.

Consulting firms flourish. A variety of military operations research firms make a good living off weapons analysis; specialist firms such as Fair Isaac have found a niche in credit evaluation; McKinsey dispenses generalized operations research and management science under a variety of names. RAND, Stanford Research Institute, Los Alamos and many others may not be in their heyday as they were when operations research was young and 100% improvement in performance almost anywhere was to be expected; but they still produce. Small groups of academic consultants provide consulting services in the design of auctions or in the structuring of games to study market structure.

I believe to this day that eventually a good planning department of any major corporation should utilize a simulation of the corporation and its market as a device to organize perceptions and data, to help formulate questions and to facilitate communication among practitioners.

To be continued...I didn't make it to the celebration, BTW, my train was late and missed the connection. Furthermore, the SNCF only refunded the unused portion of my ticket, in effect charging me for riding in their train which did not take me where I wanted to go but wasted my time.

StumbleUpon Toolbar Stumble It!
Comments: Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?