THOMAS C. SCHELLING

Lucius N. Littaluer Professor of Political Economy, Emeritus, Harvard University
and Distingished University Professor, University of Maryland, USA


Excerpt from Thomas C. Schelling's interview:
2. Why were you initially drawn to game theory?

In the late 1940s and early 1950s I was a participant in international negotiations and developed some ideas that, when I left Government in 1953 to join the faculty of Yale, I thought I’d work on.  Eventually I published an “Essay on Bargaining” in which I explored the concept of “commitment,” how one may (or may fail to) adopt a bargaining position that he or she is obliged or motivated to adhere to and—crucially important—communicate that obligation or motivation in credible manner to another party.  I explored promises, threats, and bargaining tactics, looked at contract, reputation, appeals to a deity, physical positioning (burning bridges), uses of agents. 

I also looked at coordination when payoffs to two or more parties were identical but equilibria were multiple (or infinite), and when payoffs were asymmetrical but dependent on coordination.

I was convinced that coordination of expectations was often crucial to the completion of overt negotiation. 

I had just enough acquaintance with game theory to realize that what I was doing might be construed as game theory, but not enough to be tempted to formalize any of my ideas as game theory.  I late 1957, after I had finished that work, I came upon Games and Decisions (1957), by R. Duncan Luce and Howard Raiffa, and spent a hundred hours learning game theory. 

Immediately after, I spent eight months in London on a fellowship and worked further on the same kinds of ideas, with some intention to relate them explicitly to game theory, having absorbed enough from the Luce-Raiffa book to feel sure of the connection.  I felt that game theory, as I had come across it, was more abstract than it needed to be and could usefully be enlarged in scope to encompass strategies and tactics of bargaining, becoming empirical and historical as well as logical and mathematical.  I was presumptuous enough to subtitle my article, as published in the Journal of Conflict Resolution (1958), “Prospectus for a Reorientation of Game Theory.”  (Of course, nobody took it up.)  In that article I used a few payoff matrices, mostly 2x2, without thinking that that was what made it game theory. 

It was during my stay in London that I worked on a problem that intrigued me, which I called the “reciprocal fear of surprise attack,” and began to realize that the reciprocal deterrence between the USA and the USSR was the highest priority for my interests.  I had agreed to join the RAND Corporation for a year, and during 1958-59 I began a professional interest in nuclear-weapons policy and arms control that preoccupied me for the next decade.  I did, at RAND, produce a couple of articles (included in my 1960 book, The Strategy of Conflict) that were explicitly game theory.  Mainly I learned about nuclear weapons technology and policy. 

I then produced many articles and two books on nuclear defense strategy and arms control.  None of that work looked like “game theory,” although it was of the same nature as my earlier work that used some game-theoretical terminology and simple matrices.  I also worked on criminal coercion and “organized crime,” on somewhat game-theoretic strategies of self-management or self control, on racial segregation, and on multi-person interactions like self-confirming expectations, multi-person prisoners’ dilemmas, dual-equilibrium binary multi-person choices, and situations embodying the “fallacy of composition.”  I think the latter work can be construed as many-person game theory but in no way depended on formal game theory. 

I was surprised and somewhat perplexed when the Nobel selection committee for economic sciences awarded me the Bank of Sweden Prize in Memory of Alfred Nobel in 2005 “for having deepened our understanding of conflict and cooperation through game-theoretic analysis.”   I thought I had contributed to an understanding of conflict and cooperation;  I thought in 1958 my work might usefully be construed as game-theoretic analysis;  I thought in 2005 that, excepting possibly my development of coordination theory and “focal points,” what I did was not recognizable as game theory.  Maybe the committee was trying to redefine game theory by incorporating my work, much as I had futilely tried in 1958.  

Still, if game theory is to be identified with the formal logic of rationally identifying and choosing strategies in equilibrium, and its attendant definition of payoffs and use of matrix notation, I am a user of game theory, not a creator.  In most disciplines there is a distinction between “theorists” and professionals, or practitioners.  There are economists but also economic theorists;  there are sociologists but also sociological theorists;  there are statisticians and statistical theorists, even physicists and theoretical physicists.  But game theory, unlike economics, or sociology, or statistics or physics, has “theory” in its name.  We don’t have a term like “gameist” for the one who uses game theory, the way economists use economic theory without necessarily producing theory as economic “theorists” do. 

I believe I can distinguish what I do from what game theorists do in the following two ways.  One is that most game theory is concerned with identifying rational choice when the optimal choice depends on the choice, or choices, that another is, or others are, anticipated to make.  Except for my work on coordination theory, I have been, I believe, almost entirely concerned with how individuals rationally attempt to influence, not to anticipate, the choices of others.  And, second, while I have tried to identify the logic of tactics of influence—unilateral promises, reciprocal promises, threats, commitments, the elimination of options, hostages, contracts, appeals to higher authority, etc.—I have been mainly concerned with empirical (or sometimes fictional) and historical evidence of behavior.  I have been more “descriptive” than “normative”.

Most game theory considers such things as commitments, threats, promises, contracts, etc., to be either enforceable or not enforceable;  I have been mainly concerned with how and where and by whom in what institutional environments threats, promises, and commitments can be successfully incurred, or successfully bluffed, or successfully countered.  I am more social scientists than logician. 

Read the remaining part of Thomas C. Schelling's interview in Game Theory: 5 Questions, edited by Vincent F. Hendricks and Pelle Guldborg Hansen. The book is released in April 2007 by Automatic Press / VIP.

ISBN 87-991013-4-3
(paperback)
248 pages / $26 / £16

Available on Amazon:

AMAZON   US  CA  UK  DE

Check also for availability:


© 2006, Automatic Press / VIP