2010-08-31

Aumann's View on Science and Game Theory

I was impressed by Robert Aumann, when I met him at the game theory conference in Brazil (link). And, after coming back to Japan, I got impressed again to see what he has written on the preface of the volume of his "Collected Papers." His view on science and game theory is very close to mine (perhaps, I have been unconsciously affected by his view or similar idea spread among theorists).

[A]ll the papers in the collection concern game theory, its applications and its tools. Beyond the subject matter, they also share a common methodological theme: they deal with relationships. Science is often characterized as a quest for truth, where truth is something absolute, which exists outside of the observer. But I view science more as a quest for understanding, where the understanding is that of the observer, the scientist. Such understanding is best gained by studying relations - relations between different ideas, relations between different phenomena, relations between ideas and phenomena.
(...)
Indeed, the idea of relationship is fundamental to game theory. Disciplines like economics or political science use disparate models to analyze monopoly, oligopoly, perfect competition, public goods, elections, coalition formation, and so on. In contrast, game theory uses the same tools in all these applications. (...) Perhaps the most exciting advance in game theory in recent years has been the connection with evolution: The realization that when properly interpreted, the fundamental notion of Nash equilibrium, which a priori reflects the behavior of consciously maximizing agents, is the same as an equilibrium of populations that reproduce blindly without regard to maximizing anything.
Aumann's message forward to Two-Sided Matching: A Study in Game-Theoretic Modeling and Analysis by Roth and Sotomayor (1990), which he describes as a book chronicles one of the outstanding success stories of the theory of games,  is also insightful. I again share his view on evaluating the good "matching" of theory and practice.

The theoretical part of the story begins in 1962, with the publication of the famous Gale-Shapley paper, "College Admissions and the Stability of Marriage." Since then, a large theoretical literature has grown from this paper, which is thoroughly covered in this book. But the most dramatic development came in 1984, when Roth published his discovery that the Gale-Shapley algorithm had in fact been in practical use already since 1951 for the assignment of interns to hospitals in the United States; it had evolved by a tirial-and-error process that spanned more than half a century.

No comments: