Wednesday, October 12, 2005

Game Theory

Michael Mandel (who is interesting despite the Business Week affiliation) writes:
Game theory is no doubt wonderful for telling stories. However, it flunks the main test of any scientific theory: The ability to make empirically testable predictions. In most real-life situations, many different outcomes -- from full cooperation to near-disastrous conflict -- are consistent with the game-theory version of rationality.
Okay, that's a fair criticism. Tyler Cowen responds, but I like what Russell Roberts has to say more:
Game theory generates no predictions about the real world but it is a useful way of organizing your thinking about various real-world phenomena. It's a language that helps avoid mistakes or confusion.
And I like Daniel Drezner's view even more:
[G]ame theory has the wrong name. It is a theoretical tool rather than a theory in and of itself. Because of this, Mandel is correct that it is possible to devise game-theoretic models that lead to contrasting predictions. However, the virtue of game theory is that the differences made in starting assumptions, institutional rules, and causal processes are laid bare. One can then argue about how realistic the assumptions, rules, and processes are.
Combine the two and you've got it exactly right. Game theory is not so much a theory of how people play games, or even necessarily how people should play games, but rather a way to formulate and talk about problems involving strategic interactions. Of course you can make a game with any solution, but why should that matter? You can develop lots of different models of spacetime but (presumably) only one fits reality. Mandel also writes:
Instead, the real progress in economics these days is coming not from game theory, which has been around for 60 or more years, but from the much newer fields of behavioral and experimental economics. Behavioral and experimental economics don't start with the assumption of rationality used by game theory.
But even behavioral economics is formulated largely in terms of game theory (see, for example, this Matthew Rabin paper). Of course some of the standard assumptions of game theory are dropped (rationality, common knowledge) but the framework of players, strategies and payoffs is kept. It's that framework that allows economic (and biological, and political) ideas to be phrased in a very precise manner, which is the real benefit of game theory.


Anonymous Battlepanda said...

From John Kay's "The Truth about Markets" p209:
"An economist specializing in game theory is in the wilderness with a friend when they see a bear approaching. The theorist pulls out his laptop computer and starts to compute an optimal strategy. His friend cries out in alarm, 'Run, there is no time to waste.' The economist smiles complacently. 'Don't worry,' he says, 'the bear has to work it out too.'"

The problem with game theory is how often it is misapplied or overapplied, especially with the old chestnut, the prisoner's dilemma. It leads people to feel like there is an inevitability to any course of events it is employed to explain, despite the fact that as you correctly pointed out, it can be used to explain everything from full-cooperation to the opposite.

8:31 AM  

Post a Comment

<< Home