Game theory seems to set up a moral versus practical dichotomy. The typical Prisoner's Dilemma game is set up so that "each player pursuing his own self-interest leads both players to be worse off than had they not pursued their own self-interests (Wikipedia: Game_theory)."
There, of course, is strategical value in trying to understand what the other person will do next, but what is the proper approach?
I would assume first consider the context, determine would could be done morally, and consider those options along with the corresponding actions of the other party but am hoping for more detail.
A valid approach for what? In what context? Game theory is an abstract mathematical model of interactions between humans. There are lots of insights that you can get from analyzing a problem from a very theoretical abstract perspective, that's what game theory does, it allows you to analyze human interactions from the ivory tower of mathematics. In that sense it is valid, but one should not confuse the map with the territory. Using frictionless pulleys to learn the principles of physics, or to analyze pulley systems abstractly is valid. This does not mean that one should treat actual pulleys as if they were frictionless in all aspects of our interaction with them. The same applies to game theory, if modeling an interaction between humans as a game theory situation gives you insight, then it is useful and valid, but if you start treating humans as if they were the one dimensional utility maximizing agents of game theory, then that is an improper use of a model.
As an aside game theory is not limited to the prisoner's dilemma, it studies many different situations where two utility maximizing agents (which may or may not be humans) are engaged in competition or cooperation with each other, and are trying to maximize their own individual outcome based on what they think the other agent is going to do. The prisoner's dilemma is only one of those situations. Also note that the context of the prisoner's dilemma is a very contrived situation which is very unlikely in reality. In reality you know the people you interact with, you communicate with them, you are likely to interact many times with them, they will talk to their friends and business associates about their interactions with you, etc. So the situation of the prisoner's dilemma almost never applies in reality.
answered Oct 06 '10 at 16:21
The book, Evolution of Cooperation describes several historical oddities that can be explained through the application of the iterated prisoner's dilemma. In particular, he analyzes the static status of trench warfare in World War one, among several other examples.
The prisoner's dilemma is a bit of a red herring in that it's not really a mathematical problem. The mathematical "optimal" solution is obvious. Yet, we all know examples where that would not be the chosen strategy. The reason is the assumptions made on the opposing player. If we trust the opposing player, we'll operate completely differently than if we don't trust the opposing player. And the nature of that trust will be a very important factor as to how the "game" is played. Bottom line, in order to "solve" the Prisoner's Dilemma, you need more context than just the pure mathematical/game theoretic formulation. This is due in part to the fact that game theory typically analyzes games with respect to a min/max strategy. (That is minimize all possible losses first, then maximize all possible gains.) If for some reason that is not the best strategy (due, for instance to additional information not captured in the game theoretic formulation, or knowledge that you have that your "opponent" will in fact be cooperating with you) then most standard game theory results will give you the wrong answer. As with any tool, if game theory is applied to an appropriate problem it can produce great insights into the problem. But, if the problem is a moral problem, a mathematical model is not going to help you out very much. This is a limitation of the theory, but every theory has limitations, and the knowledgeable thinker will take this into account when choosing a theory to tackle a problem.
Depending on your definition of game theory, it has huge applications. Differential game theory is used to analyze interactions between surface warfare ships, and to identify tactics and strategies appropriate for various tactical situations. Differential game theory coupled with genetic algorithms was used to identify new tactics for the F-22 Raptor when it was first built.
answered Oct 12 '10 at 22:38
John Hoffman ♦