Sunday, March 11, 2007
You Know, If Egoism Means I Do What I Prefer To Do, It's Vacuously True
Repeat slowly after me: game theory does not have to come with a set of claims about human nature attached, since it calculates strategies for interacting with other players once preferences and outcomes are held fixed, which means that you have to have some other claims about the preferences people have, the structure of institutions the players are interacting in and the pay-offs attached to the relevant outcomes within those institutions. Therefore, game theory cannot, by itself, show anything about the world: it depends on judgments about people's preferences, the institutions they are working within and the pay-offs that those institutions assign to outcomes - as well as, of course, the claim that people will or ought to rationally utility maximize - in order to have any predictive or normative power. Therefore, if Adam Curtis thinks that game theory by itself could legitimise the view that social systems are best organised by self-interest's invisible hand, he is mistaken, since what's doing the work there is the view that humans have self-interested preferences and that our interactions are such that they go better when we act self-interestedly, none of which bears any relationship to the claim that we can model strategies for interaction within institutions once we know people's preferences and the pay-off structure they face. I stopped watching after about twenty minutes of talking over the television.