Hacker News Reader

Comment by suprememoocow | Hacker News Reader

For a contrary view on game theory, I highly recommend watching the excellent BBC documentary series "The Trap" by film maker Alan Curtis.

http://en.wikipedia.org/wiki/The_Trap_(television_documentar...

In the first episode, Curtis examines the rise of game theory during the Cold War. It discusses John Nash's paranoid schizophrenia and shows footage Nash in his later years acknowledging that his paranoid views of other people at the time were false.

View on HN

There are two sides of Nash's work. One is the actual math, which can be evaluated without considering the credibility of the author. The other is Nash's attempt to use it to describe the real world, which necessarily involves assumptions and simplifications. This second side is where the obvious problems are.

Game theory remains popular because people can easily take the underlying math and apply it with their own assumptions. Of course, if you make too many simplifications you risk achieving a near-useless comic-book view of reality (literally, as in the blog post here).

> Of course, if you make too many simplifications you risk achieving a near-useless comic-book view of reality (literally, as in the blog post here).

For a moment there, I thought you were going to invoke Rorschach as an example.

Game theory doesn't require that people be selfish bastards. For example, in the Prisoner's Dilemma, it's perfectly fine to assume that each prisoner places a high value on the other person remaining free. That changes the payoff matrix, and the resulting matrix can still be analyzed by game theory (which will then predict that the prisoners will cooperate, since in game theory terms it won't be the "prisoner's dilemma" anymore).

The common mistake people make on game theory is the assumption that people will always use a hill climbing algorithm to find equibrilium, instead of considering the global optima. Hofstadter discusses this with regard to the "hyperrational" solution of cooperation-cooperation in the Prisoners dilemma.

I did a little googling, looks interesting but I couldn't find Hofstadter's actual work on this...where do I need to look?

Reply on news.ycombinator.com