Axelrod’s famous book "The Evolution of Cooperation" has been one of my all-time favorites for a long time.
SmartMobs now reports via Wired that a new successful strategy has been found for the iterated prisoners’ dilemma that is different from Tit-for-Tat.
This strategy basically steps out of the original system by having prior knowledge of the strategies of other entrants, by attempting to recognize other entrants during the game, and then behaving differently. Clever! and obviously much closer to reality than the original setup.