Review: The Evolution of Cooperation by Robert Axelrod
Axelrod had a very nice idea for an experiment. Take a much-studied behavioural model – the (iterated) Prisoner’s Dilemma from game theory – and run a computer tournament between alternative strategies to try to understand the mechanisms driving cooperation.
Programs of various degrees of sophistication and complexity were submitted from experts in the field and the laurels were awarded to perhaps the least sophisticated and least complex among them. The entire strategy is clear from its name: TIT FOR TAT.
The book describes the experiment, shows how the Prisoner’s Dilemma is a reasonable model for many kinds of human interactions and explores the reasons why TIT FOR TAT was so successful. It shows, for example, soldiers discovering the strategy for themselves during trench warfare in World War One without any face-to-face negotiation.
When run in populations, it was shown to be evolutionarily stable. In other words, when discovered and entered even in small numbers into the population, it grows in influence, driving out more selfish strategies.
A second book delved deeper and an almost-as-simple program – TIT FOR TWO TATS – emerged as an interesting alternative strategy.