Computer Tournaments in N-person Stochastic Iterated Prisoner's Dilemma

This paper generalized the strategy selection problem in 2-person iterated prisoner's dilemma to n-person stochastic cases. The concept of n-person stochastic iterated prisoner's dilemma was proposed to accommodate to new game problem caused by technology advancement. Every player's history information of choice could be seen by all other players. At each round, two players were selected randomly to play the prisoner's dilemma game. This paper designed a computer tournament to investigate which is the best strategy and described in detail the design and experiments of the tournament. The experimental results showed that TIT-FOR-TAT is not always the best strategy, and the best strategy is closely correlated with the repeated game round and the strategy distribution