Evolving diverse populations of Prisoner’s Dilemma strategies

It is common for evolved populations of Iterated prisonerpsilas dilemma to become homogenous with most of the strategies either identical or similar to each other. As fitness is usually based on play with other members of the population, this favors the evolution of strategies which score well when playing themselves or close mutants of themselves. Also, populations tend to change considerably over time. New strategies arise and take over. A population consisting entirely of a highly cooperative strategy like tit-for-tat can become a population consisting entirely of a highly uncooperative strategy like always-defect. This study uses an experimental setup which incorporates geography in an attempt to evolve a diversity of coexisting strategies. The resulting populations are analyzed using prisonerpsilas dilemma fingerprints and found to be both diverse and ldquostablerdquo in the sense that they remain highly cooperative over time.