The simplest strategy in a famous contest was just four lines of code.
Two prisoners each do better by betraying the other, yet both end up worse off than if they had kept quiet. That's the prisoner's dilemma. When Robert Axelrod ran a tournament of strategies for playing it again and again, the winner was the shortest program entered: cooperate first, then copy your opponent.
The game was devised in 1950 by Merrill Flood and Melvin Dresher at the RAND Corporation, and Albert Tucker later gave it its prison story. Two gang members are held separately. If both stay silent, each gets a year. If one testifies and the other stays silent, the witness walks free and the other gets three years. If both testify, both get two.
The logic is brutal. Whatever your partner does, you do better by testifying: freedom beats one year, and two years beats three. So a rational, self-interested player always defects, and when both reason this way they each serve two years, worse than the one year they would have got by cooperating. Mutual defection is the game's only stable outcome, even though everyone would prefer mutual silence. The same structure appears whenever cooperation pays but is hard to guarantee.
Things change when the game repeats. In the very first informal trial at RAND, two colleagues played a hundred rounds and often cooperated, and John Nash remarked that behaviour in a repeated game can differ from a single round. Later, Robert Axelrod invited academics to submit computer strategies for an iterated tournament. Greedy strategies tended to do badly over time. The winner was Anatol Rapoport's tit for tat, the simplest entry at four lines of BASIC: cooperate on the first move, then do whatever the other player did last time. Axelrod described the results in his 1984 book The Evolution of Cooperation.
Axelrod argued this shows how cooperation can arise from purely selfish starting points, through repetition and reciprocity. A variant that occasionally forgives a betrayal can do even better, because it escapes endless cycles of revenge. Much later, in 2004, a team from the University of Southampton topped a new tournament by entering 60 programs that recognised and colluded with each other. The deeper lesson holds: trust is easier to build when you expect to meet again.
Source: Wikipedia — Prisoner's dilemma · Text summarised from Wikipedia (CC BY-SA 4.0)
