The Ethical Spectacle September 1995

Introduction: We Are All Prisoners

The prisoner's dilemma is a game invented at Princeton's Institute of Advanced Science in the 1950's. In the basic scenario after which it is named, two prisoners who the police know to have committed crime A, but whom they wish to convict of the more serious crime B, are held in separate cells and offered a deal:

Thus there are two choices--usually known as to cooperate, in this scenario remain silent, or to defect, which here means to confess. And there are four possible outcomes, depending on your partner's move: you may serve 0, 1, 2 or 3 years in prison.

Cooperation either means you serve one or three years. The results of defection straddle this: you may serve 0 or 2 years. Because you do not know whether you can trust your partner (there is no opportunity to communicate when deciding your move), most rational players will choose to defect in order to maximize the upside (0 years) and minimize the downside (only 2 years instead of 3). Yet the outcome consistently is better for two cooperating players than for two defecting players.

However, in a sequence of games (an "iterated prisoner's dilemma") something different may happen. One or both players may fall into a pattern called "Tit for Tat", in which cooperation is rewarded and defection punished. Effectively, this means doing on this move whatever your partner did on the last. In a computer tournament of programs playing the prisoner's dilemma against one another, held by political scientist Robert Axelrod in 1980, a four line program playing "Tit for Tat" beat out much more complex and sophisticated programs. Yet "Tit for Tat" can only draw; it can never score more points in the game (fewer years, in this scenario) than its partner. On the other hand, a player who, out of moral obligation or naivete, cooperates on every move no matter what the partner does (the All C strategy) will be ignominiously defeated. His partner has no incentive to cooperate, but can defect and earn the greater payoff on every move. The moral: cooperation is best, but only if defection is immediately punished. Axelrod coined the phrase "shadow of the future" to describe the force that keeps a player cooperating. Someone who knows he will never meet you again may have nothing to lose by betraying you; someone who will have to deal with you many times more may be deterred, for fear of retaliation. Thus the future has a longer shadow in the second case.

Human nature being what it is, in some prisoner's dilemmas. both parties will always defect (all D strategy), thus scoring worse than they would if they always cooperated.

The prisoner's dilemma is a simple but powerful idea; once you have hold of it, you see its applicability to every walk of life and all human experience. The prisoner's dilemma has been used to analyze problems in nuclear warfare, anthropology, biology and evolution. In the following essays, I discuss its applicability to love, business, law, politics and software development; introduce some variations on the theme, such as the Gandhi game and the scorpion player; and finally raise the question of whether it is possible to base an ethical system on the prisoner's dilemma.