Following is the next in Richard Rapp’s series of articles on compensation. – Bruce
Economists and evolutionary biologists share an interest in a model of incentives known as “the prisoner’s dilemma.” It is a game for two players with the payoffs arranged so that each player does better by behaving selfishly but the two would collectively be better off if each behaved altruistically. Two crooks are questioned separately about a crime they committed together. Each one gets a reduced sentence by ratting out the other (“defecting”) but if they both keep their mouths shut (“cooperating”) they avoid jail completely. When played repeatedly by the same players, “iterated prisoner’s dilemma” (IPD) enables learning by experimentation as each player reacts in the next game to the other player’s decisions in past games between them.
Since the 1950s the prisoner’s dilemma model has been a mainstay of game-theoretic research in an extraordinarily wide range of settings. Some examples: Early in its history, the prisoner’s dilemma was a strategic metaphor for the thermonuclear arms race of that era. “Defection” in this context meant building an H-bomb arsenal and “cooperation” meant resisting that temptation.
Later on, biologists began to see IPD as a means for thinking about problems like the evolution of symbiosis (Crocodile: “Should I dine on that bird that’s cleaning my teeth?”) and altruistic behavior in animals. Economists wonder about the prisoner’s dilemma-like incentives of an adversarial legal system in which hiring lawyers is somewhat like building H-bombs.
In IPD (as distinct from the one-shot version) players are thinking about how current moves may affect an opponent’s future moves: “Can I teach him to cooperate by cooperating after he does?” In the one-shot game, defection is the dominant strategy – if your opponent defects when you do not, you’re sunk. But when the game is played repeatedly, if both parties can develop a habit for cooperation they will achieve the best long-term outcome.
For the past thirty years is has been widely understood that the winning strategy in IPD is “tit-for-tat” – a cooperative strategy, notwithstanding its spiteful-sounding name. “Tit-for-tat” means a strategy of cooperation, retaliation, forgiveness and consistency in response to the other player’s actions. In experimental IPD tournaments, tit-for-tat proved superior and this widely-reported outcome has been a source of encouragement to those natural scientists and social scientists that see cooperation and altruism rather than selfishness as the successful evolutionary path in both nature and human relations.
Trouble is, it turns out that’s wrong.