The prisoner's dilemma is a game in game theory in which two players each have an incentive to act in their own interest, yet when both do so they end up worse off than if both had cooperated. It is the best-known example of how individually rational choices can produce a collectively poor outcome, and it is used to analyze a cartel, arms races, advertising battles, trade wars, climate agreements and many other situations.
The game was devised around 1950 by Merrill Flood and Melvin Dresher at the RAND Corporation, and the mathematician Albert Tucker gave it its story of two prisoners and its name. It is closely tied to the Nash equilibrium, formulated by John Nash in 1950, because its equilibrium is an outcome in which neither player can gain by changing strategy alone, even though both would prefer a different outcome.
The game and its payoff matrix
In the classic story, two suspects are held in separate cells and cannot communicate. Each is offered the same deal: confess and testify against the other, or stay silent. If both stay silent, each serves 1 year on a minor charge. If one confesses and the other stays silent, the confessor goes free and the silent one serves 10 years. If both confess, each serves 5 years. The table shows years in prison for Player A and Player B, in that order:
| B stays silent | B confesses | |
|---|---|---|
| A stays silent | 1, 1 | 10, 0 |
| A confesses | 0, 10 | 5, 5 |
Staying silent is called cooperating (with the other prisoner) and confessing is called defecting. The payoffs follow a fixed ranking that defines the game: the temptation of defecting against a cooperator (0 years) is better than mutual cooperation (1 year), which is better than mutual defection (5 years), which is better than being the lone cooperator (10 years).
Dominant strategies and Nash equilibrium
Each prisoner reasons the same way. If the other stays silent, confessing gives 0 years instead of 1. If the other confesses, confessing gives 5 years instead of 10. Whatever the other does, confessing is better, so confessing is a dominant strategy for both players (see strategies, payoffs and best responses). When both follow it, each serves 5 years: this is the game's unique Nash equilibrium.
Yet both would be better off staying silent and serving 1 year each. The cooperative outcome is a Pareto improvement on the equilibrium, since it makes both players better off, but it is not stable: each player, expecting the other to cooperate, gains by defecting. The dilemma lies in that gap between individual and collective rationality. Combined, the two prisoners serve 10 years at the equilibrium against 2 years under mutual cooperation.
Applications in economics and finance
- Cartels. Members of a cartel such as OPEC jointly gain by restricting output, but each has an incentive to cheat by producing more at the high price, which makes cartels unstable without enforcement.
- Pricing and advertising. Two firms in an oligopoly may each cut prices or spend heavily on advertising to win market share, leaving both with lower profits than if neither had.
- Arms races and trade wars. Countries may each build weapons or raise tariffs to gain an edge, ending with higher costs for all.
- Bank runs. Depositors who would all be safe if none withdrew may each rush to withdraw if they fear others will, causing a bank run; strictly this is a coordination game rather than a prisoner's dilemma, but it shows the same clash between private and collective interest.
Repeated games and tit-for-tat
The bleak result applies to a game played once. When the same players meet repeatedly with no known final round, cooperation can be sustained, because a player who defects today can be punished tomorrow. Around 1980 the political scientist Robert Axelrod ran computer tournaments in which strategies submitted by researchers played a repeated prisoner's dilemma against one another (see types of games in game theory). The winner, submitted by Anatol Rapoport, was tit-for-tat: cooperate in the first round, then copy whatever the other player did in the previous round. It never defected first, punished defection at once, forgave quickly and was easy for opponents to read.
Repeated interaction helps explain why reputation, long-term business relationships and trust matter in markets, why tacit collusion can emerge among a few firms that compete repeatedly, and why antitrust law treats agreements between competitors as serious offenses. If the number of rounds is known and finite, standard theory predicts defection in the last round and, by backward reasoning, in every round, although people in experiments often cooperate for a while anyway.
A social dilemma is the many-person version of the prisoner's dilemma: a situation in which each individual's best choice, taken alone, leads to an outcome that is worse for everyone than if all had acted differently. Each person gains by contributing less or taking more, but if all do so, the group loses.
Public goods are a central case. Each person benefits from clean air, national defense or a well-maintained open-source project whether or not they help pay, so the free-rider problem leads to underprovision. Common resources such as fisheries and groundwater face the tragedy of the commons, in which each user's extra catch or extraction depletes the stock for all. Climate change is the largest modern example: each country bears the full cost of its own emission cuts but receives only a small share of the benefit. Field studies, notably the work of Elinor Ostrom on community management of common resources, show that social dilemmas are often solved in practice through communication, monitoring, graduated sanctions and the reciprocity and norms described under social preferences and fairness, as well as through taxes, regulation and international agreements.
Ref 1100001727
Social Dilemma