Repeated Games
Repeated games are sequential interactions in which the same players face the same strategic situation multiple times, with the outcomes of each round influencing the incentives and possibilities of all subsequent rounds. Unlike one-shot games, where defection is often the dominant strategy and cooperation is fragile, repeated games create a shadow of the future — the possibility of future retaliation or reward that can sustain behavior that would be irrational in a single encounter.
The framework was formalized in the 1950s and 1960s by mathematicians including John Nash, Robert Aumann, and Lloyd Shapley, who recognized that repetition transforms the strategic structure of a game in ways that one-shot analysis cannot capture. The central insight is that repetition converts a game of conflict into a game of trust: the threat of future punishment can enforce cooperation today, even when no external enforcement mechanism exists.
The Folk Theorem: Cooperation as a Equilibrium
The Folk Theorem is the foundational result of repeated game theory. It states that in an infinitely repeated game, virtually any outcome that is individually rational for all players can be sustained as a Nash equilibrium — provided that players value the future sufficiently. The theorem is called 'folk' because it was understood informally by game theorists long before it was formally proven.
The logic is disarmingly simple. Consider the Prisoner's Dilemma played repeatedly. In a one-shot game, defection is dominant: each player does better by defecting regardless of what the other does. But in an infinite repetition, a player who defects today can be punished by the other player defecting forever after. If the short-term gain from defection is smaller than the long-term loss from perpetual mutual defection, then mutual cooperation becomes self-enforcing. No external authority is needed. The equilibrium is maintained by the Grim Trigger strategy — cooperate until the opponent defects, then defect forever.
The Folk Theorem reveals a deep connection between time, trust, and cooperation. It shows that the problem of social order is not primarily a problem of altruism or shared values. It is a problem of discount factors — of how much players value the future relative to the present. When the future matters enough, cooperation emerges not despite self-interest but because of it.
Strategies and Their Dynamics
The most studied strategies in repeated games are reciprocity-based. Tit for Tat, as demonstrated in Robert Axelrod's tournaments, cooperates on the first move and then mirrors the opponent's previous action. It is nice, provocable, forgiving, and clear — properties that make it both effective and evolutionarily stable. The Grim Trigger strategy is less forgiving: it cooperates until the first defection, then punishes forever. Grim trigger sustains cooperation under weaker conditions than tit for tat but is fragile in noisy environments, where a single misperceived move can lock both players into perpetual defection.
More sophisticated strategies use higher-order reasoning, attempting to model the opponent's beliefs and intentions. These strategies perform well in laboratory settings but face a computational burden: modeling another agent's mental state requires recursion, and the depth of recursion is limited by both cognitive capacity and the opacity of real-world interaction. In practice, simple reciprocity strategies often outperform more complex ones because they are legible — opponents can learn to cooperate with them.
The dynamics of strategy selection depend critically on the discount factor, which measures how much a player values future payoffs relative to current ones. When the discount factor is high — when the future matters — cooperation is stable. When it is low — when players are impatient or the probability of continued interaction is small — cooperation collapses. This is why cooperation is easier to sustain in small, stable communities than in large, anonymous markets; in long-term employment relationships than in gig work; in diplomatic alliances than in one-off negotiations.
Applications and Extensions
Repeated games model a vast range of social phenomena. Repeated interaction in markets sustains reputation effects: a firm that cheats its customers today loses future business. In international relations, the threat of future retaliation sustains arms control agreements. In biology, repeated encounters between the same animals sustain reciprocal altruism. The Ultimatum Game, when played repeatedly, produces different outcomes than in its one-shot form: proposers learn what responders will accept, and responders learn what proposers will offer.
The framework extends beyond two-player games. In public goods games with repeated interaction, free-riding can be suppressed by the threat of social exclusion. In network games, the topology of interaction determines which pairs of agents interact repeatedly and which do not, and this topology shapes the global pattern of cooperation. The same game, different network, different outcome.
The Folk Theorem is often read as a mathematical curiosity — a technical result about Nash equilibria in infinite games. This is a failure of imagination. The theorem is a structural explanation of how civilization is possible. Every handshake agreement, every reputation system, every unwritten rule of fair dealing is a folk theorem in action. The insight is not that cooperation is mathematically possible. It is that cooperation requires no miracle — only repetition, patience, and the credible threat of consequences. The shadow of the future is not a metaphor. It is the infrastructure of social order.