Robert Axelrod — The Evolution of Cooperation (1984)
Axelrod invited game theorists and social scientists to submit strategies for a computer tournament of iterated Prisoner’s Dilemma — a scenario where two actors independently choose to cooperate or defect, with mutual cooperation producing the best collective outcome but individual defection offering a short-term advantage. The question: what strategy produces the most cooperation over repeated interactions?
The winner across two tournaments was tit-for-tat: cooperate on the first move, then mirror whatever the other player did last round. Despite its simplicity, tit-for-tat outperformed every more elaborate strategy. Axelrod’s analysis identifies why: it is nice (never defects first), retaliatory (immediately punishes defection), forgiving (returns to cooperation once the other player does), and clear (legible enough that other players can learn to cooperate with it).
From this result, Axelrod draws a general theory of how cooperation evolves without authority:
- Shadow of the future: cooperation is more likely when actors expect to interact repeatedly — the anticipated value of future cooperation outweighs the short-term gain from defection
- Reciprocity: sustained cooperation requires that defection be punishable and cooperation be rewarded — tit-for-tat structures this automatically
- Clustering: cooperative strategies can invade populations of defectors if cooperators can form clusters — they benefit from each other without being exploited
- Robustness: once established, cooperation based on reciprocity is stable against invasion by defectors
The book’s larger claim is that cooperation can emerge bottom-up, through iterated interaction and reciprocity, without requiring a Leviathan to enforce it — a result with significant implications for political philosophy, evolutionary biology, and institutional design.
For OM: The Evolution of Cooperation establishes the game-theoretic foundation that the Moloch critique engages. OM’s point is not that cooperation theory is wrong — tit-for-tat is real and important — but that it cannot address failures whose source is the game board itself rather than the players’ strategies. When the rules of the game encode ontological violence (as in white supremacy, analyzed through Afropessimist frameworks), better coordination mechanisms cannot resolve it. The question shifts from “how do we cooperate within this game?” to “how do we change the conditions under which the game is played?” This is the limit cooperation theory cannot cross. See also: Six Advances in Cooperation Theory (2000) for Axelrod’s later synthesis.