The Evolution of Cooperation
Why enemies choose to cooperate
Description
During the First World War, on stretches of the Western Front, something strange kept happening between men who were supposed to be killing each other. Units dug in across a few hundred yards of mud would fall into unspoken truces: artillery fired at predictable times so the other side could take cover, snipers aimed to miss, ration parties were left alone at breakfast. Officers on both sides tried to stamp it out, rotating troops and ordering raids. But the pattern kept reasserting itself, because the same men faced the same men day after day, and both sides worked out that a shell not fired was a shell not received. Enemies, with every incentive to destroy each other, had quietly decided to get along.
That episode sits near the heart of a small, much-discussed book the political scientist Robert Axelrod published in 1984, The Evolution of Cooperation. Axelrod wanted to know something that runs under superpower arms races, business rivalries, and ordinary human quarrels alike: how can cooperation ever emerge among self-interested parties when there is no authority above them to enforce a deal? No referee, no police, no court. Just two sides, each better off if the other cooperates and each tempted to betray. On paper, the cold logic says betray. In the trenches, and in a lot of other places, it doesn't.
To get a real answer rather than a hopeful one, Axelrod did something unusual for a political scientist. He turned the problem into a game, invited the world's sharpest strategists to compete at it, and let a computer run the results thousands of times over. What came back was a winner nobody would have bet on — and a set of conditions that explain why the shooting sometimes stops on its own.
The question we’re asking : Can genuine, unforced cooperation emerge among self-interested rivals when no authority stands over them to enforce it?What we’ll see : A famous computer tournament, an almost embarrassingly simple winning strategy, and the conditions that let cooperation take hold and hold its ground.
Table of contents
01Chapter 1 — The trenches that stopped shooting
Axelrod builds his whole inquiry on one clean piece of logic that game theorists call the Prisoner's Dilemma. Two players each choose, at the same time, to cooperate or defect. If both cooperate, both do reasonably well. If both defect, both do badly. But if one defects while the other cooperates, the defector does best of all and the cooperator worst of all. Work through it and the trap snaps shut: whatever the other player does, each one is individually better off defecting. So two rational players both defect — and land in the outcome they'd both have preferred to avoid. That is the puzzle in miniature. Self-interest, followed faithfully, makes everyone poorer.
The trench truces of 1914-1918 are the Dilemma played out in flesh and mud. Each side would gain, in the short term, by firing while the other held back. Yet firing invited return fire, and the men knew it. What changed the arithmetic was not conscience but repetition. These were not strangers meeting once; the same regiments stared at each other across the same ground for weeks. A betrayal today would be answered tomorrow, and the day after that. Under those conditions, restraint quietly paid.
02Chapter 2 — A tournament for selfishness
Axelrod's idea was disarmingly direct. He would hold a tournament for strategies playing the iterated Prisoner's Dilemma, and he would invite experts — game theorists, psychologists, economists, mathematicians — to submit programs. Each program would encode a strategy: a rule for deciding, on each move, whether to cooperate or defect, based on the history of the game so far. Every program would be matched against every other, against a twin of itself, and against a program playing at random, over two hundred moves per encounter. The scores would be added up. No morality, no instructions to be kind — just accumulate as many points as possible.
Fourteen programs came in for the first tournament, some of them elaborate. There were strategies that tried to model the opponent, strategies that probed for weakness by slipping in the occasional defection, strategies with long memories and clever contingencies. And then there was the shortest program of the lot, submitted by the mathematical psychologist Anatol Rapoport. It was called Tit for Tat, and it did almost nothing. On the first move it cooperated. After that, it simply copied whatever the other player had done on the previous move.
03Chapter 3 — Why nice beats clever
Axelrod sifted the results and pulled out the traits that separated the winners from the losers. The first was being what he called nice: never being the first to defect. The top eight strategies in the first tournament were all nice, and every strategy below them defected first at some point. It sounds soft, almost naive. It was in fact the single best predictor of a high score, because a nice strategy meeting another nice strategy locks into mutual cooperation and stays there, round after profitable round, while the clever defectors keep triggering retaliation and bleeding points.
But niceness alone is not enough — it would be a doormat. The second trait was being provocable, or retaliatory: responding to a defection promptly, so no one could exploit you for free. Tit for Tat answers a betrayal on the very next move. That immediate response teaches an opponent that cheating does not pay, without any need for escalation or grudge. A strategy too forgiving invites relentless exploitation; a strategy too eager to punish starts wars it cannot end.
04Chapter 4 — The shape cooperation needs
The deeper thing Axelrod is arguing is that cooperation among egoists does not require goodness, government, or even friendship. It requires the right structure — and above all what he calls the shadow of the future. The reason Tit for Tat works, the reason the trenches went quiet, is that the same parties expect to meet again, and the value of future exchanges outweighs the one-off gain from betrayal today. Shrink that shadow — make the end of the relationship visible, or the next encounter unlikely — and defection floods back. Cooperation is not a trait people have. It is a thing that certain conditions produce.
This reframes the whole question the book opened with. We tend to assume that getting selfish parties to cooperate takes a strong hand above them: a state, a contract, a policeman. Axelrod's tournaments suggest that under the right conditions, self-interest does the work by itself. Two parties who expect a long relationship, who can recognize each other and remember past moves, will drift toward cooperation even with no one enforcing it, because betrayal simply costs more than it earns over time. The enforcer, in a sense, is the future.
05Conclusion
The soldiers who stopped shooting were not pacifists, and they were not disobeying the logic of their situation — they were following it. Facing the same enemy day after day, with the certainty of tomorrow's fire hanging over today's restraint, they arrived at the same answer a stripped-down computer program arrived at in Axelrod's tournaments: cooperate first, answer betrayal, forgive quickly, keep it simple. What broke the truces was not the return of good sense but the removal of the future, when the high command rotated the men and forced the raids that made continuity impossible.
