On September 3, 1949, an American weather monitoring plane collected air samples over Japan and found something that changed everything: traces of radioactive material that could only have come from a Soviet nuclear explosion. The age of American military supremacy was over, and two superpowers now faced a choice that game theorists would spend decades trying to solve. What followed was not just a Cold War arms race but a living laboratory for one of the most consequential puzzles in human history, a puzzle whose answer turns out to be disarmingly simple.
The game nobody realized they were playing
In 1950, researchers at the RAND Corporation were searching for a framework to understand the US-Soviet standoff. That same year, two mathematicians there invented a new game that, without anyone quite realizing it at the time, mirrored the nuclear conflict almost exactly. The prisoner’s dilemma works like this: a banker offers you and one other player a choice, cooperate or defect. If you both cooperate, you each receive three coins. If one of you defects while the other cooperates, the defector takes five coins and the cooperator walks away with nothing. If you both defect, you each receive one coin. The cruel logic of the game is that no matter what your opponent does, your best individual move is always to defect. And yet when both players follow that logic, they end up with one coin each instead of three. The US and Soviet Union spent roughly $10 trillion developing nuclear arsenals that neither could use, arriving at exactly that outcome.
The dilemma gets even more interesting when the game repeats. Impalas on the African savannah groom each other to remove ticks, a behavior that costs saliva, electrolytes, and precious attention under a sky full of predators. Each impala faces the same temptation to free-ride. But because these animals see each other day after day, the single-game logic breaks down. Your choices today follow you into tomorrow.
A tournament that rewrote the rules
In 1980, political scientist Robert Axelrod decided to find out which strategy actually wins the repeated version of this game. He invited leading game theorists from multiple disciplines to submit computer programs that would compete against each other across 200 rounds, with every strategy facing every other strategy and a copy of itself. The 14 submitted strategies ranged from elaborate scripts with 77 lines of code to devious programs like Joss, which copies its opponent’s last move but then secretly defects about 10 percent of the time. There was Friedman, which cooperates until betrayed once and then defects for the rest of the game without mercy. Axelrod added a 15th entry called random, which cooperates or defects with equal probability.
The winner used four lines of logic. Tit for Tat, submitted by peace researcher Anatol Rapoport, starts by cooperating and then mirrors exactly what its opponent did on the previous move. Axelrod later noted, with some amusement, that Rapoport had not been entirely confident in the submission: ‘I just wanna be clear that I’m not sure that this is really such a good idea.’ Rapoport’s own instincts leaned toward being even more forgiving and less quick to retaliate. He submitted it anyway.
Axelrod found that the top-performing strategies shared four qualities. They were nice, meaning they never defected first. They were forgiving, meaning they did not hold grudges past the last move. They were retaliatory, striking back immediately when betrayed rather than absorbing punishment indefinitely. And they were clear, transparent enough that opponents could recognize a pattern and trust it. As Steven Strogatz, who joined Axelrod to discuss the findings, put it: the strategies that were too opaque ‘ended up more or less defaulting to thinking every turn is like the last time I’m gonna see you, so I might as well defect.’
A second tournament with 62 entries confirmed the results. Nice strategies again dominated the top 15; nasty ones clustered at the bottom. Even when Axelrod introduced noise, simulating real-world errors like the 1983 Soviet early-warning system that mistook sunlight reflecting off high-altitude clouds for an incoming American missile, the core insight held. A slightly more forgiving version of Tit for Tat, one that absorbs roughly one defection in ten without retaliating, performed best in those messier conditions.
A detail that never quite settled
Tit for Two Tats, a strategy that only retaliates after two consecutive defections, would have won the first tournament had anyone submitted it. It finished 24th in the second.
From 1986 onward, the US and Soviet Union began reducing their nuclear stockpiles not through a single sweeping agreement but through incremental steps, disarming a small number of weapons each year and then verifying that the other side had done the same before proceeding. It was, structurally, a repeated prisoner’s dilemma played out in slow motion across decades, by two rivals who had quietly learned what a computer tournament had already demonstrated.
Wake up to Only Happy News!
