Iterated Games and Tit for Tat
The one-shot prisoner’s dilemma is bleak: two rational players, each better off defecting no matter what the other does, both defect, and both end up worse than if they’d cooperated. But almost no real relationship is one-shot. We deal with the same people, firms, and countries again and again. So around 1980 the political scientist Robert Axelrod asked a better question — what happens when the dilemma repeats? — and ran it as a tournament. He invited game theorists to submit computer strategies that would play the iterated prisoner’s dilemma against each other, hundreds of moves at a time.
The winner was the shortest program anyone sent. Anatol Rapoport submitted four lines of code called Tit for Tat: cooperate on the first move, then simply do whatever your opponent did last time. It won. Axelrod ran a second, bigger tournament, told everyone the first result so they could try to beat it, and Tit for Tat won again.
Axelrod pulled out why such a simple strategy was so robust, and the four properties read like a character sketch. It was nice — never the first to defect. It was retaliatory — it punished a defection immediately, so it couldn’t be exploited. It was forgiving — one cooperative move and it went right back to cooperating, refusing to hold grudges. And it was clear — so legible that opponents could quickly learn they were better off cooperating with it. Niceness without retaliation gets exploited; retaliation without forgiveness spirals into mutual punishment. Tit for Tat held all four in balance.
The deeper condition Axelrod identified is what he called the shadow of the future. Cooperation becomes rational only when the game is expected to continue — when you’ll meet again, so today’s defection gets punished tomorrow. “The future,” he wrote, “can cast a shadow back upon the present.” Lengthen the shadow and selfish players cooperate; shorten it — make it the last round, or a one-time deal — and cooperation collapses.
This is one of the most load-bearing results in social science, because it explains cooperation without requiring anyone to be good. It connects trust, symbiosis in biology, and gift economies to a single mechanism: repetition plus memory plus consequences. Reputation works because interactions repeat. The reason people are more honest in a small town, or a tight industry, or any place they’ll be seen again, isn’t superior virtue — it’s a longer shadow.
The practical lessons fall right out. To get cooperation, extend the shadow of the future: turn one-off transactions into ongoing relationships, make reputations visible and durable, ensure people expect to meet again. And the personal strategy is hard to improve on — be nice, but not a pushover; punish betrayal, but don’t nurse it; stay legible enough that others can see cooperating with you pays. Watch the last round, though. When the future stops casting a shadow — the final deal, the exit, the lame-duck term — the logic that held cooperation together quietly lets go.
Go Deeper
Books
- The Evolution of Cooperation by Robert Axelrod — The 1984 book on the tournaments, Tit for Tat, and the shadow of the future.