Cobra Effect · Decisions and strategy
The prisoner’s dilemma
Why two sensible people can both choose the worst outcome.
7 cards, read aloud in 3:05, with a test and sources.
Two suspects, two cells, and a deal on the table.
The police have enough to jail both for a year on a small charge. They want a confession for the big one. So each prisoner is told, alone: turn on your partner and you walk free, and he does three years. If you both talk, two years each. If you both stay silent, one year each.
Now think it through as one of them.
If your partner talks, you should talk. Two years beats three. If your partner stays silent, you should still talk. Walking free beats one year. Whatever he does, talking is better for you. He reasons the same way. So you both talk, and both serve two years, when silence would have cost you one.
That is the prisoner’s dilemma.
Two people, each choosing what is best for themselves, arrive somewhere worse for both. Neither is stupid. Neither is wicked. Merrill Flood and Melvin Dresher found it in 1950, running games at the RAND Corporation. Albert Tucker, a mathematician, wrapped it in prisoners a few months later, to explain it to a room of psychologists.
You are playing it every day. You just aren’t in a cell.
Two shops in a price war. Two countries building missiles neither wants to pay for. Two athletes deciding whether to cheat, when each suspects the other will. Every fishing boat taking one more catch from a shrinking sea. In each, the sensible move for one is the ruinous move for all.
The way out is to play again.
In 1980 the political scientist Robert Axelrod invited experts to send in computer programs to play the game 200 times in a row. The winner was the simplest entry. Four lines of code, from Anatol Rapoport. Cooperate on the first move, then do whatever your opponent did last. It never beat a single opponent. It won the tournament, because everyone who played it did well.
Axelrod drew four lessons from it.
Be nice. Never defect first. Hit back. Answer betrayal at once, or it pays. Forgive. Go back to cooperating as soon as they do. Be clear. Let the other side see exactly what you will do. Cooperation grew wherever the players knew they would meet again.
So when you find yourself in one, change the game, not the player.
Ask: will we meet again, and does he know it? Make the deal repeat. Make defection visible. Make promises expensive to break. Trust is not naivety. It is the strategy that won. And if it really is one shot, with a stranger, expect the two years.
Sources
- Prisoner’s dilemma, Wikipedia. The 1950 origin at RAND, Tucker’s prisoners, the payoff table, the repeated version and Axelrod’s tournaments.
- The Evolution of Cooperation, Robert Axelrod, 1984. The tournaments, tit for tat, and the four lessons, with a chapter on how enemy soldiers in the trenches of 1914 learned to cooperate across no man’s land.
- Prisoner’s Dilemma, Stanford Encyclopedia of Philosophy. The serious survey. Every variation of the game, what rational players should do in each, and why the answer changes when the game repeats.
Nearby ideas
- The tragedy of the commons. Why shared things get used up, and how communities stop it.
- Braess’s paradox. Why adding a road can make everyone’s journey slower.
- Moral hazard. Shield people from a risk and they take more of it.
- The Overton window. Ideas move from unthinkable to normal before politicians touch them.
- Satisficing. Good enough, chosen quickly, often beats the endless hunt for best.
- Schelling points. Without talking, people meet at whatever stands out.
- Two-way doors. Decide reversible things fast, and save care for what you cannot undo.
- The Pareto principle. A few causes usually produce most of the results.