The original trust problem
Prisoner's Dilemma
You and a stranger are arrested. The prosecutor talks to each of you separately. Stay silent (cooperate) and you both get a short sentence. Inform on the other (defect) while they stay silent, and you walk — they take the fall. If you both inform, you both get a medium sentence. You cannot talk. The clock is ticking.
The Prisoner's Dilemma is the canonical social dilemma: two people would both be better off cooperating, but each has a private incentive to cheat. Economists, biologists, and political scientists use it as a stripped-down model of cartels, arms races, climate agreements, and everyday trust.
In the one-shot game, informing strictly dominates staying silent. Repeat the encounter and the math changes. Strategies that start nice and punish defection — Tit-for-Tat is the famous example from Axelrod's tournaments — can sustain cooperation without anyone being a saint.
This site lets you play it against bots that implement those strategies, so the textbook matrix becomes a sequence of choices you can feel. After a match, you will see your score next to theirs — and a short recap of the theory.
What theory predicts
Informing is a dominant strategy: it pays more no matter what the other person does. The unique Nash equilibrium is both inform, even though both would be better off staying silent. That gap is the dilemma.
What people actually do
In the lab, many people still stay silent, especially when the game repeats. Tit-for-Tat — start nice, copy whatever they just did — is a famous simple strategy that does surprisingly well.