Two suspects are held in separate rooms, unable to communicate, each offered the same deal: betray the other and go free while your partner takes the full sentence, or stay silent and hope your partner does too. Work through the payoffs carefully, and each suspect's individually rational choice — betray — leads both of them to a worse outcome than if they'd both stayed silent. That's the Prisoner's Dilemma, one of the most famous constructs in game theory, and it captures a genuinely uncomfortable fact: doing what's individually rational doesn't guarantee a good collective outcome, even when every player plays perfectly.
Working out why betrayal wins, even though silence is better for both
From each individual suspect's perspective, betraying is the better choice regardless of what the other person does. If your partner stays silent, betraying gets you freedom instead of a lesser sentence. If your partner betrays, betraying at least avoids the worst outcome. Betrayal dominates silence no matter what the other player picks — which is exactly why both rational suspects end up betraying each other, landing in a worse spot than the mutual-silence outcome that was available to them the whole time, just unreachable without trust or coordination.
A Nash equilibrium is stable, not necessarily good
The economist John Nash's key contribution to game theory was defining a stable outcome — now called a Nash equilibrium — where no single player can improve their own result by changing their choice alone, given what everyone else is doing. Mutual betrayal in the Prisoner's Dilemma is exactly this kind of equilibrium: stable, because neither suspect benefits by unilaterally switching to silence while the other keeps betraying, but not remotely optimal, because the mutual-silence outcome, better for both, was sitting right there, just unreachable by individually rational choices alone.
What we're still unsure about
The one-shot version of the dilemma reliably predicts mutual betrayal, but real-world interactions are rarely one-shot — they repeat, with reputations and future consequences in play. Repeated-game research, building on strategies like "tit-for-tat," which performs well across repeated tournaments, shows cooperation can emerge and stabilise under the right conditions, even among purely self-interested players. Exactly which real-world conditions — how many repetitions, how visible reputations are, how the incentives are structured — reliably tip a group from betrayal toward sustained cooperation is still an active area of research in game theory and experimental economics, not a single settled formula.
This sits inside Game Theory & Strategic Behaviour, one of eight topics in Microeconomics, one of five domains in Economics, one of seventeen subjects the app can quiz you on.