Two suspects, two locked rooms
In the first guide of this rung you learned the moving parts of any game: the players, the moves each can make, and the payoff matrix that records what everyone gets for every combination of moves. Now we put those parts to work on one specific, devilishly simple game — the prisoner's dilemma — because it is the single most illuminating story in the whole field. Almost everything strategy has to teach about why cooperation is hard is hidden inside it.
Here is the classic setup. Two partners in crime, Ada and Ben, are arrested and locked in separate rooms; they cannot talk to each other. A prosecutor offers each the same deal. If you betray your partner (confess) and they stay silent, you walk free and they get ten years. If you both stay silent, the prosecutor can only pin a minor charge on each of you — one year apiece. But if you both betray, you each get eight years. Each must decide, alone, with no idea what the other is choosing.
Notice the cruel shape of it before we touch any numbers. *Together* they are best off staying silent — one year each. Yet each is privately tempted to betray, because betrayal might set them free. The whole drama lives in that gap between what is good for the pair and what is tempting for the individual. Let us lay the four outcomes out plainly and follow each prisoner's reasoning step by step.
The dominant strategy: betrayal wins no matter what
Years in prison (Ada's, Ben's) -- fewer is better
BEN stays silent BEN betrays
ADA stays silent 1 , 1 10 , 0
ADA betrays 0 , 10 8 , 8
Ada's view, reading down each of Ben's choices:
If Ben is SILENT: silent=1yr vs betray=0yr -> betray is better
If Ben BETRAYS : silent=10yr vs betray=8yr -> betray is betterPut yourself in Ada's chair and reason it through. She cannot control Ben, so she considers his two possibilities in turn. *Suppose Ben stays silent.* Then silence gets her one year, but betraying gets her zero — betrayal wins. *Suppose instead Ben betrays.* Then silence gets her ten years, but betraying gets her eight — betrayal wins again. In both cases, whatever Ben might do, Ada is better off betraying. A move that beats every alternative no matter what the other player chooses has a name: a dominant strategy.
The dilemma's punch comes from the symmetry. The game looks identical from Ben's side, so betrayal is *his* dominant strategy too. Two careful, self-interested thinkers therefore both betray — and land in the bottom-right cell, eight years each. But look back at the matrix: had they both stayed silent, they would have served just one year apiece. Their individually flawless logic delivered them, together, to an outcome that is plainly worse for both. That gut-punch is the prisoner's dilemma, and the surprise never quite wears off.
Why the trap holds: a stable, lousy equilibrium
You might protest: surely, once they both see they could each get one year, they would just... cooperate? Here is the merciless part. Imagine they somehow agreed beforehand to stay silent. Now Ada sits in her locked room and thinks: "Ben promised silence. If he keeps it, I can betray and walk free — zero instead of one year." The promise itself *increases* her temptation to break it. And Ben, in his room, is thinking the exact same thing about her. The cooperative deal is not stable: each has a private reason to cheat on it.
The both-betray outcome, by contrast, is rock-solid. Sitting there having betrayed, Ada asks: "Given that Ben betrayed, would I have done better staying silent?" No — silence would have given her ten years instead of eight. She has no regret, no reason to change her move alone, and neither does Ben. An outcome where no player can improve by changing only their own choice is a Nash equilibrium — the central solution concept you met in the previous guide. The tragedy of this game is that its only Nash equilibrium is the outcome both players would least have chosen together.
The same trap, dressed as the economy
Strip away the prison and the structure reappears all over the economy. Picture two airlines on the same route, deciding whether to charge a high fare or a low one. If both keep fares high, they share a fat profit. But each is tempted to undercut: drop your price a little and you steal the rival's customers and earn more — *as long as the rival holds high*. So both cut. Fares collapse, customers cheer, and the two airlines end up in the both-betray cell: low fares, thin profits, exactly where neither wanted to be. That is a price war, and it is a prisoner's dilemma wearing a business suit.
This is exactly why a cartel is so prone to breaking down. A handful of producers in an oligopoly — say, an oil cartel — agree to hold output low so the price stays high, like all prisoners agreeing to stay silent. But the deal hands each member a private incentive to cheat: secretly pump a bit *extra* and sell it at the high price the others are propping up. Every member feels that same pull, so quotas get quietly broken, output creeps up, and the high price the cartel was built to defend sags toward the competitive level. The cartel is not betrayed by enemies; it is dissolved by the ordinary arithmetic of self-interest.
The same logic empties the sea. Imagine many fishing crews sharing one ocean — a common-pool resource. If everyone fished modestly, the stock would replenish and feed all of them for generations. But each skipper reasons: "The fish I leave swimming will just be netted by someone else, so I may as well take them now." Every skipper thinks this, all over-fish, and the shared stock collapses — leaving everyone worse off than restraint would have. It is the prisoner's dilemma scaled up from two players to a whole fleet, and we will return to this many-player version in a later guide.
Escapes from the trap — and their honest limits
If the dilemma were inescapable, cooperation would be impossible — yet airlines do sometimes hold prices, cartels sometimes endure, and fisheries are sometimes saved. So how do players climb out? The single most important escape is repetition. The one-shot version above is rigged for betrayal precisely because there is no tomorrow. But play the same game over and over with the same partner, and a new force appears: today's betrayal can be punished by retaliation in every round to come.
In these repeated games, a strikingly simple strategy often works: "tit for tat" — start by cooperating, then in each round just copy what your opponent did last time. Cooperate with me and I cooperate back; betray me and I betray you next round. When players value the future enough, the looming threat of lost future cooperation can outweigh the one-time gain from cheating, and stable cooperation becomes possible. This is the formal heart of "why repeated dealings build trust," and it explains why long-running rivals and neighbours often cooperate where strangers would not.
One last guard-rail. The neat 0/1/8/10 numbers are a model, not a measurement; change the payoffs and the dilemma can dissolve entirely. And real people are not the icy calculators the story assumes — in experiments many cooperate far more than pure self-interest predicts, swayed by fairness, guilt, anger, or simply trusting the other person. The prisoner's dilemma is so valuable not because it describes how everyone always behaves, but because it isolates, with brutal clarity, the precise reason cooperation is fragile: when private incentives and the common good point in opposite directions, good outcomes need more than good intentions to survive.