Why Rational People Choose Badly in the Prisoner's Dilemma

The Prisoner's Dilemma: Why Rational People Choose Badly

The Setup

The prisoner's dilemma is arguably the most famous thought experiment in all of social science. It showed up in 1950, created by Merrill Flood and Melvin Dresher at the RAND Corporation, and then mathematician Albert Tucker wrapped it in the dramatic prison story we still use today.

Here's the scenario. Two suspects — let's call them Ada and Bob — are arrested for committing a crime together. The police don't have enough evidence for anything serious, so they need at least one of them to flip. The detectives separate them and offer each the same deal:

• If you betray your partner and they stay silent: You walk free; they serve 5 years.
• If you both stay silent: You each serve 1 year on a minor charge.
• If you both betray each other: You each serve 3 years.
• If you stay silent while they betray you: You serve 5 years; they go free.

Let's map this out in a payoff matrix (using negative numbers for years in prison — higher is better):

Now walk through Ada's logic. If Bob stays silent: Ada can stay silent and get -1, or betray and get 0. Betraying is better. If Bob betrays: Ada can stay silent and get -5, or betray and get -3. Betraying is still better. Betray is Ada's dominant strategy — it wins no matter what Bob does. And by the exact same logic, betray is Bob's dominant strategy too.

So both players betray, and they each serve 3 years. Yet if they'd both stayed silent, they'd have each served just 1 year. Rational logic, followed to its conclusion, traps them in a worse outcome.

Understanding Dominant Strategies and Nash Equilibrium

What makes this dilemma so powerful is what's really happening underneath. The fact that "betray" is a dominant strategy for both players is actually remarkable: neither player needs to guess what the other will do to decide. Betraying wins regardless.

This leads straight to a Nash Equilibrium at (Betray, Betray) — the outcome where each player is doing the best they can given what the other is doing. Ada has no reason to switch if Bob is betraying. Bob has no reason to switch if Ada is betraying. They're locked in place. This is the "stable" outcome from a game theory perspective.

But here's where it gets weird: there's another outcome — (Silent, Silent) — that's better for both of them than (Betray, Betray). Both get -1 instead of -3. So why don't they just do that? Because each player faces a personal incentive not to. If Ada thinks Bob will stay silent, Ada can break that trust and walk free (getting 0, which beats -1). The mutual benefit of cooperating is constantly undermined by the individual payoff for defecting.

This structure — where the best outcome for everyone is actually unstable because of what's best for each person individually — is the whole dilemma. It's not about stupidity or being irrational. It's about the gap between what's individually optimal and what's socially optimal.

Why This Is Everywhere

The prisoner's dilemma isn't just a thought experiment for textbooks — it's a template for an astonishing range of real situations:

Nuclear arms races: During the Cold War, if the U.S. and Soviet Union both built enormous arsenals (both "betray"), they spent ruinously and faced mutual destruction. If both had disarmed (both "cooperate"), they'd save money and reduce existential risk. But each nation faced an individual incentive to build while the other disarmed. And the fear that the other side might do so forced both to arm. The Cold War was, in many respects, one long prisoner's dilemma. Both superpowers pursued nuclear buildup not out of desire but from mutual fear. They would have both been better off at a lower equilibrium, yet neither could safely reduce their arsenal without risking annihilation. This played out over decades, consuming billions of dollars and creating constant existential terror.

Environmental regulation: Every country would actually prefer a world where others reduce emissions while they keep polluting (defection) compared to everyone cutting emissions together (cooperation). The individually rational play for each country is to free-ride on others' sacrifices — which is why international climate agreements fall apart so spectacularly. When the Paris Agreement was negotiated, every nation committed to reducing carbon emissions, yet each had an incentive to emit more (to power their economy) while hoping others would bear the burden. That's why enforcement mechanisms and verification are so crucial to environmental treaties: they're trying to change the structure of the game itself.

Price competition: Two competing airlines could both keep prices high (cooperation) and earn healthy profits. But each has an incentive to undercut the other slightly to steal customers — leading to fare wars where both end up with thinner margins. In the 1970s, many airlines maintained high fares through implicit agreements and industry norms. But once the regulatory environment shifted, price competition intensified dramatically. Routes that had generated strong profits for both airlines suddenly saw fares collapse as each player chose defection. Sometimes entire industries spend years oscillating between periods of cooperation (high prices, good margins) and defection spirals (discount wars, razor-thin margins).

Antibiotic resistance: Each individual patient wants their doctor to prescribe antibiotics for any illness. But if everyone overuses antibiotics, resistance evolves and the drugs stop working. Individual rationality produces collective disaster. Your personal benefit from taking an antibiotic for a minor infection is clear and immediate. The cost (contributing to resistance) is distributed globally and shows up years later. This time-lag mismatch makes the dilemma especially hard to manage — people don't feel directly punished for their individual defection.

Doping in sports: Each athlete would prefer a world without doping, but faces the temptation that if others are doping while they're clean, they'll lose. So everyone dopes, and the sport reverts to approximately the same competitive landscape as before — just with everyone's health at greater risk. Surveys of Olympic athletes have shown many express real frustration with doping: they'd prefer a clean sport, but can't afford to be clean unilaterally. The prisoner's dilemma structure is unmistakable.

The Collective Action Problem

What the prisoner's dilemma really reveals is a deep tension in human social life: individual rationality and collective rationality can come apart. The term collective action problem (which political scientist Mancur Olson made famous in his influential book The Logic of Collective Action) describes situations where individuals acting in their own interest produce outcomes that nobody actually wants.

The prisoner's dilemma is the simplest, starkest example. But once you see it, you see it everywhere — in communities that can't maintain public spaces, in industries that can't self-regulate, in countries that can't cooperate on shared threats. Much of political science, sociology, and behavioral economics is, at some level, the study of how humans manage to solve (or fail to solve) these problems.

graph TD
    A["Each Individual Pursues<br/>Self-Interest"] -->|Rational Individual Action| B["Collectively Suboptimal<br/>Outcome"]
    C["Collective Benefit of<br/>Cooperation"] -->|Unstable: Individual<br/>Incentive to Defect| B
    B -->|Everyone Worse Off<br/>Than Possible| D["Collective Action<br/>Problem"]
    style B fill:#ffcccc
    style D fill:#ffcccc

A Common Misconception: "The Dilemma Proves Humans Are Irrational"

It's tempting to misinterpret the prisoner's dilemma as a critique of rationality itself — as if it proves that rational people inevitably make bad choices. But that's getting it backwards. The dilemma doesn't show that rationality is flawed; it shows that individual rationality doesn't automatically produce collective good outcomes. Actually more subtle and more important.

A player who defects in the one-shot prisoner's dilemma is behaving rationally given the payoff structure. The problem isn't irrationality — it's that the incentive structure itself is misaligned. What we really need are ways to change the structure (through monitoring, punishment, institutions, trust-building) rather than ways to make people "less rational."

This distinction matters because it points us toward actual solutions. If the dilemma were about irrationality, we'd be stuck with no good answers. But if it's about structure, we can redesign institutions, laws, and agreements to create incentives for cooperation.

Crucially: One-Shot vs. Repeated

The classic one-shot prisoner's dilemma — where the game is played exactly once and players never interact again — offers no escape: rational players must defect, and the dilemma persists. This is theoretically clean and deeply depressing.

But change the structure even slightly — allow the game to be played repeatedly, allow communication, change what information players have, or introduce reputation concerns — and cooperation becomes not just possible but sometimes inevitable. That's why understanding the basic structure of the dilemma is just the beginning. We'll explore the iterated version in depth in the next section, and you'll see how repeated interaction fundamentally transforms the strategic landscape.

This video by Veritasium does an extraordinary job of exploring the dilemma and the iterated version: