The Prisoner's Dilemma

Why did the 2024 Nobel Prize in Economics go to researchers who study why some countries prosper with strong institutions while others don’t? When I looked into the question, the answer turned out to be connected to a famous social game called the Prisoner’s Dilemma.

I had encountered the Prisoner’s Dilemma before as a single-game thought experiment, but the Nobel Prize work focuses on repeated interactions. When the same dilemma plays out over and over, questions like "Why should I pay taxes anyway?" or "Why should my country cut emissions if others won't?" become surprisingly complex strategic problems.

To explore the dilemma, I wanted to work through its basic mechanics using a scenario that makes the challenges clear. This is why I chose to embed the discussion into the context of Breaking Bad, which felt like a perfect example.

Setting the scene with Breaking Bad

The characters are: Walter White (high school chemistry teacher turned meth cook), Jesse Pinkman (Walter’s former student and business partner), and Hank Schrader (DEA agent).

We can now consider these three characters in a standard scenario for prisoners: Walter White and Jesse Pinkman have been arrested by the DEA and are being interrogated separately. Agent Hank Schrader offers each the same deal: testify against your partner to get immunity, or stay silent and face whatever charges can be proven.

This setup—two people who must choose between cooperation and defection without being able to communicate—provides a clean framework for analyzing why mutually beneficial cooperation often breaks down.

The Breaking Bad scenario

This is the perfect Prisoner’s Dilemma scenario. Walter and Jesse each have exactly two choices:

• Cooperate (C): Stay loyal to your partner - "I don’t know this person, we’ve never met."
• Defect (D): Blame your partner - "He forced me into this, he’s the real criminal."

The sentences depend on what both choose:

• Both cooperate (C, C): Each gets 3 years. The DEA has some evidence but not enough for major charges without testimony.
• One blames the other (D, C): The betrayer walks free with immunity, the cooperator gets 15 years for "being the mastermind."
• Both blame each other (D, D): Each gets 5 years. Their contradictory stories help neither case.

The Dilemma Unfolds

Now try putting yourself in Walter’s position. You’re sitting across from DEA Agent Hank Schrader, knowing Jesse is in the next room facing the exact same choice. What goes through your mind?

The interactive game above reveals something unsettling: there’s a strong pull toward betrayal. You might find yourself thinking, "If I blame Jesse, I could walk free. What do I owe him anyway?" This gut reaction points to the heart of the dilemma.

Two Ways to Think About It

The Team Perspective: If Walter and Jesse could step back and ask "What’s best for us together?", the answer is clear. Both cooperating (3 years each) beats both betraying (5 years each). Total time served: 6 years versus 10 years. They’re partners against the DEA, not enemies.

The Individual Perspective: But here’s where Walter’s mind would really go: "What’s best for me?" And that’s where the mathematics becomes crucial—and troubling.

Walter’s Cold Calculation

Walter is, at heart, a high school chemistry teacher who thinks systematically. He’d probably work through the logic like this: "Let me figure out what Jesse might do, then decide accordingly."

Suppose Walter estimates there's a probability p that Jesse will betray him. That means there's a probability (1-p) that Jesse cooperates.

Walter's expected sentence if he cooperates:

• If Jesse cooperates (probability 1-p): 3 years
• If Jesse betrays (probability p): 15 years
• Expected sentence: 3(1-p) + 15p = 3 + 12p years

Walter's expected sentence if he betrays Jesse:

• If Jesse cooperates (probability 1-p): 0 years (Walter goes free)
• If Jesse betrays (probability p): 5 years
• Expected sentence: 0(1-p) + 5p = 5p years

For Walter to prefer cooperating, we'd need: 3 + 12p < 5p

Solving this: 3 < 5p - 12p = -7p

This gives us: p < -3/7

Since probabilities can't be negative, this condition is impossible to satisfy. Walter should always betray Jesse, regardless of what he thinks Jesse will do.

This is the mathematical heart of the Prisoner's Dilemma: individual rationality leads both players to a worse outcome than cooperation would provide. The outcome where both players betray is what game theorists call a "Nash equilibrium" - a stable situation where neither player can improve their payoff by unilaterally changing their strategy. Even though both would prefer mutual cooperation, neither wants to be the "sucker" who cooperates while the other defects.

When Cooperation Becomes Rational

Walter's calculation revealed a stark conclusion: betrayal dominates in his specific situation. But this raises a deeper question: are there any circumstances where cooperation becomes the rational choice in a prisoner's dilemma?

Let's step back from Walter's specific numbers and examine how this logic applies to any prisoner's dilemma. Game theorists use a standard notation to describe these situations:

• T = Temptation payoff (what you get for defecting when your partner cooperates)
• R = Reward for mutual cooperation (what both get when both cooperate)
• P = Punishment for mutual defection (what both get when both defect)
• S = Sucker's payoff (what you get for cooperating when your partner defects)

In Walter's case: T = 0 years (immunity), R = 3 years, P = 5 years, S = 15 years.

For any situation to be a true prisoner's dilemma, we need T < R < P and S is the worst outcome. Using this general framework, let's rewrite Walter's expected value calculation:

Expected sentence when cooperating:

$E[\text{Cooperate}] = R(1-p) + Sp = R + (S-R)p$

Expected sentence when defecting:

$E[\text{Defect}] = T(1-p) + Pp = T + (P-T)p$

For cooperation to be rational, we'd need:

This gives us the condition (where we assume T=0 for simplicity):

We can directly see that this equation only makes sense if S < R+P, as we would be dealing with negative probabilities otherwise. This replicates the situation we encountered previously and hence confirms our earlier conclusion that Walter should betray Jesse.

But notice what happens if we change the payoffs. We know that $p<1$, which directly implies that cooperation becomes a viable option for $S < P$. The condition above tells us exactly when cooperation becomes rational, and the interactive plot below lets you explore how different scenarios affect this critical threshold. Try adjusting the payoff values to see how they change the probability at which cooperation becomes individually rational.

As you can see from experimenting with the plot, the condition for rational cooperation depends critically on the relationship between the payoffs. When the punishment for mutual defection becomes severe enough relative to being betrayed, cooperation can emerge as the individually rational choice.

Real-World Applications:

The mathematical analysis reveals something surprising: Walter should only cooperate if the punishment for mutual defection (P) exceeds the punishment for being betrayed while cooperating (S). In our scenario, mutual defection gives 5 years each, while being the "sucker" gives 15 years—so defection dominates.

But imagine if Walter and Jesse were part of a larger criminal organization. If both betray each other, they don't just get 5 years in prison—they also face execution by the organization for breaking the code of silence. Suddenly, mutual defection might carry a sentence equivalent to 20+ years (or death), while being betrayed only gets you the original 15 years.

This explains why organized crime groups, military units, and tight-knit communities develop strong codes of loyalty: they artificially raise the cost of mutual defection to make cooperation individually rational.

The Single Game Conclusion

In our DEA interrogation room, Walter's rational analysis points to one conclusion: betray Jesse. The general mathematical framework confirms this isn't unique to Walter's situation—it's a fundamental feature of prisoner's dilemmas where the sucker's payoff is worse than mutual defection.

But Walter and Jesse don't just interact once—they're partners in a long-running operation. What happens when the same dilemma repeats over multiple "episodes"? This is where the mathematics becomes more hopeful, and where we can explore whether reputation, trust, and reciprocity can overcome the pull of individual rationality.

When Breaking Bad Becomes a Series: Repeated Interactions

Walter's rational analysis came to a depressing conclusion: in a single encounter, defection dominates. But this raises an important question: if everyone follows this logic, how does cooperation ever emerge in the real world?

The Power of Repetition

The key insight is that repetition fundamentally changes the game. When Walter and Jesse know they'll face similar situations again, each decision carries weight beyond the immediate outcome. Future interactions create new incentives that can make cooperation rational even when it wouldn't be in a single encounter.

Consider Walter's strategic calculus now: "If I betray Jesse today, he'll remember that when we face our next crisis together. But if I stay loyal, maybe he'll reciprocate, and we'll both benefit in the long run." This forward-looking reasoning can make cooperation rational even when defection would be better in isolation.

The mathematical foundation is surprisingly simple: if both players value future payoffs enough (not discounting them too heavily), cooperative strategies can become Nash equilibria in repeated games. The exact threshold depends on the payoffs, but the principle holds broadly.

Four Strategic Approaches

In repeated prisoner's dilemmas, players typically adopt consistent behavioral rules rather than making isolated decisions. Let's examine four fundamental strategies that capture different approaches to the cooperation-defection dilemma:

1. Always Cooperate (Season 1 Jesse approach)

• Rule: Stay loyal to your partner regardless of their past behavior
• Logic: Trust and loyalty should be unconditional
• Risk: Vulnerable to exploitation by defectors

2. Always Defect (Season 5 Walter approach)

• Rule: Always prioritize self-interest, minimize your own sentence
• Logic: Others can't be trusted, so protect yourself first
• Advantage: Cannot be exploited, guarantees reasonable outcomes

3. Tit-for-Tat (Reciprocal relationship)

• Rule: Start by cooperating, then match whatever your partner did in the previous interaction
• Logic: Reward cooperation with cooperation, punish defection with defection
• Balance: Enables mutual benefit while deterring exploitation

4. Random (Unpredictable behavior)

• Rule: Make decisions based on emotions, circumstances, or chance
• Logic: Unpredictability might confuse opponents
• Problem: Cannot build trust or sustain beneficial patterns

Interactive Strategy Exploration

These strategies aren't just theoretical—they reflect real behavioral patterns we see in ongoing relationships. Let's explore how they perform when Walter and Jesse face repeated crises:

The simulation reveals key insights about repeated interactions:

• Always cooperating works well with trustworthy partners but becomes costly against defectors
• Always defecting provides security but forecloses opportunities for mutual benefit
• Tit-for-tat balances cooperation and protection: starts nice, retaliates against betrayal, and forgives when partners return to cooperation
• Random strategies perform poorly because they can't build the consistent patterns needed for trust

💡 Key Insight: There's no universally "best" strategy—success depends on what your opponent does.

Systematic Strategy Comparison

To move beyond individual simulations, let's examine how these strategies perform when they face each other systematically. This approach mirrors the famous research by political scientist Robert Axelrod, who ran computer tournaments in the 1970s to identify the most successful strategies in repeated prisoner's dilemmas.

Axelrod's key finding was that "nice" strategies (those that never defect first) often outperformed "nasty" strategies, but only when the population included enough other cooperative players. The winning strategy in his tournament was tit-for-tat—exactly the reciprocal approach we've been examining.

Let's run our own tournament to see how the Breaking Bad strategies fare against each other:

Interpreting the Tournament Results

The tournament results reveal a crucial insight about repeated interactions: the dominance of defection in single games doesn't necessarily carry over to repeated games. Here's what the data typically shows:

When Strategies Face Each Other:

• Tit-for-Tat vs. Always Cooperate: Mutual cooperation emerges, both players benefit
• Always Defect vs. Always Cooperate: Defection exploits cooperation, creating large payoff differences
• Tit-for-Tat vs. Always Defect: Initial cooperation is punished, leading to mutual defection
• Random vs. Any Strategy: Unpredictability prevents stable cooperation patterns

Key Findings from Game Theory Research:

1. Nice strategies (never defect first) can outperform nasty ones when there are enough cooperative players
2. Retaliatory strategies (punish defection) prevent exploitation better than unconditional cooperation
3. Forgiving strategies (return to cooperation after punishment) enable recovery from conflicts
4. Clear strategies (predictable rules) build trust better than complex or random approaches

This connects directly to Robert Axelrod's famous finding: tit-for-tat succeeded because it was nice, retaliatory, forgiving, and clear. It could cooperate with other cooperative strategies while defending against exploitative ones.

Conclusion: What Walter and Jesse Taught Us

Walter and Jesse's predicament in that DEA interrogation room captures something profound about human cooperation. In a single interaction, both would be better off if they could trust each other to stay loyal (3 years each vs. 5 years each), yet individual rationality pushes them toward mutual betrayal (5 years each).

But as we've seen, repetition changes everything. When the same players expect to face similar situations again, strategies like tit-for-tat can sustain cooperation by making betrayal costly in the long run. The prospect of future interactions makes present cooperation rational.

However, our analysis also reveals why cooperation remains fragile. Always-defect strategies are surprisingly robust—they can't be exploited, even if they forgo mutual benefits. This explains why selfish behavior persists even in repeated interactions, and why building cooperative institutions requires careful design.

This Walter-Jesse dynamic isn't unique to criminal partnerships—similar patterns appear throughout society whenever individual incentives conflict with group welfare:

• Team Projects: Each member faces the repeated choice between contributing effort or free-riding on others' work. Successful teams develop tit-for-tat norms: contribute when others do, withdraw effort when others slack.

• Neighborhood Maintenance: Each homeowner decides whether to maintain their property, knowing it affects everyone's property values. Communities with strong reciprocal norms (contribute to neighborhood appearance and receive social approval) maintain higher standards.

• Climate Action: Nations face repeated decisions about emissions, with each country's choices affecting all others. Successful climate agreements attempt to create tit-for-tat dynamics through monitoring and graduated sanctions.

• International Trade: Countries must repeatedly decide whether to honor agreements or pursue short-term advantages. Trade relationships prosper when nations adopt reciprocal strategies—cooperation for cooperation, retaliation for cheating.

The mathematical insight is that cooperation can emerge from purely self-interested behavior when interactions repeat and players value future outcomes. We don't need to change human nature—we need to create institutional structures that make cooperation rational for self-interested individuals.

This brings us full circle to the Nobel Prize research that started our exploration. When institutions create repeated-game dynamics with clear rules, predictable enforcement, and expectations of ongoing interaction, even selfish individuals find cooperation profitable. When institutions fail to create these conditions, societies can become trapped in cycles of mistrust and mutual defection—exactly what we see in countries with weak governance and extractive institutions.