Accessible text version of a twelve-slide presentation. Each heading below is one slide, in order.
LER 565, Week 3: Cooperation, Defection and the Prisoner’s Dilemma. Part of HRM and Strategy. Instructor: Ryan Lamare.
Recall from Week 2: when the pie is fixed, that is zero-sum, you find your dominant strategy and use it. This week we go non-zero-sum: the pie can grow or shrink, and cooperation can pay. Our focus all week is the prisoner’s dilemma, the most famous game in the social sciences. We will read the prisoner’s dilemma matrix, see why “rational” play backfires, and how to escape it.
Two suspects are arrested for a crime and interrogated in separate rooms, with no chance to coordinate. Each can confess, betraying the other, or stay silent, which in effect cooperates with the other. If you confess while your partner stays silent, you walk free and your partner gets about twenty-five years. If both confess, three years each. If both stay silent, just one year each. Each must choose at the same time, not knowing the other’s choice. So what should each do? Related media: “Coffee Wars” from Curb Your Enthusiasm, a price war neither shop can stop.
Suspect B is the column player, choosing Stay Silent or Confess. Suspect A is the row player, choosing Stay Silent or Confess. Each cell shows the pair of outcomes as Suspect A, then Suspect B. If both stay silent: one year, one year. If A stays silent and B confesses: twenty-five years for A, free for B. If A confesses and B stays silent: free for A, twenty-five years for B. If both confess: three years, three years. Fewer years is the better payoff. Both choose simultaneously, blind to the other; rows are A’s options, columns are B’s.
Confess is a dominant strategy for both players, so they land on the outcome where both confess, getting three years each. That choice is individually rational but collectively worse than both staying silent, which would have been one year each.
Generalize the game. You and the other side each choose to Cooperate or Defect. If both cooperate, both win. If you cooperate and they defect, you lose big. If you defect and they cooperate, you win big. If both defect, both lose. Defecting always tempts you, and if you both give in you land on “both lose,” which is worse than if you had cooperated. Any situation with that set of incentives is a prisoner’s dilemma.
Three workplace examples. Contract standoffs: a team and a player each move to dodge their own worst case, so both defect and the deal collapses, even though cooperating was cheaper; see The Economist, “How to Lose Quarterbacks and Alienate People: Contracts in American Football” (2017). Because NFL contracts are a repeated game, reputation makes cooperation the smarter play. Train your own, or poach? Why build talent if rivals just poach it, so firms buy instead of build, and the talent pool grows scarce and expensive; see Peter Cappelli, “Talent Management for the Twenty-First Century,” Harvard Business Review. Workplace surveillance: “see something, say something” systems reward snitching until everyone informs on everyone; see the Workplace Spying reading.
States and cities compete to attract companies by offering subsidies — a prisoner’s dilemma among governments, paid for by taxpayers. Example: when Amazon, headquartered in Seattle, Washington, sought a site for a second headquarters, cities and states across the country bid against each other. Among them, New York offered incentives of about 1.5 billion dollars and Virginia about 573 million dollars. Each state, fearing it will lose the jobs to a rival, offers more — so the company wins and taxpayers pay.
Prisoner’s dilemmas are not always bad; it depends on who is outside the game. Two firms choose whether to poach each other’s workers. Left to themselves, firms prefer the cooperative no-poach pact: wages stay suppressed, which is good for them but bad for workers. That pact is collusion. Regulation, through antitrust enforcement, forces firms off the pact and into competition: both poach, and wages get bid up. This is shown as two copies of the same payoff matrix — the first with the cooperative no-poach cell highlighted, the second with the mutual-poach cell highlighted in red to mark what regulation forces. So a cooperative outcome can harm society, and forcing players to defect can be the better result. Related media: Talent Wars in Silicon Valley, a no-poach pact that hurt workers.
The main route out is Tit for Tat: cooperate first, then mirror the other side, and only if the game repeats. It works because it is nice, never defecting first; forgiving, returning to cooperation; retaliatory, punishing defection at once; and clear, simple enough to trust. These four traits are why Tit for Tat won Robert Axelrod’s 1980 prisoner’s-dilemma tournaments, where the simplest strategy beat far more complex ones. Related media: How to Outsmart the Prisoner’s Dilemma, on repeated play and tit for tat with the discount factor approaching one.
Conditions for cooperation to hold. Detect defection fast: cooperation holds only if cheating is spotted quickly. Repeat the game: a “shadow of the future” makes today’s defection cost you tomorrow. Make punishment credible and clear: everyone knows what counts as cheating, and that it draws a response. Keep it proportionate: unbelievable threats don’t deter, and forgiving the odd mistake stops a spiral. In practice, parties can try to incentivize cooperation, but it’s always fragile.
The prisoner’s dilemma: each side’s rational move, defect, leaves everyone in the game worse off. Escape routes: repetition and reciprocity through Tit for Tat, regulation, and enforceable agreements. Most workplace “dilemmas” have cooperative gains on the table, if you can detect defection and make cooperation stick. Next week: coordination games, when several strategies are equally viable and the trick is agreeing on one, using focal points.