Monday, April 1, 2013

The Prisoner's Dilemma

If there were one idea I could drill into the head of everyone on earth, it would be the prisoner's dilemma. I am complete serious in saying that if you don't know what I mean, that you should read the rest of this post, and then spend the next couple hours further studying the topic. The prisoner's dilemma, or more broadly game theory, is an absolute necessary component of understanding politics, economics, biology, and human relations.

So what is a prisoner's dilemma? Essentially, it is any scenario where two people or groups interact in such a way that


If they cooperate, both do well

If one cooperates but the other double-crosses the cooperator, the cooperator is screwed and the betrayer makes out very well

If both double-cross the other, both do poorly


The problem with a prisoner's dilemma is thus: no matter what the other party does, you are better off if you betray them. If they cooperate, you do very well, rather than merely well, if you cheat. If they betray you, you are screwed if you cooperate but make out merely poorly if you cheat as well. So if you are the self-interested person of the economics textbook, you cheat. Unfortunately, the other party faces the exact same incentives you do, so if they are self-interested, they cheat as well. Hence, everyone winds up doing poorly, rather than everyone doing well, as would happen if they cooperated. Game theory is simply the extension of this to more complex systems, with repeated encounters, memory, multi-party, etc. The key thing to note here is that depending on the payouts, self-interest does not lead to optimal outcomes.

So is the prisoner's dilemma relevant to the real world? Absolutely. An oft-cited example is the observation that male peacocks have huge tails that both attract females and make them vunerable to predators. Why don't they all cut (or evolve) their tails in half, maintaining their pecking order but making all of them safer? Because it's a prisoner's dilemma. Why can't the US and China agree to cut carbon emissions? It is a prisoner's dilemma. Why do states keep offering corporations "incentives" in order to poach jobs from one another? It is a prisoner's dilemma. Why do athletes dope? It is a prisoner's dilemma.

Note that a key element of most prisoner's dilemmas is ranking. Prisoner's dilemmas often arise when what matters is not ones absolute performance, but relative performance. Any improvement in your performance lowers everyone elses' relative performance, forcing them to do more just to get back to where they were. But once they do, everyone is putting in more effort but in the same place they were before. Variants on this theme riddle every aspect of human (and biological) endeavor, and to understand and identify them as prisoners' dilemmas brings some level of clarity to the issue.

Unfortunately, prisoners' dilemmas are hard to resolve, requiring repeated rounds of trust-building, acts of faith, and a few well-timed (but clearly explained) sticks being used to push the laggards from behind. However, if you find yourself in a prisoner's dilemma and are arguing for defection, you really need to reflect on what the end game is. Sure, defecting might work in the short term and you might even be able to drag forth evidence supporting such. In the long run, however, defection only leads to ruin for all. It is always a good idea to examine your political beliefs, identify any lurking prisoner's dilemmas, and ask yourself if you are arguing to defect. If you are, you are probably wrong.