tioned his game theory interests. He was asked unexpectedly to present a seminar, so he quickly conjured up the scenario of two criminals captured by the police and separately interrogated.20

You know the story. The police have enough evidence to convict two criminal conspirators on a lesser offense, but need one or the other to rat out his accomplice to make an armed robbery charge stick. So if both keep mum, both will get a year in prison. But whoever agrees to testify goes free. If only one squeals, the partner gets five years. If both sing like a canary, then both get three years (a two-year reduction for copping a plea).

Years in prison for Bob, Alice

If you look at this game matrix, you can easily see where the Nash equilibrium is. There’s only one combination of choices where neither player has any incentive to change strategies—they both rat each other out. Think about it. Let’s say our game theory experts Alice and Bob have decided to turn to crime, but the police catch them. The police shine a light in Bob’s face and spell out the terms of the game. He has to decide right away. He ponders what Alice might do. If Alice rats him out—a distinct possibility, knowing Alice—his best choice is to rat her out, too, thereby getting only three years instead of five. But suppose Alice keeps mum. Then Bob’s best choice is still to rat her out, as he’ll then get off free. No matter which strategy Alice chooses, Bob’s best choice is betrayal, just as Poe’s detective had intuited. And Alice, obviously, must rea-

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement