On Sheepdogs

On Sheepdogs

My previous post about oxytocin mentioned the “Prisoner’s Dilemma” game that figured in some recent oxytocin research, and included a link to an article about Anatol Rapoport, the game theorist whose winning entry in a cybernetic “Prisoner’s Dilemma” tournament employed a combination of clear-sighted retaliation and sweet-natured trust and forgiveness. If the Prisoner’s Dilemma forces players to decide whether to be sheep or wolves, Rapoport’s winning strategy could be said to convert its player into a sheepdog: someone who never hurts the sheep but is primed for ruthlessness against the wolf. But how to know for sure which one your opponent is? What if he might be either one, at different times, or even depending on how you treat him?

Before Rapoport’s contribution, people were drawing sour conclusions about human conflict from the established fact that the optimum solution to the “Prisoner’s Dilemma” from the point of view of one player in a single game was to choose a betrayal (wolf) strategy. This was a frustrating conclusion, given that the optimum solution from the point of view of both players considered together was mutual trust (sheep). The problem, of course, is that one player has no way of knowing whether the other player will take the first player’s well-being into account, a welcome development that would convert the two players into a cohesive unit for which the game’s results can be optimized.

Rapoport’s genius was to consider that people don’t always engage in single, isolated conflicts with strangers. More often (unless they’re engaged in a species-ending paroxysm) they need strategies for addressing repeated conflicts with people about whom they can learn something, and to whom they can impart information about themselves. They live in a world where each party to the conflict may learn from mistakes, build a reputation for trustworthiness, and use effective sanctions against predatory behavior: become sheepdogs.

Rapoport was a man with many generous tools for conflict-resolution in his box. According to Daniel Dennett, he

once promulgated a list of rules for how to write a successful critical commentary on an opponent’s work. First, he said, you must attempt to re-express your opponent’s position so clearly, vividly and fairly that your opponent says “Thanks, I wish I’d thought of putting it that way.” Then, you should list any points of agreement (especially if they are not matters of general or widespread agreement), and third, you should mention anything you have learned from your opponent. Only then are you permitted to say so much as a word of rebuttal or criticism.

Excellent advice, not (unfortunately) followed by its admiring but irritable quoter. Dennett’s review left me thinking that some books promise to be so unpleasant in their style of argument that I can do without buying and reading them. But this Rapoport guy – he looks like someone worth knowing more about.

His winning “tit for tat” strategy is said to be an “exceptionally effective sanction” for selfish behavior, in that the punishment lasts only as long as the selfish behavior lasts, whereas cooperative behavior is rewarded immediately in kind. Rapoport’s “tit for tat” strategy can yield even better empirical results in the "tit for tat with forgiveness" variant, in which the first play occasionally, and unpredictably, “turns the other cheek” by declining to respond to a betrayal in one game with his own act of betrayal in the next. This promises both players an exit from a disastrous vicious cycle of retaliation without exposing them to permanent exploitation by dyed-in-the-wool predators. In other words, if the early work on the Prisoner’s Dilemma suggested Leviticus 24:19-21, the work of this mathematical Russian Jew suggested an empirically successful fusion of that hard old law with Matthew 18:22.

No comments: