Tuesday, March 30, 2010

Game-Theoretic Obama



If you scroll to 10:45 you will see Obama ridiculing critics of health care reform. A few days after he declared the debate on the new law over, he does this. In the same speech he dares the Republicans to try to repeal the health care plan. According to this article, this is the new Obama: more in-your-face, even less conciliatory than before. The author thinks the in-your-face health bill will be followed by a when-I-slap-you-you'll-take-it-and-like-it securities industry regulation bill, or possibly a f**k-you education bill.

If that is so, why does Obama think this is a good idea?

The really cool thing about this article is that the WSJ reporter who wrote it actually interviewed the game theorist Robert Axelrod for a sound bite, one that offers a possible explanation:
In classic game theory, confrontation is sometimes necessary when cooperation breaks down to present a credible potential threat and get the two sides to re-engage, said Robert Axelrod, a University of Michigan political scientist and author of the game-theory book, "The Evolution of Cooperation."
In Axelrod's famous series of experiments, he had his subjects participate in a series of "iterated prisoners' dilemma" games. In each PD encounter, players have two choices: either benefit both yourself and the other player a modest amount in the short run ("cooperate"), or benefit yourself more while also hurting the other player in the short run ("defect"). Since this is an iterated game, you will encounter the other player again, and they will remember what you did to/for them last time. Thus you must think about the long run as well as the short. This is where "strategic behavior" comes in: you make choices in order to elicit desired behavior (cooperation) from the other person.

Axelrod found that a very successful "strategy" (in this sense) is "tit for tat": you cooperate if the other player cooperated in your last encounter and defect ("punish" them) if they defected last time. Both behaviors can (though will not necessarily) elicit cooperation from the other side. In the above quote, Axelrod seems to be saying that in-your-face-Obama makes game-theoretic sense as a tit for tat punishment.

Amazingly, a sound bite from a Republican aide uses concepts right out of Axelrod's book:
The Senate doesn't work the way game theorists think, said Antonia Ferrier, an aide to Republican Sen. Orrin Hatch of Utah. A body built on personal relationships is likely to spiral into endless tit-for-tat retaliations in the face of Mr. Obama's new turn, she said.
Even though she denied the applicability ofgame theory to the Senate, I think Ferrier is using Axelrod's own theory better than he does himself in this case.

The punishment strategy only works if the other player has reason to think that you really are playing tit for tat, and are not just attacking them for short-term gain. This means: you must in the past have given them real evidence (behavior, not words on a teleprompter!) that you will cooperate in the future. In Senatorial terms this means: putting forth a bill that is based on both Democratic principles and Republican principles, ie., a bill that is ideologically adulterated and impure from a Democratic point of view. The problem is that, as I predicted they would long ago, BHO and the Democrats have given absolutely no evidence that they will ever do such a thing, and overwhelming evidence to the contrary. The health bill could have included Republican approaches like tort reform and interstate health insurance competition, yet such things were never even on the table, despite Republican attempts to put them there.

The Republicans have to assume that any "cooperate" behavior on their part will be followed by "defect" behavior by Obama and company.

In a way, I can't blame the Democrats for acting this way. For a long time, they thought they didn't need a single Republican vote, and thus had nothing to gain by cooperating. Also, after eight years of the (on some key issues) arrogantly non-conciliatory Bush, they were in no mood to behave in any other way themselves. But that just reinforces my point: game-theoretically speaking, the Republicans would be damn fools to trust Obama to cooperate in the future.

In view of this, the only game-theoretically rational thing for the Republicans to do is to assume that this even-more-in-your-face stuff is just another attack. In other words, they ought to defect at every turn: filibuster the crap out of every single Democratic legislative proposal that they find ideologically repugnant.

This is where things stand.

2 comments:

Mark M said...

I read that article and also took note of the game theory inclusion - stood out as some thoughtful reporting.

Lester Hunt said...

Yes. It might actually be evidence for the Flynn effect.