A Charles Krauthammer column was the topic of debate between some republican friends and me this weekend. It’s convincing, at first, but fails to consider one obvious possibility: That liberals are stupid AND conservatives are evil.

Allow me to explain. Your daily interactions with other people are in fact little positive sum games, or prisoner dilemmas. If you go into these games trusting the other party at all times (the conservative criticism of liberals), the other party will quickly learn that it can take advantage of you by making a selfish choice rather than the cooperative choice. You lose, and you are stupid for it.

If you go into these games making the selfish choice at all times, you preempt any chance of building a history of trust and benefiting from cooperation, and instead you head for a race to the bottom, friendless and unloved for initiating a Hobbesian nightmare of a world. You are evil, indeed.

In fact, the smartest tactic lies between the two, and it has been documented by evolutionary psychologists as the dominant form in which primates (and humans) interact in a social context. Because you play lots and lots of positive sum games with the individuals that you come into daily contact with, there is plenty of opportunity to reward cooperation and punish selfish defections. What you do is you trust the other party the first time; all subsequent times, you make the same choice (cooperative or selfish) that the other party did the last time. If the other party was selfish last time (because he/she is a conservative), you are selfish back the next time. If the other party was a liberal, you continue trusting them and behaving cooperatively. Let’s call it enlightened liberalism.

Interestingly, this tactic allows for educating conservatives, because they will soon learn that if they stop behaving selfishly, you will trust them again, and you will both benefit from the extra rewards brought on by cooperation. Yep, conservatives may be evil, but they’re not stupid.

[Thu, Aug 01 2002 – 14:30] Matthew (email) stefan, you truly are an idiot/liberal.

1) you fall into the dreadful trap of equating conservatism with selfishness which is a nasty liberal trope if ever i heard one.

2) on your logic–which is pretty flawed–everyone having been burned once would abandon their naivity and turn into meanies/conservatives.

3) the question more revolves around whether a state is a better regulator of human interaction (simplistic liberal view) or whether human interaction is a better regulator of human interaction (simplistic conservative view). there’s no game playing in the liberal scenario, unless you consider being told what to do a game.

4) game theory doesn’t explain everything. in fact, it hardly explains anything, apart from the possbile exception of games.

[Thu, Aug 01 2002 – 14:46] Kim the other problem with this is it assumes that to make any decision that is in your interest is to be “selfish.” We all make hundreds of decision every day to do what is best for us, rather than what is worst for us. that is called being rational and intelligent, not selfish. only liberals could think that making rational decision constituted a hobbesian race to the bottom. probably because they are stupid.

[Thu, Aug 01 2002 – 16:35] Charles Kenny (www) (email)

Have to agree with Kimthew on this one. Just because a schizophrenic mathmo says its true doesn’t make it so.

Game theory, like most good social science theories, is broad enough to allow any outcome to fit, which makes it pretty damn useless in practice (see also another one-time Stefan favourite, memetics). The theory as you describe manages to make the stunning prediction that animals sometimes cooperate and sometimes don’t (although it doesn’t help explain when they will and when they won’t). Nobel-worthy stuff, indeed.

Meanwhile, game theorists’ big success was being allowed to design mobile auctions, and we all know how that ended (not that I mind greedy and stupid capitalist pigs taking a bath, of course –pigs, whatever the new conventional wisdom says, are dirty and smell. But when my pension fund goes down too, its a case of the baby being thrown out with the capitalist and the bathwater).

The ‘new left’ version of game theory that you cite is just silly from the start. Game theory assumes rational actors. In the game as you have described it, if both actors know they will be punished if they don’t cooperate, both actors know that the payout over multiple rounds will from the start be higher from cooperating than from non-cooperative or sometimes-cooperative strategies. Therefore, both players will always cooperate. Doesn’t really help to explain non-cooperative behavior, then, does it.

You (Stefan) are too nice a guy to always punish someone who lets you down –even if you might trust them a bit less the next time. Yet, levels of trust don’t come into your scheme. Why do levels of trust matter? because I don’t go into all relationships with total strangers by trusting them, or choose a stranger when it comes to advice, looking after my keys and wallet, etc. like the theory suggests (and me a nice liberal, too).

What maeks it worse for liberals is that some societies are (a bit) more trusting than others from the get-go (call it an assumption of good will). Perhaps that’s because in general people reckon they won’t be done over on average the first time they interact in some small way with someone, and perhaps that’s a self-fulfilling prophecy. But those societies can be, perhaps more often are, ‘conservative’ (in a Bill Bennett sense, at least). Shame, that.

[Thu, Aug 01 2002 – 18:28] Felix (www) (email) I don’t know that more trusting societies are more conservative, even with a small c. Stefan’s adoptive home of Sweden comes to mind.

I’ve just come back from a week in the tiny and very trusting town of Seal Harbor, ME, where it seems that no one locks their houses or cars. And I think that if I found a gold watch on the sidewalk in Seal Harbor, I’d be more likely to hand it in than if I found it on the sidewalk in New York. (We’re talking probabilities here: as a good liberal, I’d be much more likely to hand it in than to hold onto it in either location.) Anyway, I think that Seal Harbor, for all its Rockefellers, is actually a pretty liberal town in a Northeastern Intelligentsia kind of way.

Of course, conservatives can be trusting in this sense too, but I don’t think they’re trusting more often than liberals are.

[Fri, Aug 02 2002 – 09:28] mATTHEw (email) actually, the watch example is interesting and backs up kennythew’s point.

your decision to hand over or not hand over the watch stems from some calculus based on rational self-interest. in the context of seal harbor, a small community where you have multiple interactions with the same few people, you’re more likely to be personally rewarded either through money or kudos if you find the watch’s owner. you’re more likely to find the owner, too. that all plays into why small communities can have stronger civic values than cities, for example (although for the same reason those values can break down pretty easily, too).

in new york, you may get some personal gratification from “doing the right thing” in some generic sense, but you probably also know you’ll get nothing else. worse, the watch will probably be pocketed by some corrupt cop, so why not keep it yourself? you’re happier and the watch-loser wouldn’t have gotten it back anyway.

neither example has much to do with being liberal or conserative and more to do with your own value sets. and guess what? you’re a much better judge of what you value than some self-appointed arbiter of public taste, like stefan, for example.

[Fri, Aug 02 2002 – 10:32] Kim small note on seal harbor trustiness. of course nobody locks their houses or cars there. they are all so enormously rich, why would they steal from each other? it isn’t as though david rockefeller is going to drive his jag down to martha stewart’s house so that he can steal her souped up SUV. but the fact they “trust” each other probably has very little to do with their liberal philosophies

[Fri, Aug 02 2002 – 10:32] Stefan Geens (www) (email) God, you’re all so STUPID.

Matthew, you completely miss the point. The whole point of the exercise is to tell conservatives that if they begin cooperating, you will cooperate back. You’re teaching them a lesson, literally.

Kim, you purposelfully muddle the word “selfish”, which I use as a term of art. Of course we are all selfish, all the time. But cooperating is a more enlightened form of selfishness, because the payout for both of us is higher than if we had defected.

Charles, I think you last looked at game theory when it stood at version 1.0. Rationality is only assumed if you play one game. But play many games, and participants can learn, evolving from emotionally stunted misanthropic conservatives to trusting, emotionally secure liberals.

But yes, the more rational you are, the faster you will learn to cooperate. Hence the excellent track record of the Nordic countries.

[Fri, Aug 02 2002 – 10:40] Kim small note on seal harbor trustiness. of course nobody locks their houses or cars there. they are all so enormously rich, why would they steal from each other? it isn’t as though david rockefeller is going to drive his jag down to martha stewart’s house so that he can steal her souped up SUV. but the fact they “trust” each other probably has very little to do with their liberal philosophies

[Fri, Aug 02 2002 – 14:20] Matthew (email) are you allowed to comment on your own blog? surely that’s breaking the rules of blogdom.

[Mon, Aug 05 2002 – 14:29] Charles Kenny (www) (email) Nope, rationality is always assumed (you have to assume rationality, otherwise you can’t predict outcomes. Example –every time Stefan makes a stupid point, 15 people jump on his blog site to correct him. In rational, game theoretic world, Stefan learns from his mistakes, and makes fewer stupid points and/or turns off the comment feature. Only in messy, real, non-rational world does Stefan do what the model could never predict –go on to make even more stupid comments). Learning and irrationality are not opposites, as you seem to suggest.

What is sometimes not assumed is perfect knowledge, but in the grossly over-simplistic game you outlined, there was no mention of imperfect knowledge. In Kim’s and Matthew’s responses, knowledge does come in. They suggest, I think rightly, that in an environment where it is more likely that your good deed goes recorded, the real owner of the watch is more likely to be known, etc, (i.e. there’s more knowledge around), doing the right thing pays off better. This is, after all, the idea behind freedom of information and the rest of it.

[Mon, Aug 05 2002 – 16:42] Matthew (email) which only goes to show how stefan, despite clinging to seemingly objective scientific language such as “rational” and “game,” is a living example of what krauthammer pointed out in his column: that the language of the liberal left is self-fulfilling and circular, and by itself not a convinving argument for anything. liberals are “rational” and “cooperative” and therefore conservatives are irrational and uncooperative. neither description is necessarily true nor are they particularly helpful ways to understand human behavior.

[Mon, Aug 05 2002 – 16:53] Matthew (email) also, just noticed something else on stefan’s original blog. he says life is full of “positive sum games, or prisoner dilemmas.” err, please someone correct me if i’m misremembering, but the prisioner’s dilemma is a very specific game in which the outcome is almost always zero-sum; meaning, one prisioner loses if the other wins or both lose. there are very few likely outcomes, because they can’t collude, in which they both win.

[Mon, Aug 05 2002 – 16:55] Matthew (email) also, just noticed something else on stefan’s original blog. he says life is full of “positive sum games, or prisoner dilemmas.” err, please someone correct me if i’m misremembering, but the prisioner’s dilemma is a very specific game in which the outcome is almost always zero-sum; meaning, one prisioner loses if the other wins or both lose. there are very few likely outcomes, because they can’t collude, in which they both win.

[Fri, Aug 09 2002 – 05:44] eurof (email) WRONGGGG! Silly Matthew. Stefan is right, you are SO stupid. “Prisoners dilemmas” have 4 outcomes: 1)prisoners keep stumm, neither can be convicted, both walk free. 2)one prisoner squeals, other stays stumm, he gets 5 years, the other gets 10 years, 3) the same, other way round, 4) both squeal, both get 10 years. So they can both win, and it’s not in any way a zero sum game, where one can only win by making another lose. The point is, it often turns out zero sum, because the first time people play it they generally both squeal. If they play it again, they tend to try not squealing even if they aren’t able to collude beforehand, which works, and they generally go free. They keep with this winning strategy if they play a 3rd time.

So stefan is largely correct i think. Not so sure there’s a political application, but it’s an interesting idea i guess.