Arnold Kling  

Pacifism and Repeated Prisoner's Dilemma

PRINT
Jail a Policeman?... Think on the Margin...

Tyler Cowen argues with Bryan on pacifism.


There is also a Lucas critique issue of how the bad guys start behaving once they figure out that the good guys are pacifist, and I don't see him discussing that either.

I do not know how relevant this is, but I believe there is a result that in tournaments of repeated-Prisoner's-dilemma games, the "tit-for-tat" strategy works the best. I don't have a reference at my fingertips, but I think I recall reading this in Howard Rheingold's Smart Mobs, and he would have had references.

These are games in which the best outcome is "cooperate/cooperate," but for each individual the best outcome comes from defecting when the other person cooperates. In repeated games, one strategy is to play "tit for tat." When the other player cooperates in this round, next round you cooperate. When the opponent defects this round, next round you defect. As I understand it, this strategy works best against opponents, either human or computers, who follow a variety of strategies.

I am not sure what tit-for-tat means in the context of international relations. But perhap one example of a state that plays that way is Switzerland. Famously neutral if you leave it alone, but well armed if you want to attack it. As Tyler says, that has worked well.

As a practical matter, I think that a Swiss approach for the U.S. would be, like pacifism, quite a departure from the way that the U.S. has conducted itself for much of (all?) its history.


Comments and Sharing





COMMENTS (24 to date)
Lint writes:

That was the result of the tournaments from The Evolution of Cooperation by David Axelrod.

John Thacker writes:

There are also some results indicating that if you allow for some percentage of miscommunication (someone thinks that you defected when you meant to cooperate), then "tit for tat with some chance of forgiveness" ends up winning.

Bret writes:

Tit for tat, eh?

What if "tit" is total annihilation? Then how do you pull off "tat"?

"Tit" and "tat" are rarely symmetric in the real world.

Scott Wentland writes:

I think Bret is right. I don't see a close tie between repeated prisoner's dilemmas and war. It's much closer to a one-shot game.

So Bryan is still on the hook for explaining his "cooperative" pacifism in the context of a one shot prisoner's dilemma.

frankcross writes:

Well, I believe the Berger Commission found that Switzerland refused to accept Jewish refugees and made interest free loans to the Nazis, among other things. Is that your tit for tat? In exchange for you not invading, we will decline to be humanitarian and give you money?

Emerson White writes:

This sounds a lot like when advocates of Abstinece only sex education hold up the benefits of an abstinence plus sex education with reduced pregnancy and STI rates. Tit-for-Tat is not pacifism, always cooperate is pacifism. Tit-for-tat is retaliatory.

Howard writes:

Switzerland is a free rider that benefits from peace bought by the blood of others, or the franc ensured by stability fought for by other nations (including the stability enforced in the Middle East). Theoretical arguments like the one made by Kling above (who I like to read otherwise) lose the smell test right out of the bat. "well armed: Swiss, indeed. That's a hoot.

Aeon J. Skoble writes:

Actually, tit-for-tat has been superseded (sp?) by a stragety called "Pavlov" or "win-stay/lose-shift," which provides for benevolence + retalitation + forgiveness. See Martin Nowak and Karl Sigmund, "A Strategy of win-stay, lose-shift that outperforms tit-for-tat in the Prisoner's Dilemma game," Nature, vol. 364, 1 July 1993. (Or my book, where I discuss Pavlov in relation to Axelrod's tit-for-tat.)

Pandaemoni writes:

Wouldn't a repeated Hawk-Dove game be more apt? My recollection is that pacifism did not work there either as a evolutionary stable strategy, but that game does not model two important aspects of Caplan's premise — namely, (1) that fighting back damages innocent civilians, not just the other hawk and (2) that fighting back encourages greater aggression in the future.

I don't think Caplan is opposed to any self-defense, but rather defense that injures others who are not otherwise directly involved. Two person game theory is likely not a good model for those dynamics. You could build them into the payouts for these games (i.e. if being an aggressive hawk gives you a certain benefit, discount that by the expected future costs that the aggression will foist upon you). I don't know that doing so would lead to any analytically interesting insights, though.

Kevin Dick writes:

Axelrod was certainly who I had in mind when I commented about evolutionarily stable strategies in Bryan's last post. But I assume there has been significantly more work since I read Axelrod in 1990.

Actually, tit-for-tat in an international relations context is easy to imagine and I recall it being one of the primary areas of application in the seminar where we read Axelrod.

When an adversary makes an aggressive move, such as occupying an island or moving troops to the border, you make an aggressive counter move. Rarely does a country simply roll tanks across the border with no warning whatsoever.

Dave writes:
Jim Glass writes:

I believe there is a result that in tournaments of repeated-Prisoner's-dilemma games, the "tit-for-tat" strategy works the best.

Correct. But nobody has to go into such theoretical niceties to understand the incentives to behavior directing different kinds of societies when they meet.

The main problem with pacifism as a method for a libertarian society to use in interacting with the rest of the world is it assumes that pacifism is, for the rest of the world, the course of rational self-interest, so the rest of the world will follow it too.

This is totally false. A libertarian society -- heck, any advanced civilization, even as *non*-libertarian as the 3rd Century Roman Empire -- is highly advanced in relying on peaceful co-operation in coordinating its economy and institutions to build its high wealth (relative to the rest of the world).

For the poorer rest of the world, as *amply* documented in all history all the world over -- Sargon's east, Persia, Greece, the Vandals and Goths taking Roman territory, Saxons taking Britain, The Russes, the Golden Horde ... Stalin taking Eastern Europe, Saddam rolling into Kuwait, etc. -- the true policy of rational self interest is "you have more stuff than I have, I have more power than you have, I'm taking your stuff." People follow incentives! The incentives there are very clear.

Pacifism may be a credible policy within a libertarian society that has culturally developed a sufficiently high level of internal cooperative behavior as its social norm. Internally. Well, someday it may be (we can hope).

But the external *border* of that society had better be damn well protected, strongly enough to make "I'm taking your stuff" a clearly non-rational policy for all outside the border who consider it. As Clausewitz said, if you want peace the best policy is to be very strong.

AMW writes:

As Clausewitz said, if you want peace the best policy is to be very strong.

Actually, that runs completely counter to your prior argument. Because a strong nation has an incentive to take the stuff of weaker nations. True, a strong military makes it relatively expensive for others to try to impose their will on your country. But it also makes it relatively inexpensive for your country to impose its will on others. And that means plenty of wars, though they may be against fairly impotent foes.

If you want peace, the best policy is to have little power to project force but great power to repel force. (Admittedly, in a nuclear age maybe you need some ICBMs. But they should be the *only* credible way you can strike someone a continent away.)

ajb writes:

It is problematic because Axelrod's work on Tit for Tat showed that retaliatory strategies worked well only in tournaments he ran. But theoretically Tit for Tat isn't stable (I believe Nachbar still has the standard theory paper on this in dynamic games). In that world you either get All Hawks or All Doves, but that mixes are not stable. [Or more accurately, you can get a lot of different equilibria in infinitely repeated games but Tit for Tat doesn't stand out.]
Certainly not Doves with a few Hawks. Of course, you can dismiss the theory then you're rejecting arguing from first principles. Whereas Caplan wants to dismiss the empirical relevance of the daily newspaper (as MR points out) while arguing from his rather blinkered notions of cost benefit and morality.

Kevin Dick writes:

@ajb

I vaguely remembered the explanation for the difference between the Nachbar and the Axelrod results was interesting. So I looked it up.

Nachbar's result appears to apply to finite repeated games. Axelrod's simulation was also finite, but he attempted to induce the perception that it was infinite.

Do you have a pointer on the theoretical analysis of the infinitely repeated version? I'm curious how all this stuff worked itself out in the last 20 years.

Kevin Dick writes:

I found this review of how Axelrod's work has stood up, complete with references.

http://jasss.soc.surrey.ac.uk/3/2/forum/1.html

ajb writes:

I don't know this lit at all. But isn't the infinite game just the Folk Theorem stuff?

Adam writes:

I vote for Jim Glass's argument. AMW objects since the strong country would then find it easy to 'take the stuff' of the weak. But any wealthy, strong and rational people realizes that wealth cannot be taken by force, but only created by trade. Force destroys the high value stuff. It's better for all to trade.

Remember Adam Smith? Long ago he argued that the British monarchy was better off to grant independence to its colonies rather than pay the continuing expense of military subjugation.

Jim Glass writes:

Actually, tit-for-tat has been superseded (sp?) by a stragety called "Pavlov" or "win-stay/lose-shift," which provides for benevolence + retalitation + forgiveness. See Martin Nowak and Karl Sigmund,

Yes, but as far as pacifism as a strategy against external powers goes, one doubts that even this strategy if available would have deterred the Mongols from pillaging Kiev. The game played by one-time pillagers, and one-time permanent conquerors, is a different one.
~~~~~~~~

"As Clausewitz said, if you want peace the best policy is to be very strong".
Actually, that runs completely counter to your prior argument. Because a strong nation has an incentive to take the stuff of weaker nations... If you want peace, the best policy is to have little power to project force but great power to repel force.

Which may be why I said:

the external *border* of that society had better be damn well protected
But beyond that, the larger point is that even if Pacifismland has as a libertarian society developed to the high point where by its advanced values it would truly *never* attack another, it *still* needs to be able to project force sufficiently to be able to defend itself, until all the rest of the world has reached the same high point.

Else all the rewarding fruit that it has developed for its own people via its social advancement will be very low-hanging and ripe for the plucking by everybody else.

Don Levit writes:

Jim:
I agree with your assessment of needing sufficient force to defend ourselves.
Unfortunately, if we get desperate enough, for example, for oil, we become just like the lesser thugs you are describing.
In addition, we need to be concerned about the internal enemy, the disparity in income and wealth.
Why would our society be any more civil than others in history?
Well, I guess you could say we were "civilized" during the Great Depression.
Don Levit

B.B. writes:

Silly old me.

I keep remembering that Belgium declared itself neutral in World War One and in World War Two. How did that work out?

The Netherlands declared itself neutral when World War Two broke out. How did that work out?

I guess neutrality is not as easy as it seems. It may be that the Swiss Alps are easier to defend that the open plains of northern Europe.

The US was a declared neutral in both World Wars, if anyone wants to bother with history. How did that work out? The Axis powers declared war on the US in both circumstances. I have to wonder whether if the US had had a much larger and stronger military prior to those wars, would the Axis powers have wisely decided to leave the US alone? If you want peace, prepare for war.

And if a strong military means you want to conquer your neighbors, why are Canada and Cuba still independent? You miss the difference between a strong democracy and a strong autocracy.

As for the Cold War, a strong military and tit-for-tat meant that the US and USSR never went directly to war, and the US outlasted the USSR.

AMW writes:

I vote for Jim Glass's argument. AMW objects since the strong country would then find it easy to 'take the stuff' of the weak. But any wealthy, strong and rational people realizes that wealth cannot be taken by force, but only created by trade.

The European powers were all wealthy, strong and rational. And they fought each other like hell.

Silly old me.

I keep remembering that Belgium declared itself neutral in World War One and in World War Two. How did that work out?

The Netherlands declared itself neutral when World War Two broke out. How did that work out?

Arguably, much better for them than for Russia and Britain.

The US was a declared neutral in both World Wars, if anyone wants to bother with history. How did that work out?

In the case of WWI it's laughable to think that American losses from U-boats sinking our merchant ships was anything like an existential crisis, or anywhere near as costly in blood and treasure as the actual war. In the case of WWII the United States may have declared neutrality, but its actions clearly showed that it was not.

And if a strong military means you want to conquer your neighbors, why are Canada and Cuba still independent? You miss the difference between a strong democracy and a strong autocracy.

You'll get no argument from me that democracies tend to be less belligerent than autocracies. But a brief perusal of American military interventions post WWII clearly demonstrates that a democracy with a massive military can get itself into a lot of conflicts.

Adam writes:

AMW: Yes, the Europeans fought each other back and forth for years. Of course they did!--the dominant idea was mercantilism, so the only way to get richer was to steal the wealth of one's neighbor.

The idea of 'mutual gains from trade' remains unrecognized by most of the world as the alternative to theft and war. Mercantilism fed by envy is an all to common theme in everyday talk, in political speech and in formal state policy. We hopeful libertarian economists have huge educational task to accomplish.

Tracy W writes:

AMW: Re-answer that question about the outcome for the Netherlands bearing in mind what happened to Dutch Jews compared with British Jews.

And also note that the Nazis, being racists, were planning on, after killing all the Jews, to kill tens of millions of Slavs, to expand living space for Germans. The outcome for Russians would have been far worse than for the Dutch if they had chosen not to fight.

And finally, the Nazis' economic policies were crazy, even from the point of view of Aryan Germans, let alone from the point of view of anyone else. Submitting to the Nazis meant significantly increasing your risk of dying of poverty.

Comments for this entry have been closed
Return to top