Wednesday, April 6, 2011

Mythology Meets Game Theory

Game theory is used quite often in economics. The "prisoner's dilemma" is used to analyze many phenomena like collusion between oligopolists and trade negotiations.

What appears below is an excerpt from the 1991 book titled Prisoner's Dilemma by William Poundstone. When he refers to "Axelrod's findings" he means what Robert Axelrod found from staging a prisoner's dilemma tournament.

It turns out that the best strategy is called "tit-for-tat," meaning a player should cooperate or try to help the other player in the beginning. If the other player "defects" or tries to take advantage of you, then you punish him by defecting. If they start cooperating again, so do you. Without getting into too much detail, this is the best strategy in a multi-player tournament.

The passage below relates to international conflicts and the part about myth-making is in red. It refers to "chauvinist mythmaking" and how these local or national myths make it hard for the tit-for-tat strategy to work in international relations. I can't help recalling Joseph Campbell and his call for a world myth that everyone can believe in.


Many have expressed hope that Axelrod's findings might be applied
to human conflicts. One would like to think that statesmen and mili-
tary leaders would take a course in "practical TIT FOR TAT" and
suddenly much of the world's problems would be solved.

Axelrod himself downplays the idea. When I asked him if he
thought his findings could be translated into advice for statesmen, he
insisted that wasn't the goal. "I think the goal is to help people see
things more clearly, which is different. The value of any formal model,
including game theory, is that you can see some of the principles that
are operating more clearly than you could without the model. But it's
only some of the principles. You have to leave off a lot of things, some
of which are bound to be important."

Part of the problem with advising anyone to start using TIT FOR
TAT in foreign relations is that, in a sense, most reasonable people
already do it without knowing it. Responsible leaders don't start trou-
ble, and are provocable. The practical difficulty is not so much in
knowing when to cooperate or defect but to decide what is going on. In
the real world, it is not always obvious whether someone has acted
cooperatively or defected. Actions can fall somewhere between the two
extremes, and it is frequently unclear what one's adversary has done.
When one cannot tell what the other player has done, it is impossible
to use any conditional strategy.

Since the 1950s, the United States and the Soviet Union have used
a "tit-for-tat" policy-often called by that name, which predates Axel-
rod's studies-in granting travel permits to citizens of one nation resi-
dent in the other. This appears to be a genuine case of a TIT FOR TAT
strategy evolving spontaneously. In 1990, after a American diplomat
in Leningrad was denied permission to travel to Lithuania, the U.S.
State Department revoked a permit for Gennady Zolotov, Soviet dep-
uty consul general in San Francisco, to travel to give an unrelated and
uncontroversial speech at a small college in Nevada. State Depart-
ment spokesman Chuck Steiner explained, "It wasn't retaliation. It
was just an in-kind response. They denied our request, so we denied
theirs. It's been a long-held rule between the two countries."

In an uncertain world, TIT FOR TAT-like strategies may be as
much a part of the problem as the solution. It is an all too familiar
phenomenon of real conflicts that both sides claim the other started it
and that they were just giving a tit for a tat. Conflicts escalate mutu-
ally. A war of words leads to a war of gunfire and then of air raids.
Each side can truthfully cast the other as the side to cross the thresh-
old of war provided it gets to decide where that threshold lies. Ber-
trand Russell claimed that there was only one war in the history ofthe
world for which people knew the cause: the Trojan War. That was over
a beautiful woman; the wars since have lacked rational explanation
and have produced nothing, Russell said. In Axelrod's abstract game,
there is never any question about who was first to defect, and in this
sense it is unrealistic.

Writing in World Politics (October 1985) Stephen Van Evera consid-
ered whether TIT FOR TAT or a similar strategy might have pre-
vented World War 1. He concluded it could not. He said:

Tit-for-Tat strategies require that both sides believe essen- tially the same history; otherwise the players may be locked into an endless echo of retaliations as each side punishes the other's latest "unprovoked" transgression. Because states seldom believe the same history, however, the utility of Tit-for-Tat strategies is severely limited in international affairs. Strategies to promote in- ternational cooperation through reciprocity may therefore require parallel action to control the chauvinist mythmaking that often distorts a nation's view of its past .... In sum, because conditions required for successful application of a Tit-for-Tat strategy were missing in 1914, Europe was an infertile ground for Tit-for-Tat strategies. These conditions are often absent in international affairs; the syndromes of 1914 were merely pronounced varieties of common national maladies. It fol- lows that we will fail to foster cooperation, and may create greater conflict, if we rely on Tit-for- Tat strategies without first establish- ing the conditions required for their success.
How much is game theory presently used in diplomacy? The answer
appears to be very little. Axelrod speculated that "I think you can say
that [Thomas] Schelling's work is well known and was probably help-
ful in establishing some of the ideas we have on arms control. But at
the very top, you probably cannot find a secretary of state who can tell
you what a prisoner's dilemma is."

Axelrod finds game theory's influence more diffuse: "Some of the
ideas of game theory are in the public domain very much now, so that
somebody can be influenced by them. I think everybody really does
know what a non-zero-sum game is. You can use that term in News-
and not even explain it anymore. Just that is a major intellectual
advance because we're so prone to think in zero-sum terms."


  1. This comment has been removed by the author.

  2. I saw an interesting documentary about game theory, how it was developed and how it was used, and what the consequences could be.
    First, the game theory was developed by people that were suspicious by nature. They played it safe.
    Second, people's natural state of mind, is in stead to trust. Therefore, you will not see a secretary playing a rational game like the tit-for-tat. They will always collaborate, no matter what. Imagine them playing game theory with their boss. Will he know the rules of the games or will he only take it as an insult.
    Third, studies have shown that only certain people make rational choices: psychopaths and economists.

    Curtis documentary, "F*ck you body" is a great reference, nicely put.:)

  3. Here is the link.

  4. Carmen

    Thanks for reading and commenting. I will have to take a look at that video.

    When you say only economists and pyschopaths make rational decisions, is that all the time or just in the prisoner's dilemma game?

    It might be right that people are naturally trusting. That seems to be what neuroeconomics is finding (although I don't know that much about it).

    Do you think secretaries would keep collaborating even after the other side continually cheats them? I think eventually they would stop collaborating.


  5. Carmen

    I hope you are doing well, too.


  6. Thank you Cyril. I am fine.. one could say. I sure hope that you are well too. Sorry for not being around that much, i have too many webpages and it's difficult to keep track of everything.

    I can't back up this one, but I guess that economists will play tit for tat mostly in their strategies. They are more aware then the rest of the people about losses and gains. They have a greater understanding. In real life, those things are more subtle, and the tendency to forgive and forget is greater. You must be very calculating and cold in order to remember what somebody has done to you many years ago and to act according to tit-for-tat.

    I remember reading many years ago about the iterated prisoner's dilemma as well. How to you handle a situation where nobody will ever collaborate with a "target"?

    In this case, a rational person will never collaborate back, provided that they knew about it. Yet, people will do that anyway, because people in general do not make rational choices. Applied to for example the Holocaust, the Jews, that mostly collaborated in their history, being the best strategy for surviving, in the Holocaust situation, collaborating with the Nazi forces was a clear mistake, because they ended up dead. The best strategy would have been to revolt and choose non-collaboration at any stage. Many knew that they would end up in concentration camps, yet they still hoped that collaboration would be rewarded.

    I think that a secretary would collaborate to great lengths with a superior. According to the famous Milgram's "Obedience experiment", people would do many things for authority and in it's name.

  7. No need to apologize. I'm flattered that you take the time to read my blog and comment.

    That video is interesting but it tries to cover alot in 10 minutes in does not go deeply into anything.

    Yes, I agree that economists "more aware then the rest of the people about losses and gains." I think this plays out in government policies. Lots of people propose new programs, citing their benefits. But it is usually the economists that remind people that they all have costs and that if you produce more of one good you will have to give up some of another good.

    I don't know if tit-for-tat applies over a long time period. It says that if a person defected, you punish them by defecting, too. But if they start playing nice, you forget about it and start playing nice again yourself.

    I think tit-for-tat says that if the other person does not collaborate (or cooperate), then you don't either. It will keep going on. But no one, not just game theorists, know how to make a mean person (psychopath) into a nice person.

    I don't know much about Jewish history. My guess is that if they had tried to fight back against the Nazis, they all would have been killed.

    And I don't know if the prisoner's dilemma applies to a supervisor-employee relationship. If a boss is mean, an employee can try to find another job.

  8. It's a pleasure actually to read your blog and everything that you have written!

  9. Game theory applies everywhere and there are many many "game types"... it depends a little on how the actors are in the game, and not the game itself. The game is mostly used as a way of understanding social interaction (that is what I understood from what I read), and yes, I guess that people play it withot being aware of that.