Wednesday, October 12, 2005

previous entry | main | next entry | TrackBack (0)

Thomas Schelling gets his due from Sweden -- but not from Slate

My favorite class to teach in recent years has been Classics in International Relations Theory. This is a great books course, starting with Thucydides' History of the Peloponnesian War and ending with Thomas Schelling's Strategy of Conflict.

The reason this is my favorite course is the effect it has on the grad students, who consume a very steady diet of literature that is supposed to be "cutting edge." They are therefore shocked to discover that the modern version of democratic peace theory bears little relationship to Kant’s original formulation, for example. However, they are always stunned to learn that whole careers in international relations have been built out of codifying a few sentences in Schelling. [Oh yeah, and you're not guilty of this?--ed. I'll plead not guilty on Schelling, but nolo contendre with regard to another Nobel-worthy economist.]

So it's wonderful news to read that Schelling has co-won (with Robert Aumann) The Bank of Sweden Prize in Economic Sciences in Memory of Alfred Nobel "for having enhanced our understanding of conflict and cooperation through game-theory analysis." Kieran Healy has a good post up detailing the relative contributions of Schelling and Aumann. Tyler Cowen has a lovely post up (one of many) about his old Ph.D. advisor.

In Slate, Fred Kaplan tries to throw some cold water on Schelling's Nobel, pointing out:

Today's papers note his ingenious applications of "game theory" to labor negotiations, business transactions, and arms-control agreements. But what they don't note—what is little-known in general—is the crucial role he played in formulating the strategies of "controlled escalation" and "punitive bombing" that plunged our country into the war in Vietnam.

This dark side of Tom Schelling is also the dark side of social science—the brash assumption that neat theories not only reflect the real world but can change it as well, and in ways that can be precisely measured. And it's a legacy that can be detected all too clearly in our current imbroglio in Iraq.

Alas, Kaplan commits the very sin he accuses Schelling of making -- providing an overly neat theory of how Schelling contributed to U.S. policy in Vietnam. Kaplan's own description of Schelling's role in Vietnam contradicts his claim:

[Assistant Secretary of Defense John] McNaughton came to see [Schelling]. He outlined the administration's interest in escalating the conflict in order to intimidate the North Vietnamese. Air power seemed the logical instrument, but what sort of bombing campaign did Schelling think would best ensure that the North would pick up on the signals and respond accordingly? More broadly, what should the United States want the North to do or stop doing; how would bombing convince them to obey; how would we know that they had obeyed; and how could we ensure that they wouldn't simply resume after the bombing had ceased?

Schelling and McNaughton pondered the problem for more than an hour. In the end, they failed to come up with a single plausible answer to these most basic questions. So assured when writing about sending signals with force and inflicting pain to make an opponent behave, Tom Schelling, when faced with a real-life war, was stumped.

He did leave McNaughton with one piece of advice: Whatever kind of bombing campaign you end up launching, it shouldn't last more than three weeks. It will either succeed by then—or it will never succeed.

The bombing campaign—called Operation Rolling Thunder—commenced on March 2, 1965. It didn't alter the behavior of the North Vietnamese or Viet Cong in the slightest. Either they didn't read the signals—or the signals had no effect.

In this description, there's not a whole hell of a lot of brashness -- indeed, Schelling's recommendation was not to escalate Rolling Thunder if the initial bombing didn't work. In Kaplan's passage, Schelling appears to be acutely aware of the difficulties of measurement in applying his theory of compellence to Vietnam. He made a recommendation, but with none of the hubris Kaplan associates with social science (Kaplan also elides Schelling's leadership in a subsequent attempt to convince then-NSC adviser Henry Kissinger to withdraw from Vietnam in the early days of the Nixon administration).

Kaplan's essay contains a grain of truth about the dangers of social science. Too often, theorists come up with great models of the world by assuming away petty inconveniences like bureaucratic politics, implementation with incomplete information, or the effects of rhetorical blowback. But before he throws out the baby with the bathwater, Kaplan might want to ask himself the following question: if policymakers choose not to rely on social science theories to wend their way through a complex world, what navigational aid would Kaplan suggest in its stead? Policymakers across the political spectrum always like to poke fun at explicit theorizing about international relations. The problem is that they usually rely on historical analogies instead -- which are, in every way, worse than the use of explicit theories.

UPDATE: Tyler Cowen quotes Business Week's Michael Mandel on the drawbacks of game theory:

Game theory is no doubt wonderful for telling stories. However, it flunks the main test of any scientific theory: The ability to make empirically testable predictions. In most real-life situations, many different outcomes -- from full cooperation to near-disastrous conflict -- are consistent with the game-theory version of rationality.

To put it a different way: If the world had been blown up during the Cuban Missile Crisis of 1962, game theorists could have explained that as an unfortunate outcome -- but one that was just as rational as what actually happened. Similarly, an industry that collapses into run-amok competition, like the airlines, can be explained rationally by game theorists as easily as one where cooperation is the norm.

Tyler has a number of responses (to which Mandel responds) but mine is simple: game theory has the wrong name. It is a theoretical tool rather than a theory in and of itself. Because of this, Mandel is correct that it is possible to devise game-theoretic models that lead to contrasting predictions. However, the virtue of game theory is that the differences made in starting assumptions, institutional rules, and causal processes are laid bare. One can then argue about how realistic the assumptions, rules, and processes are.

ANOTHER UPDATE: Mark Kleiman points out and explains why the blogosphere is united in its high regard for Schelling.

posted by Dan on 10.12.05 at 11:34 AM


Terrific post. As someone who also teaches international relations and includes a large Schelling component, I think the award is well deserved. Your final thoughts can be summed up by the first rule of wing-walking (don't let go of anything until you have something more useful to hang on to).

All of this begs the question of HOW policymakers should use social science theories (of which historical analogies are one category, see Khong's Analogies at War.). Here the ivory tower dwellers make the mistake all too often of ignoring/hiding/moving the boundary conditions within which their models are useful.

posted by: anon on 10.12.05 at 11:34 AM [permalink]

Kaplan at Slate is almost a complete waste of time. The man's background is a jazz writer, for Pete's sake.

posted by: Don Mynack on 10.12.05 at 11:34 AM [permalink]

Excellent post. Kaplan criticizes the approach of the political scientist, but concludes with:

"If Donald Rumsfeld and Paul Wolfowitz had studied history better, they, too, might have appreciated those limits before chasing their delusional dreams into the wilds of Mesopotamia."

Instead of the social science approach, Kaplan recommends applying lessons learned from past human experiences to current problems. What is social science itself but a systematic application of this very principle!? Furthermore, Kaplan's final conclusion is certainly misguided. There seems to be little that Wolfowitz and Rumsfeld could have learned from Operation Rolling Thunder. Iraqi Freedom was/is primarily a ground force operation and would not entail the same considerations as a punitive bombing campaign. If anything, the historical record would have cautioned against the U.S./NATO punitive bombing of Yugoslavia.

As a sidenote to Mr. Mynack: while I too am often disappointed with Fred Kaplan's posts to Slate, reducing his resume to jazz writing is an oversimplification. He spent considerable time with the Boston Globe as its military correspondent, Moscow bureau chief, and New York bureau chief.

posted by: Joe Rose on 10.12.05 at 11:34 AM [permalink]

"[W]hat navigational aid would Kaplan suggest in its stead?"

Why, himself, of course!!! (Unless he had to take any responsibility.)

posted by: lancer on 10.12.05 at 11:34 AM [permalink]

"It is a theoretical tool rather than a theory in and of itself."

Yes, much in the same way that statistics is not a theory, just a tool to formulate, and/or test hypotheses, or even to help guide us to fruitful places to look for causal pathways.

posted by: Robin on 10.12.05 at 11:34 AM [permalink]

I thought Kaplan was unfair too - well-informed, but unfair. If anyone is looking for a counterpoint I have a short piece praising Schelling in today's Financial Times:

posted by: Tim Harford on 10.12.05 at 11:34 AM [permalink]

Kaplan at Slate is almost a complete waste of time. The man's background is a jazz writer, for Pete's sake.

To be fair he does have a PhD in Political Science from MIT and wrote Wizards of Armageddon--although I thoroughly disagree with his analysis...

posted by: bp32 on 10.12.05 at 11:34 AM [permalink]

Count me one less than overwhelmed. I've read a lot of game theory, from von Neumann & Morgenstern through Luce and Raiffa and into the more recent literature. I read a good bit of Schelling and found him less than convincing.

posted by: anonymous on 10.12.05 at 11:34 AM [permalink]

Congratulations to Schelling, and great commentary on Kaplan.
As a simple 'bang for buck' military practitioner of the art, I found best of all is your reminder that Game Theory is just a tool in the artist's toolbox, albeit an invaluable one.

posted by: vexare on 10.12.05 at 11:34 AM [permalink]

Kaplan is infamous for great details and illogical and inconsistant conclusions (as you point out). He could be a great researcher it he would only keep his own ideas out of the way. His views in "Wizards " in most instances are just plain wrong.

posted by: jim on 10.12.05 at 11:34 AM [permalink]

Its unfair to single out games theory for its lack of predictive power. Or more accurately its over-abundence of conflicting predictions. Any science associated with human behavior suffers from the same problem.

Games theory relies on rational players to create definitive predictions, and we all know that is not always the case with human society. It also often relies on the players being aware of the results of their decisions many moves ahead, which again is unrealistic.

Worst of all, Games Theory sometimes presents scenarios where the most beneficial results come from both parties acting 'irrationally' (or at least against their certain best interest). Humans understand this instinctivley, while logical formulations require some serious contortions to get there. Cooperation begs for betrayal, but ultimately their could be no human society as we know it without cooperation.
I would say that games theory teaches us more about why we are the way we are than what we can do to take advantage of same. Or to put it in Sun Tsu's language, it allows us to understand ourselves rather than understand our enemy. That perhaps is why it fails when applied to 'how can we coerce our opponent'.

posted by: Mark Buehner on 10.12.05 at 11:34 AM [permalink]

It would be useful if those using Vietnam as an example of how the military can't solve political problems actually knew the history of the war.

By 1973 the war was won except for one small problem. The South Vietnamese had an inadequate air force and depended on promised American support to make up for that lack.

In 1975 the Democrat US Congress forbid the US military from offering that support (and resupply).

The short version: the Democrats gave the South to the Communists after Nixon's policies had won the war and removed most American ground troops from South Vietnam.

It seems that our Copperheads (Democrats then too) are still at it.

posted by: M. Simon on 10.12.05 at 11:34 AM [permalink]

Similarly, an industry that collapses into run-amok competition, like the airlines, can be explained rationally by game theorists as easily as one where cooperation is the norm.

Yes, competition CAN be explained far more rationally than government directed trusts or cartels. I don't think it would be "rational" to go back to the good-old cooperative days of the Civil Aeronautics Board, plums and dogs.

posted by: HA on 10.12.05 at 11:34 AM [permalink]

Games Theory is a tool. Where inputs are qualitative, e.g. when dealing with macro-economics while ignoring "behavioural" (non-rational) aspects, the tool posits plausible outcomes-- in fact, not meaningful, because participants' views of basic inputs are essentially random. Non-definitive, they can go either way.

Where inputs are quantitative, especially in "real-time", the games-theoretical tool contributes great advantage. The classic case was Israel's destruction of Syria's entire Air Force in twenty minutes (early 1980s), assisted by American AWACS platforms orbiting in the Eastern Mediterranean. Israeli fighters vectored in on the Syrian formations, double-teaming as AWACS programs indicated... in his final days, Leonid Brezhnev was informed that unless his Soviets somehow dealt with AWACS "games", Communism's aggressive salients would be forever curbed. End of Evil Empire.

So "prisoner's dilemma" scenarios are mere adjuncts to Games Theory. In proper context, its "multiple simultaneities" may prove invaluable. Out-of-context, which means in most specific cases, von Neuman's baby can kill as well as cure.

posted by: John Blake on 10.12.05 at 11:34 AM [permalink]

Combined with your prior "little finger" post, I am constantly enraged by the continuing Anti-(Vietnam) War sentiment, as if "anti-war" was good.

Anti-war meant the US failed to bring democracy to S. Vietnam (while succeeding, slowly, with S. Korea) -- and S. Vietnamese lost "their" country, and many their lives.

M. Simon noted it -- the US Democrats voted to let commies win; after having been told there would be a bloodbath. There was a bloodbath.

Whether this was a Game Theory result or not, the ability of pointy heads to ignore the actual results of the anti-war policy, when the policy is actually tried, such ignoring of reality is terrible.

Despite your own final support for the implicitly pro-commie, pro-genocide Kerry, I'm sorry you're not getting tenure; and don't know why.

I don't know of any game theories comparing questionable competence with good payoffs in policy versus definite competence with questionable payoffs in policy. (How many SE Asians would have to be killed before the US leaving Vietnam is a "mistake"?)

posted by: Tom Grey on 10.12.05 at 11:34 AM [permalink]

Isn't one of the arguments of people influenced by Schelling that the way you end a war is as important as the way it's started? That an ounce of threat is worth a ton of force? And that you don't want to destroy the enemy before negotiating the peace, lest you're stuck without anyone to negotiate with? See Freddy Lee, aka Fred Ikle, for this.

posted by: bat on 10.12.05 at 11:34 AM [permalink]

Kaplan says "And it's a legacy that can be detected all too clearly in our current imbroglio in Iraq."

How is it that any initiative the US takes is an "imbroglio" unless it lasts only a day? Why is it never that our enemies are in a current imbroglio instead. Surely even Kaplan understands that the war against the terrorists is not of our choosing and cannot be ended by our choosing to do so. The enemy has made it clear that they will wage this war everywhere, regardless of what we do. Invading their home turf to force its waging there was a stroke of brilliance and not the delusional dreams of Rumsfield and Wolofitz. The terrorists must fight us there at all costs, using all the manpower and resources they can muster. If they fail to fight, the two things they fear most, democracy and freedom will take root and force them out. To continue the fight they must dedicate adequate resources to do so and thus are imbroiled in an imbroglio for as long as we wish them to be so.

posted by: willis on 10.12.05 at 11:34 AM [permalink]

One of the things that I appreciate about Schelling's study of game theory (and one thing that is all too often lost with those who have used it in Poli Sci since) is that Schelling appreciated the idea that rationality occured within a social structure. You have to build a set of rules in which a rational actor can pursue bargaining strategies. For instance, take his famous discusson of Focal Points. The idea of a focal point as the rational point of convergence is completley contingent on a shared set of cultural understandings. Only within that shared culture does it make sense. Schelling related the story of how he came up with the idea in an NPR interview the other day-- its all about him and his friends figuring out an emergency meeting point. They needed to pick generic points in a city-- a place that every city had, had only one of, was accessible, and that they might already go given a promise they had made to their parents earlier.

Schelling had a fine appreciation of the cultural rules that constructed rationality.
Within those rules, game theory produces what is almost an "ideal type" to understand social interactions.
I think too many contemporary game theorists forget this.

posted by: peter on 10.12.05 at 11:34 AM [permalink]

The main point of game theory is not to predict, but to give participants a strategy that optimizes their own outcomes. A lot of pro poker players are using it these days.

@Mark Buehner: Apply game theory to situations with continuing relationships, and "tit for tat" strategies become optimal...and if everybody follows it, you end up with high cooperation.

That said, humans do cooperate more than game theory recommends...why they do this is an active area of research.

posted by: Dennis on 10.12.05 at 11:34 AM [permalink]

Sidebar: Rolling Thunder makes for good reading in these books:

Going Downtown by F-105 pilot Jack Broughton

and any of the Vietnam war novels of former F-4 Phantom pilot Mark Berent.

posted by: The Sanity Inspector on 10.12.05 at 11:34 AM [permalink]

I see the usual irredentist fanatics are out in force to defend the Vietnam War.

M. Simon says that "The South Vietnamese had an inadequate air force and depended on promised American support to make up for that lack." That is precisely the point. South Vietnam was never a viable state. It always depended on American aid to survive. As soon as that aid was cut off, it collapsed. Recall that we had been pouring our blood and treasure down that rathole for 25 years by this point. How much longer were we supposed to keep it up? Blaming "Congressional Democrats" won't do. These Democrats were doing the will of the American people, who were sick and tired of the war by that point. They gave our politicians every chance to win the war, but it had become clear that it could never be won - no matter what we did, the corrupt and kleptocratic "government" of South Vietnam would never be self-sufficient. If not 1975, when? Should we have kept sending our young men there to fight and die until the USSR collapsed? What if the North Vietnamese still hadn't given up? Should we still be there?

The Vietnam War, like its successor in Iraq, was a form of lunacy, and those who defend either war have forfeited all claim to be taken seriously in political debate.

posted by: Firebug on 10.12.05 at 11:34 AM [permalink]

Firebug relies on the usual shibboleths to support anti-war's mantra of Vietnam as the failed war. I suppose he has been a vociferous critic of troops in the Korean DMZ? After all, at a minimmum, the corrupt, autocratic leaders of South Korea would never change. And the Korean war could never be won. It sure is a shame that we never left South Korea since it would never be self-sufficient.

The fact is America's abandonment and yes betrayal of Vietnam caused the millions of Vietnamese boatpeople, death of millions in Cambodia and the devestation of Southeast Asia. Par for the course, he certainly advocates the abandonment of Iraq, heedless of the consequences.

While it is debatable whether we should have become involved in a war in the first place, once we commit to action, we have responsibility to weigh the consequences of leaving and be prepared to accept the moral outcome of our actions. Something Firebug and his ilk will never understand.

posted by: Jay on 10.12.05 at 11:34 AM [permalink]

Kaplan's comments are alot of sound and fury signifying nothing. Schelling won the Nobel prize in economics not the Nobel peace prize! It is not the purview of the committee adjudicating on the winner of the prize in economics to consider non-economic writings of candidates. Kaplan obviously knows nothing about the immense contributions Schelling has made to game theory.

In short, we shouldn't narrow the Nobel candidate pool in any field to include only saints. After all, the prize only exists because Nobel filled his coffers with the invention of dynamite.

posted by: Borgia on 10.12.05 at 11:34 AM [permalink]

Game Theory is definitely not a social science theory, but then its name comes not from the social sciences, but from mathematics, where it originated.

In the same sense that Probability Theory is the mathematical theory that describes probability, Game Theory is the mathematical theory that describes games.

posted by: Mycroft on 10.12.05 at 11:34 AM [permalink]

Lame attack on Mandel, in my opinion. It's pretty clear that we've avoided armaggedon in spite of game theory, not because of it.

posted by: Deb on 10.12.05 at 11:34 AM [permalink]

Pol Pot took over in Cambodia because our bombing and other interventions there destabilized the Sihanouk regime and killed probably hundreds of thousands of rural peasants, infuriating the population. It was the Vietnamese who eventually overthrew the Khmer Rouge.

The end result of our leaving Vietnam was that Vietnam is today eagerly welcoming international business investment, the predicted toppling of "dominoes" in Asia never happened, and Chinese influence in the area was reduced after the Vietnamese defeated them militarily.

If you can't see how silly the Vietnam war was then there is something wrong with your capacity for common sense.

posted by: MQ on 10.12.05 at 11:34 AM [permalink]

Asking whether Game Theory can be falsified is like asking whether differential calculus can be falsified. It can not since it is a mathematical theory. Models constructed with it as an aide can be tested however.

posted by: Johan Richter on 10.12.05 at 11:34 AM [permalink]

What I like about Thomas Schelling's work is the importance he places upon intuition when it comes to establishing the unique qualities of a given focal point. Conventional game theory really disregards subjective factors on the understanding that a common problem can be resolved by examining various alternative solutions in the abstract.

It could be argued that an intuitive faculty (I hesitate to use the term "psychic") comes into play when one moves beyond what is definitely known in order to designate a given focal point. The factors that point to uniqueness are to some extent subjective, and so suggest a mode of communication that transcends the merely rational.

Schelling's solution to the problem of coordination without communication, is a solution that has long been used without achieving the status of "a theory". I'm quite certain that the hunters and warriors of earlier epochs operated in a similar fashion when looking for a kill and/or a victory.

posted by: Aidan Maconachy on 10.12.05 at 11:34 AM [permalink]

Second the comment above re Kapalan at Slate. Yeah jazz is about his speed.

posted by: Aidan Maconachy on 10.12.05 at 11:34 AM [permalink]


You see the post-Vietnam War outcome as a good result because Vietnam, after 30 years of communist/totalitarian rule and the exodus of millions of boat people is now open to international investment, and no dominoes toppling in Asia (aside from Cambodia and Laos)? I'd hate to think what you consider a bad result.

posted by: Jay on 10.12.05 at 11:34 AM [permalink]

Game theory would not have predicted the Japanese attack on Pearl Harbor because everyone knew (including the Japanese) that such an attack would be suicide.

Yamamoto (their most trusted Admiral) said he would have the run of the Pacific for six months, a year at most. Then crunch time.

He got almost exactly six months and we made him fight for that.

Game theory is no good unless you get the payoffs right. Like how much is honor and pride worth in the payoff matrix? And then - is it worth two extra carrier battle groups in terms of morale? Or does it cost you two from carelessness?

That is where the whole deal founders in the real world. No one knows "what it is worth".

posted by: M. Simon on 10.12.05 at 11:34 AM [permalink]

No, Pol Pot's reign of terror, came about mostly because of the 'happy, jazz-playing prince' (as the late Spaulding Gray, put it. Prince Sihanouk,
for reasons, only known to him, let the VM have
sanctuary, in Cambodia. Their supply lines to the
Minh, precipitated the bombing, which led to the
coup, while he was away, Which led to the invasion. It was the Chinese sponsorship of Prince
Sihanouk's,umbrella group, that led to Pol Pot's pre-eminence, before the rise of Year Zero.This ranks up their with Lebanese Prime Minister
Franjieh's observance of the Cairo Declaration, that gave the PLO free reign to attack Israel that led to the rise of other, predominantly Christian Militias that led to the Civil WAr,
and ultimately Syrian intervention, the Israeli
invasion, and the rise of Hezbollah


posted by: narciso on 10.12.05 at 11:34 AM [permalink]

Post a Comment:


Email Address:



Remember your info?