Yeah, so I missed a couple of Ward Wilson Wednesdays. Suck it.

Ward’s final contribution contains an elaborated version of my favorite joke about the epistemology of deterrence.  Like the others, it’s an enjoyable read.  It’s been a pleasure hosting Ward these past few weeks. In case you missed them, here are links to parts 1, 2, and 3.  And remember, Five Myths About Nuclear Weapons is available at a fine bookstore near you.


Doubts About Nuclear Deterrence, Part IV: The Squirrel Man
by Ward Wilson

So I ran into a guy on a street corner in New York. He was dressed a little strangely and he was whirling something over his head. I looked closer and it looked like he had a dead squirrel that he’d spray painted bright, day-glo pink. Holding it by the tail whirling it like a lasso. You see all kinds of things in New York and I didn’t think anything of it, but there was a tourist from England standing nearby and he accosted the squirrel man. “I say old chap, what is it that you’re doing there?”

Squirrel man says, “I’m whirling a squirrel over my head.” (Like: what’s it to you?)

Englishman: “Yes, yes, I see that. I see that very plainly. But why are you twirling a squirrel over your head, my good man?”

Squirrel man: (patiently) “To keep away the elephants.”

Englishman: (obviously–but still politely–incredulous) “Elephants?”

Squirrel man: (tired of this conversation) “Yep.”

Englishman: “I say, you do know that there are no elephants within four thousand miles of this place?”

Squirrel man: (knowingly) “Yep. It works.”

This is a deterrence joke. The problem with the squirrel man’s answer (“Yep. It works.”) is the same problem I have with nuclear deterrence. Just because there’s been an absence of nuclear war doesn’t mean nuclear deterrence is doing its job. After all, just because there had never been a hurricane of the force and size of Sandy over the last 100 years didn’t prove that Sandy could never happen.

Deterrence is by definition a non-observable phenomenon. It goes on inside someone’s head and can’t be measured or recorded. A leader who is deterred has a strong interest in denying that deterrence had anything to do with his decision afterward. You can’t observe deterrence and you can’t necessarily rely on the testimony of participants afterward. Which makes it a tricky subject to pin down.

Deterrence is usually proved by absence, the way the squirrel man proved his day-glo squirrel was working. I threaten you, telling you not to do something. Then when you don’t do it, I claim success. But the fact that there’s been no nuclear war doesn’t prove that we’ve got an ironclad method for preventing it.

One way to see if nuclear deterrence is working is to set a fairly clear set of criteria for success and then go back over the evidence and see if it looks like the criteria have been met. Since nuclear deterrence proponents generally argue that nuclear weapons induce caution, I’ve gone back to look for instances of rashness and aggression. And found a surprising number. It’s unnerving.

Moving the goal posts

One way that proponents of nuclear deterrence answer these objections is to say that nuclear deterrence worked, it just didn’t work in the way I’ve defined it. The problem here is that this seems to me like moving the goal posts.

For example, I think John Lewis Gaddis moves the goal posts in the Berlin crisis. The Berlin crisis of 1948 was the one that led to the famous Berlin Airlift, in which West Berlin was supplied by air after the Soviets cut off ground and rail access to the city. The crisis began in April 1948 and in June word “leaked” to the press that B-29 bombers had been redeployed to England. The B-29 was the bomber that dropped nuclear weapons on Japan and it was widely assumed (and the Truman administration intended it to be widely assumed) that the bombers in England were nuclear-capable. They weren’t. Only one squadron had been specially modified to be able to dropped the hefty nuclear bombs of the day, and it was still in New Mexico. But everyone took it to be a nuclear threat.

The threat was clearly a failure. On the face of it, it did not resolve the crisis. The redeployment was made in July 1948, the crisis didn’t resolve itself until May 1949–11 months later. It would be ludicrous to try to argue that the Soviets ignored the danger of nuclear war for 11 months and then suddenly were so overwhelmed with fear that they capitulated and ended the blockade. So it looks pretty clearly as if nuclear deterrence failed.

But the nuclear threat is described by Gaddis as a success. He sees it as a success, rather than a failure, because he’s redefined the purpose of the threat. The threat’s purpose wasn’t to resolve the crisis, the threat’s purpose was to prevent the crisis from getting worse. Truman moved bombers to England in order to keep the Soviets from, for example, shooting down the US supply planes that were flying into Berlin. And, according to Gaddis, it worked. The crisis did not escalate.

Of course, Gaddis could be right. No one in the Truman administration ever said what the purpose of the threat was–it was an entirely implied threat: no public statement was ever made defining the goal of the redeployment.

But it makes little sense to think of it as an attempt to head off escalation. If you’re in a crisis, and there’s a problem, and you’re thinking of threatening nuclear war, you do it in order to solve the crisis, not to keep the crisis from getting worse. How often have you thought, “Wow. Things are bad. I’d better take extraordinary measures to keep them from getting worse”?

Most tellingly is the attitude at the time toward nuclear weapons. In 1948 nuclear weapons were new. No one knew what their capabilities were. No one had ever tried to use them in a crisis. Initial assessments in 1945 had been that the ability of nuclear weapons to influence events was limitless. It would make sense, knowing what we know now about the failures of nuclear deterrence to influence events, to imagine that the Truman Administration assigned an appropriately limited objective to their nuclear threat. They just wanted it to keep things from getting worse. But imagining that the Truman administration understood the limits of nuclear deterrence is to imagine that they understood the multiple failures that we know about now. They had no such knowledge. In their minds nuclear weapons were the miracle weapon that had suddenly and unexpectedly forced Japan to surrender. Why wouldn’t you try to use a miracle weapon to resolve the crisis?

But even if you accept Gaddis’s redefinition of the purpose of the threat, there’s a further problem. Because in 1950, when the Korean war broke out, the United States repeated the trick with the B-29 bombers. They shifted bombers to Asian bases and then made the information about that redeployment public. Some people argue that the move was designed to keep the Soviets from entering the war–and it worked. But I look at the move and notice that it didn’t really prevent the war from getting worse. True, the Soviets didn’t come in on the side of the North Koreans. But the Chinese did. Which seems like a failure to me.

There is a fallback to explain this failure, though. I have heard people argue that the redeployment was actually not intended to keep the Chinese from entering the war–there was no way to stop them from entering the war–it was intended to keep them from invading Formosa and Taiwan when they did. And therefore it worked. This is like the witch doctor whose patient dies and the angry relations say, “You said the potion would cure him!” And he says, “Yes, but the potion prevented a great deal of pain.” If you’re adroit at redefining success, magic always works.

Because nuclear threats are rarely entirely explicit, it’s always possible to change somewhat the purpose of the threat. It feels to me as if proponents of nuclear deterrence are always moving the goal posts in order to make nuclear deterrence successful. Deterrence failures are more frequent and less talked about than anyone would like to admit, I think. I talk about this more in my book. (Yeah, a little crass attempt at book selling there.)

Goal post movers, instead of setting clear criteria and then looking for fair tests in the evidence of history, redefine nuclear deterrence as that magic-whatever-it-is-process that prevents nuclear war. And then, since there’s been no nuclear war, claim that it obviously works. Unfortunately, there are enough instances where nuclear war was avoided by luck that this notion of a smoothly functioning doctrine doesn’t hold up. The evidence from the Cuban Missile Crisis is particularly clear: we avoided nuclear war by sheer, dumb luck. If the U.S. fighters had happened to run into the Soviet fighters while both were looking for the lost U-2, nuclear weapons would have been detonated over the Soviet Union. And odds are that that would have led to nuclear war. Claiming that nuclear war was averted during the Cuban Missile crisis because deterrence worked especially well is to ignore the facts.


Batting .333 in the Gulf War

Last but not least, let’s look at what happened in the Gulf War. This episode matters in part because General Kevin Chilton (at one time commander of all U.S. nuclear forces) argued in Strategic Studies Quarterly that it was proof that deterrence worked. Remember this threat? After Iraqi forces had invaded and occupied Kuwait in 2000 (but before Coalition Forces had begun their counterattack) U.S. Secretary of State James Baker delivered a letter to Saddam Hussein. In it President Bush (senior) told Saddam not to use chemical or biological weapons in the coming conflict. If he did, the United States would respond with “the strongest response possible.” And the Iraqis didn’t use chembio weapons. Chilton points to this episode as proof that nuclear deterrence works.

But if you go back and read the letter Baker delivered, you find that it actually draws three red lines in the sand: don’t use chembio, don’t set the oil wells on fire, and don’t make terroristic attacks against our friends and allies (Israel). And as we all know, the Iraqis did two out of three of those: they set the oil wells on fire and they launched scud missile attacks against Israeli civilians. So are we arguing that nuclear deterrence only works one third of the time?

In baseball, if you bat .333 you’re doing really well. You can make a lot of money if you get a hit every third time you come up to the plate. But nuclear deterrence is not baseball. If it only works one third of the time, that is not nearly good enough. I’m not claiming here that based on this one episode we can deduce how often nuclear deterrence works. I’m just trying to figure out how reliable nuclear deterrence is? That’s the sixty-four dollar question.

There’s no question that ordinary deterrence (the kind that deters children in everyday life or people from committing crimes) works some of the time. Even though there are still murders (some people are not deterred by the threat of capital punishment), still, capital punishment probably does deter some people some of the time. And there’s little doubt that nuclear deterrence probably works some of the time. Nuclear war is pretty scary. But “some of the time” isn’t good enough. Because any failure of nuclear deterrence could lead to a catastrophic nuclear war. When the stakes are so high, nuclear deterrence has to be 100 percent reliable. As Martin Hellman is fond of pointing out, even if the risk of a catastrophic outcome is relatively small, we still have to take it very, very seriously. And a catastrophic nuclear war would be devastating in a way that it is difficult for us even to imagine. You could say that with nuclear deterrence, failure is not an option. But the evidence seems to show that the risk of nuclear deterrence failing is much greater than we have complacently told ourselves.

Some people argue that over time proof by absence becomes more persuasive. Every day that the sun rises again is further proof that it will rise tomorrow. They look at the last sixty years, count the number of days we haven’t had a nuclear war and conclude that that’s pretty good evidence that deterrence is working. But I’m a historian and we historians take a somewhat longer view. Human history–events we have some real record of–goes back about 6,000 years. Across the course of that span of time, war has shown itself to be a remarkably durable and stubborn accompaniment to human civilization. If you take 6,000 years as the time frame of human history, drawing conclusions about the likelihood of war from evidence over the last 60 years is drawing conclusions based on 1 percent of the evidence. Doesn’t strike me as too large a sample, actually.


The evidence of clear failures of nuclear deterrence throughout the Cold War record is undeniable. None of those failures led to nuclear war, but there were plenty of times when the risk of nuclear war did not cause leaders to pull back and be cautious. Far from being perfect, there were, on the contrary, a significant number of times leaders acted aggressively and made matters worse despite the risk of nuclear war. Which raises the question: How can we rely for our safety and security on a process that doesn’t seem very reliable? Particularly where the costs of failure are so high?

If we are to base our safety and security on nuclear deterrence, it has to be a doctrine that works, not a joke about pink squirrels.t