1. Nuclear deterrence and hunches

(The first in a series of occasional short pieces designed to challenge accepted assumptions about nuclear weapons.)

I gave a talk at the A-10 Directorate at the Pentagon once. The A-10 Directorate is the Air Force’s office in the Pentagon for policy, planning, and strategy on nuclear weapons and is full of people who take their jobs seriously and who, on the whole, are the kind of smart, dedicated, patriotic people you’d want to have working on such an important issue. We had an interesting, stimulating discussion. We talked facts and the history of war, and they seemed to genuinely enjoy the pragmatic challenges I was putting to U.S. nuclear policy. But when I argued that there is very little we actually know about nuclear deterrence, one of the civilian defense intellectuals smiled a polite but gently scornful smile. He pointed out that nuclear-armed states have tested nuclear weapons more than two thousand times in deserts and other remote parts of the world. “I think we have enough information about the capabilities of nuclear weapons to know all we need to know about nuclear deterrence,” he said. He thought he’d made a powerful rebuttal. But in fact all he had done was to illustrate how loosely and incautiously even experts tend to think about the foundations of nuclear deterrence.

Deterrence, I’m sure you know, is a type of threat, and threats are psychological. (Stimson said, “the atomic bomb was more than a weapon of terrible destruction, it was a psychological weapon.”) In order for nuclear deterrence threats to work, something has to happen inside the mind of your adversary. The thing about “inside the mind,” though, is that it is a place we don’t know very much about. Truth, remember, isn’t objective unless you can measure it, test it, and other people can reproduce the results. But you can’t get a ruler inside someone’s head to measure what’s happening in there.

It’s true: nuclear-armed states have tested nuclear weapons more than two thousand times in deserts and remote parts of the world. This gives the illusion of scientific certainty. The physics of nuclear weapons is very well understood. But the physics is largely beside the point. The important question is not “what does an explosion do on the ground?” The important question is, “what does an explosion do in the minds of leaders?” When I blow up your city, does that make you surrender? When I threaten to blow up your cities, does that make you back down? Nuclear explosions may create so much overpressure at such and such a distance from ground zero. They may ignite so many secondary fires or spread so much radiation. But all of that is really beside the point when it comes to deterrence. What matters for deterrence is what the explosions do inside a leader’s mind. What is their symbolic effect? The well-studied physical characteristics of nuclear weapons are, for all intents and purposes, irrelevant. It is their psychological characteristics that matter.

We have good evidence about what nuclear weapons do to buildings and cars and factories and people’s bodies. Despite the belief and gentle scorn of the civilian expert at the A-10 Directorate, we don’t have good evidence about what they do inside people’s heads. The problem is that there isn’t any way to scientifically measure inside someone’s head.

Despite those 2,000 tests, the important part of nuclear deterrence—its psychological characteristics—is, for the most part, unexplored. There is no machine that can measure the psychological impact of the threat to use (or the actual use of) nuclear weapons. Think about all the components of responding to a nuclear threat—fear and threat evaluation, emotions, doctrines, theories, assessments of seriousness, instincts of self-preservation, intuitions about your adversary’s intentions, and so on. No machine exists that can objectively measure all those components of the process happening inside someone’s head. In fact, there isn’t a machine that can measure even just one of those components. And even if one day such a machine were invented (imagine some large helmet sort-of-thing with wires attached and thousands of sensors), no leader has ever been monitored by such a machine—or is ever likely to allow himself or herself to be monitored—during a crisis. We will never have direct, objective evidence about how—or how well—nuclear deterrence works.

What this means is that our “knowledge” about nuclear deterrence is entirely speculative. It is based on intuitions about human behavior. Nuclear deterrence, for all our scholarship, is essentially based on hunches.

Should we really risk millions of lives based on hunches?