8. The dataset

Imagine you were going to defend you dissertation. And you wake up in a panic the morning of your orals to find that somehow you don’t really have that much data to support your thesis. In fact, what you’ve got is one real data point and ten or so related events that kind of bear on the point. It’s a terrifying thought, no? You can, I’m sure, picture yourself desperately flipping through all the books you spent your PhD stipend on, searching your notes again and again, and pulling your hair. With only about eleven or twelve pieces of real factual evidence to support your ideas, you think, you’ll never get your PhD. Your uncle John, who’s been aiming sly digs at you for five years during Thanksgiving dinner, is going to have a field day. How will you ever pay off your student loans? You’re more likely to wake up and realize it was all a bad dream than for anything like this to actually happen in the real world.

But perhaps surprisingly, that is exactly the situation we’re in when it comes to nuclear weapons. There is hardly any real data to build respectable theories on. The people who teach about nuclear deterrence and nuclear weapons in general don’t stress this point, but although the theories are taught with the kind of authority and self-assurance you’d expect for a subject dealing with the safety of literally billions of people, there’s almost no reliable data from which to build a coherent theory in the nuclear weapons field. A weapon with only one real field test? A series of threats that are all, in one way or another, discouragingly ambiguous? It doesn’t sound very promising.

Of course some scholars have proposed that this problem can be solved by including virtually everything that has happened while nuclear weapons were in existence in the dataset. Are the threats in crisis A different from the threats in crisis B? Doesn’t matter, they say, nuclear weapons were in existence in the background. Was the size of the arsenal different in crisis A than in crisis B? Immaterial, they say. And so on. By shifting the criterion from actual use or actual threats and instead looking at simple existence, they try to multiply the number of events available to support theories about nuclear weapons. But it won’t do. If you aren’t required to establish a link between an object and events, then my father’s lucky penny that he’s had since 1945 could as easily be the source of a stable world order. It’s not enough to claim that there’s a relationship between an object and events. At some point you have to say what you think that relationship consists of, show how it could potentially have worked, and then—oh, right—prove it. And as I’ve discussed elsewhere in these essays, proving that nuclear weapons effect events is a non-trivial problem.

In some ways the field of nuclear weapons is like early cosmology. The astronomers that came up with our first model of the solar system also built that system out of very few facts. Seven, when you stop and think about it. They looked at the heavens and saw the sun, the moon, and five visible planets (against the background of the unchanging stars.) From this data they constructed the Ptolemaic model of the universe with the earth at the center and the sun, moon, and planets revolving around it. It was a congenial view of things for most people to imagine the universe revolving around our own planet earth. The Ptolemaic model was taught at the most august institutions of learning, accepted throughout governing establishments, and dominated thinking about the heavens for more than a thousand years.

But Copernicus eventually came along and showed that the Ptolemaic system was wrong. The sun was at the center of the solar system, not the earth. The earth was only one of a handful of planets that revolved around the sun. If Copernicus were here today he might warn nuclear weapons experts that systems built from vary few facts can sometimes turn out to be wrong. And not just sort of wrong, but completely, turn-on-their-heads wrong. It’s dangerous to build systems of thought out of very few facts.

Of course, one difference between early cosmology and nuclear weapons theory is that if you’re wrong about cosmology the consequences are not very severe. But with nuclear weapons, unfortunately, the lives of hundreds of millions of people are at stake.

Reality is not theoretical. The nuclear believers assert the truth of nuclear deterrence theory with confidence and certainty. But if nuclear deterrence theory is to be reality based, it is essential to return to the few facts we have and examine them carefully. Realism means relying on facts, not theories.