I’ll never shake the sense memory of that numbed shiver. It was November 2001, and I was reading about rumors that were rippling through Washington: a nuclear bomb had been smuggled into the city, and would be detonated that weekend.

It was utterly believable, after all. In those dreary, surreal post-9/11 months — the tip of Manhattan a smoldering crater, anthrax spores dancing in the air — anything seemed possible.

Of course, anything is possible. But in his coolly analytical new book, On Nuclear Terrorism (Harvard University Press), Michael Levi, a fellow for science and technology at the Council on Foreign Relations, explains how recognizing that fact — how understanding the full range of terrorist capabilities, rather than obsessing over terrifying worst-case scenarios — is key to effecting a successful defense against nuclear attack.

This means, first of all, strengthening an integrated web of weapons and materials security, intelligence, diplomacy, law enforcement, and border patrols. “We need to think about a whole wide range,” says Levi, “from the most likely but least threatening, to the least likely but most threatening.”

Moreover, from a terrorist’s point of view, the idea that anything can happen becomes a serious obstacle to pulling off a catastrophic attack. In Levi’s book, the IRA’s chilling admonition upon failing to assassinate Margaret Thatcher in a 1984 UK hotel bombing — “Today we were unlucky, but remember; we only have to be lucky once. You will have to be lucky always.” — is turned on its head. A nuclear plot is so complex that each successive step must be pulled off with perfection.

Levi’s argument is that thinking broadly about various permutations of threat and mounting a similarly multi-layered defense puts the onus on the bad guys to succeed. That’s comforting, in a way. But, again, anything is possible. The Phoenix asked Levi recently how scared we should be.

Is the threat of nuclear terrorism overblown or just misunderstood?
Misunderstood. Some people overestimate it, some people underestimate it. What most have in common is that they have a single view of exactly what the threat is — when in reality it can come in a lot of different forms.

Like what?
Different groups have different capabilities. Some may be very good at stealing what they need to put into a weapon. Others may be very good at building [bombs]. Some may be skilled at moving materials around the world. Some will be very willing to take risks, others won’t. A group might try to make an illicit purchase. It might try to team up with a state. A group could try to bring materials through an official border crossing. Or bring them through a remote, unmonitored point. There are enormous variations a nuclear-terrorist plot can take, and in thinking about how to defend, we have to look at that whole spectrum.

Is that the key? Exploiting the obstacles they have to overcome?
Our best hope is to exploit the fact that a group needs to cross a large number of hurdles, even if many of them are relatively small. And to exploit an understanding of how groups might shape their plots in response to what we do. These are thinking, adapting groups, but that doesn’t mean that they’ll all respond in the same way to a particular defense.

What are some ways those defenses can improve?
Fairly obvious things, like improving security at the source of nuclear weapons and materials, or training people who protect borders to recognize telltale signs of nuclear plots. But there are also overarching things we can do. We need to make sure that policy makers and strategists understand the threat properly. We need to make sure that planners look at the entire system of defense as a whole, not just as a set of isolated parts. We’re also going to need to rethink our approach to secrecy a little bit. A lot of the most powerful elements of a broad defense are at a local level or state level, not in the traditional national security establishment. And they need information in order to be able to identify plots.

No defense is perfect. But describe, in a nutshell, the optimal defense.
I don’t think it’s possible to describe an optimal defense in a few broad strokes. A big part of our problem has been an insistence on having a defense that we can describe in two sentences. That’s always going to have big problems. During the Cold War, there were two defenses that could be described in one sentence. One was Mutual Assured Destruction. The other was Star Wars. Both had big problems. The more sensible things didn’t fit on a bumper sticker. There are a couple principles you can get down in one sentence: don’t fixate on the worst case, and look at the whole defense as an integrated system.