The short answer is: it depends on what you think. The long answer is, well, a bit longer. 

Matt makes a strong case for emphasizing the centrality of security for nuclear weapons and materials to any defense against nuclear terrorism – and it would be fair to say that he’s contributed more than anyone else to creating the intellectual foundations for getting that right. His argument backs up Bill’s assertion that the bulk of our money should be invested there.

Indeed, as I write on page five of my book, “the most powerful tools for preventing nuclear terrorism are those that directly deny nuclear materials and weapons to terrorists.” Moreover, most of the examples I give of how defense can work as a system examine cases where such security empowers the balance of the defense.

All that said, let’s look at the specifics.

Some of our disagreements are largely semantic. I write in the book that the traditional notion that terrorists must succeed only once “is true from plot to plot.” When Matt points out that groups can try, try, and try again to acquire nuclear materials, I consider each of those tries, often separated by years and involving different players, to be separate plots. Once the language is sorted out, we pretty much agree.

That said, when Matt argues that “for many stages of the plot, the evidence of al Qaeda’s nuclear efforts seems to be the cost of failure is small and the group repeatedly tries again,” I’m not sure the evidence is as clear-cut as he suggests. In none of the incidents he cites is it clear that the group came close to success. (I’ll admit that the 2003 episode is much murkier.) Were we to see much more ambitious efforts from al Qaeda – and ones that came much closer to success – it would be much easier for us to come down hard. (And it’s not entirely clear that the cost of failure has been small – after all, we’ve significantly stepped up broad efforts against al Qaeda over several years.) My book argues that we have many opportunities to stop plots, not that we’re currently exploiting all of them. What Matt has done is identify multiple opportunities – thwarted acquisition attempts, failed recruiting efforts – that we could target.

A quick point on criticality accidents. Yes, they might not expose a group, though again, it’s not obvious to me that terrorists could easily regroup. In particular, accidents could easily kill some of their most skilled and critical experts. That would be a big blow.

The well-known OTA judgment is interesting in two ways. It describes one extreme, assessing what a group “could possibly” do. That’s an important bit of information, and indeed, I often point out in the book that a broad defense is relatively ineffective when faced with plausible worst-case scenarios. But it comes with a big qualification. It says nothing about what groups would do, nor does it tell us about what less capable groups could do. If we only focus on this worst case, we’re going to miss useful opportunities to improve defenses against lesser threats.

I’m afraid I won’t be able to give an adequate response here to Matt’s assertion about the weakness of the later lines of defense. My argument is built on how small pieces add up, and it requires more than a couple paragraphs to persuasively pile up those examples. (Indeed, it requires a book.) I’ll make two specific points and give one example, though.

Matt focuses on what terrorist groups could in principle do, while I focus on what they might do in practice. As I note in the book, anyone who starts from the assumption that terrorist groups will seize any opportunity that exists will find my arguments less compelling than those who agree that groups are often more conservative.

The other point is technical: only in the worst cases will what terrorists try to smuggle be of “small size and weak radioactivity,” and as I note in the book, we’re pretty much out of luck there. As I explore at length, though, there are myriad variations where that is simply not the case. We should not reject defensive options simply because they are not universally effective – we need to ask, instead, whether they improve our defense.

As for the example I promised, let’s look at Matt’s invitation that “anyone who thinks we can put really effective controls in place to keep out a suitcase-sized box of nuclear material to go for a canoe trip in boundary waters between the United States and Canada.” Say we accept that. It strikes me as an argument for working with Canada to keep terrorists with dangerous materials out of there in the first place, not just as a way of dismissing

A simple note on pages 117-118, which Matt says “show it being very difficult for the terrorists to reduce the detection system’s effectiveness to below 75%.” That’s not what the book argues. Those pages make the case that if detectors are effective against single pieces of nuclear material, they will also be effective if a group breaks that material into smaller pieces in an attempt to foil the defense. As Matt says, current detectors aren’t great. But again, I’m not arguing that we’re in great shape now – indeed I wrote the book precisely to point out opportunities for improvement. I also agree, and argue in the book, that detecting modestly shielded pure weapons grade uranium metal (but not all HEU, as Matt suggests), is basically impossible. (I’ve had long arguments with people who think they have solutions to this.)

I wouldn’t argue, by a long shot, that “9/11 teaches us that al Qaeda uses ordinary legal border crossings.” I do argue that shifting from tried and true approaches can make terrorist groups nervous, and that if a standard approach is to use legal crossings, shifting to other ones may not be as attractive as it might otherwise seem. But again, I’m not trying to say that these things are impossible – I’m only arguing that they’re more complicated than might appear to be the case at first blush.

I’ll move on to Matt’s second post, as well as to Mike and Bill’s other points, in a moment. But I’ll conclude by largely agreeing with Matt’s closing note:

"My bottom line: we do need some investment in these later lines of defense, and we do need the kind of system-level approach to thinking about the effectiveness of the total defense that Mike describes. But we need to understand that securing material at the source provides far higher leverage in reducing the risk, per unit money or effort invested, than any other policy tool we have."

I agree with everything here (especially the nice things). But there’s a caveat: it’s wrong to assess each other policy tool in isolation. Take a look at a smart paper that Matt wrote late last year, in which he estimates the likelihood of a successful nuclear terrorist attack each year. If you break down his numbers, they say that a group has a 20% chance of acquiring enough nuclear materials for a weapon, and a 28% chance of clearing the remaining hurdles involved in an attack. This seems discordant with the assertion in his post that the later parts of a plot are fundamentally easier than the first part. (He acknowledges that the numbers are notional, and indeed I’d argue that the relative value of first line of defense is actually bigger than his numbers suggest.) He comes to that conclusion by compounding multiple small changes of terrorist failure in the later stages of a plot – exactly what I argue one should do in assessing the problem.