Michael Shellenberger recently wrote an article over at Forbes arguing for the abolition of the nuclear nonproliferation regime. To be sure, he is not as radical as some proponents of proliferation, such as Kenneth Waltz, but he argues that any state that feels it needs nuclear weapons for deterrence should be able to acquire them unmolested. Shellenberger, whom I don’t know, was kind enough to engage me on Twitter. However, I feel compelled to flesh out my rebuttal here because Twitter debates are so messy and scattered. Frankly, I think Shellenberger is dead wrong. While his arguments are intuitive and certainly not crazy, I think they reveal a lack of familiarity with the literature and relevant empirical facts. Indeed, in many points throughout the article, I feel like I’m just reading extended excerpts from some of Waltz’s work instead of a measured and thorough treatment of the literature.
His article starts off with an explanation of deterrence. To clarify the concept, Shellenberger recounts the opening scene of the fantastic Tarantino film Inglorious Basterds. According to Shellenberger, Nazi Germany invaded France because the French lacked a credible deterrent. While plausible on the face of it, this analogy actually makes very little sense and conflates several things. First, France did have a deterrent, two in fact. The first was its large, well-equipped army and strong support from capable British forces. The second was the Maginot Line, an impressive defensive fortification along its eastern border with Germany. You see, there are two basic types of deterrence: deterrence by denial and deterrence by punishment. Both seek to raise the cost of aggression to unacceptable levels, but they function in distinct ways. Deterrence by denial hardens targets, making them difficult to defeat and thus raising the costs of victory. Deterrence by punishment uses retaliatory measures to inflict so much damage on an aggressor that the costs of attacking outweigh the benefits. But the important point is that there are multiple ways to deter, and retaliatory capabilities are, therefore, not necessarily required.
His analogy is also silly because it presupposes that nuclear weapons are the only effective tool for enabling a strategy of deterrence by punishment. In Shellenberger’s telling, deterrence was impossible before 1945, but this is obviously absurd. During the interwar period, aircraft were viewed in a similar way as nukes are today. They were novel, seemingly unstoppable, and could simply bypass the frontlines to rain destruction down on population centers. This may seem preposterous to us now, especially given the air superiority enjoyed by the U.S. and its allies, but people in the 1930s were absolutely terrified of aircraft. Indeed, it’s helpful to read the work of Douhet and Mitchell to get a sense of how powerful people assumed aircraft, particularly strategic bombers, to be. In other words, planes served the role of nuclear weapons, acting as deterrents, and governments treated them accordingly. For example, at the same time Neville Chamberlain was declaring “peace in our time,” he was ordering the Royal Air Force to massively increase its size and capability just in case Hitler turned out to be as megalomaniacal as he in fact did. While I’m less familiar with the French case, and the Armée de l’Air was certainly inferior to the Luftwaffe, the aggregate power of French and British forces still seemed overwhelming. Indeed, the German High Command thought Hitler was a lunatic and strongly cautioned against a war with the Allies. Thus, it is simply wrong to say France lacked a deterrent because, quite frankly, German commanders were convinced that they would suffer egregious losses if they fought the French. The deterrent was ineffective because it failed to convince Hitler, but it did exist. This is important because it highlights the fact that, no matter how credible the deterrent, deterrence can still break down in the face of a reckless leader. And this is true regardless of whether nuclear weapons are involved. Obviously France lacked a nuclear deterrent, as nuclear weapons hadn’t yet been developed, but it did have a conventional deterrent, which in many cases can be just as effective. After all, North Korea’s conventional artillery pieces aimed at Seoul have kept the peninsula quiet for decades, and this is true despite continued provocations from the North and massive force asymmetry in favor of the U.N. Combined Forces Command.
Shellenberger then provides a very brief account of France’s quest for nuclear weapons, highlighting de Gaulle’s lack of faith in American extended deterrence guarantees. Extended deterrence is certainly more difficult than primary deterrence, as it is premised on a patron state risking a nuclear attack to defend an ally. Nevertheless, it has historically been effective. After all, no country with a nuclear guarantee from the United States has suffered an invasion by an enemy country. The reason for this is simple and was noted by a British diplomat several decades ago: While the recipient of an extended deterrence pledge needs to be 90% certain that Washington will honor its commitments to be reassured, a potential adversary only needs to be 5-10% certain to be deterred. Moreover, most countries have not acquired nuclear weapons yet seem to be doing fine. During the Cold War, was it really more believable that the U.S. would risk Boston for Bonn than Pittsburg for Paris? No. So it was not, in fact, French vulnerability that led them to acquire nukes, but rather the insecurities of French national security elite. It’s also worth pointing out that France likely acquired nukes as much for its national ego as its national security; it had just been humiliated by Germany and wanted to reclaim its status as a great power.
Shellenberger then goes on to mock the notion of domino proliferation. In fact, I largely agree with him that the risks of rapid domino proliferation are overstated. However, he neglects to place the historical record in context. Proliferation was slow and measured precisely because the Americans and Soviets provided their client states with nuclear and conventional security guarantees. Moreover, proliferation was and is stymied by treaties such as the NPT, which create a normative and legal prohibition against the acquisition of nuclear weapons. But, as a matter of fact, domino proliferation has occurred. Pakistan is certainly a domino proliferator, as are Britain and France. While China is less clear, what is certain is that U.S. threats to use nuclear weapons to end the Korean War played a major role in Mao’s calculus to acquire nukes for himself. Now imagine Shellenberger’s ideal world. This is a world in which any country that wants to can, without any restrictions, acquire nuclear weapons. Suddenly the normative, legal, and military framework retarding the rapid spread of nuclear weapons is gone and, as a result, destabilizing domino proliferation and arms racing become far more likely.
The next set of arguments in his piece are conveyed via a series of bullet points that attempt to demonstrate just how effective nuclear weapons are at ensuring a country’s security. The first bullet point asserts that no nuclear powers have been invaded. I think this is a pretty misleading statistic because, as a matter of fact, most countries haven’t been invaded since 1945, and this is true irrespective of their nuclear status. The nuclear club is a small one, so it really isn’t surprising they’ve avoided invasion. Furthermore, with the exception of India and Pakistan and the allied countries of China and North Korea, no two nuclear powers share a land border. Therefore, it isn’t surprising they aren’t invading one another. But more importantly, this assertion is just wrong: Israel possessed nuclear weapons during the Yom Kippur War and India had nuclear weapons when it was invaded by Pakistan in 1999. Nuclear weapons can also lead to a phenomenon known as the stability-instability paradox. In short, by convincing leaders that a large-scale war is impossible because of their nuclear deterrents, nuclear weapons can lead to increased conflict at lower levels. So while nukes may have helped keep the peace between New Delhi and Islamabad, they also likely played a role in Pakistani-sponsored terrorists’ attacks on the Indian Parliament and Mumbai in 2001 and 2008, respectively. The second bullet point, that battle deaths worldwide have declined 95%, is correct, but it fails to prove that nuclear weapons are responsible for this trend. Other factors, such as U.S. hegemony, changes in normative views on war, increased economic interdependence, and more effective international organizations seem to be equally plausible explanations. Shellenberger may be correct that nuclear weapons are the cause of this happy trend, but, as we are frequently reminded, correlation and causation are not equivalent. It’s also worth noting that, given how peaceful and prosperous our world has become, it may be prudent to keep a steady course. After all, we seem to be trending in the right direction anyway. Why, exactly, do we need to change things? The final argument advanced by Shellenberger is that Indian and Pakistani civilian and security forces’ deaths in two disputed territories declined 95 percent after Pakistan’s first nuclear weapons test in 1998. But this, again, ignores the stability-instability paradox and doesn’t account for non-nuclear factors like the role of outside mediation and domestic politics.
Shellenberger then argues that nuclear weapons moderate state behavior because “History shows that when countries acquire the bomb, they feel increasingly vulnerable.” (quote from Waltz) This makes absolutely no sense. Either nukes ensure existential security, preventing great power intervention, or they make countries more vulnerable, but to argue that nukes simultaneously make countries more and less vulnerable is almost Trumpian in its incoherence. And sure, maybe nuclear weapons promote foreign policy moderation, but that isn’t the same thing as internal moderation: The Cultural Revolution occurred after China had nuclear weapons, after all. I’m sympathetic to Shellenberger’s appeal for greater U.S. restraint abroad, but sometimes one has to intervene to save lives. I bet Saddam and Qaddafi were terrified of coalition forces, but their people were terrified of them. Why do a strongman’s discomfort and fear outweigh that of his people? Moreover, could anyone imagine how much worse the Syrian civil war would be if Assad had nukes – which, by the way, he attempted to acquire? The same goes for Yugoslavia; is it really worth forgoing even the possibility of military intervention, even when millions are calling out for help, simply to alleviate a dictator’s sense of insecurity? Absolutely not. I think it is clear that the U.S. too frequently intervenes militarily in other countries’ internal disputes. However, it’s a big leap to go from acknowledging American hubris to preventing all humanitarian intervention by allowing strongmen to acquire weapons of mass destruction at will.
Here’s the other thing, Shellenberger presumably is only advocating for American acceptance of proliferation. After all, forcing other countries to go along with Washington is the exact kind of interference and American bullying he seems to so despise. But not every country will agree. Israel has struck nascent nuclear programs on several occasions, for example, and the Soviets almost launched an attack on the Chinese nuclear program. So, even if nuclear weapons make conflict less likely, attempting to acquire nuclear weapons actually tends to precipitate conflict as potential adversaries try desperately to stop a proliferator before it is too late. This is, after all, the reason the U.S. and its coalition partners invaded Iraq.
The final argument advanced by Shellenberger is that nuclear powers are hypocritical. They are, but that doesn’t make them wrong. Nuclear weapons are complicated systems that are difficult to secure and safeguard. The history of America’s nuclear weapons program is replete with near misses and close calls, and Soviet history is similarly concerning. Of course, the U.S. and other nuclear powers could still end up accidentally or inadvertently releasing nuclear weapons, but they have now had decades to develop protocols and standard operating procedures designed to reduce the likelihood of this occurring. A new nuclear state would need to develop many of these protocols from scratch. And while established nuclear powers might be able to offer guidance, no country concerned about an invasion by a great power like the U.S. is going to allow that same great power to develop and design its nuclear arsenal’s security features. According to Shellenberger, it is now incredibly easy and intuitive to design safe arsenals because nukes have been around for some time and the U.S. gave some money to Russia back in the 90s to secure the Soviet arsenal. I’m sorry, but the conclusion simply doesn’t follow from the premise. Horses have been around for a long time, but you better believe riding them is not intuitive to someone like me. And nuclear weapons are far more complex than horseback riding. As for Russia, who cares? The fact that the Nunn-Lugar Cooperative Threat Reduction Program worked to secure the Soviet arsenal is interesting, but it hasn’t prevented nuclear theft from the former USSR and is totally irrelevant in the context of any country other than the former Soviet republics. It’s also worth noting that nuclear weapons, in and of themselves, are insufficient to deter an attack from the U.S. (see also here). If possessing a handful of weapons was all it took, the Soviets and Americans wouldn’t have continued to build increasingly complex and large arsenals during the Cold War, and the Indians and Pakistanis wouldn’t be doing so now. Shellenberger admits as much, noting that states are currently expanding their arsenals. Correct! But that just proves that nukes aren’t nearly as effective as he seems to assume. And the more complex and variegated the arsenal, the more difficult it is to secure and safeguard.
So yes, some countries want nuclear weapons. And yes, there is a degree of hypocrisy in the way the nonproliferation regime is organized. But that absolutely does not mean we should tear up the NPT. Instead, the U.S. needs to work hard to reassure allies and encourage them to continue to build up their conventional and civil defense capabilities. Does this guarantee their security? No. But it has worked so far, and there is little reason to believe that it can’t continue to work in the future.