In the previous chapter, we developed a theory about coercion in international politics. At its core, coercion is about persuading an adversary to get out of the way so that a state can achieve its foreign policy objectives without war. This endeavor is more likely to work if the coercer can convincingly say, “If you don't get out of the way, I'll either push you out of the way myself or cause you so much pain that you'll acquiesce just to make it stop.” A coercive target is more likely to concede without a fight if it is faced with inevitable defeat, or if the price of victory is too much to bear.

We argued that nuclear weapons fall short as instruments of coercion. For military conquest, nuclear weapons have limited utility. Nuclear weapons can accomplish some missions better than conventional weapons (such as destroying hardened targets), but those missions are few in number and extremely rare. More importantly, they tend not to be the kinds of missions required for successful coercion. Coercion often involves seizing objects: taking territory, rescuing hostages, or liberating victims of conquest. Nuclear weapons are not useful for any of these tasks. Instead, nuclear weapons are best for inflicting pain: they destroy, irradiate, and kill indiscriminately. But their extraordinary lethality is what limits their usability in coercive contexts. For self-defense, nuclear weapons can be credible and persuasive tools of punishment. For coercion, however – when the challenger's survival is rarely at stake – they are simply too costly to be credible.

We now turn to the empirical record to evaluate these claims. Does the evidence support our argument about the irrelevance of nuclear weapons in coercion? In this chapter, we examine coercion in its most explicit form. We analyze more than 200 instances in which one state made an explicit compellent threat against another state, and we assess the effect of nuclear weapons on the outcomes of these threats. The evidence is clear: states that possess nuclear weapons enjoy no more success when making compellent threats, even when they enjoy nuclear superiority over their opponent. Moreover, we find that nuclear states are not “manipulating risk” in the way that 1960s-era nuclear strategists and some contemporary game theorists expect them to.