Breaking down the risk of nuclear deterrence failure

In the international debate over nuclear disarmament, one long-running bone of contention is the effectiveness of nuclear deterrence—that is, using the threat of nuclear retaliation to prevent another state from going to war. Nuclear-armed states claim that maintaining (or only gradually reducing) the large arsenals required for deterrence is the safer path. Many non-nuclear states claim that nuclear deterrents make the world less safe, and that therefore rapid disarmament is required.

Nuclear arsenals do clearly have some deterrent effect, but how do we know whether disarmament or nuclear deterrence will make the world safer? It’s difficult to precisely calculate the risk of any kind of war, under any circumstances. In fact, though, it is possible to apply a rough risk analysis to the question without knowing exact probabilities or potential severity. And doing so strongly suggests that disarmament is far preferable to continued nuclear deterrence as a means of protecting humanity.

On the surface, the disagreement between nuclear and non-nuclear weapon states may look intractable. People even speak of a crisis in the Nuclear Non-Proliferation Treaty, with its “grand bargain” unraveling due to different views on disarmament. The non-nuclear weapon states have held up their end of the bargain by not building atomic weapons. But many believe that the nuclear weapon states have not done their part, which is to shrink their nuclear arsenals. Frustration about the glacial pace of disarmament has fueled the rise of the humanitarian initiative in recent years, a coalition of non-nuclear-weapon states (such as Austria) and civil society groups (such as ICAN) that are pushing for arms reduction by drawing attention to nuclear weapons’ potential humanitarian impact. The initiative offers a counter-narrative to nuclear weapon states’ claims that the weapons play a positive role as deterrents.

Upon closer inspection, though, the international divide might not be so large after all—or at least it shouldn’t be when the whole story about deterrence failure is taken into account. The competing claims differ mainly in emphasis. There seems to be basic agreement on which principles to use in evaluating the possibility of deterrence failure, and on what the evidence says. By taking a closer look at both the principles and the evidence, a consensus on nuclear deterrence and disarmament may be within reach.

Evaluating the possibility of deterrence failure should be based on the same principles always called for when making decisions under risky conditions. Everyone agrees that the risk of nuclear war is bad; if all else were equal, we would rather not have this risk. But all else is not equal.

The crucial step is to clarify what decision is being faced. In this case, we are not deciding between the risk of nuclear deterrence failure and no risk at all. If that were the case, it would be clear that there should be no nuclear weapons. But the actual decision we’re facing is whether to rapidly disarm. Complete and permanent nuclear disarmament would end the possibility of nuclear war, but not the possibility of war. Thus, the key comparison should be between the risk of deterrence failure—and therefore war—with nuclear weapons and the risk of deterrence failure without them.

Which risk is larger? This is where the empirical details come in. Risk can be quantified as the probability of some harm times the severity of that harm if it were to occur. That means we are comparing four quantities: the probability of war with nuclear weapons, the probability without them, the severity of war with them, and the severity without them.

The probability of any war is a very difficult number to pin down. The causes of war are complex and contingent on many case-specific factors, and the historical record is murky and limited. But at least one thing is clear: The probability of nuclear war is not zero. Nuclear deterrence can fail. It is a fallacy to presume that just because no nuclear war has occurred since the post-World War II advent of nuclear deterrence, therefore it will never happen. The historical record contains several near-misses in which nuclear war was narrowly avoided due in no small part to luck. When these points are raised in disarmament debates, there is no significant international divide. When pushed, everyone seems to agree that nuclear deterrence can fail.

Like nuclear weapons, conventional weapons can also act as deterrents, discouraging states from going to war by presenting a counter-threat. So how does the probability of deterrence failure with nuclear weapons compare to the probability without them? This is a much harder question, and not one commonly considered in the debates. Several studies have attempted to answer it by looking at data on the history of war between states that do and do not have nuclear weapons. A 2009 study by Robert Rauchhaus found a lower probability of war between states that both have nuclear weapons but a higher probability when only one has them. However, a 2015 study by Mark Bell and Nicholas L. Miller finds no significant difference in the relative probabilities. These studies are helpful but inconclusive. Such an important policy decision should ideally rest on more robust research. (I would welcome any reader comments on these two studies or suggestions for additional ones.)

The relative severity of nuclear and non-nuclear war is also complex and uncertain, but easier to compare. In principle, either war could result in few or many deaths depending on how it proceeded. In practice, there is reason to believe that nuclear war would be vastly more severe. Non-nuclear war could result in many millions of deaths, as in the two World Wars, but the unparalleled explosive force of nuclear weapons makes great harm much easier to cause and therefore more probable.

By far the biggest difference between nuclear and non-nuclear war would be that the former would likely result in nuclear winter. Nuclear explosions send smoke up past the clouds, into the stratosphere, which causes global environmental problems for years. Crop failures could ensue in every country, no matter where the war occurred. The survival of human civilization could be at stake. Nuclear winter could astronomically dwarf the potential harm of a non-nuclear war, unless the latter involved some other weapon of global destruction, such as contagious biological weapons. (Those are already banned by treaty and not in active deployment).

The relative severity of nuclear and non-nuclear war is likewise not a point of international disagreement. While awareness of nuclear winter remains low, all sides concur that the impacts of the former would be catastrophic, and the impacts of the latter not as bad. Taking nuclear winter into account, the ambiguity about the probabilities of whether war will occur becomes less important. That is, even if nuclear weapons significantly improve deterrence, nuclear disarmament still reduces the risk of war. Thus, while it is understandable that nuclear-armed states would want to avoid disarmament in order to avoid war, the world will ultimately be safer without nuclear weapons. In other words, nuclear disarmament should proceed rapidly.

Of course, this still leaves the practical challenge of achieving rapid disarmament. It will require progress in several fronts, especially with respect to Russia, which has made repeated nuclear threats since the onset of the Ukraine crisis. But when all the risks are considered, the argument that nuclear deterrence makes the world a safer place is not persuasive. It should not be used to resist rapid disarmament.

The views presented here are the author’s alone, and not those of the Global Catastrophic Risk Institute.