Low-yield nuclear weapons are being manufactured and deployed again in the United States, for the first time since the end of the Cold War. They are not the lowest-yield nuclear weapons ever manufactured, and they still can destroy a city. The justification for this step, as for many others, is that it enhances deterrence. A detailed look at that argument shows how polymorphous and poorly supported arguments on deterrence can be.
The advocates of those lower-yield nuclear weapons rely on firm assertions about deterrence, a foundation of nuclear war strategy—which is mostly concerned with averting it. Deterrence balances adversaries’ judgments of unacceptable harm to themselves against the gains they might make. The calculation includes conventional military power as well as nuclear. It depends on the opponent’s state of mind.
Arguments based on deterrence are necessarily deductive. The world has never fought a nuclear war, so we don’t know how one would play out. The entire history of nuclear weapon use in anger is the two bombs dropped on Japan by the United States in 1945. At that time, no other nation possessed nuclear weapons, so there was no possibility of retaliation or escalation.
Game theory and the history of devastating battles provide hints, but analyses of nuclear war are in a different domain than other military theory. Additionally, much of the strategy of nuclear war was developed in the 1960s and 1970s, when the belief in a rational economic man was firmer than it is today. Decisions to balance hundreds of thousands of deaths against a national interest would be made dispassionately as the missiles flew. Those decisions could be gamed out in advance for guidance if the worst happened. The decision-makers were men of principle.
There are a few negative examples. The United States and the Soviet Union faced off over Cuba in 1962, and both nations found a way not to use nuclear weapons—just as India and Pakistan explicitly rejected first use of nuclear weapons in their crisis of 2001 to 2002. It is possible that India and Pakistan were influenced by the Americans and Soviets backing away from the brink. A few false alarms teach us some things about human reactions—an essential part of a strategy fundamentally built on ideas about psychology, especially the psychology of adversaries. It’s worth asking how much of that foundation of deterrence thinking dates back to 1950s psychology and hasn’t been updated.
All this renders deterrence a deeply uncertain art, not a carefully formulated science. Nevertheless, advocates assure us that those lower-yield nuclear weapons are an essential step to stabilize deterrence by matching weapons. A Russian nuclear cruise missile, promoted by President Vladimir Putin as being able to stay aloft for long periods of time and real enough to have killed seven people when it was tested, means the United States must have nuclear cruise missiles; if Russia is developing maneuverable hypersonic missiles to avoid missile defenses, the United States must have hypersonic missiles too. If the United States matches every Russian weapon, if every move is potentially countered, advocates argue, deterrence exists. Former U.S. Secretary of Defense Robert McNamara recognized the tendency to shrink nuclear strategy to comparison of weapons. Thus the military theorist Carl von Clausewitz’s classical postulate is turned on its head; weapons rather than politics drive preparations for war.
The doctrine that is to be countered with lower-yield nuclear weapons is called “escalate to de-escalate”—using a small nuclear weapon as a warning to an opponent to stand down as a larger conventional conflict appears to be going badly. Advocates for lower-yield nuclear weapons such as Matthew Kroenig and Elbridge Colby assert that escalate to de-escalate is part of Russian war plans. The basis for this belief is a single sentence in an obscure part of a Russian strategy document, along with Russia’s large number of smaller nuclear weapons.
The advocates argue that Americans would hesitate to respond to escalate to de-escalate with the nuclear weapons currently in the U.S. arsenal. Their term for this is “self-deterrence.” However, they ignore already-existing nuclear weapons that offer options for yields in the sub- and low-kiloton range, such as already-deployed B61 bombs whose yield can be adjusted down to 0.3 kilotons, or air-launched cruise missiles with variable yields down to 5 kilotons.
Whether escalate to de-escalate in fact is Russian doctrine is disputed (see: to research by Amy Woolf, Olga Oliker, Michael Krepon, and Kristin Ven Bruusgaard), but it is the easiest way to develop a scenario in which a conventional war goes nuclear. The assumptions of how a conventional war might develop are equally shaky. For example, here’s an assessment of Russian decision-making in a Nordic crisis. Going by that assessment and others, it’s hard to see how a Russian attack comes about, much less escalate to de-escalate.
But say that Russia’s strategy does include escalate to de-escalate—or even go beyond the comparison of hardware to the strategic moves involved.
Escalate to de-escalate would come after a war in progress presents a danger to the Russian homeland. For the United States to be involved, that war would most likely be in Europe—such as a Ukraine-like intrusion into Estonia or Latvia, or a conflict over Russian activity in the Swedish or Finnish archipelagoes, or perhaps an Arctic confrontation between Russia and Norway or Finland. All of those possibilities involve NATO nations or nations cooperating with NATO.
One way or another, Russia would arrive at a decision for an escalate to de-escalate strike. It would strike Amari Air Base in Estonia, where NATO planes are stationed. Amari is fairly isolated, but even with a lower-yield nuclear weapon, there would be civilian casualties along with military. Fallout over Tallinn could be serious.
The use of a nuclear weapon, NATO casualties, and the obvious possibility of further Russian incursion into Estonia would be considered threatening in Washington, Brussels, and other European capitals. Pressure would mount for retaliation.
In response, NATO would use a nuclear-armed cruise missile to destroy the Seshcha Air Base, also isolated and somewhat equivalent in absolute size, although much less important relatively.
Now two nuclear weapons would have been dropped. Emotions would be high across the Northern Hemisphere. Would a country willing to drop the first nuclear weapon in an escalate to de-escalate scenario stand down after a nuclear response? Would NATO stand back if Russia overran the Baltic States? Or if Moscow grabbed critical islands in the archipelagoes to control shipping to Stockholm and Helsinki?
Or would Russia strike Warsaw, Poland, population 1.7 million? Hundreds of thousands of people would be killed or maimed. And then would NATO choose Rostov-on-Don, Russia, population 1.1 million? At that point, everyone would be on hair-trigger alert, increasing the probability of mistakes.
Has an escalation ladder been gamed out for the use of these weapons? Do we know that the two sides take the meanings intended in each strike?
Civilians don’t know. Even those involved in nuclear planning know only their own side. Yet the advocates of these weapons have provided no answers beyond the assumption that being able to respond in kind to an escalate to de-escalate strike will deter the first strike from happening at all. But deterrence is far broader than this. We need to know whether this is in fact Russian doctrine, how Russia values situations that could lead to war, and how they think about existing American weapons and defenses. The nonproliferation expert Jeffrey Lewis has done a partial analysis of such factors. A justification for introducing new nuclear weapons should go beyond his initial take.
Another argument for more deterrence lies in the speed of communications. That speed, via technological advances, erodes the capability for the devastating second strikes on which the logic of deterrence currently rests. This argument also devolves to technology versus technology.
Once again, the logic is taken to false extremes. Thus, a dead-hand system with decision-making by artificial intelligence after human decision-makers are killed has been proposed, citing the implied arguments against such practices in the movies Dr. Strangelove, WarGames, and Terminator, but forging ahead anyway. A more detailed refutation of both the idea that speed of communications is an issue and the use of artificial intelligence can be found here. Nations abstain from war for many reasons. The concept of deterrence has come to represent that abstention, while at the same time shrinking it to the matching of weapons, one for one. Although the arguments for that sort of deterrence are flimsy, the magic of the term justifies spending enormous sums—for purely theoretical gains, and the likely ratcheting up of nuclear danger.
Anne I. Harrington and Marty Pfeiffer contributed to this piece.