Russia’s invasion of Ukraine has revived many questions about nuclear deterrence. Whatever the outcome of what could be a long war, the issues it has raised won’t go away.

In 1994, Ukraine surrendered the nuclear weapons it had inherited from the Soviet Union in return for security guarantees from the United States, the United Kingdom and Russia. But those guarantees turned out to be worthless, and because Ukraine isn’t a member of NATO, it’s not covered by the extended deterrence of the US nuclear umbrella.

What about the former Soviet republics that have joined NATO? Would US extended deterrence actually work for Estonia, Latvia and Lithuania, or for America’s allies in Asia? For deterrence to be credible, nuclear weapons must be usable. But if they are too usable, an accident or misjudgement could easily lead to a disastrous nuclear war.

To achieve an effective balance, we must consider the appropriate mix of nuclear, conventional and other instruments, and then reduce the nuclear component whenever possible. For example, whatever the appropriate response to North Korea’s growing nuclear arsenal may be, it shouldn’t include a reintroduction of the tactical nuclear weapons that President George H.W. Bush removed ^[1] from the Korean peninsula in 1991.

Similarly, for Japan, the credibility of US extended deterrence hinges on the stationing of American troops there, not on the presence of nuclear weapons. By sharing the vulnerability that Japanese troops face, the US establishes a community of fate that reduces its allies’ fear of abandonment. While sceptics used to point out that the small contingent of US troops in Berlin couldn’t possibly defend that city against the Soviet Union, America’s physical presence nonetheless proved to be essential to deterrence and a peaceful outcome to the Cold War. (There was also a time when the US had nuclear artillery stationed in Europe; however, because of the risks to command and control, it was removed.)

As the US and other countries have continued to modernise their forces, the usability debate has persisted. Deterrence depends on psychology, and some analysts argue that perceived superiority in usable weapons can make a difference during crises. Others, like the late Columbia University political scientist Robert Jervis, argue ^[2] that all measures of nuclear balance are too crude to be useful in reaching such conclusions. Mutual assured destruction is a condition, not a policy.

In fact, history has shown that one doesn’t need a high probability of use to create existential deterrence. Despite the overwhelming superiority of America’s nuclear arsenal, President John F. Kennedy still felt deterred by even a small risk of escalation during the Cuban missile crisis. Today, small, accurate nuclear weapons seem so usable that we have come to treat them as normal. But the dangers of escalation remain, and the location of some military targets near cities means the dangers will persist. Avoiding catastrophe depends more on reducing the risks of nuclear war—both deliberate and inadvertent—than on changes in targeting doctrines.

Following a risk-reduction maxim, we can reject some policies outright. For example, a launch-on-warning protocol delegating nuclear launch authority to battlefield commanders may enhance deterrence, but it also raises the risk of unnecessary provocation. Defence hawks sometimes forget that deterrence depends on the opponent’s psychology, not just their own.

On the other hand, defence doves’ proposals to escape the usability dilemma and appease adversaries may create an impression of weakness, thereby tempting adversaries to take more risks. Dovish nuclear strategists are sometimes too clever by half when they devise elaborate strategies based solely on calculations rather than on experience.

Representing a middle ground between hawks and doves, defence owls place a premium on risk reduction. Whereas hawks have a hair trigger and doves have a sticky holster, owls offer a reliable safety catch.

Russia’s invasion of Ukraine reminds us that we are still living in a world with nuclear weapons, and that we should be endeavouring to reduce (though not abolish) stockpiles over the long run. As the physicist Richard Garwin once noted ^[3], ‘If the probability of nuclear war this year is one percent, and if we manage each year to reduce the probability to only 80 percent of what it was the previous year, then the cumulative probability of nuclear war for all time will be 5 percent.’

The psychological effect of nuclear deterrence on our moral lives is another significant long-term consequence to consider. The theologian Paul Ramsey once likened ^[4] nuclear deterrence to tying babies to the bumpers of cars as a means of slowing traffic and reducing the number of lives lost to road accidents. But while that metaphor helps incite moral repugnance, it’s not an accurate depiction, because people today simply don’t suffer from the kind of anxiety that one would expect to see in Ramsey’s scenario. A lack of anxiety doesn’t warrant complacency, of course; rather, it vindicates ‘just deterrence’ (an extension of just war theory), combined with a long-term focus on reducing nuclear risk.

Although any effort to predict long-term change will almost certainly be frustrated, we can still sketch rough outlines of plausible future scenarios, while always remaining prepared for surprises—both technological and political. In the past, technological improvements in accuracy made it possible to reduce the yield and volume of nuclear weapons. However, a whole new set of problems has come with the rise of cyberattacks on command-and-control systems, laser attacks on satellites, and autonomous weapon systems. These are the types of risks that we must seek to anticipate, understand and reduce.

Politics, too, will change. During the Cold War, the ideological antagonists slowly developed a regime of tacit and explicit rules of the road, because each recognised that it had an interest in avoiding nuclear war. Today’s strategic competition with China and Russia could take any number of turns in the future. As we adjust to changes and surprises, we must continue to consider how our decisions will affect the long-term goal of reducing the risk of nuclear war.