Introduction

“In developing this strategy, the Department considered the risks stemming from inaccurate predictions, including unforeseen shocks in the security environment.”—U.S. Department of Defense, 2022 National Defense Strategy¹

“I am now going to ask the reader for an unpleasant feat of imagination.”
—Herman Kahn, On Thermonuclear War²

Nuclear weapons are back. Not that they ever went anywhere. But while most people, including policymakers, were attending to other matters, the atomic landscape morphed. With Russia’s 2022 invasion of Ukraine, it shifted tectonically, focusing American officials and their constituents on nuclear threats in a way not seen since the collapse of the Soviet Union.

Polls last year showed that roughly three-quarters of Americans feared the conflict would prompt a Russian nuclear strike on the United States. In October 2022, President Biden, one of the few remaining U.S. elected officials with deep nuclear experience, said, “We have not faced the prospect of Armageddon since Kennedy and the Cuban missile crisis.”³ This spring, nuclear experts moved the Doomsday Clock, that iconic symbol of existential danger, to a mere 90 seconds from midnight—the closest it has ever been.⁴

These fears may be overstated—Metaculus, a forecasting platform, has consistently put the probability of Russian nuclear use in Ukraine in 2023 at less than 0.5 percent⁵—but Putin’s provocations are not the only nuclear danger the United States confronts. The erosion of the U.S.-Russian arms control regime, the growth of China’s nuclear forces, and the persistence of geopolitical flashpoints—to say nothing of North Korea’s solid-fuel intercontinental ballistic missiles (ICBMs), Iran’s growing stocks of enriched uranium, and the seeming proliferation of personality-driven regimes—all herald a “new nuclear age” whose dynamics are more complex than the bipolarity of the Cold War.

A new nuclear age would seem to demand new ideas about how best to deter America’s adversaries and what to do if deterrence fails. Yet for all the recent talk of “integrated deterrence”—the holistic approach to discouraging enemy aggression articulated in the 2022 National Defense Strategy—when it comes to nuclear weapons specifically, the novelty of the current moment seems mostly to have revived old arguments about old ideas.⁶ Once again, we are debating the utility of lower-yield devices, the wisdom of a no-first-use posture, and the proper role of nuclear weapons in national security generally.

Admittedly, “new” is poor proxy for “better.” The quality of an idea relies on how well it reflects the way the world works, just as the quality of a map depends on its geographical accuracy. Many old ideas remain estimable to the extent that the assumptions underlying them still accord with reality. The central challenge of nuclear policy is that the nature of reality is elusive. The relationships between cause and effect are obscure. This makes it difficult to be certain that we are pursuing our goals most effectively.

We can ascribe some of our ignorance to the limits of causal inference, but we also lack data. The Cold War left many questions unanswered because it remained cold, affording few opportunities to test hypotheses. So, although the superpowers may have avoided direct conflict because they feared the high costs of nuclear war, in accordance with deterrence theory, it is also possible that we just got lucky. The ratio of good strategy to good fortune remains unknown. Seventy-eight years after the United States dropped atomic bombs on Hiroshima and Nagasaki, we cannot say whether nuclear deterrence is delicate or robust. We cannot say if “tactical” weapons strengthen deterrence by increasing the credibility of nuclear threats, or whether they weaken it by making nuclear weapons more “usable.” We cannot say, if such weapons were used, whether conflict could remain controlled or whether battlefield use would inevitably escalate to a strategic exchange.

Yet we do say these things. Often with great conviction. The problem is not that we have failed to surmount the limits of causal inference. We have not because we cannot. The problem is that we often act as if we have. The changing nature of the nuclear threat has prompted less theoretical introspection than warranted. In many cases, we are simply reprising Cold War arguments about the military utility of nuclear weapons and the value of diplomatic constraints on them. Confronted with the challenge of traversing new territory, we have simply dusted off outdated maps, increasing the risk that we will head in the wrong direction and wind up somewhere we don’t want to be.

The goal of this project was not to map the new nuclear age—the landscape is changing too rapidly for a static picture—but to renew interest in cartography. With apologies to Eisenhower, maps are useless, but mapping is essential, particularly in a time of dynamic change. Every policy is a prediction, and many nuclear policies rest on implicit long-term predictions. For example, per its nuclear modernization plans, the United States is building the arsenal policymakers think the country will need decades from now based on existing beliefs and expectations. Yet the far future is likely to surprise us, so we must learn to keep up.

Many defense thinkers nod to the difficulties posed by the uncertainty of the future—the 2022 Nuclear Posture Review notes the need for resilience and adaptability in the face of “significant uncertainties and unanticipated challenges”⁷—but how can we prepare for the unanticipated? As the economist Thomas Schelling once wrote, “One thing a person cannot do, no matter how rigorous his analysis or heroic his imagination, is to draw up a list of the things that would never occur to him.”⁸ As a solution, Herman Kahn, the infamous defense strategist, suggested several “strange aids to thought,” including scenarios—narratives of plausible alternative futures that he used to imagine various paths to and outcomes of nuclear conflict.

In that tradition, this project surveyed and convened subject-matter experts to generate plausible visions of nuclear dangers in the year 2043. This exercise in scenario-generation was not an effort to predict the future, but an attempt to highlight how little we know about it—a thought experiment to help the nuclear-security community consider the uncertainty of the unthinkable.