Hedgehogs on a Treadmill

International relations is what decision-making scholars call a “low-validity domain.”¹ Politics among nations is marked by murky causal relationships and frequently vague, often delayed feedback, all of which inhibit learning. Despite the popularity of the metaphor, geopolitics is not chess. There are few rules, meaning that experience does not necessarily lead to expertise and that mastery is illusory. Policymakers may benefit from their years of service and from the knowledge generated by political scientists and their academic kin, but high-stakes national security decisions often come down to judgment, a faculty that can fail the best of us at the worst of times.

Good judgment is particularly hard to come by when the future is murkier than usual, and as publications from Financial Times and Foreign Affairs to Harvard Business Review have all declared, we are living in an “age of uncertainty.”² Last year, a United Nations report noted that political polarization, broad societal transformation, and climate change have combined to generate a “new uncertainty complex.”³ And a recent analysis found that global uncertainty has increased dramatically over the past decade and that the outbreak of the COVID-19 pandemic produced more uncertainty in the United States than any other event since 1952—a remarkable finding considering the many wars, crises, and disruptions of the past 70 years.⁴

This high degree of uncertainty suggests that we should embrace epistemic humility—the recognition that we often know far less than we think we do.⁵ Yet instead of analytical restraint among political experts, scholars find opinionated overconfidence. Psychologist Philip Tetlock’s early research on expert political judgment found that events which experts deemed 100 percent certainties occurred only 80 percent of the time. More generally, their expectations of the future tended to be no more accurate than those of well-informed amateurs or simple extrapolation algorithms. Their judgment was hit-or-miss, to put it kindly.⁶

Although many took these findings as an indictment of the very notion of expertise, Tetlock found significant variance among specialists with different cognitive styles—specifically between “hedgehogs” and “foxes.” In Tetlock’s formulation, hedgehogs cling tightly to parsimonious worldviews, while foxes nimbly switch among many, recognizing that all models are wrong but that some are useful—sometimes. Forecasters benefit from acknowledging what they don’t know and continually asking, “How might I be wrong?” Epistemic humility is a potent judgmental tonic.

Such humility would seem particularly beneficial for nuclear experts, if for no other reason than the stakes of being wrong could well be civilizational. But the greatest barrier to demonstrably effective nuclear policy is a lack of experience. As scholar Francis Gavin has noted, we have only three fundamental data points: two, nine, and zero.⁷ Two atomic bombs have been used, nine states possess nuclear weapons, and there have been zero nuclear wars. That zero is both welcome and problematic: We can take comfort from the fact that no one has employed a nuclear weapon in war since 1945, and we can infer that they have not done so because the costs of nuclear war would be unacceptably high, as deterrence theory suggests. But we cannot prove it, and there are plausible competing or complementary explanations, leaving a substantial analytical challenge. Determining why something happened is difficult. Determining why nothing happened is far more so.

Nevertheless, arguments about nuclear deterrence have often employed a deductive logic based on rationalist assumptions that have little basis in evidence because no one has ever fought a nuclear war. Which makes many of the beliefs about nuclear deterrence just that: beliefs. And, absent the ability to test hypotheses, it is difficult to say that one belief is more accurate than another. We cannot, a priori, distinguish good strategy from bad strategy. We do not know what would cause deterrence to fail, and if it failed, we do not know how a nuclear war would unfold. Despite decades of scholarship by some of the nation’s top thinkers, there is not a science of deterrence, and theories of nuclear conflict remain contested.⁸

“Determining why something happened is difficult. Determining why nothing happened is far more so.”

The nuclear age does hold lessons, notably in the terrifying (but empirically useful) history of crises and near misses—times when that zero threatened to become nonzero.⁹ Unfortunately, the usual challenges of learning from history—wherein we must confront the mixed lessons of experience, assess the value of competing narratives, and continuously update beliefs in the face of new evidence—are compounded by the closeness with which the U.S. government guards its nuclear secrets and the difficulty of accessing Soviet archives. This is why, 60 years later, our understanding of the Cuban Missile Crisis continues to evolve.¹⁰ Even if we understood perfectly what happened during those 13 days in October 1962, we could only apply those lessons cautiously because a future crisis will differ in fundamental ways. We must always handle historical analogies with care.¹¹

Given the dearth of experience and given the challenges of extracting knowledge from the experience we do have, one might expect nuclear experts to be like Tetlock’s foxes, displaying modesty in the face of existential threats. Instead, the field is rife with hedgehogs—and it has been for a long time. In the late 1980s, one RAND Corporation report complained: “The American debate over the use and control of nuclear weapons tends toward the theological. … [S]ince little evidence is available, the debaters’ assertions are untested and untestable. The dearth of test data has prevented the strategic debate from changing very much.”¹²

It was that very debate in the 1980s that prompted Tetlock’s original research on expert political judgment. To oversimplify the dispute: Doves were convinced that Ronald Reagan’s arms buildup would prompt nuclear apocalypse, while hawks were certain that only Reagan’s arms buildup would prevent it. How to determine who was right? As Tetlock wrote: “It was not clear how the classic methods of clarifying causality, experimental and statistical control, could even be applied to explain the nonoccurrence of an event (nuclear war) that qualified as sui generis if one ever did.”¹³ This epistemological challenge did little to temper the confidence of either hawks or doves, but it did preclude convergence. Instead of synthesis, we got silos.

More than three decades later, the arguments persist. As archives have opened and scholars have applied new research methods to larger data sets, we know more than ever, but wisdom remains elusive. The barriers that obscure the truth are still in place, yet many nuclear experts are convinced that they know how nuclear dynamics work and therefore that they can identify the optimal course of action. As Gavin has written: “I have had more than one important scholar in this field tell me we know all [we] need to know about how nuclear deterrence works and, by extension, why we have never had a thermonuclear war.”¹⁴ Political scientist Paul Avey reports similar experiences, writing that, by the end of the Cold War, “there was a sense that we understood the major contours of the nuclear world.”¹⁵

“As archives have opened and scholars have applied new research methods to larger data sets, we know more than ever, but wisdom remains elusive.”

That was not true then, and it is not true now if for no other reason than the topography of the nuclear landscape has shifted radically in the last 30 years. Just the past few years have seen dramatic change. North Korea, which has enough fissile material for several dozen nuclear weapons, has flight-tested a solid-fuel ICBM capable of reaching the United States.¹⁶ Iran has enriched uranium to the point where it could have enough for a weapon within a couple of weeks and produce an actual weapon within a year.¹⁷ China is expanding and modernizing its strategic nuclear forces, and a recent Pentagon report estimates it could achieve rough parity with the United States by 2035, suggesting that it might be abandoning its minimum deterrent posture.¹⁸ And of course Russia has reemerged as an adversary. Vladimir Putin and other top Russian officials have repeatedly threated to use nuclear weapons in Ukraine, and Russia has deployed nuclear weapons in Belarus. Meanwhile, Moscow has suspended participation in the New Strategic Arms Reduction Treaty (New START) after undermining the Intermediate-Range Nuclear Forces Treaty to the point of unsustainability, bringing the U.S.-Russian arms control regime to the brink of collapse.

We have entered what a number of analysts have termed a “new nuclear age,” and unfortunately we are cognitively ill-equipped to navigate it.¹⁹ As Vipin Narang and Scott Sagan have written, “This new nuclear age demands new thinking and analysis about the challenges generated by the continued existence and spread of nuclear weapons.”²⁰ The growth of China’s arsenal alone has replaced the relative (if often overstated) stability of Cold War bipolarity, replacing it with a dynamic that scholars and practitioners have compared to the three-body problem in celestial mechanics.²¹ Whereas Newtonian physics allows scientists to model the gravitational interplay of two objects, the behavior of three is unpredictable. One nuclear analyst drew a more colorful analogy: “I think of giant Calder mobiles. You push on a thing in one place and the whole thing starts swinging in a really unpredictable, wacky way.”²² Nevertheless, the schools of thought to which most American experts belong today are familiar:

• Dominance: Nuclear weapons did not change the nature of war, which remains an extension of politics, and deterrence requires robust offenses and defenses because they enhance credibility and will limit damage if deterrence fails. In that case, the United States must be prepared to meet and defeat the enemy at every rung of the escalatory ladder, thereby ending the conflict on favorable terms. More extreme adherents believe that arms control advocates are merely useful idiots who legitimize America’s adversaries, constrain its freedom of action, and codify the perversity of “mutual assured destruction.”
• Flexible Response: The primary goal of nuclear weapons is to deter an attack on the United States and its allies by maintaining a secure second-strike force. Arms control can provide transparency and predictability among adversaries, enhancing strategic stability. Some adherents advocate a no-first-use pledge, but others argue that the United States might face a conventionally superior foe who could only be stopped with nuclear weapons. In a crisis, the president needs options, and adversaries with secure retaliatory forces may dismiss U.S. threats of a strategic nuclear strike. Lower-yield weapons preserve a credible response, convincing the adversary that victory is impossible.
• Minimum Deterrence: The only credible role for nuclear weapons is to deter a strategic attack on the homeland. Deterrence is robust because even a handful of survivable weapons ensure a retaliatory capability that could inflict unacceptable damage on an adversary. A large arsenal that includes “more usable” weapons only increases the risk of accident, miscalculation, and inadvertent escalation. Some advocates argue that the weapons themselves are not the problem, meaning that, although potentially useful, arms control is no panacea. Peace requires reconciling the competing interests that lead to conflict.
• Abolition: Deterrence provides a precarious peace at best, and a nuclear war would be apocalyptic because escalation is not controllable. What’s more, nuclear weapons are morally abhorrent, and the fact that the president has sole launch authority is reckless. Nuclear disarmament—a commitment already enshrined in the Treaty on the Non-Proliferation of Nuclear Weapons—is the only sensible goal. In the interim, the United States must take all diplomatic steps possible to reduce the risk of nuclear war, and it should eschew both “tactical” nuclear weapons and strategic missile defenses because they threaten stability.

Nuclear experts I interviewed were blunt in appraising the similarities between Cold War and post-Cold War thinking. One said, “I think deterrence theory hasn’t changed in 70 years.” Another agreed: “Nuclear thinking is largely stuck.” A third just said: “It’s fucking Groundhog Day.”

To some, the situation teeters between the satirical and the soporific. In an interview last year, Middlebury’s Jeffrey Lewis lamented the predictability of discussions among nuclear experts: “I almost never go to talks in my own field anymore, because I know what people are going to say. It’s like the old joke about the comedians’ convention where somebody says, ‘Number 47!’ and everyone laughs—that’s what it’s like.”²³ Remarkably, this quip parallels a comment that nuclear scholar Janne Nolan made to me some 20 years ago during a particularly tedious panel discussion we attended: “They should just hand us all auction paddles with numbers representing different arguments. It’d go a lot faster.”²⁴

Perhaps the greatest change has been the revival of the abolition movement,²⁵ whose profile surged in 2007 when former statesmen George Shultz, Sam Nunn, Henry Kissinger, and Bill Perry published an op-ed calling on the United States to reassert the “vision of a world free of nuclear weapons.”²⁶ Two years later, President Barack Obama, who had thought about the perils of Cold War deterrence since his undergraduate days,²⁷ said, “I state clearly and with conviction America’s commitment to seek the peace and security of a world without nuclear weapons.”²⁸ A few months later, the Nobel Committee awarded him its Peace Prize, emphasizing his “vision of and work for a world without nuclear weapons.”²⁹ But the following year Obama signed off on a nuclear posture that fit within the flexible-response camp, demonstrating the stubbornness of the status quo.

One way to understand this stubbornness is as an example of a calcified mental model. Mental models consist of beliefs about cause and effect that allow us to test courses of action before actually committing to one.³⁰ They help us make sense of a world that overwhelms with information. Unfortunately, we tend to filter information in a way that supports rather than updates our beliefs. As a result, our mental models remain sticky even in the face of disconfirming evidence. They turn stale when exposed to change, yet we remain confident in their reliability. The problem is not only that static models produce delusions—beliefs about reality that do not accurately reflect reality—but also that those delusions feed decisions that can have unintended consequences. In a rapidly changing world, cognitive stasis is deadly. Our beliefs must shift as the operating environment shifts.

If the problem is that our mental models are stuck, then the obvious course of action is to unstick them. We don’t need a new idea. We need the ability to generate new ideas as appropriate. We don’t need a new map. We need the ability to map. We need the fox’s epistemic humility and the intellectual flexibility that it promotes. Unfortunately, the national security community does not reward such things. Instead, we mistake confidence for competence, we disdain “wishy-washiness,” and we embrace simplistic analyses that make for punchier headlines, pithier bottom lines, and snappier soundbites. As MIT’s Heather Williams put it, “Moderation isn’t sexy, particularly when it comes to nuclear weapons.”³¹ In the wild, foxes hunt hedgehogs. In Washington, the opposite is true.

“In a rapidly changing world, cognitive stasis is deadly. Our beliefs must shift as the operating environment shifts.”

This project was an attempt to challenge the hegemony of the hedgehog by allowing nuclear experts the opportunity to play with ideas and causal relationships in a consequence-free environment, generating a set of alternative futures, or scenarios. Scenarios were first used by Herman Kahn to explore the uncertainty of the nuclear revolution—to deal with the lack of experience. As he wrote, “Nuclear war is still (and hopefully will remain) so far from our experience, that it is difficult to reason from, or illustrate arguments by, analogies from history.”³² While his erstwhile colleagues at the RAND Corporation turned to systems analysis, game theory, and other quantitative methods, Kahn turned to imagination and narrative. His cure for the dearth of experience was “ersatz experience.” If we didn’t have analogies, we would simply make them up.

Via Kahn, scenarios became a popular tool for breaking up staid ways of thinking. Pierre Wack, a Royal Dutch Shell executive, first encountered scenarios in the 1970s at the Hudson Institute, which Kahn had co-founded.³³ He subsequently used the method to challenge his colleagues’ persistent assumption that the company’s access to Middle East oil would remain open, despite signs of incipient disruption. Static mental models might work fine in stable times, he later wrote. “In times of rapid change and increased complexity, however, the manager’s mental model becomes a dangerously mixed bag: Rich detail and understanding can coexist with dubious assumptions, selective inattention to alternative ways of interpreting evidence, and illusory projections.” Engaging with alternative plausible futures shone a light on these cognitive failings. “Scenarios give managers something very precious: the ability to reperceive reality.”³⁴ That re-perception, in turn, facilitates new thinking, or what Kees van der Heijden, another Shell scenario practitioner, called “the art of strategic conversation.”³⁵

This exercise was an effort to encourage that art.