Illustration by Molly Crabapple

Seventy-five years ago, at 8:16 on the clear morning of August 6, the world changed forever. A blast equivalent to more than 12,000 tons of TNT, unimaginably larger than that of any previous weapon, blew apart the Japanese city of Hiroshima, igniting a massive firestorm. Within minutes, between 70,000 and 80,000 died and as many were injured. Hospitals were destroyed or badly damaged, and more than 90 percent of the city’s doctors and nurses were killed or wounded. By the end of the year, thousands more had died from burns and radiation poisoning—a total of 40 percent of the city’s population.

The mushroom cloud became a universal symbol of horror. As Michael D. Gordin and G. John Ikenberry, the editors of The Age of Hiroshima, describe, entirely new ways of thinking about war and peace had to be invented, together with a new understanding of global interconnectedness. “Very few aspects of life,” geopolitical, technological, or cultural, they write, “have been left untouched,” not just among the superpowers but worldwide.

In part because of effective deterrence, fear of their destructiveness, and a growing taboo against their use, and in part because of dumb luck, nearly a century has passed without nuclear weapons being used again in conflict. The US and the Soviet Union survived the cold war, living on a knife edge of fear that drove each to accumulate more than 30,000 nuclear weapons, enough to destroy all life on the planet many times over. In retrospect, as documents are declassified and participants speak and write about their experiences, and as brilliantly chronicled by Fred Kaplan in The Bomb, the competition emerges, on the US side at least, as a largely mindless cycle of more and larger weapons aimed at ever more targets, and more and more targets deemed to require ever more weapons, the whole enterprise impervious to the efforts of administration after administration to define saner policies.

Kaplan tells the story of how, two weeks into the Kennedy administration, Secretary of Defense Robert McNamara traveled to Strategic Air Command (SAC) headquarters in Omaha for his first briefing on nuclear war’s holy text, the Single Integrated Operational Plan (SIOP). One of its thousands of targets, he learned, was an air defense radar station in Albania. The bomb slated to destroy it was—by then only a few years into the arms race—roughly three hundred times larger than the bomb that destroyed Hiroshima. “Mr. Secretary,” said the commanding general, “I hope you don’t have any friends or relations in Albania, because we’re going to have to wipe it out.” Albania, a tiny country, was Communist but politically independent of Moscow.

Decades later, the same thinking—if that’s what it should be called—still prevailed. A Carter administration effort to reduce the consequences of nuclear war added “leadership” targets to the list of those to be hit in the belief that it would effectively deter Soviet leaders. The SIOP was accordingly revised to include not only government ministries but the homes and vacation dachas of every government minister, not just in Moscow but in every oblast across Russia. The use of megaton bombs to kill individuals meant, of course, that many hundreds of thousands of other people would also be killed.

The cold war ended peacefully, and the deployed nuclear arsenals of the US and Russia have been reduced by nearly 90 percent, but we are not safer today—quite the reverse. After decades of building just enough weapons to deter attack, China is now aggressively modernizing and enlarging its small nuclear arsenal. Russia and the US are modernizing theirs as well with entire menus of new weapons. Activities in space are enlarging the global battlefield. Advances in missile technology and conventional weapons “entangle” scenarios of nuclear and nonnuclear war, making outcomes highly unpredictable. The risk of cyberattacks on command and control systems adds another layer of uncertainty, as does research on artificial intelligence that increases the prospect of accidents and the unintentional use of nuclear weapons. Arms control agreements that significantly limited the US–Soviet arms race are being discarded one by one. And from Russian efforts to destabilize America through social media attacks on its democracy, to Chinese bellicosity in the South China Sea and clampdown on Hong Kong, to erratic lunges in US foreign policy, there is deep and growing distrust among the great powers.

Yet the public isn’t scared. Indeed, people are unaware that a second nuclear arms race has begun—one that could be more dangerous than the first. Decades of fearing a nuclear war that didn’t happen may have induced an unwarranted complacency that this threat belongs to the past. A million people gathered in New York’s Central Park in 1982 to call for an end to the arms race in the largest political demonstration in US history. Today the prospect of nuclear disaster is barely noticed.


In the US, the nuclear age has been a fruitless, decades-long search for answers to three linked questions. The most basic is: What is our goal in a nuclear war? The military has a definite answer: “to prevail.” Civilian leaders’ answers have varied widely. President Eisenhower favored nuclear weapons because they were less expensive than conventional forces, yet he nevertheless told the Joint Chiefs that our aim in a general nuclear war should be not “to lose any worse than we have to.” Defense Secretary Harold Brown, reaching for a formula to satisfy President Carter, described the goal as ending a war “on acceptable terms that are as favorable as practical,” leaving “acceptable” and “practical” undefined. Ronald Reagan wrote in his memoir that he thought that those who claimed nuclear war was “winnable” were “crazy,” apparently forgetting that he had signed a nuclear policy document that stated the US “must prevail.”

What winning might look like is what makes this seemingly simple question so hard to answer. In the early 1960s, SAC was asked how many Russians, Chinese, and Eastern Europeans would die from its all-out attack plan. The answer was a nearly inconceivable 275 million, just from the bombs’ blasts. (Heat, fire, smoke, and radiation would kill tens of millions more, but the numbers would vary depending on wind and weather, so SAC did not count them.) Presidents and their advisers found it difficult if not impossible to imagine the conditions under which they would launch such a holocaust. Only in the basement at SAC headquarters—where targeters sat, day after day, assigning weapons to targets in a policy-free environment—did it make sense. “Look,” yelled the SAC commander General Thomas Power at a nagging policy analyst from Washington who was arguing for a war plan with fewer casualties, “at the end of the war, if there are two Americans and one Russian, we win!”

The second question concerns deterrence: What weapons and force structure are needed to deter an enemy or enemies from attacking us? Unfortunately there are no metrics to measure what makes a deterrent credible. Answers are entirely in the eye of the beholder, and arguments can almost always be contrived to justify the need for more weapons. What can be said with certainty, however, is that the threshold the US judges necessary to deter the enemy is always set immensely higher than what has actually deterred the US. In The Button, former defense secretary William J. Perry writes that at the time of the Cuban missile crisis the US had about five thousand warheads to the Soviets’ three hundred, but “even with this seventeen-to-one numerical superiority, the Kennedy administration did not believe it had the capability to launch a successful first strike.” Notwithstanding the enormous gap between the two arsenals (which has never again been anywhere near as large), Washington was deterred by the risk of a Soviet counterstrike.

The third question, closely tied to what it takes to create deterrence, asks what happens if deterrence fails. Can nuclear weapons then be useful instruments for fighting, as opposed to preventing, a war? Understandably, presidents demand all kinds of flexibility—weapons and war plans suited to a general war and to regional aggressions in different settings of greater or lesser geopolitical importance. The problem is that weapons and plans tailored to every situation, especially smaller weapons and plans for limited nuclear war, may be understood by the enemy (and by domestic opponents) as preparations for going to war. “The logic,” writes Kaplan,

involved convincing adversaries that you really would use the bomb in response to aggression; part of that involved convincing yourself that you would use it, which required building certain types of missiles, and devising certain plans, that would enable you to use them—and, before you knew it, a strategy to deter nuclear war became synonymous with a strategy to fight nuclear war.

Many plans for limited nuclear war have been created on paper, but they immediately raise yet another critical question: Can there really be such a thing? To assert that the answer is yes, one has to believe that intentions can be clearly signaled (“I’m attacking you but with much less firepower than I might have used”), accurately interpreted by the other side, and responded to not in rage or fear but with calm reasonableness (“I’m retaliating but much more lightly than I might have”). There are all kinds of technical reasons to doubt that this is more than a fantasy. For example, at one point an American analyst discovered that Russian air defense systems could pinpoint no more than two hundred incoming missiles before they merged into a blob on the radar screen. Yet at that time the SIOP’s smallest limited attack option called for launching one thousand missiles, which would therefore be indistinguishable to the Russians from an all-out attack.


The more powerful reasons to doubt that there could be a limited nuclear war, to my mind, are those that emerge from any study of history, a knowledge of how humans act under pressure, or experience in government. In his “speculative novel” The 2020 Commission Report on the North Korean Nuclear Attacks Against the United States (2018), the nuclear analyst Jeffrey Lewis convincingly traces the path to an unintended war. The book’s lessons are much broader than the particulars of the Korean setting. Lewis uses variations on actual events to trace a series of miscalculations, mistakes, coincidences, domestic pressures, and misreadings of others’ intentions, beginning with the mistaken shooting down of a commercial South Korean plane by North Korea and ending in a nuclear war involving both Koreas, Japan, and the US. Each step toward disaster is plausible. After a limited South Korean missile response to the downing of its plane, to which Seoul chooses not to alert its American ally in advance, North Korean leader Kim Jong-un finds that he can’t use his phone. The phone system is simply overloaded, but in the aftermath, one of his aides tells the commissioners investigating how the war had happened that the North Koreans had concluded something quite different: “We assumed it was an American cyber-attack. Wouldn’t you?”

Photographers taking pictures of an unarmed intercontinental ballistic missile being tested by the US military at Vandenberg Air Force Base

Ringo Chiu/AFP/Getty Images

Photographers taking pictures of an unarmed intercontinental ballistic missile being tested by the US military at Vandenberg Air Force Base, California, May 2017

The recent real-world version of the recurring debates about limited war and the weapons needed to fight one is the Trump administration’s decision to deploy low-yield warheads on American Trident submarines. The move was prompted by Russia’s fielding of new low-yield tactical warheads aimed at Europe. Did this mean that Moscow had detected some gap in our deterrent that such a weapon could exploit? Didn’t Washington have to respond in kind, asked proponents of the new warheads? Opponents argued that Russia had turned to tactical nukes because it feared American advances in long-range conventional weapons. The disadvantage was on their side, not ours. Moreover, the Russians would be unable to quickly distinguish one of these low-yield warheads fired by a submarine from the many megaton strategic warheads these ships carry, and hence unable to immediately distinguish a limited from an all-out attack.* Nevertheless, proponents won the day. The warheads have been deployed, strengthening the hand of those who believe that nuclear wars can be fought and won.

A decade ago, President Obama made a fateful bargain to secure Senate approval of the New START arms limitation treaty he had reached with Russia. He agreed to a major upgrade of the aging American nuclear complex, including production facilities and laboratories, with a controversial price tag nearing $100 billion. This was the seed of a modernization program that has since multiplied to include command and control systems, all the delivery vehicles of the nuclear triad—bombers, ICBMs, and submarines—refurbishment of existing warheads, and the development of a range of new warheads and weapons.

The need for modernization results partly from aging systems that require replacement and partly, in an all-too-familiar pattern, from a perceived need to keep up with the Russians. Moscow began a sweeping modernization program in the early 2000s to keep up with American advances and compensate for weakness in its conventional forces. Rose Gottemoeller, the former deputy director general of NATO and chief US negotiator of the New START Treaty, argues that the real purpose of Russia’s program, which includes exotic weapons like an underwater nuclear drone and a nuclear-propelled cruise missile, had more to do with politics than with security. These weapons are meant, she says, to signal Russia’s “continuing scientific and military prowess at a time when the country does not otherwise have much on offer.”

Unfortunately, the program coincides with an American president who loves nukes. At the disastrous briefing session arranged for Donald Trump in the summer of 2017 in the Joint Chiefs’ secure room at the Pentagon known as the “tank,” he was shown a chart illustrating US and Russian success in cutting their arsenals from more than 30,000 warheads to about 6,000 each (which in both countries includes 2,500 retired warheads waiting to be destroyed). Like everything else that awful morning, it backfired. Why aren’t we building back up to 30,000, Trump demanded in a tantrum, during which he called the assembled military and civilian leaders “dopes and babies.” Defense Secretary Mark Esper leaves no doubt that modernizing the entire strategic nuclear force is the president’s “priority number one.” The modernization plan now includes a new fleet of ballistic missile submarines, a new stealth bomber, new ICBMs, the first new warhead design in more than thirty years, a sea-launched cruise missile, and a new air-launched cruise missile. The estimated price tag over the coming twenty-five years is $1.7 trillion (assuming, against experience, no cost overruns)—seventeen times Obama’s down payment—and represents a policy that is as far as it is possible to go from Obama’s plan to “reduce the role of nuclear weapons in our national security strategy”—of which Vice President Joe Biden was a strong supporter.

Some modernization is necessary, but there is no question that the current plan goes far beyond what is needed. Contractors are now driving it forward, and there is no one with sufficient standing to say “Stop.” But there are ways to save hundreds of billions of dollars without loss to national security. For decades, the triad has been the sine qua non of nuclear force structure. The apparent need for missiles, submarines, and bombers is now so entrenched that it is difficult to remember that it emerged not out of strategic necessity but from fierce rivalry among the military services, the Air Force and Navy especially, each of which wanted its own nuclear weapons.

Of the three legs of the triad, ground-based ICBMs are both the most threatening weapons to the enemy, because of their number and huge megatonnage, and the most vulnerable, because they sit in fixed, easily targeted silos. They are therefore “use them or lose them” weapons that must be fired on warning of an attack, before they are hit by incoming missiles. This means that a president has about ten minutes—less than the time it takes to confirm an attack—to make a life-or-death decision for the country and probably for the planet. Rather than spend $150 billion or more to replace these missiles, the sensible step is to retire them. Ballistic missile submarines, backed up by bombers plus cruise and hypersonic missiles launched from ships and planes, can provide the necessary firepower and strategic depth for an ironclad deterrent and the capability for a devastating second strike.

Years from now, the Trump administration’s wholesale withdrawal from international agreements, its “unsigning” of treaties, and its weakening of international organizations will stand out from the lies, the corruption, the incompetence, and the breaking of norms as one of its most damaging features. A partial list includes the Trans-Pacific Partnership (TPP) trade deal, NAFTA, the Paris Climate Accord, the Iran nuclear deal, the Arms Trade Treaty, and, most recently, the World Health Organization. Among these, withdrawals in the nuclear arena may prove to be especially harmful.

The administration’s hostile view of arms control was evident in its 2018 Nuclear Posture Review: The US “will remain receptive to future arms control negotiations if conditions permit.” These agreements have their flaws. Negotiations take years and years. Often the sides agree to give up weapons they no longer want. Violations are not uncommon and, to satisfy domestic hawks, both sides frequently build new weapons to compensate for those they negotiate away. Nonetheless, over more than three decades of painstaking effort by Republican and Democratic administrations, a set of agreements was hammered out that built trust between the West and Russia, created a degree of transparency into what the other side was doing, and banned or severely limited particularly destabilizing types of weapons, such as missile defense systems and multiple-warhead missiles. Over time the agreements slowed the arms race from a gallop to a jog. Without them, the two sides might still be holding 65,000 warheads instead of 13,000.

The dismantling of these agreements began with President George W. Bush’s withdrawal from the Anti-Ballistic Missile Treaty in 2001, but in the last few years Trump has wiped away almost everything that was still in place. In 2018 he announced that the US would withdraw from the Intermediate-Range Nuclear Forces (INF) Treaty. Russian violations of the agreement, which Moscow had refused to acknowledge over many years, made this a close—and understandable—call. Still, withdrawing from an agreement gives the other side what it wants. And prompt American testing of a missile banned under the treaty suggests that Washington was eager to dispose of it.

The administration then announced its intention of leaving the Open Skies Treaty, a 1992 multilateral agreement that allows signatories to fly unarmed observation flights over the territory of the others to collect data on military forces and activities. Though its value to the superpowers has diminished with satellite technology, it remains important to European parties and has been a significant contributor to strategic stability.

The only remaining limit on strategic arms is New START, which is due to expire two weeks after the next president is inaugurated, unless extended by mutual agreement for a further five years. The treaty limits each side to 1,550 deployed strategic warheads and 700 launchers. The US now insists it will not extend the treaty unless China is included. Since both Russia and the US have about five times as many warheads as China (which may double its arsenal in the next ten years), Beijing has absolutely no reason to become part of US–Russian arms control talks at this point and has made that clear on many occasions. Moreover, although the administration has been talking about this for two years, it has taken no diplomatic steps—plans, proposals, or drafts exchanged—to make it happen. The policy bears all the signs of a poison pill designed to force New START’s demise while obscuring the cause.

In addition, news leaked in May—perhaps purposefully—that administration officials were discussing breaking the twenty-eight-year moratorium among the major powers on nuclear testing. The stated reason was to try to use a nuclear test to pressure Russia and China to agree to Washington’s New START position. On the very long list of self-defeating moves this administration has made, breaking the moratorium belongs near the top. A nuclear test would not frighten Moscow or Beijing into doing what the US wants, it would drastically weaken global nonproliferation efforts, it would make the US an international pariah, and it would erase an important US advantage. The US has conducted more tests than any other country—more than one thousand to China’s forty-five, for example—so if testing is resumed, every other nuclear power stands to gain much more than the US.

Taken together, the loss of New START, a tease—at least—on a resumption of testing, and the vast weapons modernization plan, all enhanced by work on new cyber and space weapons, applications of AI, and a range of new weapons capable of carrying either conventional or nuclear warheads, amount to a running leap into a new arms race, this time among at least three powers, perhaps joined by North Korea, Iran, and other new nuclear states. The Trump administration seems eager for it, no matter the cost. “We know how to win these races,” said the US arms control negotiator Marshall Billingslea recently, “and we know how to spend the adversary into oblivion.”

Little can be done to reverse direction unless Donald Trump is defeated in November. Even if he loses, stopping a burgeoning arms race will have to compete for public attention with an overwhelming list of priorities: repairs to democratic governance, health care reform, racial justice, climate change, economic recovery—and enormous post-pandemic budget deficits. Only the last of these can help focus attention where it’s needed. Without public pressure the military-industrial-congressional complex will push nuclear modernization forward step by multibillion-dollar-step without attention to the $2 trillion bottom line, locking in a new generation of threats that Russia and China will feel they must counter. The deficits, however, demand a more provident approach to the ballooning defense budget (now larger than everything else in the federal discretionary budget combined). A spasm of spending on what are essentially twentieth-century weapons, without a pause to rethink, is strategically irresponsible and fiscally unsound. Congress can instead insist that appropriated dollars not be spent on nuclear weapons tests, support the new president in restoring various arms limitation agreements, and undertake a serious, nonpartisan study of the actual need for a new fleet of ICBMs.

The single step from which profound policy change could flow, domestically and internationally, would be formal endorsement by the five original nuclear powers—the US, Russia, the UK, France, and China—of the Reagan-Gorbachev principle, jointly articulated by the two leaders at their 1985 summit. It states simply, “a nuclear war cannot be won and must never be fought.” International adoption would simultaneously indicate the nuclear powers’ recognition of the rising dangers of nuclear conflict and the need to move toward nuclear forces around the world that are structured for deterrence, not war fighting. Words as principle have power. Eventually, these eleven words could underlie the next generation of arms control negotiations, strengthen the global nonproliferation regime, and help short-circuit a second nuclear arms race.

—July 22, 2020