This past summer, a bipartisan majority of Congress, with the blessing of President Biden, approved a massive military-spending bill that included $51 billion for nuclear weapons — nearly 20 percent more than allotted by the previous year’s budget, which itself broke previous records. Earlier in the spring, the Biden administration sent to congress a Nuclear Posture Review, committing to upgrade all three “legs” of the “strategic Triad” — including a new missile-launching submarine, a new bomber and a new land-based intercontinental ballistic missile — as well as a bevy of new bombs and warheads for these weapons to launch or drop. Since these weapons are still in development or the early phases of production, the costs are bound to grow; the price tag for the refurbished Triad alone is estimated at $2 trillion over the next 30 years.

The official rationale for this upgrade is that the existing subs, bombers and ICBMs are approaching obsolescence. Even if this claim were true (more about that later), it begs the question of whether the arsenal needs to be as large as it is. A serious assessment of the arsenal must begin by asking “How much is enough?” and, its corollary query, “Enough to do what?”

Yet in the debate over America’s nuclear stockpile, to the extent there is debate, these questions are going unasked. It is hard to have an informed public debate, as many of the issues are classified, esoteric or both. But even the debates in Congress and inside the executive branch tend to be shallow. Almost nobody is asking those basic questions. In fact, in the 60-plus years of the nuclear arms race, almost nobody ever has.

It’s important to examine the secret history of this race to understand not only how we got here but how to ask those questions — and how to change course.

Momentum drove the nuclear arms buildup from the beginning. There never was a decision to drop the atomic bombs on Hiroshima and Nagasaki at the end of World War II. Once it was decided to build the bombs two years earlier, it was inevitable they would be dropped. ¹ The United States had just two A-bombs in early August 1945; the Japanese emperor surrendered before a third bomb was ready to go.

After the war, as the Cold War got underway, it was assumed — no doubt correctly — that the Soviets would build A-bombs once they figured out how (they exploded their first roughly four years after Hiroshima); so the American weapons labs preemptively churned out more bombs. Once American scientists tested the much more powerful hydrogen bomb in 1952, there was little doubt that it too would be built (though many who had helped build the A-bomb protested) or that the Soviets would build their own (which they did three years later).

President Dwight Eisenhower, a retired five-star Army general and WWII commander, was not at all bloodthirsty; once he understood the power of nuclear weapons, he feared and detested them. But he also believed, as did most officials and analysts, that if the U.S. and the USSR ever locked arms, even in a “small” war over a narrow strip of territory in Europe or Asia, it would soon escalate to nuclear exchanges. So the wise policy would be to deter the Soviets from attacking in the first place, and the best way to do that, he figured, was to warn them that we’d blow them to smithereens if they did. His secretary of state, John Foster Dulles, called the policy “massive retaliation,” and the Joint Chiefs of Staff — composed of the top U.S. military officers — translated it into a war plan that italicized massive. ²

Few realized at the time, or in the years since, just how massive it was. By 1960, the U.S. war plan called for launching the entire nuclear arsenal — at the time, 3,423 weapons, exploding with the blast power of 7,847 megatons — against 1,043 targets in the Soviet Union, its satellite countries in Eastern Europe and Communist China. ³ This was not a plan to strike back if the Soviets launched a nuclear attack on the U.S.; it was a plan to strike first if the Soviets mounted a non-nuclear invasion against U.S. allies.

Some in Washington asked how many people such an attack would kill. The answer that came back from those who devised the war plan at Strategic Air Command (SAC) in Omaha was 275 million. ⁴ Such a figure had previously been inconceivable. No one could imagine a war aim that required killing so many civilians.

What’s striking is that, even so, no one among the few officials privy to this plan questioned its validity or how the numbers were calculated. They never asked whether such a massive arsenal, or such a cataclysmic attack, was necessary for national security.

The plan was founded, in large part, on the basis of self-interest. SAC — the branch of the now-independent U.S. Air Force that controlled nuclear plans and operations — had set up a unit called the Joint Strategic Target Planning Staff (JSTPS). Its job was to find every plausible target inside the Soviet empire, then assign U.S. nuclear weapons to destroy each one. As JSTPS found more targets, SAC had a rationale to request more weapons. As the Soviets matched the U.S. arms buildup, they created more targets — thus driving the rationale for still more U.S. bombs and warheads.

As an added twist, JSTPS declared that some of these targets were deemed more “high-value” than others. At the time, no single nuclear bomb had as much as an 80 percent chance of destroying a particular target on the other side of the world, partly because they were inaccurate, partly because some might be duds. So more than one weapon — in some cases three or even more weapons — would have to be aimed at each of those targets in order to achieve the required level of destruction. In the first formal war plan in 1960, known as the SIOP (for Single Integrated Operational Plan), JSTPS designated targets and specified for each a minimum probability that it would be destroyed. Seven of the most important, hard-to-hit targets needed to be targeted with a probability of 97 percent, 213 targets with 95 percent, 592 targets with 90 percent and 715 targets with 80 percent. ⁵ Raise the required level of destruction — that raised the required number of weapons.

All of this was more highly classified than anything else in government. These decisions and calculations — which crucially affected how many weapons the government would build — were made with no input from any officials, or even military officers, in Washington. No part of the plan was declassified until decades later. ⁶ Still, through the decades, the Pentagon and Congress routinely approved SAC’s requests for more weapons.

As this self-propelled nuclear arms race hurtled into motion, two obstacles reared their heads. First, as the Soviets started producing and deploying their own nuclear weapons, “massive retaliation” became a suicide pact. If the Soviets invaded West Germany and the president clobbered Moscow and other Russian cities with nukes, the Kremlin would clobber New York and other American cities right back.

The second obstacle came in the form of a bureaucratic battle inside the Pentagon. At the time, nuclear weapons dominated the military budget (much of the Army, which fought World War II and the Korean War, had been demobilized). The Air Force controlled the arsenal, and so dominated the budget. But in the late 1950s, the Navy, whose battleships had been eclipsed by the Air Force in the budget wars, designed and built the Polaris submarine — a vessel that could carry 16 nuclear missiles and launch them while moving, undetected, underwater. Analysts had observed that Air Force bombers and land-based missiles would soon be vulnerable to a Soviet first strike; the Air Force cited this claim as another excuse to request funds for more bombers and missiles. But analysts in the Navy’s official think tank came up with a new strategic idea: finite deterrence. All the U.S. needed, they argued, was enough nuclear weapons to destroy the 100 largest cities in the Soviet Union; this could be done with a mere 640 missiles in 40 submarines. ⁷ If the Soviets built more weapons, it wouldn’t matter; as long as the U.S. had the submarines, which the Soviets couldn’t track, it could destroy those cities and deter the Soviets from starting a war to begin with. Unlike the Air Force, the Navy could ensure deterrence without setting off an arms race.

Air Force generals were panicked, so they turned to their own think tank, the RAND Corporation, where analysts had devised a counter-idea: If the Soviets invaded Western Europe, the U.S. should respond not by firing all its nukes at once but by launching a limited nuclear attack only against the Soviets’ ICBMs, bomber bases and submarine ports. They could then tell the Kremlin, “If you continue your aggression, we will use our remaining arsenal to destroy your cities.” The idea was to stop the war before it careened out of control.

At first, the idea did not appeal to the Air Force generals: The whole point of a nuclear bomb, to them, was that it inflicted massive damage; they had no interest in war plans that put the arsenal’s power on a leash. But the RAND idea required bombs that were accurate enough to hit an airstrip or a missile base without damaging a nearby city. Bombs dropped from Air Force bombers could do that; Polaris missiles, which were notoriously inaccurate, could not. So the generals adopted the strategy, known as “counterforce” — at least rhetorically — as a way of beating back the Navy’s challenge. (The RAND memorandum outlining the strategy was titled “The Polaris Problem.”) ⁸

Robert McNamara, secretary of defense under President John F. Kennedy, also liked the idea, but for different reasons. Days after he entered the Pentagon, he was briefed on SAC’s nuclear war plan — which would launch the entire U.S. arsenal, as quickly as possible, at every target in the Communist world — and was appalled by its inhumanity. He and his aides ordered SAC to revise the plan to give the president “options” to launch various limited attacks. ⁹

Within a few years, though, McNamara discovered a downside to his order. If the plan entailed destroying all of the Soviets’ nuclear weapons, then for every new nuke that the Soviets deployed, the U.S. would need to buy more than one new nuke. In other words, an unending arms race was a consequence of the policy — and the Soviets were now starting to deploy a fair number of nukes. So, in 1964, McNamara declared a new nuclear policy, stating — in very similar terms to the Navy’s strategy of a few years earlier — that the U.S. needed only enough weapons to kill roughly 30 percent of the Soviet Union’s population and half of its industrial capacity. At that point, all of the major cities would be destroyed; launching any more weapons would inflict only marginally more damage.

McNamara called this new policy “assured destruction.” (A critic of the idea, who supported counterforce, lampooned it as “mutual assured destruction” in order to come up with the acronym MAD.) ¹⁰ But McNamara was being disingenuous. In a top secret memo to President Lyndon B. Johnson, McNamara wrote that MAD did not reflect how the U.S. would actually use nuclear weapons in a war. In fact, only a few hundred bombs and warheads out of the arsenal’s several thousand were aimed at “urban-industrial” targets; the rest were aimed at military targets (though many were near or inside cities, so tens of millions of civilians would still die).

Counterforce was still the strategy; a plan for nuclear war fighting was always — secretly — an element of credible nuclear deterrence. MAD — the threat to destroy civilian targets in response to a Soviet first strike — was a political device to curb the Air Force’s appetite. ¹¹

The generals, of course, were livid. McNamara left their strategy in place but, in their eyes, didn’t give them enough weapons to execute it. They’d asked for 2,000 ICBMs, but McNamara capped the number at 1,000. So, the generals came up with their own clever scheme. NASA scientists were designing a rocket that, once in outer space, could unfurl several satellites into different orbits. Air Force scientists saw that they could adapt the technology so that an ICBM could fire several warheads at different targets. They called the program the MIRV (for Multiple Independently targetable Reentry Vehicle), and proposed installing MIRVs on an upgrade of the Minuteman missile, called the Minuteman III.

The generals told McNamara that they would accede to 1,000 Minuteman missiles if McNamara approved MIRV. McNamara took the deal. ¹² The generals had outsmarted him. Over the next decade, they built 550 Minuteman IIIs, each with three warheads; along with their 500 single-warhead missiles, that gave them 2,150 ICBM warheads. In other words, the generals got what they wanted, plus some.

Soon, the Navy put MIRVs on its submarine-launched missiles — and, a bit later, they too built missiles accurate enough to destroy Soviet military targets, as a result of which they abandoned their “finite deterrence” philosophy.

All of this was highly classified. Most people, even those versed in nuclear strategy and history, thought that the U.S. policy was MAD. In 1974, during Gerald Ford’s brief presidency, when Secretary of Defense James Schlesinger publicly announced a policy of “limited nuclear options,” aiming nuclear weapons at Soviet military targets while avoiding cities, many politicians and analysts thought this was a new and dangerous departure from the philosophy of deterrence. But it was not at all new (nor was it understood that Schlesinger’s most limited options would still involve firing a couple hundred nuclear weapons). The real policy was never MAD — it was always counterforce; Schlesinger’s policy was a refinement of that policy. As long as an arms race was on, counterforce extended the race in perpetuity: More nukes meant more targets, more targets required more nukes.

But this is not to say that the impulse to build more was entirely rational. ¹³ In 1989, soon after George H.W. Bush was sworn in as president, his secretary of defense, Dick Cheney, was briefed on the latest version of the nuclear war plan. Cheney asked his assistant on strategic issues, a civilian analyst named Franklin Miller, to sit in. Miller had perused the array of classified documents reciting the rationales for limited nuclear options. Yet, he noticed, the briefing said nothing about such options.

Cheney and Miller were also struck by one detail in the war plan: It called for hitting the Soviet transportation network with 725 nuclear weapons. Cheney asked the briefer, a SAC general, why. The general shrugged and said he’d get back to him on that. (He never did.) After the meeting, Cheney told Miller to go out to Strategic Air Command’s headquarters, in Omaha, and conduct a thorough review of the war plan; he alerted the officers at SAC that Miller should have full authority to look at everything.

What Miller discovered made the term “overkill” seem a gross understatement. For example, just outside Moscow, the Soviets had an ancient anti-ballistic missile system holding 68 interceptors. After the Cold War, U.S. inspectors discovered that the system was completely useless. But the war plan specified that the site had to be destroyed with near-total certainty. SAC intelligence estimated (incorrectly) that each of the Soviet interceptors had a high probability of shooting down an incoming American warhead. So, JSTPS — Omaha’s nuclear targeting agency — assigned 69 warheads to hit the site, to make absolutely certain that at least one of the warheads got through.

Another jaw-dropping example: One part of the nuclear war plan called for destroying the Soviet tank army. As a result, JSTPS aimed a lot of weapons at not only the tanks themselves, but also the factory that produced the tanks, the steel mill that supplied the factory, the ore-processing facility that supplied the steel mill, and the mine that furnished the ore. Miller and his staff learned that some SAC analysts had already pointed out the excesses. A branch of math called nodal analysis suggested that, as long as the central links of a supply system were destroyed, there was no need to destroy every single piece; in many cases, just a few warheads, aimed at the right targets, would cripple the system. Gradually, Miller realized that the entire war plan was like this — a senseless aggregate of compartmentalized calculations.

Then came the key revelation. At this point the Bush administration was negotiating the Strategic Arms Reduction Treaty with the Soviets. During one of his trips, one of Miller’s assistants asked a JSTPS officer whether the treaty’s prospective cuts would affect SAC’s ability to fulfill its mission — whether the U.S. could continue to deter nuclear war and limit damage if deterrence failed. The officer replied that he didn’t do that sort of analysis. JSTPS, he went on, was prohibited from setting requirements or analyzing whether a certain kind of attack, with a certain number of weapons, would be militarily effective. When asked what the JSTPS actually did, the officer explained that they take all the weapons that are assigned to SAC and aim them at all the targets on the list.

The code was unlocked. It turned out that the war plan was based on supply, not demand — on how many weapons SAC happened to have, not on how many were needed.

Cheney ordered Miller and some officers in the Pentagon’s Joint Chiefs of Staff — who were just as staggered by this revelation as the civilians — to go over every single weapon in SAC’s arsenal and every single target on the JSTPS list with an eye toward figuring out how many nukes were really needed, even if the policy didn’t change. This was the first time that any civilians — in fact, any official from Washington — truly scrutinized the nuclear war plan.

In the end, they calculated that, without any changes in the war plan’s broad aims or policies, the arsenal could be cut in half — from roughly 12,000 to 5,888. After the Soviet Union imploded, they further reduced this figure, to around 2,200. (Much of this stemmed from removing targets in Eastern Europe, which now consisted of independent states, not Soviet allies.) As it happened, the Strategic Arms Reduction Talks (START) set 2,200 as the maximum number of weapons that each country could deploy. U.S. negotiators went into the talks knowing that the military chiefs — who ordinarily might protest such deep cuts — would be fine with them; and the chiefs were fine with such deep cuts because of the analysis forced by (of all people) Dick Cheney.

For the first couple of decades of the post-Cold War era, few people thought, much less worried, about nuclear war. Presidents Bill Clinton and George W. Bush each had tangles with North Korea, but they didn’t rise beyond the realm of rhetoric. Clinton and his team devoted time to pulling Boris Yeltsin’s new Russia out of various crises, helping to secure the disheveled country’s loose nukes and forging relations with the newly independent states of Eastern Europe, some of which were eager to join NATO. But tensions between the U.S. and Russia had abated; existing arms control treaties held firm. Remarkably, the size and structure of each side’s nuclear arsenals — and their basic attitude toward nuclear weapons — held firm as well. President Barack Obama did revisit the arsenal, yet even he held back from the bold steps that his analysis suggested were possible.

In 2010, Obama asked his top national-security staffers to conduct an analysis similar to Cheney’s. ¹⁴ He had just signed the New START treaty with Russian President Dmitri Medvedev, which cut the arsenals further to 1,550. Eager to push ahead with still deeper cuts (one of his passionate aims, upon taking office, was to “reduce the role of nuclear weapons”), Obama wanted to know how deep the cuts could safely go.

Every couple of weeks for four months, a group of senior officials from the Pentagon and the National Security Council met with Gen. Robert Kehler, the head of Strategic Command (the new name for what had been SAC, referred to as StratCom), to go over, once again, every single weapon and every single target. They broke down the targets into categories and asked whether StratCom needed to hit every target or just some — and whether fewer weapons would be adequate to hit each target.

The leader of the Washington team, a Pentagon civilian named Jim Miller, was well aware of what Frank Miller (no relation) had done. He understood the practice, dating back to 1960, of pushing up the “required” number of weapons by elevating the certainty with which certain targets be destroyed. Kehler understood this too. If the war plan required that Target X be destroyed with 90 percent certainty, that meant the U.S. would need to launch two warheads against the target. If you’re willing to reduce this probability to 75 percent, Kehler told the staffers, then one warhead would be sufficient. It was up to them, the political decision makers.

Even so, after all the paring down, Miller concluded — and Kehler concurred — that the U.S. could safely eliminate one-third of its nuclear arsenal; in other words, that New START’s limits of 1,550 bombs and warheads could be further cut to 1,000. This too was a compromise. Miller reflected that even 1,000 left in place a lot of overkill; in the event of war, it meant merely that “the rubble would bounce one less time.”

Then Kehler and the Joint Chiefs threw in a caveat: They would not publicly endorse a reduction of this magnitude unless the Russians cut their arsenal by the same amount in a follow-on treaty to New START. Obama agreed. Though he understood that the nation would be perfectly secure with one-third fewer nuclear weapons, regardless of what Moscow did, he saw no point in making unilateral cuts. He wanted to negotiate deeper cuts with the Russians, who might see no need to go along if Washington made deep cuts on its own. He also knew he would get no support for unilateral cuts from Congress. Obama was a visionary, but when it came to policy, he was pragmatic. As one of his aides put it, he liked to “paint within the lines.”

Not only did Obama make no further cuts, he got trapped into appearing to do the opposite. As part of a deal to get the Senate to ratify New START (which required a two-thirds majority), Obama signed a letter pledging that he would request funds to “modernize or replace” all three legs of the Triad (emphasis added). ¹⁵ This was a carefully written note, as “modernize” could mean upgrading a missile’s software or installing new communications gear. Obama did not regard the letter as a promise to buy any new weapons; nor did he attach a dollar figure to the pledge.

However, the Republican hawks in Congress rolled out a list of all-new weapons and calculated that the package would cost $1.3 trillion over the next 30 years. (The estimate has since been revised to $2 trillion.) More than that, after the 2016 election, when Donald Trump came to office, newly appointed officials in the Pentagon, some of whom had worked for Republican senators, referred to this list as “the Obama program of record,” thus ensuring that anyone who proposed a less ambitious plan would be tagged “more dovish than Obama.”

Getting agreement on this required some sleight of hand. Trump’s first secretary of defense, Jim Mattis, was a retired Marine general and therefore had little interest in nuclear weapons. (The Marines retired their short-range nukes decades ago and never built long-range weapons.) In fact, two years earlier, Mattis testified as a private citizen that it might be a good idea to get rid of the land-based ICBMs. ¹⁶ He had come under the influence of arms-control analysts, who argued that such weapons were inherently “destabilizing.” They were at once the most accurate and the most vulnerable weapons; therefore, in a crisis, their very existence could provide an enemy with incentives or excuses to launch a preemptive first strike. Many years earlier, ICBMs had served a specific function: They were the only weapons that could quickly knock out blast-hardened targets, such as ICBM silos. However, since 1990, the Navy’s Trident II submarine-launched ballistic missiles were accurate enough to destroy hardened targets as well. Land-based ICBMs had no purpose whatsoever.

But then someone (I have been unable to trace who) came up with a new rationale: the sponge theory. ¹⁷ The idea was that, if some president dismantled all the land-based ICBMs, there would be only five strategic targets in the continental U.S. — three bomber bases and two submarine ports. The Russians could launch an attack on those five targets with just one or two multiple independently targetable reentry-vehicle missiles — and in a crisis, they might think they could pull it off. The U.S. would be left with only the half dozen or so submarines constantly at sea, and the president might not launch those missiles, knowing that if he or she did, the Russians could strike back with much more. On the other hand, the theorists went on, if the U.S. kept its 400 ICBMs, the Russians would have to fire 800 warheads to destroy them — and that would constitute a “major” attack. Any American president would have to retaliate, and the certainty of that prospect would deter the Russians from attacking in the first place.

This was a bizarre theory for three reasons. First, nuking just those five U.S. bases would kill hundreds of thousands, possibly millions, of American civilians; a president couldn’t be counted on to do nothing. Second, during the Reagan years, some of these hawks argued that the Soviets would launch 2,000 warheads in order to destroy the 1,000 ICBMs that the U.S. had at the time, without blinking an eye, even though such an attack would kill tens of millions of Americans. Finally, allowing that the sponge theory might have some logic, it is extreme to contend that the U.S. needs 400 missiles to keep the Russians at bay; the sponge effect of 100 or even a dozen ICBMs would require the Russians to launch 200 or two dozen warheads respectively — certainly a “major” attack by any measure.

In any case, however many sponges we might need, it is doubtful that a whole new missile is necessary for the task. The present Minuteman missiles have been modified several times over the past 50 years, and they could be modified several times more.

The same is true of the B-52 and B-1 bombers, which can stay aloft and fire air-launched cruise missiles — which have a range of 1,500 miles, making it unnecessary for the airplanes to penetrate Russian airspace, where they might be shot down by air-defense missiles. A brand-new bomber isn’t needed to launch cruise missiles.

New missile-launching submarines will be needed sometime in the future. Submarines are undetectable and, therefore, as long as nuclear weapons exist, they are a stabilizing force in the balance of terror. It is also a good idea to keep improving command-control-communications technology, to make sure that the weapons launch if the president wants to launch them (if just to maintain credible deterrence) and to minimize the chance of an unauthorized launch.

But in Congress, the White House and the civilian-run corridors of the Pentagon, the rationales for revamping all three legs of the Triad — and preserving the present numbers of bombers, missiles and warheads — gained support.

There was a brief time, shortly before and just after the 2020 election, when some legislators, even some conservatives, began to think about putting some clamps on the nuclear juggernaut. COVID-19 and its economic side effects were soaking up hundreds of billions of tax dollars; a movement was afoot to rethink priorities, to wonder whether cuts were possible in other programs, including nuclear programs. But then came the upswing in Putin’s aggressiveness in his decision to invade Ukraine, building on his interference with U.S. elections and his moves on Syria. Tensions with China escalated as well. To argue against building more nuclear weapons might be seen as being “soft on defense.”

At that point, perceptions overtook the slight steps toward objective analysis. This happened frequently during the Cold War. Whenever someone challenged the need to build new nuclear weapons, a general would reply that the U.S. cannot be “perceived” to be behind the enemy, even if objectively it wouldn’t matter. The main argument these days for new nukes — and, to some, for a larger arsenal of nukes — is that they’re needed because the Russians continue to upgrade their arsenal and the Chinese are expanding theirs. But just because the Russians and Chinese are wasting their money on more nuclear weapons than they need doesn’t mean the Americans have to follow suit.

Adm. Charles Richard, the head of Strategic Command, said this past March that the U.S. needs to rethink the whole concept of deterrence because it now needs to deter two “near peer powers” — Russia and China — simultaneously. ¹⁸ This is questionable. First, the U.S. nuclear war plan has always taken into account the possibility of a war with more than one nuclear-armed country. Second, as Richard explicitly noted, China isn’t quite a “peer power”; it has about one-tenth as many nuclear bombs and warheads as the United States or Russia.

China is a “peer power” in the sense that, if attacked with nuclear weapons, it could — even with its much smaller arsenal — retaliate with devastating consequences. But that only bolsters the argument that the U.S. could get by with far fewer weapons than it presently has. Back in the mid-1960s, the Kremlin considered attacking China, but Mao Zedong’s paltry arsenal at the time — about a half dozen A-bombs — deterred Leonid Brezhnev from doing so. A few American presidents in recent decades have thought about attacking North Korea — but the Kim regime’s handful of nukes has kept Washington at bay. The U.S. probably needs more than a few dozen nuclear weapons — but does it need a lot more than that? And if so, for what?

We would be better off mounting a full rebuttal of the vague assertions about “perceptions” (no one argues that some wasteful anti-poverty program should be funded anyway, so that it looks like we’re fighting poverty). It’s time for a return to first principles, the basic questions that nobody has been asking: How much is enough? And enough to do what? Enough material has been declassified over the last few decades that informed citizens can hold hearings and stage debates, even if members of Congress won’t. At long last, let the hearings and debates begin.

*Some of this article is based on research for my two books on nuclear issues, The Wizards of Armageddon (Simon & Schuster, 1983) and The Bomb (Simon & Schuster, 2020).