The Doomsday Machine Summary

Introductory Summary from The Doomsday Machine by Daniel Ellsberg

  • The basic elements of American readiness for nuclear war remain today what they were almost sixty years ago: Thousands of nuclear weapons remain on hair-trigger alert, aimed mainly at Russian military targets including command and control, many in or near cities. The declared official rationale for such a system has always been primarily the supposed need to deter—or if necessary respond to—an aggressive Russian nuclear first strike against the United States. That widely believed public rationale is a deliberate deception. Deterring a surprise Soviet nuclear attack—or responding to such an attack—has never been the only or even the primary purpose of our nuclear plans and preparations. The nature, scale, and posture of our strategic nuclear forces has always been shaped by the requirements of quite different purposes: to attempt to limit the damage to the United States from Soviet or Russian retaliation to a U.S. first strike against the USSR or Russia. This capability is, in particular, intended to strengthen the credibility of U.S. threats to initiate limited nuclear attacks, or escalate them—U.S. threats of “first use”—to prevail in regional, initially non-nuclear conflicts involving Soviet or Russian forces or their allies.


  • The required U.S. strategic capabilities have always been for a first-strike force: not, under any president, for a U.S. surprise attack, unprovoked or “a bolt out of the blue,” but not, either, with an aim of striking “second” under any circumstances, if that can be avoided by preemption. Though officially denied, preemptive “launch on warning” (LOW)—either on tactical warning of an incoming attack or strategic warning that nuclear escalation is probably impending—has always been at the heart of our strategic alert.


  • During the 2016 presidential campaign, Donald J. Trump was reported to have asked a foreign policy advisor, about nuclear weapons, “If we have them, why can’t we use them?” Correct answer: We do. Contrary to the cliché that “no nuclear weapons have been used since Hiroshima and Nagasaki,” U.S. presidents have used our nuclear weapons dozens of times in “crises,” mostly in secret from the American public (though not from adversaries). They have used them in the precise way that a gun is used when it is pointed at someone in a confrontation, whether or not the trigger is pulled. To get one’s way without pulling the trigger is a major purpose for owning the gun. (See chapter 20.)

Moreover, our “extended deterrence” over allies in Europe or Japan rests on our preparedness and our frequently reiterated readiness to carry out threats of first use (initiation of limited nuclear attacks with short-range tactical weapons) and/or, implicitly, to carry out a disarming first strike on the homeland of the USSR or Russia, mostly with long-range strategic weapons, in response to large non-nuclear attacks by its conventional forces or those of its allies.

As candidate in 2016, now President Donald J. Trump repeatedly asserted his unwillingness to take nuclear first-use threats “off the table” in any conflict, including with ISIS, or in Europe. (He also said that he would be “the last to use nuclear weapons”—unless, evidently, he were the first.17) In the first debate of the presidential campaign, he was asked: “On nuclear weapons, President Obama reportedly considered changing the nation’s long-standing policy [i.e., changing it to no-first-use]. Do you support the current policy?”

Given two minutes to answer, Trump said, among other things: “I would like everybody to end it, just get rid of it. But I would certainly not do first strike.b I think that once the nuclear alternative happens, it’s over. At the same time, we have to be prepared. I can’t take anything off the table.”

In her two minutes, Hillary Clinton managed not to repeat Trump’s words about the table, or to respond to the question at all except to “reassure our allies … that we have mutual defense treaties and we will honor them.” But clearly if she had been pressed, the former secretary of state would have given substantially the same answer as Trump did in all his interviews. Our mutual defense treaties have never excluded U.S. first use of nuclear weapons. (As a candidate in 2008, rebuking Senator Barack Obama for saying he would not use nuclear weapons against Pakistan, she said that no president should ever say what weapons he or she would or would not use.)

In the meantime, up through 2016, President Obama, under pressure to reject a no-first-use policy from his secretaries of defense, state, and energy as well as U.S. allies, complied with such advice both in his 2010 Nuclear Posture Review and in his last year in office. He was continuing a policy of threatening possible American initiation of nuclear war that has, outside public awareness, characterized every American administration since Truman’s. Inheriting this policy and reiterating it, President Donald J. Trump continues to apply what Richard Nixon called the “madman theory,” with, as some see it with unease, more plausibility than some of his predecessors.


  • Posing as it does the threat of nuclear attack by the United States to every state that might potentially be in conflict with us (like North Korea), this persistent rejection by the United States of a no-first-use commitment has always precluded an effective nonproliferation campaign. So it does at this time under President Trump. Indeed, it has encouraged proliferation in states hoping either to counter these American threats or to imitate them. But other aspects of U.S. nuclear policy as well have the same outcome, effectively promoting proliferation. Of course, our insistence on maintaining an arsenal of thousands of weapons, many on alert, a quarter century into the post–Cold War era, nullifies our advice to most other states in the world that they “have no need” or justification for producing a single nuclear weapon.


  • With respect to deliberate, authorized U.S. strategic attacks, the system has always been designed to be triggered by a far wider range of events than the public has ever imagined. Moreover, the hand authorized to pull the trigger on U.S. nuclear forces has never been exclusively that of the president, nor even his highest military officials. (See chapters 3 and 7.)

As I discovered in my command and control research in the late fifties, President Eisenhower had secretly delegated authority to initiate nuclear attacks to his theater commanders under various circumstances, including the outage of communications with Washington (a daily occurrence in the Pacific) or a presidential incapacitation (which Eisenhower suffered twice). And with his authorization, they had in turn delegated this initiative, under comparable crisis conditions, to subordinate commanders.

To my surprise, after I had alerted the Kennedy White House to this policy and its dangers, President Kennedy continued it (rather than reverse the decision of the “great commander” who had preceded him). So did Presidents Johnson, Nixon, and Carter. So, almost certainly, has every subsequent president to this day, even though in the past several decades there may have been at least nominal “devolution” to some civilian outside Washington. This delegation has been one of our highest national secrets.

The same was true for the Soviet Union, now Russia. Public discussion of American plans for “decapitation” of Soviet command and control led to the institution and maintenance of a “Dead Hand”19 system of delegation that would assure retaliation to an American attack that destroyed Moscow and other command centers. This, too, has been treated as a state secret: paradoxically, since on both sides the secrecy and denial diminish deterrence of a decapitating attack against it (see chapter 9).

An urgent reason for enlightening the world’s public on this reality of the nuclear era is that it is virtually certain that this same secret delegation exists in every nuclear state, including the new ones: Israel, India, Pakistan, and North Korea. How many fingers are on Pakistani nuclear buttons? Probably not even the president of Pakistan knows reliably. Meanwhile, frequent leaked reports in the American press throughout 2016 and 2017 of U.S. contingency plans and exercises aimed in crucial part at decapitating North Korean leadership and command structure have, in my opinion, very probably had the effect in that country of creating a Soviet-like Dead Hand system for assuring retaliation to such an attack.


  • Thanks to revelations from the former Soviet Union, there has been growing appreciation of the extreme dangers posed by the Cuban missile crisis. Yet my own highly classified study in 1964—following my high-level staff participation in the crisis—unearthed never-before-revealed details that, together with the new data, demonstrate that the risks were even higher than any previous account has concluded. Despite what I believe was the determination of both leaders to avoid nuclear war, events spiraled out of control, coming within a handbreadth of triggering our plans for general nuclear war. (See chapters 12 and 13.)


  • The strategic nuclear system is more prone to false alarms, accidents, and unauthorized launches than the public (and even most high officials) has ever been aware. This was my special focus of classified investigation in 1958–61. Later studies have confirmed22 the persistence of these risks, with particularly serious false alarms in 1979, 1980, 1983,23 and 1995. The chance that this system could explode “by mistake” or unauthorized action in a crisis—as well as by the deliberate execution of nuclear threats—taking much of the world with it, has always been an unconscionable risk imposed by the superpowers upon the population of the world.


  • Potentially catastrophic dangers such as these have been systematically concealed from the public. In 1961 I had learned as an insider that our secret nuclear decision-making, policy, plans, and practices for general nuclear war endangered, by the JCS estimate, hundreds of millions of people, perhaps a third of the earth’s population. What none of us knew at that time—not the Joint Chiefs, not the president or his science advisors, not anyone else for the next two decades, until 1983—were the phenomena of nuclear winter24 and nuclear famine, which meant that a large nuclear war of the kind we prepared for then or later would kill nearly every human on earth (along with most other large species). (See chapter 18.)

It is the smoke, after all (not the fallout, which would remain mostly limited to the northern hemisphere), that would do it worldwide: smoke and soot lofted by fierce firestorms in hundreds of burning cities into the stratosphere, where it would not rain out and would remain for a decade or more, enveloping the globe and blocking most sunlight, lowering annual global temperatures to the level of the last Ice Age, and killing all harvests worldwide, causing near-universal starvation within a year or two.

U.S. plans for thermonuclear war in the early sixties, if carried out in the Berlin or Cuban missile crises, would have killed many times more than the six hundred million people predicted by the JCS. They would have caused nuclear winter that would have starved to death nearly everyone then living: at that time three billion.

The numbers of warheads on both sides have since declined greatly—by over 80 percent!—from their highest levels in the sixties. Yet by the most recent scientific calculations—confirming and even strengthening the initial warnings of more than thirty years ago—even a fraction of the existing smaller arsenals would be more than enough to cause nuclear winter today, on the basis of existing plans that target command and control centers and other objectives in or near cities. In other words, first-strike nuclear attacks by either side very much smaller than were planned in the sixties and seventies—and which are still prepared for instant execution in both Russia and America—would still kill by loss of sunlight and resulting starvation nearly all the humans on earth, now over seven billion.

There would be no limiting of damage to the superpower attacker—or to its allies, or the “enemy” population or that of neutrals throughout the globe—by its superpower adversary striking first rather than second (even without suffering retaliation), or by its preemption, “counterforce,” or “decapitation” attacks, in short by any of the missions the great bulk of its weapons are specifically designed and intended to do. Damage to itself, and to everyone else, from its own first strike would be total, unlimited.

There is no sign that the findings of the latest scientific peer-reviewed studies of climatic consequences of nuclear war over the past decade have penetrated the consciousness of U.S. officials or Russian officials or have influenced in any way their nuclear deployments or arms-control negotiations.

There is good reason to doubt that either George W. Bush or Barack Obama—or, for that matter, George H. W. Bush or Bill Clinton in the previous twenty years since the original studies—was ever, once, briefed on the scale of this result of the large “options” he was presented with in nuclear command exercises. (Gorbachev has reported that he was strongly influenced by Soviet studies of this phenomenon, which underlay his desire to seek massive reductions and even the elimination of nuclear weapons in his discussions with Reagan, who made a similar attribution.

Whether or not President Donald Trump has been briefed on this (almost surely not), both he and several of his cabinet officials, along with leaders of the Republican majority in Congress, are famous deniers of the scientific authority of such findings, based as they are on the most advanced climate models.

*   * *

At the conclusion of his famous satirical film of 1964, Dr. Strangelove, Stanley Kubrick introduced the concept of a “Doomsday Machine”—designed to deter nuclear attack on the Soviet Union by destroying all human life as an automatized response to such an attack. His Russian leader had fatefully installed the system before he had revealed it to the world, and it was now subject to being triggered by a single nuclear explosion from an American B-52 sent off by a rogue commander without presidential authorization.

Kubrick had borrowed the name and the very concept of such a hypothetical machine from my former colleague Herman Kahn, a RAND physicist with whom he had discussed it. In his 1960 book On Thermonuclear War and in popular articles in 1961, Kahn had said he was sure he could design such a device. It could be produced within ten years and would be relatively cheap, one of its main attractions as a deterrent system. It would cost closer to ten than to a hundred billion dollars, he guessed—only a fraction of the current budget for strategic weapons—since it could be emplaced in one’s own country or in the ocean. It would not depend on sending warheads halfway around the world by expensive planes and missiles that would have to penetrate enemy defenses.

But, he said, it was obviously undesirable. It would be too uncontrollable—too inflexible and automatic—and it might fail to deter, and its failure “kills too many people”: in fact, everyone, a result that the philosopher John Somerville later termed “omnicide.” Kahn was sure in 1961 that no such system had been built, nor would it be, by either the United States or the Soviet Union.

The physicist Edward Teller, known as the “father of the H-bomb,” went further to deny that omnicide—a concept he derided—was remotely feasible. In answer to a question I posed to him as late as 1982, he said emphatically it was “impossible” to kill by any imaginable use of thermonuclear weapons that he had co-invented “more than a quarter of the earth’s population.”

At the time, I thought of this assurance, ironically, as his perception of “the glass being three-quarters full.” (Teller was, along with Kahn, Henry Kissinger, and the former Nazi missile designer Wernher von Braun, one of Kubrick’s inspirations for the character of Dr. Strangelove.) And Teller’s estimate was closely in line with what the JCS actually planned to do in 1961, though a better estimate (allowing for the direct effects of fire, which JSC calculations have always omitted) would have been closer to one-third to one-half of total omnicide.

But the JCS were mistaken in 1961, and so was Herman Kahn in 1960, and so was Teller in 1982. Nobody’s perfect. Just one year after Teller had made this negative assertion (at a hearing of the California state legislature which we both addressed, on the Bilateral Nuclear Weapons Freeze Initiative), the first papers appeared on the nuclear-winter effects of smoke injected into the stratosphere by firestorms generated by a thousand or more of the fifty thousand existing H-bombs used on cities. Contrary to Kahn and Teller, an American Doomsday Machine already existed in 19