The Doomsday Machine

by Daniel Ellsberg

Yes, but the whole point of the doomsday machine is lost if you keep it a secret! Why didn’t you tell the world, eh?

  —Dr. Strangelove

  When Daniel Ford, the former executive director of the Union of Concerned Scientists, published his well-researched book The Button in 1985, he could get no official comments on the question implied by his title: “How many fingers are on the nuclear button?” Did official pre-delegation exist—as logic suggested and various people had hinted or speculated—or did it not? From Donald Latham, assistant secretary of defense under Reagan for command and control systems, he received this answer: “Well, there are contingency plans, I just really can’t discuss them.” That, he noted, was about all the Pentagon would say about the subject. Ford then quotes Desmond Ball, the extraordinarily well-informed Australian defense analyst, as commenting to him, “This is probably one of the most closely kept secrets.”236

  Indeed. Going back to 1960, a quarter century before Ford’s investigation, the answer to his query was given to me as a highly sensitive secret—perhaps the most highly guarded secret in the American military system—in confidence that I would not reveal it to the U.S. public or the world, thereby contradicting a decade of explicit denials by the highest American authorities that any such pre-delegation existed. That confidence in my discretion was justified at that time.

  But there was, and still is, a stunning paradox here. Why has it been kept secret at all, especially from our adversaries?

  After all, the most compelling and legitimate purpose of delegation by the president has always been to assure that the Soviets (or now, Russians) could not paralyze our retaliatory forces by a “decapitating” attack on Washington, D.C., or by attacking the president wherever he might be. But even more important than establishing that reality is to make sure that the adversary also understands, recognizes, and believes that reality beyond any doubt. Otherwise, in a crisis or faced with a (possibly false) warning of a U.S. attack, any enemy uncertainty about presidential delegation could nourish hope that its best chance—perhaps its only chance—of survival was precisely to launch a decapitating attack against the U.S. capital and our major known command posts. To deter such a reckless action, surely nothing could be more important than convincing our adversary that such hopes were futile, that destroying our leadership would not prevent or even reduce the devastation they should expect. Secrecy about this, denial of it, refusal to confirm rumors about it, could have only the opposite effect.

  Of course, there was no way that the Soviets would have been certain that such delegation existed. Even presidential statements that it did exist could be disbelieved. But a declaratory policy that it did not exist (as the American public was repeatedly told) and that only the president himself, or a successor brought into possession of his “football,” could launch nuclear strikes, would only strengthen the hopes of Soviet planners that there was, after all, a form of preemptive attack that could allow them to survive and even “prevail.” A decapitating attack, in the absence of any U.S. delegation, might actually paralyze U.S. retaliation, or at least cause a significant delay. After all, that was the logic of our own secret military planning, with its emphasis on Moscow as the highest-priority target.

  Such Soviet planning was not just conjectural. As Ford put it, “Soviet strategists have written extensively about the need to bring the ‘disorganization of [the enemy’s] state and military command control,’ ” particularly for strategic weapons. He quotes a Soviet article describing this objective in detail in 1966, a time when the Soviets still had too few missiles to target all our Minuteman missiles, and yet were deploying SS-9 missiles with twenty-megaton warheads,237 clearly intended for the one hundred highly hardened Minuteman control centers. With such an approach, it was certain that they would also target the high civil-military command in the Pentagon and D.C. area.

  Secrecy about any U.S. delegation not only encouraged such planning but also could stimulate desperate hopes in the midst of a crisis that it was actually better to execute these plans than to await a possible U.S. attack on their own command system. In other words, such secrecy lowered deterrence of a Soviet decapitating strike in a crisis.

  But the situation became much worse than that under Carter and Reagan. The longstanding JCS and SAC desire to attack Moscow and the whole Soviet command and control system had remained a tightly guarded secret from the era of Eisenhower through Ford. But as early as 1977, and especially in 1978–80, there were frequent leaks and official statements that the special focus of strategic nuclear planning under President Carter—pressed by his national security assistant Zbigniew Brzezinski—was decapitation238 of the Soviet command system. The Reagan administration continued239 both the emphasis and the openness about it. In other words, secrecy about our plans to nullify Soviet decapitating attacks by delegation (which have persisted, up to the present) was now joined by publicity about our intentions to decapitate the Soviets. In fact, it was in the late Carter years that the term “decapitation” first acquired public currency as an official objective.

  Leon Sloss, a Pentagon official, was given the task at the beginning of the Carter administration in 1977 of updating the guidance on nuclear operations. He told me years later that “The first thing I did” was to pull out of a Top Secret file-safe in the Pentagon my old 1961 draft guidance to planning for general nuclear war, finished on my thirtieth birthday. He had remembered it; he said it was the starting point for his own work. If so, he soon departed from it radically. According to that earlier guidance, an essential part of a coercive strategy, aimed at ending the war short of total annihilation on both sides, was to withhold attacks on the opponent’s command structure. But as a consequence of this review, as Sloss wrote later, “increased emphasis in U.S. nuclear [weapons] employment policy240 was given to the targeting of enemy military forces and political-military leadership … the Soviet command structure.”

  Ford quotes General Bruce K. Holloway, the former commander in chief of SAC, as writing in 1980 that U.S. war aims included “prevention of the loss of our way of life,” “damage limitation,” and the “degradation of the Soviet State and its control apparatus to such an extent as to make successful negotiation possible.” In achieving these objectives, “the importance of crippling the [Soviet] command241 and the control system … assumes extraordinary proportions.”

  Nothing could so decisively preclude “successful negotiations” than to destroy at the outset the opposing command authorities. With whom would these “negotiations” be carried out? What ability would we have left them to control their operations, implement any “deal,” or terminate their own attacks? Those were questions I had raised in 1961. Their logic was probably never accepted by SAC for a moment; certainly not by General Holloway, whose memo in 1980 Ford quotes further: “Degradation of the overall political and military control apparatus must be the primary targeting objective. Irrespective of whether we strike first or respond to a Soviet strike (presumably counterforce), it assumes the importance of absolute priority planning. Striking first would offer a tremendous advantage, and would emphasize degrading the highest political and military control to the greatest possible degree.”

  Obviously, the success of this would depend on the Soviets’ eschewing any delegation—unlike ourselves—that would assure a devastating response to a U.S. attack on their high command. Holloway indicated explicitly his confidence that the Soviets would be more conservative in this sense than the U.S. “I am convinced that in the Soviet system242 there is such centralized control that it would be possible to degrade very seriously their military effectiveness for nuclear or any other kind of war if the command control system were severely disrupted. Major damage would be difficult to achieve and would require better intelligence than now possible (better reconnaissance and better clandestine inputs) but it can be done. Moreover, it must be done, because there is no other targeting strategy that can achieve the war aims that underwrite survival.”

  In other
words, the compelling incentive—in the eyes of the former CINCSAC, and certainly not only for him—to “decapitate” the Soviet system at the outset of hostilities reflects the hope that it might paralyze Soviet forces to the extent that the U.S. could survive a nuclear war, and the belief that no other approach could do this. Every other strategy is seen (realistically) as a “no-win” option or, more seriously, a “no-survival” approach. Nevertheless, in planning to deal with a desperate situation (and in supporting Air Force and Navy budgeting for lots of accurate missiles for an illusory goal of winning or surviving a thermonuclear war), the supposed possibility of averting “otherwise-certain” annihilation, no matter how slim that chance, has an attraction that tends to be irresistible.

  This would not have been a big surprise to the Soviets. They undoubtedly took it for granted in our planning. McNamara had announced publicly the possibility of withholding an attack on Moscow in his Ann Arbor commencement speech of 1962, but that was, after all, only an option. Soviet military planners probably regarded it with a good deal of incredulity. Thus, at the same time in the late sixties as they were preparing for SS-9 attacks on Minuteman control centers, they were building two thousand underground bunkers243 for Soviet military officials and Communist Party leaders (more than one hundred thousand of them) with seventy-five relocation centers in Moscow, some of them several hundred feet deep. (Since we had not been as thorough in multiplying underground centers, there was indeed what General Buck Turgidson had feared in Dr. Strangelove, “a mine-shaft gap.”)

  But the new publicity a decade later about a renewed U.S. preoccupation with destroying these very shelters had to undermine whatever confidence Soviet leaders had in their own survival, even for long enough to order retaliation. Indeed, that was proclaimed as the very purpose of our plans and emerging military capabilities: large numbers of ICBMs each with multiple independently targetable reentry vehicles (MIRVs), allowing for a number for warheads with separate targets on each missile, with larger yield and greatly increased accuracy, for our Minuteman and sub-launched missiles (SLBMs) precisely to attack the large number of Soviet underground command posts as well as the increasing number of hardened missile sites. Moreover, the increasing capability of multiple-target,244 high-yield, and very accurate Trident submarine missiles meant that attacks on most of the Soviet command structure, as well as its missiles, could be launched so close to the Soviet Union as to give no warning, or almost none, to either the high command or the missile launch centers.

  Indeed, a major rationale then and now245 (see below) for buying and deploying hosts of such warheads capable of destroying super-hardened underground command centers has been to deter Soviet (now Russian) leaders from contemplating a first strike under any circumstances by assuring them that they themselves would not survive the initial exchange, whether we struck first preemptively or not. Given these capabilities, the chance that high-level Soviet officials could actually get to their bunkers in a low-warning attack, or that the shelters would survive our attack, was extremely low. This evident fact, combined with the deliberate publicity given the Carter and Reagan administrations’ effort to achieve decapitation, could only give a desperate sense of urgency to Soviet leaders and planners to maintain a deterrent retaliatory capability. The only ways to do that were what we ourselves had done in the era of the presumed Soviet missile superiority. They would either have to delegate authority for launch to lower commanders, and/or plan for launch on warning of attack (LOW) by high commanders: if not by computers, as some military commanders such as NORAD chief General Lawrence Kuter preferred. As Herbert York recounts:

  General Kuter told me that we had to complete the BMEWS246 (Ballistic Missile Early Warning System) as soon as possible, and he urged that we expand it in order to create a highly redundant capability at each site. We must expand it in order to have an absolutely reliable early warning of a missile attack. Basically, I agreed.

  All would have been well if he had stopped there, but he didn’t. In words I can’t precisely recall, he went on to say that we had to have this redundancy and the resulting high level of reliability so that, when we finally connected the warning system directly to the launch button of our own ICBMs, there would be no false alarms.

  I was astonished. I told him flatly that we would not automate our response, that we would not connect the warning system directly to the launch button. We would not, in sum, go to a “launch on warning” strategy. [York was mistaken in this prediction.] We would, especially, not go to one that did not have the president in the decision-making loop.

  Kuter coldly replied, “In that case, we might as well surrender now.”

  When I first read of the new emphasis on decapitation during the Carter administration, I worried that this publicity could only press the Soviets into a combination of launch on warning and delegation. Strategists like Holloway were gambling that even this pressure wouldn’t lead the Bolsheviks to give up their commitment to centralized control. They were mistaken. When the Reagan administration not only continued but also reinforced this public focus, the Soviets, as I expected, worked urgently on means to counteract it. And just like the Americans, just as unaccountably from the point of view of deterrence, they kept these efforts effectively secret.

  With the ending of the Cold War and the new possibility of open interaction between American and Soviet planners and strategic analysts, Bruce Blair—a former Minuteman control officer who subsequently became an expert in command and control issues—discovered and reported that the Soviets had responded to the threat of decapitation by designing an elaborate system to assure retaliation to an American attack that destroyed Moscow headquarters. Its code name was Perimeter; it was known informally by a Russian phrase that translates as “Dead Hand.” Low-level officers in deep underground centers well away from Moscow would receive by a variety of channels several forms of evidence—seismic, electronic, infrared, radioactivity—that Moscow had suffered a nuclear explosion, along with breakage in all forms of communication to and from Moscow. In that event, they were authorized to send off ICBMs that would beep a Go signal to any ICBM sites they passed over. The Soviet rockets would not merely communicate an authorization to launch to ground officers but would actually bypass them and launch the missiles.

  In an early design of the system, the signals from Moscow would launch the emergency rockets automatically, with no need or allowance for judgment or intervention by humans at the dispersed sites. In other words, this was, at last, the total embodiment of the Doomsday Machine, the device that Herman Kahn had speculatively imagined in On Thermonuclear War, whose destructive effects would be so total as to provide the ultimate deterrent while its credibility would be achieved by automaticity. There is some disagreement about whether the Soviet system was to be in continuous operation or only activated during times of crisis, when the possibility of attack seemed higher than usual, but modernized after the end of the Cold War and up to the present.

  Whereas Kahn’s Doomsday Machine had allowed that an automatic mechanism would be triggered, perhaps, by several near-simultaneous explosions on different cities, it appears that the Perimeter system would be activated by an attack on Moscow alone. That meant that nuclear winter, which was just coming to be understood as this system went into effect, could have been made inevitable by a single explosion on Moscow.

  Stanley Kubrick presents this situation in Dr. Strangelove when the Russian leader informs the U.S. president that if the one American B-52 still on its way to a Soviet target (as the result of an unauthorized action by a squadron commander) is not successfully recalled—which in the film, as in real-life SAC operational planning, neither the president nor anyone else could do once the planes had been ordered to expend—its bombload will trigger an automatic Doomsday Machine in the Soviet Union that will destroy all life on earth. The precise reason the Russian leader gives for having wired up this system is to assure that an attack by the United States would be self-destructive even if
it successfully destroyed Soviet command posts. Dr. Strangelove points out to the Soviet ambassador—present in the war room as a translator and conveyer of the information about the Doomsday Machine—that for purposes of deterrence it would be essential for the United States to know this in advance.

  “But the … whole point of the doomsday machine … is lost … if you keep it a secret! Why didn’t you tell the world, eh?”

  The ambassador answers: “It was to be announced at the Party Congress on Monday. As you know, the premier loves surprises.”247

  Satire, of course. But when the Soviets installed Perimeter, they had no intention of announcing it, ever. And they never did, while the USSR existed. (The Russians have now acknowledged that they still maintain it, describing it as a doomsday machine: see below.)

  The designer of the secret Soviet Perimeter system,248 Valery Yarynich, regarded his system until his death in December 2012 (after decades of consultation on arms control in the United States) as safer than its alternative, which was to rely on launch on warning by high officials in Moscow only. His approach still allowed for launch on warning—as the Russian system, like ours, continues to do—in order to prevent Soviet missiles from being destroyed before they were launched and to allow them to preempt U.S. missiles promptly that had not yet been launched. But the Perimeter system was meant to remove what might present an additional pressure on Soviet commanders in Moscow to launch on warning if the warning signals seemed correct: that there might otherwise be no Soviet retaliation to an American attack because the commanders themselves were about to be destroyed.

  However, as David Hoffman,249 former Moscow bureau chief for the Washington Post who had interviewed Yarynich many times for his book The Dead Hand, reported in a tribute to him on his death:

  In later years, Yarynich expressed grave doubts about the very systems of annihilation he had devoted his career to perfecting. He once told me it was utter stupidity to keep the Dead Hand secret; such a retaliatory system was useful as a deterrent only if your adversary knew about it. More broadly, he came to doubt the wisdom of maintaining the cocked-pistols approach to nuclear deterrence, the so-called hair-trigger alert, especially after the Cold War ended. He feared it could lead to an accidental or mistaken launch. Yarynich did not keep quiet. He decided to share his insights and worries with the world.