The Doomsday Machine

by Daniel Ellsberg

  In this context, its prospects didn’t have to look good, or even “tolerable,” and they certainly didn’t to me; they looked very probably catastrophic. To be acceptable for operational planning, as I was proposing, they had only to look less terrible than any available alternatives, including the previously existing plan. The strategy seemed to offer the possibility of avoiding catastrophes that would be even worse and more certain.

  This was not a reassuring message to present to the public. Understandably, no attempt had ever been made to present it to them officially. But in 1962, it was going to have an even worse ring to it. In the context of a classified NATO audience at Athens, and in the light of the great imbalance of forces we had learned about in September 1961, McNamara was now describing this planning as a U.S. first-strike strategy, in fulfillment of our long-term commitment to NATO in response to a Soviet attack on Western Europe.

  A high-level NATO audience was used to hearing—you might say they demanded—reiterations of our intent to attack the Soviet Union in that case (though they’d never heard in such detail just how we planned to do it). But it had never been forced on the attention of the American public that a large Soviet non-nuclear attack on Europe—not on the continental United States—would almost automatically trigger a full-scale U.S. nuclear attack on the Soviet Union, with the certainty of Soviet retaliation on the United States to the full limits of its capability.

  Moreover, the U.S. public had never been given any real hint as to how limited that Soviet capability was in 1961–62, with respect to the U.S. homeland. Although the Kennedy administration had acknowledged in late 1961 that “there was no missile gap,” and the Gilpatric speech (with my input) had even implied that we were significantly superior to the Soviets in strategic nuclear power, the public had never been told either officially or unofficially just how small the Soviet ICBM force was in those years. In fact, the real terms of that disparity have never entered public consciousness to this day. A scholar as authoritative as Richard Rhodes97 was still writing in 1995 that the Soviets had over forty ICBMs in 1961, ten times more than they actually had.

  In Athens, McNamara had the intent of reassuring our military allies, on a highly classified basis, that we had a way of responding to a Soviet invasion of their countries that gave us enough assurance of surviving the war that we would indeed be willing, as we always said, to carry out our commitment of a nuclear first strike in that case. And moreover, the alliance should rely on the United States’ way of doing this, rather than encouraging the growth of independent (French) forces that would only screw up the strategy and make it infeasible by hitting Soviet cities and central command and control at the outset.

  However unlikely to work the plan probably seemed to our allies, McNamara’s assured tone in describing it—and the fact that the United States was investing billions to implement it—may also have convinced some of them that McNamara actually believed in it and would carry it out in the event. Or at least, that would be the Soviet impression, and that might well scare the Soviets enough to keep them from encroaching on Western Europe. (McNamara and Kaufmann, I thought, were surely mistaken if they thought its logic was so compelling as to dissuade the French from pursuing their force de frappe, which it certainly didn’t do.)

  But these potential benefits, however speculative, didn’t offer themselves at all to unveiling this strategy to the American public, especially in a first-strike context. The language of the Athens speech and Yarmolinsky’s draft version seemed strongly to suggest that the American government put confidence in the results of a coercive strategy in a nuclear war—avoiding Soviet cities while threatening them with reserve forces as we attacked Soviet military forces. Any such confidence was bound to look bizarre, absurd.

  I learned later that Bill Kaufmann had had exactly the same reactions as I did to the idea of presenting the substance of his classified Athens speech to the American public and the world in a public address. Yarmolinsky had asked him to do the job of declassifying his speech for this event, and he refused. He didn’t believe that should be done, for all the reasons I felt. So Yarmolinsky did the job on his own.

  After reading Yarmolinsky’s draft, I handed it back to him and told him as firmly as I could that I thought this speech must not be given. McNamara would have to find another subject for his commencement address. As I was saying this in Adam’s office, he got a call on the direct line from the secretary of defense. He said, “Yes, Bob. Well, I have Dan Ellsberg standing here with me just now, and he’s read my draft and doesn’t like it. He feels strongly that it shouldn’t be given.”

  I could hear McNamara’s voice on the other end but couldn’t make out what he saying. I was feeling, I remember, a small glow that Adam would mention my name to McNamara, as an authority, without having to remind him who I was. (I’d been out of Washington for six months, and I’d met McNamara directly only once, half a year before that.) Adam said, “OK, Bob,” and hung up. He said, “Bob says you should write it the way you think it should be.”

  Rats. That was not a job I wanted, especially after I’d spent one all-nighter commenting on the JSCP. But there was no question of turning it down. This was the first time I’d ever gotten a request directly from McNamara. The problem was that I didn’t think anything like this speech should be given to the public, and Adam had already made clear to me that McNamara wanted something along the lines of the Athens speech, and that he specifically wanted parts including the crack at the French that I was particularly doubtful about.

  Adam found me a desk in his suite of offices and I got to work. As I again compared Kaufmann’s speech with Adam’s much-revised draft, it again seemed to me that Bill’s version was better worded and its logic followed more clearly. I made another copy of each and started out replacing some of Adam’s paragraphs with Bill’s. These were the days long before personal computers, or even Selectric typewriters with automatic erasure and correction. Secretaries made corrections on good copies by using white-out fluid and typing over it. I took a scissors and cut out the parts of Bill’s draft that seemed well said and taped them where I wanted into Adam’s version, cutting that apart for insertions. I rearranged Adam’s sections and I wrote out sentences and paragraphs of transitions or exposition where I thought it was now needed.

  What I thought I was doing was a kind of mock-up that I thought presented the argument that McNamara wanted better than Adam had done. If McNamara bought it, we then had a month before the speech was due to be presented. There would be time to do a really decent job—if I couldn’t persuade them after all that this was a bad idea altogether.

  I had told Yarmolinsky (not in detail) the nature of my reservations to publicizing the Athens speech. But McNamara hadn’t heard them, and that’s not what he had asked me for. I didn’t think I had time both to write those out for him and to draft from scratch an entirely different speech that I thought would be more suitable. So I made the choice—in retrospect, a bad one—of simply editing Adam’s draft, partly with the aid of Bill’s original.

  In the end, what I had was a speech that I thought read better than Adam’s but was in fact closer to Bill’s original Athens speech than Adam’s had been. It didn’t at all cure my most fundamental objections to the tone and substance. (Maybe my lack of sleep was taking its toll on my critical sense.) I had taken out parts of the original that I thought were, as I would have put it later, too Strangelovian. Not nearly enough, it turned out.

  Still, I did cut out McNamara’s presentation of the results of studies of a hypothetical nuclear war in 1966, four years away, in which he contrasted two possible courses of events. Where both sides confined their attacks to military targets, the United States might suffer 25 million deaths, the Soviets the same, and Europe somewhat fewer. But if both sides attacked urban-industrial targets as well, the United States might incur 75 million deaths, the Soviets at least 100 million, and Europe 115 million. He said, “While both sets of figures make grim reading, the fi
rst set is preferable to the second.”

  His argument that the U.S. strategy, under centralized (U.S.) control, had the best, or only, chance of attaining the first set of outcomes rather than the second set in a future nuclear war was what underlay his prior assertion that “in our best judgment, destroying enemy forces while preserving our own societies is … a not wholly unattainable objective.”

  This last phrase, along with two sets of figures he presented, would not seem to be greatly overselling the prospects of preserving societies by this new strategy (even though it really was essentially mistaken, as we’ll see). In fact, reading these estimates (since declassified) fifty-five years later, it is hard to imagine how this presentation could have been in any way reassuring to an audience of hardened military professionals in Athens. It was clearly not suitable for the graduating class at the University of Michigan to hear, or for any other audience that had not become inured to RAND or JCS classified studies. I also omitted his too-revealing reference to Soviet cities that had not yet been hit as “hostages.”

  In the morning, I handed the cut-and-paste version with my interpolations to one of the office secretaries, who were all capable of retyping drafts at lightning speed. I gave the typed version to Adam, who said, “This is good,” and sent it to McNamara. A little later that morning—I hadn’t yet gone back to my hotel to sleep—Adam told me, “McNamara says it’s fine. He’s going to go with it.”

  I said with real concern, “Wait a minute! That’s just a rough overnight draft! We’ve got four weeks before it’s given, plenty of time to improve it.”

  Adam said, “No, he’s done with it. He’s satisfied. He doesn’t change his mind about these things. That’s the way it’s going to go.” I was unhappy but shrugged it off. I went back to my hotel to go to sleep.

  In July, it was a disaster. The French, of course, were furious that McNamara’s contemptuous dismissal of their force was aired publicly. The editorial and American public’s response: about as bad as I’d foreseen. Khrushchev’s reaction I learned only half a century later, from Aleksandr Fursenko and Timothy Naftali’s Khrushchev’s Cold War. Khrushchev had decided in May to secretly send missiles to Cuba, in part in reaction to [my] Gilpatric speech. According to Fursenko and Naftali, Soviet intelligence had apparently missed the Athens speech (surprisingly, as we generally thought that NATO was penetrated enough by the Soviets that a speech to the NATO Council was a secret channel to Moscow). But when Khrushchev read about the Ann Arbor speech,98

  what McNamara said irritated the Soviet leader99 because the secretary of defense explained that in the future NATO should consider targeting Soviet military installations instead of cities. The U.S. government was making this argument because it wanted to discourage the French, the British, and the West Germans from building their own nuclear forces, which were inefficient and hard to control and bred Soviet concerns. Only the U.S. force was technologically sophisticated enough to hit the Soviet missile silos. But what Khrushchev heard was that McNamara was somehow trying to make nuclear war seem less bloody and therefore more acceptable. Minutes after outlining a new Berlin offensive, Khrushchev railed against McNamara at the July 1 [Presidium] meeting: “Not targeting cities—how aggressive! What is their aim?” he asked. Answering his own question, as he often liked to do, Khrushchev replied, “To get the population used to the idea that nuclear war will take place.”

  Ten days later, Khrushchev attacked100 the Ann Arbor speech publicly as seeking “to legalize nuclear warfare and thereby the death of millions and millions of people.” He also said it was deceptive to the American people because bases in the United States were in or near large cities. “It will be first of all the civilian population that will fall victim to the weapons of mass annihilation.”

  Khrushchev was right. To underscore that point, just three months later when SAC planes were put on high alert in an ensuing crisis (see next chapter), many of these nuclear-loaded planes were deployed to civilian airports near major cities, making these cities high-priority targets; the same happened again in October–November 1969 under President Nixon. Meanwhile, of course, the French force de frappe did go ahead, with Moscow as its principal and immediate target, negating any possibility of a no-cities, controlled, “coercive” central war strategy. (In reality, the same remained true of SAC operational planning as well: see below).

  Nevertheless, Khrushchev was bound to hear the McNamara Ann Arbor speech, like the Gilpatric speech before it, as a first-strike threat to him (the flip side of the first-strike assurance McNamara had been giving, in secret, to the NATO allies in Athens). The new speech, which I’d helped craft, must have confirmed him in his clever, reckless response to the one I’d contributed to earlier. By the time of the Ann Arbor speech in July, the Soviet medium-range ballistic missiles that were meant, among other aims, to counter American assertions of strategic superiority and warnings of possible U.S. first use or first strike over Berlin were already on their way to the Caribbean.

  CHAPTER 12

  My Cuban Missile Crisis

  On Monday, October 22, 1962, along with most people in America, I watched President Kennedy on television announcing that the Soviets were installing “offensive” ballistic missiles in Cuba, preparing a capability to attack the United States. He said we would blockade Cuba—he called it a “quarantine”—starting Wednesday morning. Any launch of a single missile from Cuba “against any nation in the Western Hemisphere” would lead to “a full retaliatory response upon the Soviet Union.”

  That last bit sounded wild. “Full response?” That meant the SIOP—the plan for general nuclear war. I was caught by that, having written the guidance for that plan eighteen months earlier. Use it essentially as a first strike against the Soviet Union in the event someone—Cubans?—launched one missile against anyone? I wondered if the speechwriter had any idea what he was saying.

  I went to the phone—I was at home in Malibu, California—and called Harry Rowen in the Pentagon. I asked him, “Could you use some help there?”

  He said, “Why don’t you come on over here—tomorrow.” I made a reservation for early the next morning and packed a bag.

  When I got to his office late Tuesday afternoon, Harry read me into the picture quickly. The group of principals called the ExComm, for “Executive Committee of the National Security Council,” had been meeting with the president, and sometimes without him, several times a day for the past week, deciding what to do. Three or four working groups of staffers were supporting them. One, centered in the Pentagon, was coordinating plans for an air attack and invasion, probably a week away as I arrived.

  Harry said to me, “Write a memo on what thirty-eight missiles could do to our strike-back ability.” He gave me a map with the ranges of the medium-range ballistic missiles (MRBM) and intermediate-range ballistic missiles (IRBM), shown as circles on it. Both Washington and Omaha were within reach of the MRBMs, some of which were already operational. My first thought was that that meant the command posts in the Washington area and at Offutt Air Force Base in Omaha, SAC headquarters, could be hit with very short warning time: minutes—essentially no warning. That was really the most significant effect. It meant the Soviets could be confident of decapitation. But I knew what most didn’t, even in the Pentagon: that wouldn’t spare them from a full, quick retaliation from our massive surviving forces, thanks to delegation.

  This ability to conduct a land-based no-warning attack on our command centers was not an insignificant effect. But it was nothing new; they could have accomplished this with cruise missiles from submarines. Therefore, we had never counted on protecting Washington or Offutt anyway. That was why the Pentagon had designed a system of alternate command posts, including at sea and airborne as well as underground, and why Eisenhower and Kennedy had delegated authority.

  As for the threat to SAC’s second-strike ability, Harry told me that bombers had already widely dispersed, including to more than thirty civilian airfields. (So much for the
plans announced at Ann Arbor four months earlier of giving the Soviets maximum incentive to forgo targeting our cities.)

  Thirty-eight missiles meant a big expansion, relatively, of their small strategic force. In Russia, they were starting to deploy their new silo-based ICBMs, the SS-7. Perhaps sixty sites were under construction, but only about ten, Harry told me, were operational.

  Along with the four SS-6s at Plesetsk, for what they were worth, that meant that the Soviet first-strike missile force was at least doubling or expanding far more than that overnight. Yet it still didn’t mean that they would escape total devastation if they struck first. A single surviving SAC base would assure that, and well more than one would survive. Aside from our theater forces, they would also be hit by Polaris missiles and carrier forces at sea, and surviving Atlas and Titan missiles. Fifty to a hundred missiles didn’t give them a disarming first-strike force.

  Nor did the vulnerable IRBMs (which were not delivered, because of the blockade) on unhardened fixed sites do much for their ability to strike second. The mobile MRBMs, if we really couldn’t find all of them, would do more for their retaliatory capability. Of course, if the Soviets were allowed to base missiles on Cuba, they could quickly deploy a larger number of these from their current arsenal. A hundred or so IRBMs would make a big difference to their first-strike capability. Or so we calculated then, in days when it was assumed by our military that either side could “accept” tens of millions of deaths, though not hundreds of millions.

  We have an unusual record of the Cuban missile crisis101 as a result of tapes Kennedy made of meetings of the ExComm. I wasn’t surprised to read, years later when the tapes were transcribed, that McNamara had said at the second ExComm meeting one week earlier much the same as I had: that these missiles didn’t affect our security decisively, or even significantly. “I’ll be quite frank,”102 he told the president. “I don’t think there is a military problem … This is a domestic political problem.”