The Doomsday Machine

by Daniel Ellsberg

  Now I was hearing that this impression and all the official statements that led to it were false. It was not only the president who could make the decision and issue the orders, and not even (as most people probably assumed, if they thought about it) the secretary of defense or the Joint Chiefs of Staff in the Pentagon, but commanders in the field thousands of miles from Washington who thought their forces might be about to be destroyed. Similar letters, the control officer told me, had gone out to all the unified commanders with nuclear forces and to the commander of the Strategic Air Command in Omaha.

  I had come to the Pacific in the command and control study group believing what virtually all Americans believed, in or out of the government: that the president alone was authorized to decide when to launch nuclear attacks. That was why my investigations of how U.S.-initiated nuclear war might arise, as described in the preceding chapter, focused entirely on the possibility of unauthorized actions. Now I was hearing from a very credible source that I, along with everyone else, had been mistaken. The current president had, after all, delegated his authority to theater commanders, as well as the Commander in Chief of Strategic Arms Command (CINCSAC). In some circumstances, commanders of four-star rank could issue in their own name an authorized directive to undertake nuclear attack without the immediate prior involvement of the president.

  Surprising as this was to hear, the practical logic of making such a delegation was clear enough. Without it, the Soviets could paralyze any retaliation to a nuclear attack on the United States simply by destroying Washington before the president had given an Execute order, or perhaps before there was any warning at all. That could not be allowed to happen.

  A single nuclear warhead on the capital could kill not only the president but all of his legally designated successors in the cabinet and Congress (and the JCS along with the secretary of defense, the only civilian aside from the president in the military chain of command)—all of them who were in town at that moment. If nuclear deterrence were to have any substantial backing at all—if it were to be more than an empty bluff—it could not be the case that one such explosion would definitively block any authorized, coordinated nuclear response to that or any subsequent nuclear attack. That would be virtually an invitation to the Soviets in a crisis—when they had any reason to fear U.S. escalation to nuclear war—to forestall reliably either a U.S. first strike or even U.S. retaliation to a Soviet preemptive attack by delivering a single warhead on Washington, “decapitating” American political and military leadership.

  In fact, if the Soviets were confident that a small, initial “decapitating” attack would thoroughly paralyze our strategic and tactical nuclear forces, a premeditated surprise attack would look not only feasible but also safe for them. Even the best American warning system couldn’t reliably, if at all, alert authorities to the approach of just a single vehicle: in particular, a low-flying cruise missile or a short-time-of-flight medium-range ballistic from a submarine or ship, or even a “suitcase” bomb, perhaps smuggled into the capital long in advance.

  It seemed obvious once I thought about it. The public’s impression of exclusively presidential or even high-level military control, which I’d shared up until that moment, could not be valid. That applied all the more to the notion that only the president himself could “push the button.” Could a single assassin’s bullet, or a temporary separation of the “football” from the president (as has happened several times, including, later, following the shooting of President Reagan) open a window of total inability to respond to a nuclear attack?

  Not really. The theatrical device represented by the president’s moment-by-moment day-and-night access to the “football,” with its supposedly unique authorization codes, has always been exactly that: theater—essentially a hoax. Whatever the public declarations to the contrary, there has to be delegation of authority and capability to launch retaliatory strikes, not only to officials outside the Oval Office but outside Washington too, or there would be no real basis for nuclear deterrence.

  At that time, no system of Permissive Action Links (PALs)54 existed, in which a coded signal was necessary to permit the physical detonation of any nuclear weapon or the launching of a nuclear-armed missile. This would physically prevent launch or detonation without a coded “combination” from a higher authority. If such a system ever came into existence—as I, among others, hoped fervently to help bring about—the combination couldn’t be held exclusively by the president or by any individual or group of officials in Washington, D.C., or the Pentagon in Arlington, Virginia. If it were, one large bomb or device exploding on that joint target would lock up and render impotent the entire retaliatory capability of the United States.

  The most obviously necessary delegation would be to the headquarters of the Strategic Air Command at Offutt Air Force Base, in Omaha, Nebraska. But that was just as vulnerable to one large bomb as the capital, despite underground shelters in both. SAC had at that time, and throughout the Cold War, an airborne command post with a brigadier general aboard in flight at all times. That one-star Air Force officer would have to have (and, General Curtis LeMay later confirmed to me, he did have) the delegated authority to direct the execution of the SAC war plan. And what I was now being told was that delegation had extended as well to the tactical forces under theater command.

  After all, without delegation to CINCPAC, carriers and bases all over the Pacific might be precluded from launching retaliatory strikes just by atmospheric conditions that prevented an Execute message from getting through from the Pentagon to Hawaii, even if Washington had not been hit. But the same logic applied to the problem of relaying an Execute order from CINCPAC headquarters on Oahu to CINCPAC’s nuclear forces. Most of these were in “WestPac,” the Western Pacific, with the Seventh Fleet carriers or on bases in Korea, Japan, Okinawa, Taiwan, or Guam. They were as far from Oahu as Hawaii was from the continental United States. And communications from Oahu were just as subject to storms over the Pacific and other disturbances as radio signals from Washington were. On average, our study group learned, commanders in Hawaii were cut off from communications from or to Washington for some part of each day. Exactly the same was true for communications between commanders in Hawaii and Westpac.

  Therefore, the CINCPAC nuclear control officer told me, Admiral Felt had made a comparable delegation of authority to his next lower level of command, including the commander of the Seventh Fleet. Again, this was plausible, logical. And yet, like his first statement, it was startling to me. Was it really true that our practical, secret command arrangements were so sharply at odds with the policy declarations of the White House and secretary of defense? The control officer clearly believed what he was telling me in great confidence (as a member of a high-level consulting team reporting directly to Admiral Felt). But could he be right?

  I had the chance to check this out later on our visit to the cruiser St. Paul, the command ship of the Seventh Fleet. As recorded in my notes of the January 26, 1960, meeting with Vice Admiral Frederick N. Kivette, commander of the Seventh Fleet, and Vice Admiral Clarence E. Ekstrom, commander of Naval Air Forces in the Pacific, both emphasized the importance of the Navy doctrine that actual combat operations must be left to the engaged units acting with relative autonomy and with minimal attempt to control them by higher command. Even in limited war, Kivette said, “it wouldn’t matter” if communications were out between Oahu and the Seventh Fleet, or even between the Seventh Fleet and the carrier task groups: “Operations would be decentralized, I wouldn’t be interfering, unless I had some intelligence they didn’t have.”

  Kivette believed that a limited war would remain centralized only so long as political maneuvering predominated, with no shooting, as in the earlier Lebanon and Taiwan crises in 1958. (Ekstrom had commanded the Sixth Fleet in the Mediterranean during the Lebanon-Iraq crisis.) The two admirals expected and approved extreme decentralization “as soon as shooting started.” Thus, although they expected communications to be disrupted
frequently, especially in wartime but even for natural causes, they were both relaxed about the implications of this.

  They both rejected notions that preplanning could solve problems. One couldn’t plan for everything—surprises must be expected. But at the same time, they didn’t foresee or desire centralized direction during hostilities. They preferred to rely, they both said, on the judgment of the carrier task group commanders, simply providing them with objectives. And, they stressed, the commander afloat must be given great latitude in interpreting and executing his orders. “You’ve just got to trust your commander at sea.” This applied all the more, they said, to conditions of general nuclear war. They agreed that it would be “nice to know” what Air Force bases had been hit at the outset of a nuclear war or what carriers had been destroyed. But under the conditions of the CINCPAC plan for general nuclear war, “it probably wouldn’t matter anyway.”

  By this point in the discussion, our team seemed to have established some rapport with the admirals. I hadn’t indicated to them the unease that I was beginning to feel about their seeming indifference to the unreliability of communications in nuclear war, or in a non-nuclear war that could suddenly turn nuclear. So I ventured to raise the issue I’d been told about in great secrecy. I asked Admiral Kivette if he had heard of a letter from President Eisenhower to Admiral Felt delegating authority over nuclear operations if communications were out. He said, yes, he knew that Admiral Felt held such a letter.

  I asked him how common it was for his own flagship to be out of communications with Admiral Felt in Hawaii. He said, “Virtually every day, part of the day.”

  I asked him, “What if your communications with Oahu were out and you thought, for other reasons, that nuclear war might have commenced, or might be about to? What would you do?”

  To all our earlier questions, one or the other of the two admirals had responded immediately and at length. At this one, Admiral Kivette paused meaningfully, then said to me, “I stand mute.”

  It was the only question that he didn’t answer explicitly, and he drew himself up in his chair rather formally as he said it. But he was smiling, indicating—it seemed to me—that he assumed I knew the answer to my question, but that this was all he was supposed to say, and that he knew that it was, in context, an answer. Evidently, he regarded, or he wanted us to know that he was supposed to regard, Admiral Felt’s delegation of nuclear authority to him as a more sensitive matter than Eisenhower’s delegation to Felt (about which I had already revealed I knew).

  After a further pause he added, “Anyway, I just can’t believe that we could be cut off from all communications; we could get through to someone, and he would know what was happening.”

  Admiral Ekstrom added to this, “It would depend on the whole picture. What had been happening up to that moment, how ready are we, are we fueled up, etc.”

  An hour later I raised the question with Admiral Kivette’s nuclear control officer. This officer readily told me that, yes, Admiral Felt had delegated to Admiral Kivette the same authority that, he said, President Eisenhower had delegated in writing to Admiral Felt: to launch nuclear weapons at his own initiative during a crisis in case of communications outage.

  If they were right about the letter from the president, this contravened and superseded the guidance I’d read in Top Secret war planning—including the Pacific Command’s General Emergency Operations Plan (GEOP) for general nuclear war—that U.S. nuclear attacks could be initiated only by a presidential decision at the time of the attacks. The general public believed that as well, and believed further that the president would never delegate this authority under any circumstances. For once what the public had been told corresponded to the actual secret guidance written into war plans by the JCS. Yet if these officers were correct in what they were telling me, written authorizations by the president, the commander in chief, secretly contradicted this JCS guidance in the war plans. As did further sub-delegation by CINCPAC. It wasn’t only one or half a dozen four-star admirals and generals who felt authorized to initiate nuclear operations in some realistic circumstances but their far-flung three-star subordinates as well. And who knew how many others?

  I still didn’t feel certain that the alleged letters from President Eisenhower actually existed; no one had offered to show them to me, or even claimed to have seen one himself. Yet the affirmation from the Seventh Fleet nuclear control officer that CINCPAC had on his own authority made such a delegation to the Seventh Fleet commander (and perhaps others) meant to me that the belief that Eisenhower had himself formally given such authorization to CINCPAC had consequences, whether or not that belief was true.

  It was clear from both nuclear control officers’ manner in speaking about this that they were telling me something of the highest sensitivity. I refrained from asking whether they were aware of even further delegations to officers still lower in the chain of command. In light of the broadly and firmly held understanding within the military (not only the public)—explicitly confirmed in secret war plans—that authority to initiate nuclear war rested exclusively with the president, such delegations would have looked questionable or even gravely illegal to the recipients if they had not shared a secret belief that the president himself had chosen to make such a delegation to theater commanders. But given that belief—and I found it widely held in the Pacific—it was clear that the same incentives that influenced the president existed for further delegations by lower commanders.

  Each level of command had reason to worry that during a crisis, an outage of communications, whether due to atmospheric or technical difficulties or an enemy attack on that command headquarters, could paralyze the nuclear capabilities of subordinate units unless they’d been delegated authority to act under such conditions: as, the commanders apparently all believed, CINCPAC had been by the president. Indeed, CINCPAC would logically infer that he could not reliably carry out the intention of the president with respect to the actions of his theater nuclear capability in the event that Washington was attacked or out of communication unless he provided explicitly for the possibility—actually, the likelihood—that he himself would also be attacked, or might be out of communications for other, quite ordinary reasons, with his subordinate commands.

  He could provide for that only in the same way that President Eisenhower had: namely, by allowing lower commanders to exercise their own judgment in those circumstances. In any case, the two admirals found this sub-delegation totally compatible, even obligatory, in terms of naval traditions. But in this situation the logic that applied to the Navy and to CINCPAC applied as well to all the other unified and specified commanders to whom the president had allegedly delegated authority. These were the theater commanders in Europe, Alaska, the Mediterranean, the Atlantic, and the Strategic Air Command, as well as NORAD, the air defense command.

  Unless the president forbade such further delegation explicitly (and perhaps even if he did), the example of his own delegation to CINCPAC and other theater commanders seemed likely to be imitated, not only in the Pacific but also in other theaters around the world. And it was implicit in what these officers told me that the president had not explicitly forbidden these theater commanders to delegate that authority any further in the manner that CINCPAC had sub-delegated to the Seventh Fleet.

  Nevertheless, I found it hard to believe that the president would have wanted any further delegations, or that he even knew they existed. His own action of delegation—assuming these letters really existed—distributed the authority to just over half a dozen four-star generals and admirals. Further delegations multiplied the number of individuals with authority, under some conditions, to initiate nuclear war, and also drew into that circle officers of progressively lower rank, lesser experience and maturity, and narrower responsibilities and access to information.

  At some point, as one moved down the chain of command, the advantages of providing further assurance of a retaliatory response would be outweighed—it appeared to me—by the increased
risks of a wrong response. Not only were the risks progressively greater as lower units and levels of command became involved, but from the president’s perspective, the need or incentive for subsequent delegations was progressively smaller, involving smaller portions of the overall retaliatory forces.

  But to a commander at the lower level, whose mission understandably seemed to him to have transcendent importance if it involved any nuclear weapons at all, it wouldn’t look that way. He would want to be sure that “his” weapons took part in the big war—the fight for national survival and victory. If you left the decision whether to delegate further to each successive layer of command, I suspected it would be likely to go down to the bottom. In the limit, every flight commander, if not every pilot with a weapon aboard, would feel authorized, under some circumstances, to initiate nuclear war with the Communist bloc. (He might even be authorized, orally or in writing, by an immediate superior, with or without the knowledge of higher levels of command.)

  I accepted, as inescapable, the idea of Eisenhower’s delegation of authority to execute war plans to a handful of four-star admirals and generals outside Washington. But I had growing unease, to put it mildly, at the prospect that this delegation reverberated downward in a widening circle that permitted authorized launch by more and more subordinate commanders, not to mention the physical possibility of unauthorized action by control officers or by crews of alert nuclear vehicles, whether planes or submarines.