“We all muddled into war,” said Lloyd George of 1914. None of the Great Powers really wanted war; certainly none of them wanted the sort of war they got. Europe had been preparing at great expense for thirty years to fight a big war, but when the war finally commenced the leaders of Europe were amazed all the same. How the war began, and why it occasioned surprise, are among the biggest puzzles of the century. In the 1930s the exiled Kaiser Wilhelm, speaking of Hitler, told an English visitor to Doorn, “It will run away with him, as it ran away with me.” What was it that “ran away,” exactly? Why did Lloyd George describe as a “muddle” something that was planned in advance in great detail and that unfolded, in the event, precisely in accord with those plans?
The immediate problem was an Austrian quarrel with Serbia in the Balkans. The opening campaign of the war was a German invasion of Belgium a thousand miles away. It’s not hard to follow the chain of events from one to the other. Serbia was backed by Russia. Austria was backed by Germany. France had promised to come to Russia’s aid in the event of war. The prospect of a two-front war alarmed the German General Staff; they figured that their best chance for success lay in dispatching their enemies one at a time. For a number of reasons France looked to Germany like the easier of the two. Belgium offered the best route into France. Thus war in the East meant war in the West.
It all made sense, and was well understood at the time. The diplomats of Europe had spent thirty years ensuring that everything was tied to everything else. The big question was who would land the first blow. In the years before 1914 the Great Powers made elaborate plans to mobilize their armies as quickly as possible. These plans depended largely on railroad timetables. Since Russia knew war with Austria meant war with Germany, it planned to mobilize against both at the same time. But at one point in July 1914, Russia, fearful of the big war looming with Germany, toyed with the idea of mobilizing against Austria alone. This was thought to offer Russia a chance to up the ante in the war of nerves with Austria without giving Germany an excuse to jump in. General Sergei Dobrorolski, chief of the mobilization section of the Russian General Staff, however, was horrified. For technical reasons a partial mobilization would call up too few troops for war with Austria, and thoroughly gum up plans for full mobilization later on. Eventually the Czar accepted the all-or-nothing logic of the situation and approved orders for complete mobilization.
For technical reasons as well it was assumed by military men everywhere that mobilization meant war. “One has only to press the button,” said a Russian general at the time, “and the whole state begins to function automatically with the precision of a clock’s mechanism…. Once the moment has been fixed, everything is settled; there is no going back; it determines mechanically the beginning of war.” So it proved in the event. It is not far-fetched to say that the railroads of Europe carried the continent into war.
This bit of history would be of academic interest only if the situation it describes did not closely resemble current preparations for nuclear war by the United States and the Soviet Union, the principal significant difference being the length of time required for the plans to unfold—two or three days in the case of 1914, conceivably as few as twenty minutes now. The railroads of Europe have been replaced by the printed electronic circuitry of early warning devices and the systems for “command and control” of nuclear weapons, but we may legitimately think of both as wiring. The strategic nuclear forces of the United States—something on the order of nine or ten thousand warheads—are wired together by “systems of incomprehensible complexity” under the control of “fantastically complex nuclear command organizations,” which, according to Paul Bracken’s book, would find it technically difficult if not impossible to limit a nuclear war—or even to limit a nuclear alert—once begun.
This gloomy warning is the central message of The Command and Control of Nuclear Forces, probably the single most important book to be published on the subject of nuclear weapons in recent years. Bracken, who now is a professor of management at Yale University, worked as a defense analyst at the Hudson Institute. “Our intuitive, gut reaction,” he writes of the wiring, “is that the operation of anything so large and complicated must be dangerously unstable.” His book states the case for the soundness of this reaction with great persuasiveness. Most discussion of nuclear weapons concentrates on political policy, as if the danger posed by modern war systems were fully in the hands of national leaders, and we might safely depend on their prudence so long as it remains clear to reasonable men that a big war would be a big mistake. Bracken takes a different approach. He is plainly worried that our preparations for the control of a nuclear war may make one much more likely in the event of a crisis. To prove his point, Bracken provides a more complete picture of these preparations than can be found anywhere else in the open literature. The war system he describes is a kind of great microchip beast which is preternaturally sensitive and alert, with one overriding primitive impulse hard-wired into its cerebral cortex—to strike out and destroy the only other beast of its kind.
“The greatest single change in nuclear forces during the past 20 years is [the] shift from loose to tight coupling,” Bracken writes. By “coupling” he refers to the wiring between mechanisms of warning and response. Alarming news—a sudden blizzard of radio traffic between Soviet command centers and missile fields, say—would almost immediately be detected and processed by the electronic systems coordinated with American nuclear forces.
The possibility exists that each side’s warning and intelligence systems could interact with the other’s in unusual or complicated ways that are unanticipated, to produce a mutually reinforcing alert. Unfortunately, this last possibility is not a totally new phenomenon; it is precisely what happened in Europe in 1914. What is new is the technology, and the speed with which it could happen.
In short, the war systems of the United States and the Soviet Union are now so jumpy and apprehensive, so quick and potentially incontinent in response, so convinced they know what must be done, that they have the capacity to leap from crisis to war all by themselves. Such a war would be “accidental” only in the sense that it was not what the belligerents “wanted.”
The problems posed by nuclear weapons are now officially addressed by two large but separate communities with parallel but dissimilar concerns. One is made up of the analysts from think tanks and universities who use the concept of deterrence to explain why it is safe to have weapons of a kind and number that it is not safe to use. The second consists of military men, excluded by law and custom from making policy, who concern themselves with the design and deployment of nuclear weapons in peacetime, and the mechanics of managing their delivery in war. The military men do not have much to say about why a big war might start. They are not even inclined to speculate in private. For the same reason nuclear war games are notoriously difficult to get started. The players generally must be ordered to commence firing. What happens after that point—in the game, or in the real world if it comes to that—is the principal professional concern of the military planners.
Back in 1945 the mechanics were a simple matter. At the close of World War II the United States had just one assembled core for an atomic bomb, and just one squadron of specially modified B-29 bombers to deliver it in the event of a new war. An order to use the bomb probably would have come over the telephone. But the fears of the general public leaped ahead of American technical capacity in the fall of 1945. Alarmed citizens, including many scientists who had worked on the Manhattan Project, imagined any new conflict as a kind of “Buck Rogers war” run by generals in underground bunkers directing fleets of bombers and missiles.
The military commanders knew better. The Atomic Energy Commission refused to divulge just how many atomic bombs were in the arsenal, but the military officers knew there couldn’t be many. In the spring of 1946 General Curtis LeMay, back in Washington after his successful air campaign against Japan, was asked to devise a war plan against Russia using about fifty atomic bombs. The general public probably assumed there were hundreds of bombs ready and waiting, but in fact even the working figure of fifty was wishful thinking. Logistics were the hard part—getting bombs and bombers to forward bases from which they might actually reach the Soviet Union. The air force, dominated by World War II bomber generals, was so scornful of Buck Rogers notions of war that it banished the word “rocket” as smacking of science fiction, and adopted “jet” instead for what were, until then, rocket motors. Early plans for World War III called only for a larger version of the air war of World War II.
By the early 1950s the Strategic Air Command had hundreds of atomic weapons and a fleet of intercontinental bombers for their delivery. War plans called for a wholesale assault on Russian cities, with first dozens and then scores of bombs allocated to major target areas like Moscow and Leningrad. “We can’t afford to fight limited wars,” the secretary of defense, Charles Wilson, said in 1953. “We can only afford to fight a big war, and if there is one that is the kind it will be.” Missiles were added to American strategic forces before the end of the decade—crude behemoths at first, followed by smaller, better protected, more accurate missiles whose targets could more easily be shifted.
The development of these forces was matched by that of military communications and early warning systems. First there was the Distant Early Warning (DEW) line of radars to detect Soviet bombers coming over the North Pole, established in 1953 (before the Soviet Union even had bombers which could make the trip). This was followed by the Ballistic Missile Early Warning System (BMEWS) begun in 1958, and finally, during the 1970s, by early warning satellites containing infrared sensors to detect the exhaust plumes from Soviet missile launches. Other systems monitor Soviet submarine deployments and radio transmissions of all types.
Since 1957 the ever-growing torrent of information from warning systems has been forwarded to the North American Aerospace Defense Command (NORAD) in Colorado Springs, the facility which most closely resembles the popular picture of an underground command bunker from which generals at their consoles would direct a nuclear war. How long the flow of information would last in war-time is hard to guess. The vast amount of data from satellites, for example, is currently processed first at receiving stations in Denver and in Alice Springs, Australia. Both installations are “soft”—vulnerable to an atomic strike—and would likely be early targets in the event of general war. So, according to Bracken, would the command centers in the Pentagon, at Mount Weather in Pennsylvania, at SAC headquarters near Omaha, Nebraska, and even at NORAD itself, which has been buried under Cheyenne Mountain since 1965.
A former air force intelligence officer once described to me a visit to the new NORAD facility shortly after it was opened, at a time when missile accuracies were measured in miles. An enthusiastic briefing officer gave him statistics to support NORAD’s confidence that the center would survive anything but a hit within “x” yards. At this one of the party of VIP visitors, Simon Ramo, a missile builder, remarked in a musing tone, “That shouldn’t be too difficult.” The job took about ten years. Modern American missiles can be expected to land within a tenth of a nautical mile, about 600 feet, of the “Desired Ground Zero,” and technical people speak confidently of future accuracies under 100 feet. Russian missiles could come as close. It is probably fair to say that both Soviet and American missiles can destroy anything in a known location on or near the surface of the earth. This means not only soft targets like airfields and submarine ports, vulnerable for years, but hardened missile silos and the places called the “nodes” of command, control, communications, and intelligence, referred to in the professional literature as Cu3I.
War planners recognize this fact as a major problem. Communications and intelligence have always been the vulnerable part of war systems. Pearl Harbor, for example, was largely the result of a communications screw-up. Many cold war crises—in Lebanon (1958), Cuba (1962), Czechoslovakia (1968), and the October war (1973), for example—have “pinged the system” and revealed major bottlenecks or other failures of “connectivity.” Warnings arrived late or not at all, and messages failed to get through. A nuclear was would make unprecedented demands on C3I. How are the commanders to direct the war if they can’t converse reliably with the men actually conducting strikes against the enemy? More to the point, how could such a war be limited if it can’t be directed?
It is well known that US war-planning made a decisive move in the direction of “war-fighting” strategies during the Carter administration, culminating in Presidential Directive 59. The final draft of PD 59 was written by General William Odom, military adviser to Zbigniew Brzezinski on the National Security Council. Odom had read Sidney Fay’s The Origins of the World War (1928) and had been unpleasantly reminded of American preparations for World War III. In military organizations facing a crisis, Odom told a Harvard seminar in 1980, “you tend to do the things you are organized to do…. The more I thought about the way we were organized, the more it reminded me of 1914.”* The Single Integrated Operational Plan (STOP) of the early 1970s offered a president two basic, equally unpleasant choices in the event of nuclear attack—“releasing 70 to 80 percent of our nuclear megatonnage in one orgasmic whump, or just sit there and say, ‘Don’t do anything, and we will just take the incoming blow.”‘ By offering a wider range of preplanned choices, PD 59 was intended to reduce the chance of an orgasmic whump.
But rigid, tightly circumscribed war plans were not the only source of danger that any nuclear war would almost certainly end as a big nuclear war. Destruction of command and control facilities would also tend to create an all-or-nothing situation. One of the Command Post Exercises (CPXs) conducted by the Carter administration to practice command and control involved a hypothetical Soviet attack on US C3I nodes. The result was chaos. The president lost contact with his forces. Communications breakdown meant all-out war or sitting tight. This is a choice which of course must be made before the war, since the commanders of strategic forces must be informed of the choice before the breakdown in communications.
The only way to avoid the orgasmic whump, according to General Robert Rosenberg, also on the NSC staff during the Carter years and a guest at the same 1980 Harvard seminar, was somehow to maintain communications in a situation bound to include “EMP [Electromagnetic Pulse] blackout, brute force damage to systems, a heavy jamming environment and so on.” That seems a brisk description of what would amount to battlefield confusion on a scale of primordial chaos. Rosenberg was not guessing. Even in the open literature Soviet military planners frankly say that in the event of war they hope to destroy physically a third of US communications capacity, to jam a third, and to count on pure stress and message crowding to overwhelm the last third.
The Carter administration’s solution to this problem was Presidential Directive 53, a quasi-secret plan to upgrade C3I by emphasizing “endurance” rather than “survivability.” Rosenberg told the Harvard seminar, “Worrying about whether or not the President lives or dies, or whether you can dig a hole deep enough in the ground to survive a direct three-megaton blast, is just not the right way to solve the problem. It is the function that has to endure” (my italics). The emphasis on C3I has been sustained by the Reagan administration with a great deal of money and hardly any publicity.
The essence of the technical solution is to provide a communications system that can reconstitute itself piecemeal as required. When one transmitter is destroyed another comes on line. When one National Command Authority (NCA—that is, the president or his lawful successor of the moment) is dead or otherwise incommunicado, another takes his place. When satellite monitoring stations in Denver or Alice Springs are destroyed, others pick up the task. When NORAD is gone, another C3I node takes over its piece of the war. And so on down the line. Each element of the grid will be backed up in sufficient depth so the grid itself can be maintained, just as in wars past it was expected that someone would always snatch up the fallen standard on the battlefield and keep the flag flying. A man with a flag is a good target, and so is a radio transmitter; but as a nuclear war progressed the capacity of the other side to locate and destroy new nodes in the grid would become progressively weaker, thus making it easier to keep the talk flowing—in theory, at any rate.
Is this realistic? Could the Soviet and American war systems maintain their equanimity in the midst of catastrophe, and somehow conduct a measured campaign of what amounts to nuclear counterbattery fire while working their way toward a cease-fire short of the silence following an orgasmic whump? As described by Bracken, the task is clearly immense. Until recently the World Wide Military Command and Control System (WWMCCS), established by the Pentagon in the 1950s, has had only one task—to disseminate a lawful “go” order and maintain communications long enough for US forces to carry out the attack. After that, its purpose, as well as its ability to function, was expected to cease.
Two questions ought to interest us here. The first is whether communications can be reliably maintained in a nuclear environment. The answer to that, even after the expenditure of billions, is, Who knows? The second question is what US commanders in charge of strategic nuclear forces would do in the absence of reliable communications. Bomber pilots, submarine commanders, and Minuteman Launch Control Center crews might all find themselves with weapons at the ready but nobody at the other end of the line. Do they sit tight? The answer to that is a many-layered secret.
When I first started to look into the problems posed by nuclear weapons about five years ago, an academic defense expert, Michael Mandelbaum, told me there were only two real secrets in the defense community: how to build a hydrogen bomb, and who had the authority to use nuclear weapons. The latter is a deep secret indeed, and it is possible that no single person, but only the system itself as a whole, knows the answer. I once asked a military officer what would happen in the event that American NATO military officers in Europe wanted to use nuclear weapons, but could not communicate with Washington. With an unhappy expression he told me, “There is no unclassified answer to that question.”
This remark, like common sense, suggests the answer. Bracken confirms it. The war system, in the Pentagon jargon, is programmed to “fail safe” in war. In theory, nuclear weapons deter nuclear attack through a promise of devastating response, and the war system is designed to make good on that promise. As long ago as 1966 General Nathan Twining, a former chairman of the Joint Chiefs of Staff, explained in his memoirs, Neither Liberty nor Safety, that “common sense” required the president to “predelegate” his authority over the release of nuclear weapons to subordinate commanders who might then lawfully elect to shoot on their own say-so “under certain grave circumstances.” Twining spelled out those circumstances: “If the nation were under attack, and there was no Washington, D.C., left, America would fight back rather than die with its own powerful force immobilized.”
As described by Bracken, specific authority to use nuclear weapons is established by a kind of positive control system. The chain of command runs from the president through the secretary of defense to the commanders of the five unified and specified commands in charge of nuclear weapons—SAC, the North American Aerospace Command, and the Atlantic, the Pacific, and the European commands. Next in line are the actual military units in possession of nuclear weapons, followed by specific carriers (bombers and submarines) and missile units. Authority to use nuclear weapons automatically resides in the highest “live” member in the chain of command.
Imagine that early warning systems have rung all the bells and the Strategic Air Command, for example, launches its bombers assuming they can be recalled by the order of the president. So long as SAC can reach the president he is in charge. If the president’s end of the line goes dead, Bracken tells us, SAC has predelegated authority to act on its own. If the SAC command centers fall silent, then individual bomber pilots have predelegated authority to proceed. The system works the same way for missile launch crews and submarine commanders. The advantage of this approach is that it would be impossible for a Soviet attack to immobilize the US strategic forces by decapitating the system through strikes on command centers and communications systems. There is, in effect, a button, but destroying the button has the same result as pushing it.
The disadvantage of this approach is that it would make a nuclear war extremely difficult to stop. The simplest message is the easiest to get through, and the simplest is “go.” In 1980 General Odom told the Harvard seminar that he hoped the US would build “a C3I system with a very high probability of control under very adverse conditions, stressed by pretty large strikes.” Odom’s hope is now a fact. Bracken thinks it won’t work. Worse, he thinks it will fail early—perhaps even in the opening stages of an alert. In the absence of an active National Command Authority the Soviet Union, he points out, would have to negotiate a cease-fire with all five of the separate unified and specified commands. The problem would be compounded if the Soviet Union’s command system had also been fragmented.
The picture drawn by Bracken is a chilling one: a nuclear attack on the United States would destroy or silence big sections of the communications grid. This would very likely isolate elements of US strategic forces, thereby sending out predelegated “go” orders in unpredictable directions. The US response to attack, part deliberate and part automatic, would set the bells ringing in whatever remained of the Soviet war-monitoring facilities, which in turn would unleash new attacks on the United States in a reciprocal process ending finally, one imagines, in a total collapse of communications. Such a collapse would unleash all the surviving units armed with nuclear weapons to proceed as they saw fit, or had been instructed to do in advance. What those instructions might be is hard to say with confidence, but unclassified discussion of these matters stresses that the final stage of the current Single Integrated Operational Plan, as of all its predecessors, involves attacks on Soviet “recovery” targets—all those human institutions and concentrations of capital goods which might allow the Russians to rebuild. In effect, when we no longer know what’s happening, we try to bomb them back to the Stone Age.
The big question facing us is whether the nuclear arsenals built by the United States and the Soviet Union will be used in war. Just about everybody agrees this would be a catastrophe, perhaps an irreparable one. Some think it’s bound to happen if we go on as we are. Others say not. The answer depends mainly on your notion of how wars begin. If you think that war is a rational endeavor—a deliberate attempt to seize something through military force—then you might reasonably conclude no sane leader would initiate a nuclear war. But if you think of war as something that is fundamentally irrational—an unpredictable and explosive release of primitive social emotions, and a commonplace event in human history—then the mere inadvisability of unrestrained war may not strike you as a sufficient protection.
World War I has been much on the minds of historians, defense analysts, and military men in recent years, notably since an influential article, “Rumors of War: The 1914 Analogy” by Miles Kahler, appeared in the Winter 1979/80 issue of Foreign Affairs. More recently, International Security devoted most of its Summer 1984 issue to “The Great War and the Nuclear Age.” Analogies are notoriously flexible, and this one has been the object of vigorous criticism by those who dislike its implications. Still, it is not the analogy between current circumstances and the world in 1914 that is most troubling, despite the many eerie parallels cited by Kahler and others, but the evidence that World War I offers about the characteristic behavior of great powers, and the sorts of factors that can set them to fighting after a long period of preparation and fear. The Harvard defense analyst Joseph Nye has often said that the leaders of Europe would not have gone to war in 1914 if they had possessed a crystal ball to tell them what would follow. Nuclear weapons, he says, in effect provide just such a crystal ball: We can be in no doubt what will happen once the shooting starts. Bracken’s important book alerts us to a neglected danger—the inertia we have built into our command and control systems. The United States and the Soviet Union, he writes, “have built the most complex technological apparatus ever conceived, without thinking through its purpose or how to control it.” We may consult the crystal ball, but the technological apparatus does not.
Nye and other analysts are no doubt right when they argue that no sane leader would deliberately take the fatal step in starting a nuclear war. But which step is the fatal one? Bracken is trying to warn us that the fatal step may have been taken years ago when the planners tightened the coupling between early warning and command-and-control systems in an attempt to ensure that we could not be disarmed by a surprise attack. A second fatal step may have been taken when we decided to build systems that would “fail deadly” in the event commanders lost contact with their forces in wartime.
But of course these were not really “steps” at all, in the sense of being clear, fully conscious, deliberate acts. Rather they are the result of countless tinkerings with the system over four decades, each intended to solve highly particular technical problems. In a sense we have never “decided”—clearly and all at once—what to do in the event of a nuclear war, just as no one ever “decided” in 1914 that unreasonable Austrian demands on Serbia required a four-year war that would wreck Europe. In a sense, the system built itself, and only the system knows how it would respond to insult.
January 17, 1985
Incidental Paper: Seminar on Command, Control, Communications and Intelligence: Guest Presentations—Spring 1980. Center for Information Policy Research, Harvard University, p. 5. The Center’s publications on C3I are the best single introduction to the subject, and go far to explain just what sort of war the war fighters are planning to fight. ↩