The Last Warrior

by Andrew F. Krepinevich

  Their work paid off. On October 31, 1952, the United States set off the world’s first thermonuclear explosion on the Pacific atoll of Enewetok. The experimental shot, nicknamed “Mike,” had a yield equivalent to 10.4 million tons (or 10.4 megatons) of TNT—roughly a thousand times the yield of the atomic bomb dropped on Hiroshima.31 The Soviets followed suit in April 1954 with RDS-37, a two-stage, radiation implosion thermonuclear bomb. Air dropped from a Tu-16 Badger bomber, the device produced a yield of 1.6 megatons.32 Significantly the physics underlying Mike and RDS-37 did not appear to impose any obvious upper limit to the yield possible from thermonuclear weapons.* Thus was born the thermonuclear age in which both Washington and Moscow began fielding ever larger numbers of megaton-class nuclear weapons.

  Although the United States was the first of the two superpowers to test a thermonuclear device, Mike was an 82-ton experimental behemoth, not an operational weapon. Much progress in both physics and engineering had to be made before megaton-class thermonuclear warheads could be built both small and light enough to be delivered by either aircraft or intercontinental ballistic missiles (ICBMs). Nevertheless, even at this early stage it was clear that thermonuclear weapons would fundamentally change the strategic outlooks of both the United States and the Soviet Union. In their 1952 paper on the implications of large-yield nuclear weapons, Brodie, Hitch, and Ernst Plesset argued that over time the nuclear balance between the two nations would almost certainly evolve from the current period of near American monopoly and atomic scarcity to one of strategic parity and thermonuclear plenty.33 They accurately foresaw that nuclear weapons with yields of 1 to 25 megatons—or more—would eventually be available to both the United States and the Soviet Union in relatively large numbers. Once that shift in the US-Soviet nuclear balance had occurred, they warned, any large-scale thermonuclear exchange between Washington and Moscow would be an act of mutual suicide.

  While obvious in retrospect, at the time these findings were both stunning and sobering. They quickly gave rise to efforts by RAND to develop concepts and strategies to cope with the growing threat. Marshall found himself on the ground floor of these developments. After witnessing a low-yield atomic test at the Nevada Test Site with Arnold Kramish in the spring of 1952, Marshall came way awed by the sheer power of even a 10–15 kiloton atomic device.

  He and Kramish had driven from Santa Monica to Las Vegas the day before the test. After an early dinner in Las Vegas they drove out to the Nevada Test Site, checked in, and spent the night on army cots before being aroused hours before daybreak to drive to the location from which they were to observe the test. They were not even provided with protective goggles, and when the device detonated Marshall, who had covered his eyes with his hands, was able to see the bones in the thickest part of his palms from the explosion’s initial flash of light. Reflecting on this experience decades later, Marshall suggested that it would behoove the leaders of nuclear states to witness such detonations at least once in their lives to gain a firsthand appreciation of the enormous destructive power inherent in these weapons.

  Following his visit to the Nevada Test Site, Marshall happened to be in Washington on another project when the head of RAND’s Washington office, Larry Henderson, was asked to brief the Brodie-Hitch-Plesset study to the Bureau of the Budget. Henderson prevailed on Marshall to help with briefing. The project that had brought Marshall to Washington involved another RAND analyst, James Digby. He and Digby had been asked to help the US Air Forces in Europe’s (USAFE’s) intelligence directorate analyze the growing problem of securing reliable strategic warning of a Soviet nuclear attack. Marshall and Digby spent four months conducting an onsite assessment of the problem in Wiesbaden, West Germany.

  During Marshall’s early years at Project RAND, which is to say before both the United States and the Soviet Union had large numbers of nuclear weapons, it was still possible to believe that a future war between the two nations would, in many respects, resemble World War II. In planning for such a war in the late 1940s the Air Force’s initial inclination was to emphasize strategic bombing, only this time employing all available atomic weapons at the outset. The aim was the same as that which the Army Air Forces had pursued against Germany from 1943 through 1945: to destroy or otherwise neutralize the Soviet Union’s “industrial web”—the raw materials, factories, weapons and munitions, and transportation infrastructure that fueled a country’s ability to make war.34 As Marshall had discovered in 1949, the United States’ inventory of atomic weapons then consisted of only about two hundred Mark-3 and Mark-4 atomic bombs with maximum yields of 49 and 31 kilotons, respectively.35 After this small inventory of atomic bombs had been expended in an initial “atomic blitz” against the USSR’s industrial infrastructure the bombing campaign would have to fall back on attacks with conventional weapons similar to those employed during the 1943–1945 Combined Bomber Offensive against Germany in attempting to finish the job of destroying the Soviet war machine.

  As one would expect, RAND researchers in the late 1940s looked for alternative ways of using America’s small but growing atomic arsenal more effectively. One possibility was to make just a few atomic strikes on the Soviet Union and then drop leaflets encouraging the Soviet people to evacuate their cities. Such a strategy, some thought, might force the Soviet government to surrender, fearing that additional attacks by US atomic weapons held in reserve could destroy the USSR’s economic infrastructure. This and other options were explored in RAND’s “Warning and Bombing” (WARBO) study of 1949–1950.36 The study’s main idea was to find ways to control the level of violence so that it might be possible to achieve meaningful political objectives even in a nuclear war. For Bernard Brodie in particular this suggestion from WARBO appeared to offer a more rational alternative to the Air Force’s all-out “atomic blitz.”37

  Another option that surfaced within RAND during the early 1950s involved mounting a preventive war against the Soviet Union before its leaders could create an atomic arsenal of their own. The head of RAND’s mathematics division, John Williams, was a strong advocate of this view on the grounds that all-out war between the United States and the USSR was only a matter of time. If so, he argued, then it made sense to “throw the first [atomic] spear” should the Soviets refuse to relinquish their nuclear arms to international control.38 Others who advocated preventive war included Plesset, who headed RAND’s physics division, and John von Neumann, who had become an influential adviser to Project RAND due, among other things, to his help in developing the implosion (plutonium) version of the atomic bomb and his numerous contributions to theoretical and applied mathematics.39 But as Brodie pointed out during his acerbic disagreement with Williams over the preventive-war option, President Dwight Eisenhower had emphatically rejected it on the grounds that no moment could be propitious for initiating general war with the Soviet Union.40

  At this early stage in the Cold War another contingency worrying US war planners and RAND strategists alike was that, since the conventional military balance in Europe greatly favored the Soviets, the Kremlin might attempt to overrun Western Europe with conventional forces even in the face of a nuclear-armed United States. At the time of the Berlin blockade, for example, the Soviets had one and a half million troops in the sector surrounding Berlin, including the combat-experienced Third Shock and Eighth Guards armies.41 By comparison, in July 1948 there were less than 91,000 American troops in all of West Germany. Operational US forces consisted of the 12,180 men of the First Infantry Division plus two infantry and two field artillery battalions.42 During the brief period of the United States’ atomic monopoly, the prevailing wisdom among US military analysts was that a Soviet conventional assault on Western Europe would quickly overwhelm NATO’s conventional forces. Although West Germany, the Low Countries, and France would likely be lost, it was believed that NATO forces would be able to stop the Warsaw Pact’s advance at the Pyrenees Mountains along the Franco-Spanish border. As NATO ground forces retreated, the US Air Force would unleas
h its atomic blitz against the USSR’s industrial centers,43 while the United States would mobilize its own conventional forces as it had in World War II, eventually retaking Western Europe and advancing on Moscow to depose the Bolshevik regime.

  Importantly, this early vision of a future European conflict assumed that the United States had atomic weapons and the Soviets did not. As the latter began to field atomic weapons and delivery systems able to reach the United States itself, it became less and less realistic to think that one side could use nuclear weapons against the adversary’s homeland while remaining unscathed itself. To be sure, the US atomic monopoly did not end in August 1949 when the Soviets exploded Joe-1; not until the mid-1950s did Soviet Long Range Aviation begin fielding heavy bombers with sufficient range to deliver nuclear weapons against targets in the continental United States and return to Soviet bases.

  But once the Soviet Union had the bomb the handwriting was on the wall. It was increasingly evident that the strategic relationship between the USSR and the United States was rapidly evolving, and not in the United States’ favor. The 1952 Brodie-Hitch-Plesset RAND study of high-yield nuclear weapons had accurately foreseen both nations’ building up large arsenals of thermonuclear weapons. In May 1953 von Neumann assured Colonel Bernard Schriever, who went on to develop the Air Force’s first ICBMs, that by 1960 it would be possible to reduce the weight of a 1-megaton hydrogen bomb to less than a ton.44 This meant that thermonuclear warheads could be delivered by ballistic missiles against which there were no effective defenses. It was a terrifying possibility—and one that, as it emerged, would demand the intellectual talents of Marshall and other RAND strategists to address.

  By the mid-1950s Marshall had demonstrated his impressive analytic abilities across a range of subjects bearing on nuclear strategy. This led to his time being split among more projects than ever before. His insights on the limits of a nuclear bombing campaign targeting the Soviet economy, for instance, led both Air Force intelligence and the Central Intelligence Agency (CIA) to bring him onboard as an adviser to help with their projections of future Soviet nuclear forces. Marshall and Herman Kahn also continued their collaboration on developing more efficient approaches to Monte Carlo statistical methods, even after their assistance with the hydrogen bomb calculations ended. Their efforts on ways to streamline Monte Carlo computations led to a joint paper that was published in late 1953 by the Journal of Operations Research Society of America. In it the two authors offered ways of greatly reducing the large sample sizes such calculations had previously required.45

  Still yet another project found Marshall working with his RAND colleague Marc Peter to reexamine the data from the atomic attacks on Hiroshima and Nagasaki. In doing so they concluded that the vulnerability of steel-frame structures in the two cities had been overstated.46 This led the Air Force to rethink what it could expect to accomplish with its small inventory of a few hundred relatively low-yield fission bombs.*

  Marshall and Kahn also identified several other problems with RAND’s overall approach to analysis. They found RAND tended to rely too heavily on models whose abilities to provide “good-enough” approximations to reality were often suspect. Many RAND analysts also viewed the Soviets as being supremely rational in their decisions on strategic nuclear forces despite growing evidence to the contrary. And the two men agreed that RAND’s analysis was often woefully inadequate regarding the treatment of uncertainty.47

  Over time Marshall discovered that these issues were not unique to RAND, but were endemic in most individuals and organizations devoted to security studies. Whenever he encountered them, Marshall would challenge analyses that had fallen into these traps, especially the “rational-actor model” that assumed individuals and organizations made optimal strategic choices. And when at last he was put in charge of an analytic organization of his own, he would seek to purge it of these shortcomings, despite eventually admitting that “there is only so much stupidity one man can prevent.” Such limitations, however, did not prevent Marshall from trying.

  As Marshall’s reputation at RAND steadily grew, his personal life was undergoing major changes as well. In 1952 he had become engaged to be married. But when he met his bride-to-be in Paris at the end of November, she called off the wedding. Later, in the early 1960s, she married William Kaufmann, who had been one of Marshall’s RAND colleagues in the late 1950s. Returning to California, Marshall eventually recovered from the shock and developed a new romantic interest: an attractive young RAND employee named Mary Speer, who also happened to be Hitch’s secretary. Although there was no formal rule against dating other RAND staff members, Marshall was initially reluctant to approach Mary as they both worked in the same division. Eventually, however, he asked Mary whether she knew a Marjorie Speer, whose acquaintance he had made during his time in Chicago. Marjorie had recently moved to Los Angeles with her husband, Walter Richman. It turned out that Mary and Marjorie were sisters. Marjorie was invited to attend an early 1953 open house RAND held to inaugurate the think tank’s new facility on Main Avenue in Santa Monica. At the inauguration it was agreed that Andrew and Mary would come to dinner at the Richmans’ house in the San Fernando Valley, which they did. Shortly thereafter Marshall asked Mary out on a first date.

  From there things moved quickly. Marshall’s marriage to Mary Speer took place on September 12, 1953, with Herman Kahn as best man. Although neither Marshall nor his new bride could have known it at the time, their union would last for over half a century.

  Shortly after he and Mary tied the knot, Marshall took a leave of absence from RAND to return to the University of Chicago. The university now offered a PhD in statistics, and his former mentor, W. Allen Wallis, was still teaching statistics there. But much to his surprise, upon arriving in Chicago Marshall discovered that Wallis had been asked to head a Ford Foundation study to determine how best to allocate its grant-funding support to universities. As this required Wallis’s full attention, he would not be able to teach in the 1953–1954 academic year. Wallis asked Marshall to take on the classes in his absence, and he agreed.

  During two semesters at Chicago Marshall discovered two things. First, in taking over his mentor’s classes he came to understand that his knowledge of statistics was already at such an advanced level that he perhaps did not need to pursue the PhD after all. Second, Marshall found he missed both the intellectual stimulation and the sense of purpose that he had experienced at RAND. In Santa Monica he had been working on projects of great importance to his country—something that, he now realized, gave greater meaning and satisfaction to his work.

  As a result, Marshall chose not to remain in Chicago. By the spring of 1954 he had decided that tackling the real-world problems of nuclear strategy at the dawn of the thermonuclear missile age was far more interesting and rewarding than continuing his academic education. He and Mary returned to Santa Monica in April, eventually settling in a house on Kenter Avenue in the Brentwood Hills section of Los Angeles.

  No sooner had Marshall returned to Santa Monica than Hitch recruited him to join a small group of RAND’s top strategists known as the Strategic Objectives Committee (SOC). The dawning of the thermonuclear age had convinced Hitch that diagnosing its defining characteristics and strategic implications should be given high priority. Hitch formed the SOC to identify the major issues the think tank’s research on nuclear strategy and forces should pursue over the next ten years.

  One of the members of the SOC, Bernard Brodie, had argued only months after the atomic bombings of Hiroshima and Nagasaki that the sheer destructiveness of atomic weapons meant that the main purpose of the US military was no longer to win the nation’s wars but to deter them.48 Recognition that both the United States and the USSR would eventually be able to field large stockpiles of thermonuclear weapons only reinforced this view. Once the United States could no longer strike first and expect to escape Soviet nuclear retaliation, Brodie concluded, unrestricted nuclear war would be suicidal for both sides.49 Yet deterrence of
nuclear war hinged on being ready and able to inflict nuclear retaliation on the opponent. Given this dilemma, what new ideas and strategies should the think tank pursue in the years ahead to help the Air Force and the nation deal with the looming thermonuclear missile age?

  The committee members were a veritable who’s who of RAND’s leading thinkers. Besides Marshall, Brodie, and Hitch (the SOC’s first chairman), the group included Arnold Kramish; electrical engineer James Digby; Victor Hunt, a social scientist; James Lipp, an aeronautical engineer (the SOC’s second chairman); Alex Mood, a statistician; and John Williams (the SOC’s third chairman). Over time the group would also draw in Herman Kahn and economist Malcolm Hoag to support its efforts.