The various roles Marshall played in the Strategic Objectives Committee are a good example of the collaborative, multidisciplinary teamwork of many of RAND’s projects during the 1950s. John Williams was fascinated with the idea that game theory could help RAND analysts to anticipate possible Soviet responses in the event of a nuclear conflict. Toward this end, Alex Mood developed a Strategic Air Warfare game that allowed opposing teams of RAND researchers to play out US-Soviet conflicts that escalated to large-scale nuclear exchanges. Because of the natural desire of players on both sides to “win” in some sense, the teams began to judge their success based on where the front lines between NATO and Soviet forces in Europe ended up at the game’s end. Given this criterion for measuring success, opposing players tended to resort to nuclear weapons early and pay little or no attention to the damage their decisions wreaked on their respective homelands. Players on the US side found themselves sacrificing half of the American economy to Soviet nuclear attacks in order to prevent the Soviets from advancing a few more miles into West Germany and the Low Countries.
Marshall found the game neither realistic nor sensible because of the measure of military worth the players were using.50 Concentrating on where the front lines ended up in Europe ignored the damage each side’s nuclear weapons could inflict on the other’s homeland. Moreover, since 1945 American leaders had shown little inclination to resort to nuclear weapons in response to such aggressive acts as the Berlin Blockade and the invasion of South Korea. And Eisenhower had categorically rejected the option of using nuclear weapons in a preventive war to block the Soviets from fielding their own intercontinental nuclear arsenal.51
In an effort to address these problems Marshall and a colleague, Jack Hirschleifer, developed more complex measures of merit for game outcomes that included limiting the damage nuclear exchanges would inflict on either nation’s economy. This greatly improved the game’s realism, and thus its value. Despite the initial problems with Mood’s war game, Marshall felt that by playing games the researchers tended to develop good strategies so long as the games were well designed and played often enough. His conclusion was not to reject war games, but to insist that they be more carefully designed.
Another SOC project that Marshall became involved in during the second half of 1954 involved estimating the growth of the US nuclear stockpile over time based on public documents and congressional testimony. The true size of the stockpile was a closely guarded secret—or at least it was supposed to be. Kramish, who was a SOC member, knew what the actual stockpile numbers were. He insisted that someone like Marshall could develop good estimates based on open sources. In the end, Kramish proved right. Marshall took on the challenge and crafted a close estimate without any use of classified data. This would not be the last time that he was able to reach accurate conclusions with incomplete or even anecdotal data. Indeed, being able to reach sound conclusion on matters relating to national security based in minimal evidence has been one of Marshall’s main intellectual strengths, no doubt enabled at least in part by the breadth and depth of his knowledge on a wide range of issues bearing on the problem at hand.
The SOC’s final product, published at the end of December, was an internal RAND paper, “The Next Ten Years,” cowritten by Brodie, Hitch, and Marshall. In considering the lines of research on nuclear forces that RAND should emphasize through the mid-1960s, the authors began by focusing on the broad trends in weapons technology and delivery systems that would shape the US-Soviet competition in nuclear arms over the coming decade. First, they concluded that the rivals’ stockpiles would most likely grow to several thousand weapons apiece by the mid-1960s; second, that H-bombs with yields of up to 3 megatons would become light enough to be delivered by tactical aircraft and ballistic missiles; and third, barring unforeseen advances in missile defenses, within five to ten years both sides would begin fielding thermonuclear-tipped ICBMs, establishing the absolute superiority of the offense over the defense.52 In light of the anticipated Soviet nuclear buildup, the superiority the United States then enjoyed due to its virtual monopoly in long-range, or “strategic,” nuclear bombers would fade.53 Years later the value of identifying broad trends that could produce disruptive shifts in military competitions would emerge as a staple feature in the net assessments that Marshall’s Pentagon office would undertake. In light of the trends Brodie, Hitch, and Marshall saw in nuclear weaponry, “The Next Ten Years” concluded that deterring a surprise attack by the USSR would quickly become the essential, primary component of US strategic policy. In that case, a Strategic Air Command (SAC) impervious to a Soviet nuclear “Pearl Harbor” would be the keystone of the United States’ deterrence strategy.54
This finding triggered alarm bells in Santa Monica and beyond. Only a few months earlier another RAND analysis had concluded that the SAC bomber force was not, in fact, likely to be invulnerable to a surprise nuclear attack. In April 1954 Albert Wohlstetter, along with Robert Lutz, Henry Rowen, and Fred Hoffman, completed a study of SAC basing options. At the time the bulk of SAC’s bomber force was composed of approximately 1,600 medium-range B-47s and RB-47s. In addition, the Strategic Air Command had some 300 long-range B-36s and RB-36s, about a wing of B-52s, and over 700 KC-97 tankers.55 The basing scheme SAC had settled on for the period from 1956 to 1961 called for its bombers to be based in the continental United States (CONUS) in peacetime. In the event of nuclear attack, the bombers would deploy to preselected overseas bases. Wohlstetter’s Selection and Use of Strategic Air Bases (RAND report “R-266”) demonstrated that SAC’s bases in CONUS were growing increasingly vulnerable to atomic attack by Soviet bombers. The picture was even worse at SAC’s overseas operating bases, which were much closer to the Soviet Union.56 To minimize the risk to US bombers operating from forward bases, Wohlstetter recommended using these bases strictly for ground-based refueling of SAC bombers en route to their targets in the USSR.57 Unfortunately, the cost of pursuing this option for a force dominated by medium-range bombers was judged so high that it would drastically reduce SAC’s total striking power against an expanding Soviet target base.58 Fortunately, long-range B-52 bombers were entering the force. Over time they would displace the older bombers and Wohlstetter’s preferred option would become a reality.
Wohlstetter’s Selection and Use of Strategic Air Bases was one of RAND’s most influential analyses. Long before the study was completed Wohlstetter began sharing its preliminary findings with RAND’s Air Force sponsors. Starting in 1952, versions of R-266 were briefed over ninety times to SAC and the Air Staff.59 As a result, the study had an important influence on the Air Force’s strategic planning even before it was finished. By 1954 SAC was already moving away from its original plans to depend on overseas basing. Later, in 1959, the Air Force estimated that Wohlstetter’s assessment had saved the United States a billion dollars while enabling it to adopt a more secure retaliatory posture for SAC’s bomber fleet.60 In addition to Wohlstetter being an enormous asset to RAND and the US government, it would soon become clear just how crucial he was to Marshall as well.
In Wohlstetter Marshall found a mentor and friend comparable to Hitch and Kahn. Educated in mathematical logic and economics, Wohlstetter had joined RAND in 1951 as a consultant. His love of fine food, fine wine, classical music, and ballet made him one of the social pillars of the organization in the early 1950s. He and his wife, Roberta, also a member of RAND’s analytic staff, regularly hosted small groups of RAND strategists at their house in Laurel Canyon, and the Marshalls were regular guests at the Wohlstetters’ social gatherings.61
One of the enduring insights that Marshall gleaned from Wohlstetter’s work was the importance of choosing the right performance metrics, or analytic measures, to gauge the effectiveness of various courses of action. Choosing performance metrics can be exceedingly difficult, since at the outset of a study the analysts may not understand the problem well enough to separate what is important from what is not. Wohlstetter’s great “invention,” as Marshall lat
er called it, was to remain open to the possibility that an analyst’s initial intuitions about the critical factors in addressing a problem might be wrong and need revision as the problem became better understood.62 Prior to the basing study, most of RAND’s major analyses had focused on the composition of SAC’s bomber force. Edwin Paxson’s massive systems analysis, Strategic Bombing Systems Analysis (R-173), completed in March 1950, examined no fewer than 400,000 different bomb-bomber combinations, and calculated the effects of dozens of different variables, many of which were interdependent.63 The key metric of that study was to pick the most cost-effective bomber force. In contrast, Wohlstetter and his colleagues eventually realized during the course of their bomber basing study that the appropriate measures of effectiveness were not the composition or cost of SAC’s bomber force. Instead they turned out to be the distances from SAC bases to Soviet targets, the most favorable entry points for penetrating Soviet air defenses, the sources of supplies for SAC bases, and the locations from which the Soviets could attack US bases.64
Selecting the right metrics was a manifestation of what became known within RAND as the criterion problem. Marshall had already encountered this issue in Alex Mood’s Strategic Air Warfare game, in which the players’ initial metric of limiting the advance of Soviet forces into Western Europe had undermined the value of the game’s results. A more sensible metric was limiting the destruction visited on the United States by Soviet nuclear weapons. In the end, the criterion problem proved so central to RAND’s analyses that Hitch and his coauthor, the economist Roland McKean, later devoted over twenty pages to it in their influential 1960 book The Economics of Defense in the Nuclear Age.
As vital as the selection of appropriate decision criteria was in systems analysis, it had little to offer in helping Brodie, Hitch, and Marshall decide what directions RAND’s strategic research should take over the next ten years. After considering some of the requirements of containment and the central role of deterrence in US nuclear strategy, “The Next Ten Years” turned to the vexing questions of American options under the assumption that the Cold War would be an “essentially permanent” feature of America’s security environment. If, as some believed, all-out war with the Soviet Union was practically inevitable within a generation, should the United States consider a preventive war before the Soviets could field intercontinental nuclear systems? If not, then what policies should or could the United States adopt to compete more effectively and efficiently against the Soviet Union? What might US objectives be in the context of a nuclear competition based on deterrence? How might the United States prepare for the possibility of limited and peripheral wars while exploiting its own economic and technological advantages? Could the United States hold the NATO alliance together in the face of persistent Soviet efforts to fragment it?
Suggesting even tentative answers to these questions might have provided some direction for RAND’s future research on the US-Soviet competition in intercontinental nuclear forces. But Brodie, Hitch, and Marshall instead emphasized the substantial uncertainties involved in trying to answer them. Their emphasis on asking the right questions, even if good answers were not immediately apparent, would prove to be vintage Marshall. In his eyes, whether in 1954 or today, identifying the right questions—those that matter most in determining strategic advantage—is an essential first step in any comparative analysis, including net assessment.
Nevertheless, “The Next Ten Years” did offer some advice for RAND’s leadership. Contrary to much of the prevailing wisdom at RAND about Soviet decision-making, the paper emphasized that Soviet leaders were not necessarily the all-seeing, all-knowing, fearless adversaries that senior US civilian and military decision makers generally imagined them to be. “The Next Ten Years” also concluded that it was important to identify Soviet weaknesses and exploit them.65 Both were insights that Marshall would continue to develop and expand upon in the decades that followed, starting with his collaboration over the next several years with Joseph Loftus on Soviet organizational behavior.
In the end, however, “The Next Ten Years” concluded that in “policy matters even current strategies are not always easy to discover and describe; and while future strategies will somehow emerge out of present ones (and thus be affected by present habits), they will very likely do so in quite erratic fashion.”66 The best Brodie, Hitch, and Marshall could offer concerning the course RAND’s strategic analysis should pursue over the next decade was to hope that RAND’s best analysts could help shape American nuclear strategy by making “policy makers aware of the facts and implications of weapons developments and by undertaking research projects designed to throw light on strategic choices.”67 In other words, good diagnoses based on asking the right questions were key to gaining competitive advantage and helping US leaders make better strategic choices.
Marshall and his two coauthors appear to have been satisfied with the results of their efforts. John Williams, however, was not. Williams wanted more than an identification of the most important questions bearing on RAND’s future research on nuclear strategy. He wanted detailed, actionable answers. Williams saw three of the brightest, best-informed strategists in America devoting their efforts to identifying issues and questions rather than addressing and answering them.68 As he wrote in a September 1954 memo to Brodie, Hitch, and Marshall, “Chums, it seems to me that you have laid an egg—an elegant, ellipsoidal, academic egg to be sure, but nevertheless one which can serve henceforth as the model for the cipher.”69
An increasingly acerbic exchange between Williams and Brodie followed Williams’s initial criticism of “The Next Ten Years.” In the ensuing flurry of memos Brodie adamantly rejected Williams’s attraction to mounting a preventative war again the USSR. Marshall came away from the experience feeling that Williams had missed the point. The object of the exercise was not to provide answers to the US defense establishment, but rather to identify the key issues upon which RAND should focus its efforts over the coming decade. Clearly evident in “The Next Ten Years” was Marshall’s growing sense that the first priority should be identifying the right questions rather than elaborating precise or elegant answers.
For Marshall, the 1950s at RAND were a period of tremendous intellectual development, most of which came from a combination of self-education and the good counsel and encouragement he received from such men as Hitch, Goldhamer, and Wohlstetter—as well as his interactions with peers such as Herman Kahn and, later, Joseph Loftus. But as he grew intellectually Marshall also began mentoring others. Some of this occurred naturally, as colleagues began to approach him for guidance. Yet it was his remarkable generosity with his time and patience with those individuals in whom he saw promise that encouraged them to produce some of the more insightful works that emerged from the US strategic studies community during the second half of the twentieth century.
Over time the effects of Marshall’s “hidden hand” mentoring would become so pronounced that they were impossible to ignore. But in the 1950s his influence was more subtle and limited—although he did have a lasting effect on some of the people whose lives he touched during this time. One of the first to benefit from Marshall’s good counsel was Roberta Wohlstetter.
Because Marshall and the Wohlstetters were close socially as well as professionally, it was only natural that in 1957, when Roberta was searching for ideas as to what might constitute promising topics for new research, she approached Marshall for advice. The result was Pearl Harbor: Warning and Decision, which explored and assessed the causes of the US intelligence failures that enabled Japan’s successful surprise attack on Pearl Harbor in 1941. In her introduction Wohlstetter wrote: “The initial stimulus for the book came from my friend, Andrew W. Marshall.” She went on to say that she was “deeply grateful to him” for his “constant encouragement and advice through five years of research.”70 The book was published to great acclaim. In adding Pearl Harbor to its professional reading list at the time of its publication, the US National Security Agency declar
ed it “the clearest exposition of the subject that has yet been published” while concluding that it “is one of the most thorough analytical studies of the events leading up to any war and will probably become the book on the controversial question of the surprise attack on Pearl Harbor.”71
One of the many insights that emerged from the book concerned MAGIC, the process by which US code breakers were able to decode Japan’s secret diplomatic messages. Wohlstetter showed how the necessity for preserving the extreme secrecy associated with MAGIC undermined efforts to share the intelligence it produced on the possibility of a Japanese attack. She explored, in all, fifteen different intelligence signals that in retrospect clearly seemed to indicate an attack on Pearl Harbor was forthcoming. Yet the attack came as both a strategic and tactical surprise. In explaining why, Wohlstetter concluded that some intelligence failures are inevitable because of the difficulty in distinguishing the accurate intelligence “signals” from the background “noise” of many intelligence inputs.
In part Marshall had suggested strategic warning as a topic for Wohlstetter because he had done a similar study in 1952 on the strategic warning problem facing NATO in Central Europe. He later recalled that in itself Pearl Harbor didn’t influence his “own subsequent thinking very much, because the main conceptual ideas were pretty clear [to me] much earlier.”72 In fact, Wohlstetter’s book brought into the public eye Marshall’s growing concerns over the role that organizations play in decision-making as well as the need to avoid making analytically convenient but unwarranted assumptions that ignore the factor of uncertainty. As Wohlstetter wrote, “If the study of Pearl Harbor has anything to offer for the future, it is this: We have to accept the fact of uncertainty and live with it.”73
In his foreword to Wohlstetter’s book, Nobel laureate Thomas Schelling echoed her concerns about the tendency “to focus on a few vivid and oversimplified dangers.” Rather, wrote Schelling, “The planner should think in subtler and more variegated terms and allow for a wider range of contingencies.” The consequences for policy makers who fail to do so, he pointed out, “are mercilessly displayed in this superb book.”74
The Last Warrior Page 7
