The Last Warrior

by Andrew F. Krepinevich

  When Schlesinger arrived in Santa Monica he was assigned to Marshall. Schlesinger found RAND to be a “revelation,” and working with Marshall a “delight.”26 That summer, the two men spent many long hours discussing all manner of issues. In 1963 Schlesinger joined RAND, abandoning academia. Over time he and Marshall became intimate friends, both socially and intellectually. By the early 1970s both men would migrate to Washington, DC, and their friendship would continue until Schlesinger’s death in 2014.

  Once at RAND, Schlesinger quickly became engaged in the efforts of Marshall and others to develop analytic methods that transcended the limitations of systems analysis. In September 1963 Schlesinger and Richard Nelson, who with Sidney Winter later became leading figures in the revival of evolutionary economics, proposed a new long-range research program for RAND’s economics department centered on the theory of organizations.27 They acknowledged the important advances that RAND’s development of quantitative methods had achieved during the 1950s. In the case of the Defense Department, they observed, the battle for quantitative research had largely been won due to the “McNamara revolution.”28

  In a 1967 RAND paper Schlesinger reiterated the importance of quantitative methods by stating unequivocally that the “usefulness of systems analysis has been amply demonstrated.”29 Yet Schlesinger argued that it also remained true that such studies “inevitably incorporate a number of non-technical assumptions” in setting the scope of the problem, selecting the alternatives to be examined, and choosing criteria for measuring the alternatives’ effectiveness.30 Simply put, one must exercise diligence in selecting the assumptions that underlie any analysis. Recurring questions for Marshall and Schlesinger were, “What are the proper metrics to apply in this case? Have we framed the analysis to address the right questions?” To be sure, the costs of each alternative could be reduced to dollars. But what about the effectiveness of the alternatives measured against such goals as deterring a Soviet nuclear attack on the United States or its allies? “For higher-order problems,” Schlesinger observed, “the measures of effectiveness must be based on some broad strategic criteria, in which political or psychological assessments inevitably play a role, whether this be admitted or not.”31 For Schlesinger and Marshall, the framing of any systems analysis would always elude attempts at rigorous quantification and necessarily involve human judgment.

  Schlesinger, then, was “an unabashed, if qualified, defender of the value of [systems] analysis in policy formulation.” He was willing, he said, to give systems analysis two-and-a-half cheers but not three.32 By the late 1960s the limitations and distortions of systems analysis within the Pentagon were evident to others at RAND beyond Marshall and Schlesinger. Later, OSA veterans Alain Enthoven and Wayne Smith would acknowledge that systems analysis under McNamara could not claim total objectivity or infallibility and had made its share of mistakes.33 In Marshall’s opinion, if the Office of Systems Analysis had committed an original sin in the 1960s, it was in trying to extend systems analysis to decide issues that ultimately involved qualitative factors—the higher-level political and military professional judgments that are inevitably involved in strategy formulation and implementation. Indeed, Wohlstetter’s adoption of an iterative approach to his bomber-basing study, one that explicitly sought to limit and mitigate the effects of uncertainty about the future, was in Marshall’s view an example of good strategic analysis: it took into account such nonquantitative factors as the distinctive goals, resources, cultures, and strategies of the competitors; contextual elements, such as technology or climate beyond the control of the competitors; long-term trends in many of these variables; enduring asymmetries between the participants that provided the basis for identifying, developing and exploiting areas of competitive advantage; and the risks that uncertainties about the future posed, particularly in peacetime. But Marshall wanted to do more than simply catalog the limitations of systems analysis. He wanted to develop analytic tools that could better take into account the higher-level aspects of strategic choice.

  By the time that Nelson and Schlesinger proposed that RAND’s economics department initiate a program on organizational theory, Marshall had begun exploring the emerging literature on the behavior of formal organizations such as commercial business firms. Here he was struck by the seminal work of James March and Herbert Simon. The unifying construct in their 1958 classic, Organizations, was “not hierarchy but decision making, and the flow of information within organizations that instructs, informs, and supports decision making processes.”34 While it was natural to assume that an organization’s decisions generally were made at the top and flowed down its formal chain of command, in reality informal bureaucratic structures also served to process information that influenced the organization’s strategic choices.35 Such findings confirmed Marshall and Loftus’s growing doubts about the ability of any large organization to act as a unitary, rational actor.

  March collaborated with Richard Cyert in another classic work that influenced Marshall: A Behavioral Theory of the Firm, which was first published in 1963. In it the authors concluded that the decisions of senior executives often were based on relatively simple rules, rather than the product of detailed analysis.36 Moreover, they discovered that firms were composed of individuals and suborganizations whose priorities often varied, leading to competition and negotiation among them in the course of generating information and options on which decisions would be based. They also found that, in part because of these factors, firms typically engaged in “satisficing”—making decisions that were “good enough” rather than optimal.

  Simon had been pursuing the notion of bounded (limited) rationality since the early 1950s when he was a RAND consultant.37 Looking to apply this kind of thinking to the US competition with the Soviet Union, Marshall approached Simon in 1963 to undertake the kind of research into organizational behavior that Schlesinger, Nelson, and others advocated. Simon, however, declined. Marshall wanted to continue pursuing this line of research with others, but he was diverted by another RAND project.

  McNamara had requested RAND’s help in preparing for a ten-year review of NATO’s defense posture. The defense secretary’s request arose from an agreement he reached with NATO’s secretary general, Dirk Stikker. They had decided that RAND would supply analytic support for the review over an eighteen-month period stretching from late 1963 to early 1965.38 The team would be based at NATO headquarters in France.

  The small group deployed to NATO in response to McNamara’s request included Marshall, Fred Hoffman, and Oleg Hoeffding from RAND along with Peter Szanton from the Pentagon’s policy planning staff for International Security Affairs (ISA).39 Burton Klein, who had succeeded Hitch as head of RAND’s economics department in 1961, was asked to lead the group. Klein had worked on the US Strategic Bombing Survey for John Kenneth Galbraith and in 1959 had published the widely cited book Germany’s Economic Preparations for War, which dispassionately debunked the myth that Germany had prepared for total war prior to invading Poland in 1939, or even immediately thereafter.40 For the NATO mission, Klein left RAND to serve as a special assistant to McNamara.41

  Participation in the NATO mission precluded Marshall from launchinga RAND research program aimed at informing forecasts of the USSR’s future nuclear posture with the observed patterns of past Soviet organizational behavior. Not until after he returned from Paris in 1965 was Marshall able to secure Collbohm’s agreement that he would be able to pursue this line of research. In the meantime, however, the NATO mission gave the Marshalls an opportunity for an extended stay in France.

  The Marshalls had spent some time in France in the fall of 1960 enjoying French cuisine and doing some touring with the Hitches, an experience they had greatly enjoyed. Mary Marshall was therefore delighted to accompany her husband back to France in 1963. They rented an apartment close enough to NATO’s political headquarters at Porte Dauphine in Paris that Marshall could walk to work. Their life in France during these two years was filled with good f
ood, good wine, and pleasant living. As Marshall later recalled, this time may well have been the happiest of their married lives.42

  Marshall’s work at NATO headquarters proved less satisfying. The RAND contingent was asked to perform three studies, one of which Marshall ended up leading in collaboration with a succession of French colonels. Its purpose was to determine whether the United States’ NATO allies were fulfilling their military commitments to the alliance. More specifically, did they have the agreed-upon numbers of tanks and ammunition stocks? Marshall was astonished to discover that the West Germans, for example, only had three days’ worth of ammunition despite a requirement to maintain at least forty days’ worth.43 Marshall was appalled that NATO could support a large, well-paid bureaucracy and yet permit its members to essentially ignore their military obligations. He took this discovery as yet another piece of evidence that organizations more often than not fell far short of the utility-optimizing choices associated with the rational-actor model of decision-making.

  Marshall returned to Santa Monica in the spring of 1965 and renewed his efforts to apply advances in organization theory to security studies. Besides digging into the emerging literature he began looking into business strategy, thinking it could be applicable to military strategy. Good business leaders, he knew, try to exploit the strengths of their firms so as to capture markets and drive competitors out of specific product lines and businesses.

  Marshall also began reaching out to some of the leading lights at the Harvard Business School. Among his earliest contacts were the business strategists Joseph Bower and C. Roland Christensen. He had first met Bower in the summer of 1963 while he was visiting RAND.44 During 1966 and 1967 Bower would become involved with Marshall in RAND projects that dealt with organizational and bureaucratic behavior. In 1967 Marshall recruited both Bower and Christensen to participate in an effort for the Office of Systems Analysis to improve intelligence forecasts of Soviet forces based on the organizational behavior of USSR’s bureaucracy.

  Once back in Santa Monica Marshall’s discussions with Schlesinger on organizational behavior and decision-making resumed. Their conversations focused on recent research into the cognitive limits of rationality, conflicts within organizations, motivational constraints to participation in decision-making, innovation, and comparative advantage. They also became interested in ethology, the study of animal behavior under natural conditions. Among other works in this area they read and discussed were Robert Ardrey’s 1966 The Territorial Imperative: A Personal Inquiry into the Animal Origins of Property and Nations, and Konrad Lozenz’s 1963 On Aggression, which first appeared in English translation in 1966.45 One of the central themes in The Territorial Imperative was that the propensity to acquire and possess territory, which was manifest in the behavior of many animals, was also manifest in man and very likely played a role in warfare.46 Similarly, Lorenz argued that in both beast and man, aggression directed against members of the same species must be considered a part and consequence of natural selection, especially when individuals compete over resources.47

  Marshall came to view ethology as offering a more fruitful way of thinking about what drove human behavior than the utility-maximizing, rational-actor construct—homo economicus—so beloved by many economists. The nonrational aspects of human behavior, he concluded, could not be ignored in defense studies and strategy formulation. They had implications for the deterrence of nuclear war with the Soviet Union as well as for the continuing peacetime competition between the two superpowers.

  Not long after Marshall’s return from Paris, Henry Rowen approached him and suggested he meet with Richard Neustadt, a political historian who specialized in the US presidency. Neustadt was engaged in researching the way that presidents persuade and bargain with other power centers in the federal government to influence policy formulation and, especially, its implementation. He concluded that presidents must bargain to influence not only other branches of government (particularly Congress), but also various power centers within the executive branch itself. For Neustadt, “Presidential power is the power to persuade.”48

  Marshall’s meeting with Neustadt in 1965 led to creation of a group chaired by Harvard’s Ernest May, a historian specializing in US foreign policy. The May Group, as it came to be known, began meeting in the spring of 1966 to discuss the influence of organizations and bureaucracies on policy. The participants were especially interested in the gap between the intentions of government decision makers and the results of governmental actions. The May Group held a series of seminars examining the rational actor, organizational process, and governmental politics models of decision-making (Models I, II, and III respectively). In addition to Marshall, May, and Neustadt, the group included such luminaries as Morton Halperin, Fred Iklé, William Kaufmann, Don Prince, and Rowen. It also included a young student, Graham Allison, as rapporteur. Allison was searching for a dissertation topic and quickly became “hooked” on the problem of closing the gap between what policy makers desired to achieve by their decisions and how their intended goals tended to be distorted in the course of implementation. Supplied “with more ideas than I could assimilate” from the May Group seminars, Allison completed his doctoral dissertation in 1968, and began transforming it into the influential book Essence of Decision, published in 1971. In it Allison employed the 1962 Cuban missile crisis as a case study to explain how governments both make and execute decisions in ways often at odds with what, after the fact, appears to be rational behavior, both to their rivals and, in many cases, to their own leaders.

  In Essence of Decision Allison contended that many international relations studies assumed states and policy makers’ actions are based on their evaluation of all available options. He argued that this rational-actor approach (Model I) is flawed—that one must also examine the effect of organizational (Model II) and bureaucratic actors (Model III) on how decisions are made and executed. Model II—what Allison termed the “Organizational Process” view of decision-making—derives from the fact that “governments perceive problems through organizational sensors. Governments define alternatives and estimate consequences as their component organizations process information.” Allison concluded that “few important issues fall exclusively within the domain of a single organization. Thus government behavior relevant to any important problems reflects the independent output of several organizations, partially coordinated by government leaders.”49 Consequently the rational-actor model alone could not adequately explain the decisions in Washington and Moscow during the Cuban missile crisis.

  Allison cited Roberta Wohlstetter’s analysis in Pearl Harbor: Warning and Decision as an example of organizational process (Model II) at work.50 Prior to the attack on Pearl Harbor, he noted, Japan’s senior military and civilian leaders responsible for making the decision to go to war knew their country lacked the industrial capacity and military might to prevail in a protracted conflict with the United States. Yet they proceeded to attack anyway. Again referencing Wohlstetter, Allison commented that US leaders had sufficient intelligence to indicate that Japan was about to attack Pearl Harbor, but failed to understand what they had and respond accordingly. When, for instance, the War Department warned Lieutenant General Walter Short, the Army’s senior commander in Hawaii, to be prepared for “hostile activities,” Short interpreted the warning to mean acts of sabotage, and not an attack by Japan’s main forces.51

  Like most groundbreaking works Essence of Decision was far from immune from criticism. Some noted that the information required to make and execute good decisions required by Allison’s Models II and III would be so large as to be impractical during a crisis. Allison conceded as much, but also pointed out that this is hardly an argument for relying solely on the rational-actor model. In this regard he restated Wohlstetter’s (and Marshall’s) warning that one cannot simply discount key aspects of analysis simply because they cannot be easily quantified or explained.

  Essence of Decision also seconded Roberta Wohlstetter’s
concerns that the concept of mutually assured destruction might not serve as a barrier to nuclear war, and that intelligence failures similar to those that led to the surprise at Pearl Harbor might also occur in the nuclear age, “with possible consequences of an ever greater and perhaps more fatal magnitude.”52 Similarly, Allison argued that the organizational and bureaucratic models of decision-making could lead to states making irrational decisions or executing decisions in unintended ways, thereby undermining the fundamental assumption that a rational decision maker would never order a nuclear attack that would result in his or her own country’s destruction. Simply put: despite the efforts of a state’s leaders to act rationally, one could not discount the prospect that, like Imperial Japan in December 1941, they could could choose the path of nuclear war even though it would be tantamount to “committing suicide.”

  Allison cited Marshall as one of four individuals who “deserve special note for the intellectual and personal debt” he had incurred in writing Essence of Decision. He went on to state, “The impact of ideas that Andrew W. Marshall has been propagating for a decade is marked, especially in my chapter on Model II.” That said, Allison’s thinking in Essence of Decision had a somewhat different focus from Marshall’s. Whereas Allison concentrated on alternative models of decision-making in crises potentially leading to war, Marshall’s main concern was with decision-making in protracted peacetime competitions. As he stressed to RAND’s board of trustees in April 1968, a principal objective of RAND’s studies on organizational behavior was aimed at improving the United States’ ability to enhance its estimates of the USSR’s future military posture, thereby providing a firmer basis for choosing weapons and force postures to deter conflict.53 As we will see, Marshall’s emphasis on decision-making in peacetime would lead him to develop a new perspective on the competition between Washington and Moscow.