The Last Warrior

by Andrew F. Krepinevich

  These concerns became the focus of one of the panels of a 1978 Defense Science Board (DSB) “summer study” held at the Naval War College in Newport, Rhode Island. Gene Fubini, Brown’s mentor, chaired the study. Fubini was a physicist and electronics engineer who had used his insights and understanding of microwave technology to help the Army and Navy jam enemy radars during World War II. Later, starting in 1961, he had been the principal manager of the Pentagon’s $7 billion research and development budget. And from 1963 to 1965 he served as an assistant secretary of defense.

  Fubini was asked to make recommendations on two issues: improving NATO effectiveness through collaboration on armaments; and the state of the US-Soviet strategic nuclear balance. He organized the participants into four panels, one assigned to examine NATO, and the other three the strategic nuclear balance. The three nuclear panels examined requirements for nuclear systems, the nuclear-relevant technologies, and perceptions and measures of the strategic nuclear balance.

  Marshall chaired the last of these panels. His group explored US perceptions and analytic measures of the strategic nuclear competition; allied perceptions of the competition; Soviet perceptions of deterrence; and the operational performance of nuclear forces in the event that deterrence failed. Participants and contributors to Marshall’s panel included Walter Slocombe and Leon Sloss, John Battilega, Bruce Bennett and Fritz Ermarth from RAND, Paul Wolfowitz and Thomas Brown from PA&E, Allan Rehm and Sayre Stevens from the CIA, Henry Rowen (Stanford), John Steinbruner (Yale), Major General Jasper Welch (Air Force Studies and Analysis), and Paul Nitze. Besides Marshall and Roche, ONA was represented by Air Force lieutenant colonel Frederick Giessler and Peter Sharfman, who had written ONA’s 1977 assessment of the US-Soviet strategic nuclear balance.*

  Marshall’s summer study panel found that US analyses of the nuclear balance focused on comparing US and Soviet intercontinental nuclear forces in isolation from theater nuclear and conventional forces located in overseas theaters such as Central Europe. By contrast, Soviet writings had made it increasingly evident that the Soviet General Staff’s estimates of the military balance, or “correlation of forces,” were organized on geographic theaters of military action (teatry voennykh deistvii, or TVDs) and strategic directions emanating from Moscow. Thus Soviet assessments of the military balance between the two superpowers included forces that US assessments of the strategic nuclear balance did not. This was but one of many differences between the two sides’ assessments. As Ermarth concluded, “Soviet thinking about strategy and nuclear war differs in significant ways from our own.”58 This confirmed what Marshall had long believed: Soviet methods for calculating the military balance did not mirror those of their US rival. Yet, if the principal goal of US strategy was to deter Soviet acts of aggression or coercion, it was the Soviet view of the balance that mattered most. Also, as Marshall had long maintained, US objectives remained vague. Even more troubling, both static and dynamic measures of the nuclear arms race ignored important factors bearing on the balance such as command, control, communications, and intelligence (C3I) vulnerabilities, active and passive defenses, the ability to sustain a fully generated nuclear posture during a protracted crisis, and the survivability of national leaders once the nuclear threshold had been crossed. How, he argued, could one accurately assess the nuclear balance without taking these factors into consideration?

  Marshall eventually embraced Thomas Brown’s suggestion that a war-gaming approach might enable qualitative and decisional factors to be integrated with quantitative metrics. Brown was the deputy assistant secretary of defense for strategic programs in PA&E. Like Marshall, he had participated in RAND’s Strategy and Force Evaluation (SAFE) games in the 1960s, which were essentially force-posture planning exercises played out over a period of ten years. Based on this experience, Brown felt strongly that such exercises, properly crafted, could incorporate key planning factors such as a wide range of nuclear crisis scenarios, various warning times, C3I vulnerabilities, poststrike intelligence, and reconstitution capabilities, all of which traditional analyses of the nuclear balance largely neglected. Thoughtful war gaming, Brown believed, could make a valuable contribution toward better assessments of the US-Soviet competition in nuclear forces.

  Marshall and Brown were by no means the only ones in the Pentagon who were dissatisfied with existing measures and methods for assessing the strategic nuclear balance. Major General Jasper Welch, then head of Air Force Studies and Analysis, was another who found existing metrics of the nuclear balance inadequate. In fact he became so distraught over the staleness of the intellectual approach his own organization was taking to strategic force adequacy that he disbanded the whole enterprise.59

  Andrew Walter Marshall and his younger brother Frederick John (“Jack”) Marshall, Detroit. SOURCE: ANDREW MARSHALL.

  The mushroom cloud from the first successful test of an experimental thermonuclear device. Designated Ivy-Mike, the detonation took place at 07:15 A.M. local time on November 1, 1952, at the Enewetak Atoll in the Pacific. The explosion had an estimated yield equivalent to over 10 million tons of TNT and the mushroom cloud rose to an altitude of 57,000 feet in less than 90 seconds. Ivy-Mike ushered in the thermonuclear age in which both the United States and the Soviet Union were eventually able deploy thousands of thermonuclear weapons. PHOTO COURTESY OF NATIONAL NUCLEAR SECURITY ADMINISTRATION/NEVADA FIELD OFFICE.

  RAND’s headquarters at 1700 Main Street, Santa Monica, California. The building was designed by H. Roy Kelly and completed in 1953. The aerial image looks west towards the Santa Monica beach and the Pacific Ocean. The main entrance is shown in the insert (upper left). John Williams, head of the mathematics department, suggested a layout that would encourage chance interactions among RAND’s researchers. PHOTOS COURTESY OF THE SANTA MONICA PUBLIC LIBRARY IMAGE ARCHIVES.

  Left to right: Mary Beth Weicking, Mary and Andrew Marshall, and Herman Kahn. Marshall and Mary Speer were married in September 1953, just before Marshall returned to the University of Chicago to decide whether to pursue a Ph.D. in statistics. Weickling was the maid of honor and Kahn the best man. SOURCE: ANDREW MARSHALL.

  In 1979, as part of the Strategic Air Command’s readiness exercise Global Shield 79, two Minuteman III ICBMs were launched twelve seconds apart from Vandenberg Air Force Base in California toward targets in the Marshall Island’s Kwajalein Atoll. PHOTO COURTESY OF THE NATIONAL ARCHIVES, STILL PHOTO BRANCH, PHOTO IDENTIFIER 6362313.

  Six unarmed Minuteman III Mark 12 reentry vehicles are shown approaching targets near Kwajalein Atoll in the Western Pacific Ocean from the two Minuteman III ICBMs launched as part of Global Shield 79. Despite having invested over $150 billion in ballistic missile defense since President Reagan announced his Strategic Defense Initiative in 1983, US capabilities to defend against more than a handful of rudimentary ballistic missiles remain limited at best. Russian military officials claim that their road-mobile RS-24 Yars heavy ICBM can penetrate even the most sophisticated missile defenses anywhere in the world. PHOTO COURTESY OF THE NATIONAL ARCHIVES, STILL PICTURE BRANCH, PHOTO IDENTIFIER 342-B-08-16-3-K69772.

  Evening discussion at the Wohlstetters’ house, 1958. Left to right: Daniel Ellsworth, Henry Rowen, Andrew Marshall, Albert Wohlstetter (with back to the camera) and Sidney Winter. PHOTO COURTESY OF LEONARD MCCOMBE/TIME LIFE PICTURES/GETTY IMAGES.

  This photo appeared in Jay Winik’s article “Secret Weapon” in the April 1999 issue of The Washingtonian. The article discussed how members of St. Andrew’s Prep frustrated defense secretary William Cohen’s decision to move ONA from the Pentagon to the National Defense University as part of his Defense Reform Initiative. As Winik summed up the inadvisability of doing so, “Without Andy Marshall, the Pentagon would be depriving itself of its principal source of wisdom and giving license to people who would play all sorts of con games with the Secretary.” PHOTO COURTESY OF JAMES KEGLEY.

  1999 ONA summer study participants in front of the original Naval War College build
ing in Newport, Rhode Island. Standing (back row): Eliot Cohen, Lionel Tiger, Jim Roche, John Bonsell, Andrew Krepinevich, Jim Callard, Steve Rosen, Chip Pickett, Andrew May, Jeff McKitrick, Mike Vickers, David Spain. Front row: Barry Watts, Sam Tangredi, Andy Marshall, Dakota Wood. PHOTO: BARRY WATTS.

  Andy Marshall and Ann Smith Marshall at a November 2006 reception following their marriage. PHOTO: MIE AUGIER.

  Donald Rumsfeld honoring Marshall with the DoD Distinguished Public Service Award for his leadership on defense transformation. December 14, 2006. Dmitri Ponomareff reported that Marshall was the only recipient at this ceremony who received a standing ovation from those present. PHOTO: OFFICE OF THE SECRETARY OF DEFENSE.

  Dinner during a conference on the past, present and future of net assessment, March 28, 2008. Marshall largely selected the attendees for this conference. The price of admission was to give a paper during the conference. Left to right: Jim Roche, Fred Giessler, Mie Augier, Michael Pillsbury, Andrew May, Charlie Pease, Gerry Dunne, John Battilega, Andy Marshall, Mark Herman, Lionel Tiger, Dmitri Ponomareff, Jaymie Duran, Lance Lord (behind Durnan), Karl Hasslinger, Donna Hasslinger, Phillip Karber, Jan van Tol, Charlie Wolf (in front of van Tol), Barry Watts, Enders Wimbush (behind Watts), Steve Rosen, Aaron Friedberg, David Epstein, James Schlesinger, Diego Ruiz Palmer, Wick Murray, Jeff McKitrick, Chip Pickett. PHOTO: BARRY WATTS.

  President George W. Bush honored Marshall with the Presidential Citizens Medal on December 10, 2008. WHITE HOUSE PHOTO BY CHRIS GREENBER.

  Following the 1978 DSB summer study, Marshall began taking steps to implement his panel’s findings. His main thrust centered on building a consensus within the Pentagon in favor of developing more advanced war-gaming methods aimed at improving assessments of the US-Soviet nuclear competition. Marshall, along with Brown, found support among key members of the military, including Generals Richard Lawson (Joint Staff), Jasper Welch, Edward “Shy” Meyer (Army deputy chief of staff/Operations and Plans), and Admiral William Crowe (deputy chief of naval operations/Plans, Policy and Operations). The idea that emerged was to let two or three small contracts to analytic organizations such as RAND to refine thinking as to how such a gaming system might be structured, organized, and managed. To succeed the enterprise would require the defense secretary’s support. In an April 1979 memo Marshall informed Harold Brown that the effort would be a multiyear undertaking and would need some funding from the defense secretary’s special study fund.60 Brown promptly agreed, commenting that “this does need to be pushed.”61

  RAND and Science Applications International Corporation (SAIC, formerly SAI) emerged as the leading contenders to develop the new analytic tool. Marshall found RAND’s conceptual approach far more ambitious than SAIC’s in terms of incorporating state-of-the-art artificial intelligence to produce automated software agents against which human players could, if they desired, play rather than competing against human players.62 Marshall believed that SAIC’s approach would be more responsive to the project’s central aim of improving US assessments of the strategic nuclear competition.63 But most of the others involved in the selection process favored RAND. Marshall, against his better judgment, acquiesced. It was a decision he would come to regret.

  The result was the RAND Strategy Assessment System (RSAS). The RSAS was a “system for analytic war gaming that could be used as a closed model or with one or more human teams playing in a partially or completely automated environment.”64 As Marshall explained to Harold Brown,

  As I have mentioned to you in the strategic balance assessments that we have prepared in the past, our capabilities for analyzing and understanding the strategic balance are limited by the narrowness of the kinds of analyses that can now be conducted. For one thing the focus tends to be on large exchanges. Very little that happened before them (for example, crises, LNOs [limited nuclear options], escalation from regional theater war) and little about what may happen after them (recovery and reconstitution of command and control, the use of residual strategic forces) is examined. In addition, while we say that our primary objective is deterrence, the analyses that we carry out are not done in a way that can reflect the Soviet perspective. It’s likely that the Soviets do not single out the strategic forces for separate treatment in the way that we have traditionally done so. Moreover, their definition of strategic forces undoubtedly includes a number of their medium range as well as their intercontinental forces.65

  In the end, however, the RSAS drifted away from Marshall’s original goal of improving US analyses of the strategic nuclear balance. Early in the system’s development RAND began concentrating on conventional combat. By 1985 the RSAS included detailed modules for five combat theaters: Northern Europe or the USSR’s Northwestern TVD, Central Europe (Western TVD), Southeastern Europe (Southwestern TVD), Southwest Asia (Southeastern TVD), and Northeast Asia (Far Eastern TVD).66 The focus of the RSAS’ theater modules in playing regional conflicts gradually overshadowed and lost sight of the project’s original purpose. Besides the Office of Net Assessment, RSAS users eventually included two Joint Staff directorates responsible for strategy and force planning, the National Defense University, PA&E, the Air University, the Naval War College, CIA, DIA, the Naval Postgraduate School, the Army’s Concepts Analysis Agency, and the US European and Pacific Commands. Because most of these organizations also began contributing funding and were mainly interested in conventional warfare, the RSAS development was drawn increasingly toward focusing on nonnuclear regional conflicts rather than nuclear war.

  Marshall tried to get the program back on track in late 1985. In a December letter to the head of the RSAS program at RAND, Paul Davis, he reaffirmed that its central objective to develop an analytic tool that could address the problems of global nuclear warfare. “There are plenty of models of particular theaters,” he added, “but what we have lacked is both a war game, or other analytic tool, that will allow us to look effectively at global warfare. This to me implies the highest priority in the [RSAS] development is that of the strategic nuclear [warfare] and enough of the theater modeling to allow us to overcome the existing imperfections in our analysis.”67 But Marshall’s guidance had little effect in the long run. The momentum toward concentrating on the theater modules was too great for even the director of net assessment to reverse.

  Not surprisingly, the limitations of the traditional static and dynamic measures of the US-Soviet strategic nuclear balance had their parallels in Marshall’s efforts to assess the balance between NATO and Warsaw Pact forces in Europe. Once Project 186 was converted to an ONA research project under BDM’s Phil Karber, the first order of business was to develop a database on the opposing forces in which the same counting rules were applied to both sides. Karber’s team at BDM quickly learned that this seemingly straightforward task was anything but. Eventually, though, Project 186 produced “probably the most authoritative NATO–Warsaw Pact database in the West.”68 Its development owed much to Marshall’s patience and willingness to sustain Karber’s team through the end of the Cold War.

  By 1978 Project 186 had accumulated enough comparable data on NATO and Warsaw Pact forces to generate quantitative, or “bean count,” comparisons of the two sides’ manpower and major weapon systems in central Europe, and these data were displayed in ONA’s first detailed assessment of NATO–Warsaw Pact balance. The NATO forces included in this balance were those stationed in West Germany, the Netherlands, and Belgium, Denmark, and Luxembourg as well as the First French Army. The Warsaw Pact Center Region forces were those Soviet and East European forces stationed in East Germany, Poland, and Czechoslovakia.

  The theater-level force data displayed in the March 1978 assessment showed that Warsaw Pact forces in central Europe outnumbered NATO’s in most categories. The Warsaw Pact’s two-to-one advantage in main battle tanks and antitank weapons was a major concern to the NATO alliance. But, as with strategic nuclear forces, there were problems with comparing the raw numbers of weapons in various categories (main battle tanks, ar
tillery, mortars, etc.). First, as Marshall was well aware, static comparisons at a single point in time gave no sense of the trends. Second, these “raw” comparisons ignored the relative quality of individual weapons. They also ignored personnel training and readiness, logistics, tactics, doctrine, command and control, and other variables known to affect combat outcomes.

  FIGURE 6.3.NATO/Warsaw Pact Central Region Ground Forces and Ratios, 1978.

  SOURCE: OSD/NA, “The Military Balance in Europe: A Net Assessment,” March 1978, p. 46. Army Lieutenant Colonel Peter R. Bankson wrote this assessment.

  In attempting to address the issue of qualitative differences between the two sides’ weapon systems, in 1974 the Army’s Concepts Analysis Agency (CAA) published the first version of a qualitative weighting system for ground forces known as Weighted Effectiveness Indices/Weighted Unit Values (WEI/WUV, pronounced “wee-wuv”). It contained subjectively developed WEI scores for NATO and Warsaw Pact weapons in nine categories: small arms, armored personnel carriers, tanks, armored reconnaissance vehicles, antitank weapons, cannons and rockets, mortars, armed helicopters, and air defense artillery.69 While the original aim of WEI/WUV was to produce “scores suitable for use in current war gaming,”70 WEIs could also be used to weight the equipment holdings of opposing military forces, as shown in the notional WUV calculations for a US division. Once WEI values had been multiplied by the inventories in each category and the results in each category were multiplied by category weights, the results could be added up to give WUV totals for entire units such as divisions (see Figure 6–4), or for higher-level force aggregations up to and including all NATO and Warsaw Pact forces in the Central Region.