Category The Secret of Apollo

Management and the Control of Research and Development

Control. .. depends upon information and activities involving information: information processing, programming, decision, and communication.

— James Beniger, The Control Revolution, 1986

Since at least the Middle Ages, Western society’s fascination with sophisti­cated technology has demanded organizational solutions. By the middle of the nineteenth century, railroads in Europe and the United States required professional managers to run them.1 As the scale of operations increased, ex­ecutives developed “systematic management’’ to coordinate and control their midlevel personnel.2 At the beginning of the twentieth century, Frederick Winslow Taylor, publishing his major work in 1911, devised a means — by way of “scientific management’’ — of extending managerial influence to the fac­tory floors of increasingly large industrial enterprises.3 In both systematic and scientific management, information provided the levers that managers used to control their subordinates. Frequently working with engineers, managers gathered information from lower-level staff and then used that knowledge to reorganize work processes and control employees.4

Scientists and engineers eventually posed far more difficult challenges to managers. Universities trained these ‘‘knowledge workers,’’ as management consultant Peter Drucker referred to them in the late 1940s, to be dedicated to their careers and their specialties, not to their employers. They generated new ideas in an undefined process that no one could routinize, thus ruling out scientific management techniques. Their specialized knowledge placed them

beyond the competence of most managers. Even if technical personnel wanted to share their knowledge with managers (which they typically did not), they could not clearly describe their creative process. Only after the fact, it seemed, could managers control the products or the technologists who created them. Even so, managers seldom perceived research and development (R&D) man­agement as a critical issue.5 Drucker suggested a solution he called manage­ment by objectives. According to this approach, managers and professionals jointly negotiated the objectives for the agency or firm on the one hand and for the individuals on the other, each worker agreeing to the terms. Individu­als and agencies or firms would harmonize their respective goals.6

The management-by-objectives strategy worked reasonably well for man­agers overseeing individual knowledge workers, but it did little to coordi­nate the efforts of scientists and engineers on large projects, on which experts organized (or disagreed) along disciplinary lines and could form only tempo­rary committees for the exchange of information. Much like with the unique and short-lived Manhattan Project, the experience of complicated programs such as ballistic missiles demonstrated that traditional organizational schemes would not suffice. Scientists and engineers found that they needed some indi­viduals to coordinate the information flowing among working groups. These ‘‘systems engineers’’ created and maintained documents that reflected the current design, and they coordinated design changes with all those involved in the program. Perceptive managers and military officers realized that central design coordination allowed them to gain control of both the creative process and its lively if unruly knowledge workers.

This study examines how scientists and engineers created a process to coordinate large-scale technology development-systems management—and how managers and military officers modified and gained control of it. The story owes a debt to the insights of Max Weber, who noted long ago that modern organizations form standardized rules and procedures that create and sustain bureaucracies.7 Scholars since then have elaborated upon the develop­ment of these procedures as a process of ‘‘knowledge codification,’’ one that can be formally internal to individuals or informally contained in the com­munications between or among individuals.8 For organizations to learn, to adapt, and to sustain adaptations, they must have processes that are both flex­ible and durable. Recent scholarship on these so-called learning organizations has pursued and elaborated on this view, providing a perspective congenial to a historical analysis of management. By means of communication, feedback, and codification, organizations can be said to learn and retain knowledge.9

Systems management first developed in the air defense and ballistic missile programs of the 1950s, across many aerospace organizations. These programs, like any other large-scale technologies, came into being as a result of nego­tiations among various organizations, classes, and interest groups.10 Scientists typically created the core ideas behind new systems or the critical elements that made them possible or useful. Engineers developed the subsystems and integrated them into a complex vehicle. Military officers promoted these com­plex vehicles as a means of besting their Cold War foes. Managers controlled the resources required to produce the new systems. Systems management was embraced because it assigned each of these groups a standard role in the tech­nology development process. Systems management became the core process of aerospace R&D institutions, modeled largely on management techniques developed on army and air force ballistic missile programs. Methods devel­oped for air defense systems paralleled those for ballistic missiles, but in the bureaucratic battles of the early 1960s, ballistic missile officers and their meth­ods triumphed, forming the basis for the air force’s procurement regulations.11

This book thus traces a path through the literature on the history and poli­tics of aerospace development and weapons procurement.12 Instead of pro­viding another case study of a particular project or organization, it pieces together a story from elements that include military and civilian organizations in the United States and Europe. This approach has the distinct advantage of providing cross-organizational and cross-cultural perspectives on the sub­ject, as well as showing the dynamics of the transfer of management methods. NASA and the European programs encountered the same kinds of technical and social issues that the air force and the Jet Propulsion Laboratory (JPL) had previously come upon, and ultimately they looked outside of their orga­nizations to help resolve the problems. NASA looked to the air force (and to a lesser degree to JPL), and a few years later the Europeans gleaned their methods from NASA. The Apollo program became a highly visible icon of American managerial skill — the symbol of the difference between American technical prowess and European technical retardation in the 1960s and early 1970s.

European frustration reached its peak in 1969, when NASA put men on the Moon while the European Space Vehicle Launcher Development Organisation (ELDO) endured yet another failure of its launcher. ELDO only haphazardly adopted American management methods, and the lack of authority meant that those that ELDO did adopt could not be consistently implemented. The failures of ELDO ultimately proved to be the spur for the Europeans to over­come their historic hostilities and create a highly successful integrated space organization, the European Space Agency. This new agency and its predeces­sor, the European Space Research Organisation, borrowed extensively from NASA and its contractors. NASA’s management methods, when adapted to the European environment, became key ingredients in Europe’s subsequent successful space program. The air force, the army’s (and later NASA’s) JPL, NASA’s manned space programs, and the European integrated space pro­grams all learned that spending more to ensure success was less expensive than failure.

The modern aerospace industry is paradoxical. It is both innovative, as its various air and space products attest, and bureaucratic, as evidenced by the hundreds of engineers assigned to each project and the overpriced compo­nents used. How can these two characteristics coexist? The answer lies in the nature of aerospace products, which must be extraordinarily dependable and robust, and in the processes that the industry uses to ensure extraordinary dependability. Spacecraft that fail as they approach Mars cannot be repaired. Hundreds can lose their lives if an aircraft crashes. The media’s dramatization of aerospace failures is itself an indication that these failures are not the norm. In a hotly contested Cold War race for technical superiority, the extreme envi­ronment of space exacted its toll in numerous failures of extremely expensive systems. Those funding the race demanded results. In response, development organizations created what few expected and even fewer wanted—a bureau­cracy for innovation. To begin to understand this apparent contradiction in terms, we must first understand the exacting nature of space technologies and the concerns of those who create them.

ONE

The Formation of The Aerospace Corporation

From the beginning of the WDD, aircraft industry leaders complained bit­terly about R-W’s insider position. They believed that the ideal approach to weapons development was for the air force to let prime contracts to a single integration contractor, a position supported by the air force’s own regulations. These stated that the air force should hire a single prime contractor to de­velop, integrate, and test a weapon system, unless no company was qualified to perform the task. In this case, the air force itself could act as prime con­tractor. The latter position was Schriever’s justification for his approach to the ICBM program, with the important modification that the air force would in­stead hire a third party to direct technical coordination of the integration task. Industry leaders also pointed out that in R-W, the air force was creating a new, powerful competitor with close ties to air force planning and a concomitant edge in bidding.48

Normally, R-W should have been controlled by the air force in the way that any other contractor would have been. However, the air force had hired R-W to act as the air force’s technical assistant for ICBM development, in which position R-W personnel acted with virtually the same authority as the govern­ment. In 1954, Assistant Secretary of Defense for Research and Development Donald Quarles, formerly of Bell Labs, had insisted that R-W personnel be given “line” responsibility, with full authority to direct contractors, instead of “staff” status, where they would merely be advisers. This mirrored his experi­ence at Bell Labs, which acted as the technical direction authority to AT&T’s manufacturing arm, Western Electric. Bell Labs also performed this role with other contractors, sometimes on behalf of the government on high-priority military programs. This powerful position required that AT&T acquire sensi­tive data from other companies. As a regulated monopoly, AT&T could legiti­mately act in this capacity, as it essentially had no competitors.49

Caltech’s JPL and MIT’s Radiation Laboratory also acted as technical di­rection groups for the government, but these academic nonprofit institutions were little threat to industry. However, R-W was neither a nonprofit insti­tution nor a regulated monopoly, and in fact it competed for other projects against the same companies that it monitored on the ICBM program. Existing aircraft firms vigorously campaigned against the air force’s unusual relation­ship with the upstart company.

To protect his organization from criticism, Schriever enforced a hardware ban on R-W to keep it from acquiring lucrative hardware contracts on any programs in which it was the technical direction contractor. R-W ‘‘walled off’’ the technical direction work of STL from the rest of the company. Continuing concerns led R-W to establish a physically separate location for its headquar­ters — in Canoga Park, California. These measures did not satisfy industrial leaders, who continued to lobby against the company.50

Despite the clamor and the ICBM hardware ban, neither Ramo nor Wool­dridge believed that R-W could grow without manufacturing capabilities. They grasped every opportunity to expand manufacturing by aggressively pursuing hardware production products and contracts outside the ballistic missile program. These included process control computers, semiconductors, and a variety of aircraft and air-breathing missile components. Aggressive pursuit of hardware contracts paid off, as R-W received permission to build ballistic missile hardware to test ablative nose cones built by General Electric. Strongly backed by Schriever’s technical director, Col. Charles Terhune, STL then built the Able 1 lunar probe launched in August 1958 and the Pioneer 1 spacecraft launched by the National Aeronautics and Space Administration (NASA) in October 1958. These activities fomented even more severe indus­trial protests, as the hardware ban against R-W evaporated.51

Expansion on these and other ventures such as semiconductors stressed R-W’s finances. Ramo and Wooldridge leaned on their original investor, Thompson Products, for cash to expand facilities and capital equipment, and the ensuing negotiations led to an agreement that resulted in the merger of the two companies effective October 31,1958. The new combination, Thompson- Ramo-Wooldridge (TRW), became the aerospace giant that the older aircraft companies had feared.

TRW executives recognized the awkward position of STL in the new com­pany. STL handled TRW’s space business, including both the technical di­rection tasks for the air force and STL’s budding space manufacturing busi­nesses. Because of the air force connection, STL would always be vulnerable to charges of conflict of interest. To minimize this risk, TRW executives estab­lished STL as an independent subsidiary corporation with its own board of directors chaired by Jimmy Doolittle, a war hero with impeccable creden­tials and impressive ties to the air force and NASA. No TRW board member or senior manager sat on STL’s board. TRW executives recognized that they might have to divest STL, and through this reorganization they were prepared to do so.52

Although TRW was prepared to divest STL, neither Schriever nor TRW really wanted this to happen. TRW enjoyed significant profits from STL, and Schriever wanted STL’s experienced personnel directing the technical aspects of the air force’s ICBM and space programs. However, STL’s increasing in­volvement with space projects and hardware development fueled industry complaints, leading to congressional hearings in February and March 1959.

These hearings, chaired by Rep. Chet Holifield from California, featured vehement attacks against STL’s ‘‘intimate and privileged position’’ with the air force and equally strong defenses by Schriever and by TRW executives Simon Ramo and Louis Dunn. It became clear even to Schriever that as long as TRW acquired competition-sensitive technical information from other aerospace firms through STL, the clamor would continue. A plan to sell STL to pub­lic investors fell through when Air Force Secretary Douglas vetoed it on the grounds that STL would remain a problem as long as private owners used STL to make a profit. The Holifield Committee’s final report seconded this idea and urged that STL be converted into a nonprofit corporation like RAND and MITRE. Schriever reluctantly agreed, leading to the formation of The Aero­space Corporation on June 4,1960.53

At Schriever’s insistence, STL continued systems engineering and technical direction for the ballistic missile programs for the near future, but all others transferred to Aerospace. Dr. Ivan Getting became Aerospace’s first presi­dent, and a number of STL personnel transferred to the new corporation. This ended the controversy about TRW’s insider position with the air force, but as industry had feared, there was a powerful new competitor with which to contend. Aerospace became one of a growing breed of nonprofit corporations that served the air force and other military organizations.

Systems engineering, which required the coordination of all elements of the technical system, could be performed by a prime contractor for the sys­tem, by the air force itself, or by a nonprofit firm that had no interest in com­petition. The experience of R-W showed that a profit-making corporation could not act on behalf of the U. S. government to coordinate or control the efforts of its competitors. The function of systems engineering had to be con­tained within the government itself, a neutral third party hired by the gov­ernment such as Aerospace or MIT, or a prime contractor. With this contro­versy settled, the air force could now standardize systems management as its primary R&D method across all of its divisions.54

Standardizing Systems Management

By 1959, ongoing deliberations at air force headquarters were under way re­garding the applicability of Schriever’s ‘‘Inglewood model’’ to the rest of the air force’s development programs. A senior committee headed by the AMC commander, Gen. Samuel Anderson, agreed that the air force should adopt the methods used in Inglewood, with the planning and implementation of new projects on a systems, or ‘‘life cycle,’’ basis. Planning for the entire system would occur up front, and project offices would have the authority to man­age development, including funding authority. However, the committee split into three camps regarding the organization, advocating positions ranging from minor modifications to radical reorganization. In June 1960, the Air Staff selected the least ambitious plan, which did include installation of new regu­lations based on Schriever’s organizational processes, to be used on all the air force’s major development programs.55

The 375-series regulations for systems management originated with one of Schriever’s officers, Col. Ben Bellis, who headed an effort to document the procedures developed in Inglewood. After a series of reviews, the new regula­tions for systems management appeared on August 31,1960, and were applied to the air force’s major projects for missiles, space, aeronautics, and electron­ics. Subsequently revised and extended, these regulations became the institu­tional backbone of the new, Inglewood-inspired R&D system.56

Under the new regulations, the system program director gained significant authority. The air force required that the program director create and gain approval of a single document known as the System Package Program. Each System Package Program provided information on cost, schedule, manage­ment, logistics, operations, training, and security.57 The 375 regulations for­mally applied the ARDC-AMC project office concept across all air force major acquisition programs.

The more radical ‘‘Schriever Plan’’ to manage the air force’s R&D had been shelved by Anderson’s committee in 1959, but it gained new life in 1961 when Robert McNamara became secretary of defense. McNamara, trying to resolve the controversy over which service should gain the coveted military space mission, looked for evidence of managerial and organizational expertise to determine which service should lead space efforts. With several hints from the McNamara camp that the Schriever Plan would help the cause, Air Force Chief of Staff Thomas White approved it. Secretary of the Air Force Eugene Zuckert and McNamara signaled their pleasure by conferring all space research to the air force in March 1961.58

The Schriever Plan reallocated the procurement activities ofAMC to a new organization that also included the development functions of ARDC. ARDC was abolished, its place taken by Air Force Systems Command (AFSC), which came into being on April 1, 1961. Schriever, appointed the first commander of AFSC, now managed all of the air force’s major development programs in four divisions: the Ballistic Systems Division in San Bernardino, California; the Space Systems Division in El Segundo, California; the Aeronautical Systems Division in Dayton, Ohio; and the Electronics Systems Division in Lexington,

The air force’s Ballistic Systems Division and Thompson-Ramo-Wooldridge’s Space Technology Laboratory were in the center of a vast network of government and industry organizations, all of which learned aspects of systems management. ‘‘BSQ’’ represents the Ballistic Systems Division, and ‘‘SE/TD’’ stands for systems engineering and technical direction, the main function of STL. Courtesy Library of Congress.

Massachusetts. Ascending to command over all of the air force’s large acqui­sition programs, Schriever’s presence ensured the spread and enforcement of the 375 procedures.59

Standardization of R&D in AFSC went beyond the 375 regulations. By mid – 1961, Schriever’s organization molded status reporting into a highly sophisti­cated system, known as rainbow reporting because it presented each element of the system on pages of different colors in a small, brightly packaged book­let. Over the next few years, the rainbow reporting system evolved to include yearly and monthly milestone schedules, government and contractor finan­cial data, contractor manpower data, reliability data, procurement data, engi­neering qualification data, and the so-called PRESTO procedures for prob­lems needing immediate attention. They also specified acceptable formats and technologies for presentations to ensure commonality, helping the top-level managers to judge the programs on a consistent basis.60

With the establishment of AFSC, the Inglewood model of systems man­agement, including configuration management, became the dominant model for large-scale programs. In April 1961, Schriever’s authority and influence reached its apex, as he presided over all major development programs in the air force, using standardized methods of his own making.61 What Schriever and others did not foresee was that just as the air force could use systems man­agement to control contractors and its own officers, so too could the DOD use it to control the air force.

McNamara, Phased Planning, and Central Control

Within the DOD, the Office of the Secretary of Defense grew in power from 1947 to the mid-1960s. Over the years, the office progressively pulled critical decisions up the hierarchy, subordinating service interests and rivalries. Bene­fiting and exploiting this trend to the fullest was John F. Kennedy’s appointee to the office, Robert McNamara.62

McNamara trained at the University of California, Berkeley, and taught business courses for a short time at Harvard before World War II. During the war, he performed statistical analyses for army logistics, determining the quantities of replacement parts needed based upon statistical assessments of combat and operations. After the war, he joined Ford Motor Company, tagged as one of the mathematically trained ‘‘whiz kids’’ that reformed Ford’s dis­organized finances and helped turn the company around. He rose quickly, eventually becoming president.63

Famous for his faith in centralized control implemented through quantita­tive measurement, McNamara took advantage of the authority granted to the Office of the Secretary of Defense by the Defense Reorganization Act of 1958. This act gave the secretary of defense the authority to withhold funding from the services and transfer assignments between the services. Upon his appoint­ment to the office, in the spring of 1961 McNamara initiated a series of more than 100 studies known as McNamara’s 100 trombones, or the 92 labors of Sec­retary McNamara. The services readily complied with this request, expecting the novice secretary to get bogged down in conflicting piles of recommenda­tions.64

Without waiting for completion of the studies, McNamara also installed RAND chief economist Charles Hitch as the DOD comptroller. Given McNa­mara’s background as a Ford financial manager and Hitch’s qualifications as an economist, it was not surprising that they considered economic criteria to be foremost in making decisions for future weapon systems. Hitch’s Pro­gram Planning and Budgeting System required that life cycle cost estimates be performed before deciding whether to develop a new weapon system. This agreed with the result of one of McNamara’s studies — “Shortening Develop­ment Time and Reducing Development and Systems Cost’’—which claimed that ‘‘reducing lead time and cost’’ should be given the same priority as im­proving performance. It deemphasized the relentless push to higher technical performance and required that feasibility and effectiveness studies calculate technical risks and cost-to-effectiveness ratios.65

Following up on this study, in September 1961 McNamara assigned the task of improving R&D management to John Rubel, the deputy director of defense research and engineering. Rubel established model programs whose methods could then be copied throughout all of the services, starting with the air force Agena, TFX fighter, Titan III, and medium-range ballistic missile programs. Rubel required a ‘‘Phase I’’ effort to develop a preliminary design. This would ensure ‘‘that the cost estimates for the subsequent development effort’’ were ‘‘based on a solid foundation.’’66 The preliminary design effort would generate ‘‘a set of drawings and specifications and descriptive documents’’ to describe management methods, including schedules, milestones, tasks, objectives, and policies. Rubel had no reservations about forcing industrial contractors to organize and manage their projects in the way he wanted. If they wanted the job, they had to conform.67

He made clear in the request for proposals that go-ahead for Phase I did not constitute program approval. Previously, award of a preliminary design contract constituted de facto project approval for development and produc­tion. This was no longer true. Only the secretary of defense could approve a project, and he would not do so until completion of Phase I and a pro­gram review.68 According to Rubel, ‘‘The fact that improved definition is re­quired before larger-scale commitments are undertaken is neither surprising nor unique, although it is true that on most programs this definition phase has been less clearly identifiable because it has been stretched out in time and interwoven with other program activities such as development, model fabri­cation, testing and, in some cases, even production.’’ Rubel did not believe that a program definition phase would slow high-priority programs. ‘‘In fact,’’ he wrote, ‘‘our real progress should be accelerated as the result of obtaining a better focusing of our efforts.’’69

The phased approach brought several benefits to upper management. It promised better cost, schedule, and technical definition. If the contractor or agency did not provide appropriate information, management could cancel or modify the program. Organizations therefore made strenuous efforts to finalize a design and estimate program costs. The preliminary design phase provided management with a decision point before spending large sums of money, making projects easier to terminate and contractors easier to control.

By 1962, studies by Harvard and RAND economists had shown that DOD weapons projects had consistently large overruns and schedule slips, with missile programs having the worst record. The RAND study showed that for six missile projects, costs overran by more than a factor of four, with schedule slips greater than 50%. Other projects showed smaller slips, but all types aver­aged at least 70% cost overruns, and the average was more than 200% (triple the original cost estimates). The military was clearly vulnerable to criticism on cost issues, and McNamara efficiently exploited this weakness. His Program Planning and Budgeting System required that all of the services create five- year projections of programs and their costs, allocated not by specific services but rather across broad categories such as strategic offense or defense.70

Schriever sensed the change in national priorities and saw the impact of McNamara’s reforms. Replacing ‘‘concurrency,’’ ‘‘managerial reform’’ and ‘‘cost control’’ soon became the new watchwords. The immediate task facing Schriever in early 1962 was responding vigorously to the McNamara-Rubel initiatives, which he saw as cost control measures. In a February 1962 memo­randum, Schriever stated that cost overruns arose from ‘‘any one or a combi­nation of’’ factors, including deliberate underestimation, adherence to overly strict standards, too much optimism in estimating performance and sched­ules, vacillation or changes in program direction, and inadequate military or contractor management.71

One area that Schriever had to improve was cost estimation. His comp­troller’s office began by educating AFSC staff, instituting cost analysis training courses at the Air Force Institute of Technology in Dayton, Ohio. By Febru­ary 1962, the first class of 25 students graduated from this course. AFSC also developed the Program Planning Report, which allowed for improved analy­sis of cost data with respect to technical and schedule progress. He also had AFSC adopt and modify the navy’s new planning tool, PERT.72

Schriever developed other ways to improve AFSC’s management capabili­ties. He established a Management Improvement Board, ‘‘made up of Gen­eral Officers having the greatest experience in systems management matters ranging from funding, systems engineering, procurement and production, through research and development.’’ Schriever had board members exam­ine ‘‘the entire area of systems management methods to include those of the Industrial complex as well as those of the Air Force.’’ He also reinstated the Air Force Industry Advisory Group, a Board of Visitors to improve working relationships with industry, and a program of ‘‘systems management program surveys.’’ AFSC also collected ‘‘lessons learned’’ information from programs and broadcast this information through publications and industry symposia. Schriever also used this information to produce management goals for AFSC.73

AFSC also communicated systems management concepts through educa­tion. Examples included a system program management course at the Air Force Institute of Technology and the creation of a systems management newsletter within AFSC. The Air Force Institute of Technology course used case studies taught by experienced program managers such as Col. Samuel Phillips of the Minuteman program. These program managers taught about program planning and budgeting, the McNamara reforms, organizational roles in system development, systems engineering, configuration manage­ment and testing, system acquisition regulations, program management tech­niques, contracting approaches, and financial methods.74

By the mid-1960s, the combination of AFSC management initiatives and the McNamara reforms produced a mature form of systems management that is still used in the aerospace industry today. Earlier concepts and practices of

image1

Systems management phases.

concurrency contributed the detailed planning and systems engineering co­ordination necessary to rapidly develop large-scale technologies. When ICBM failures became the primary concern, engineers added change control, quality control, and reliability to the mix. Finally, the cost concerns of the early 1960s — driven by rising ICBM costs, the Vietnam War, and social issues such as the civil rights movement—contributed phased planning and configura­tion management. Both new methods provided mechanisms to better predict costs.

McNamara, duly impressed with the procedures and reforms in Schriever’s organization, used them — modified to include phased planning for central control—as the basis for the DOD’s new regulations for the development of large-scale weapon systems. In 1965, the DOD enshrined phased planning and the systems concept as the cornerstone of its R&D regulations. Having already spread to NASA, these processes moved throughout the aerospace industry. Even when the processes were not explicitly used, industry accepted the as­sumptions and ideas encompassed in these regulations.75

Smoke, Fire, and Recovery

Apollo’s troubles began in September 1965, when NAA’s second stage rup­tured during a structural test.91 Engineers pinpointed the fault, and in the process MSFC managers concluded that NAA’s management was to blame for shoddy workmanship. By October, the Industrial Operations manager, Brig. Gen. Edmund O’Connor, told von Braun, ‘‘The S-II program is out of con­trol.’’ He believed its management was to blame. O’Connor was equally blunt in a letter to Space and Information Systems Division (S&ID) President Har­rison Storms: ‘‘The continued inability or failure of S&ID to project with any reasonable accuracy their resource requirements, their inability to identify in a timely manner impending problems, and their inability to assess and re­late resource requirements and problem areas to schedule impact, can lead me to only one conclusion, that S&ID management does not have control of the Saturn S-II program.’’92

Phillips went immediately to NAA with a ‘‘tiger team’’ of nearly one hun­dred NASA personnel to ‘‘terrorize the contractor,’’93 reporting the team’s

Apollo with its major contractors identified. Apollo was perhaps the largest single R&D project of all time, integrating many contractors for its stages and requiring massive launch and operations facilities and organizations. Saturn V contractors not identified. Courtesy NASA.

findings in December 1965 in what later became known as the Phillips Re­port. While writing to NAA that ‘‘the right actions now’’ could improve the program, Phillips privately wrote Mueller that NAA’s president was too pas­sive. Storms, Phillips said, should ‘‘be removed as president of S&ID and be replaced by a man who will be able to quickly provide effective and unques­tionable leadership for the organization to bring the division out of trouble.’’94

NAA responded by placing Gen. Robert Greer, retired from the air force, in charge of the S-II program. Greer updated the management control cen­ter and ensured more rapid exchange and collection of information through Black Saturday meetings modeled after those in Bernard Schriever’s ballis­tic missile program. Greer also instituted forty-five-minute meetings every morning, eventually cutting back to twice a week. Greer’s reforms began to take hold but did not prevent the May 1966 loss of another test stage because of faulty procedures. NASA clamped down further, requiring NAA to develop better methods for managing and planning its work. In the summer of 1966, after two years of studies and preparation, NAA deployed work package man­agement for the S-II and Command and Service Module.95

Work package management extended project management to lower project levels and combined accounting and contracting procedures by creating a specific work package for each program task. The company assigned respon­sibility for each task to one person, a mini project manager for the task who accounted for performance, cost, and schedule in the same way and with the same tools as the overall project manager. Each work package was a ‘‘funda­mental building stone,’’ with specifications, plans, costs, and schedules to help managers in their monitoring. Prior to the development of work packages, ‘‘It was difficult to say what manager was responsible for a particular cost increase because there were 10 or 15 functional and subcontractor areas involved.’’96 In later versions, the work package numbering scheme matched that for cost accounting.

Grumman’s difficulties on the lunar module also attracted NASA attention. Troubles first appeared in schedule slips on its ground support equipment in the spring of 1966. Alarmed at Grumman’s growing costs, Phillips sent a management review team to Grumman that summer, prompting Grumman to sack the program manager, establish a program control office, and move Grumman’s vice president to the factory floor to monitor work. By fall, NASA pushed Grumman into adopting work package management.97 It did not im­mediately solve Grumman’s difficulties. The primary problem was a late start due to NASA’s delayed decision to use lunar orbit rendezvous. However, work package management and the new program control office found and resolved problems more quickly than before.

Despite these difficulties, Apollo moved briskly forward until its most se­vere crisis struck on January 27,1967. That day, astronauts and KSC personnel were performing tests in preparation for launch of the first manned Apollo mission. At 6:30 that evening, the three astronauts scheduled for that mis­sion, Virgil Grissom, Edward White, and Roger Chaffee, were in the spacecraft command module testing procedures. At 6:31, launch operators heard a cry from the astronauts over the radio, ‘‘There is a fire in here!’’ Those were their last words. All three astronauts died of asphyxiation before launch personnel reached them.98

KSC personnel immediately notified NASA headquarters. Administrator Webb hurriedly planned for the political fallout. He sent Seamans and Phillips to Florida, while he persuaded the president and Congress to let NASA per­form the investigation.99 NASA’s investigation concluded that the causes of the disaster were faulty wiring, a drastic underestimation of the dangers of an all-oxygen atmosphere, and a capsule design that precluded rapid escape. No one had realized how dangerous the combination was. NASA had used a pure oxygen atmosphere in all of its prior flights, as did air force pilots in their high-altitude flying. As Col. Frank Borman, one of NASA’s most experi­enced astronauts, put it during the Senate investigation, ‘‘Sir, I am certain that I can say now the spacecraft was extremely unsafe. I believe what the message I meant to imply was that at the time all the people associated and responsible for testing, flying, building, and piloting the spacecraft truly believed it was safe to undergo the test.’’100

Congress did not prove NASA to be negligent or incompetent. One of the investigation’s important results was a nonfinding. Despite searching long and hard, Congress did not find fault with Phillips’s management system. Phillips had already uncovered problems with NAA and had been working for some time to make improvements to its organization and performance. The management system used to organize the capsule design was NASA’s original

committee-based structure, upon which Phillips had superimposed configu­ration management. He and his management system came out unscathed.

Congressional investigations did uncover some of NASA’s dirty laundry, particularly problems with command module contractor NAA. Sen. Walter Mondale of Minnesota learned of the Stage II Phillips Report and confronted Webb about problems between NASA and NAA. Caught by surprise, Webb said he did not know of any such report, which at that moment he did not. After the hearing, he found out about it from Mueller and Phillips. Furious, Webb launched a ‘‘paper sweep’’ to search for more skeletons in the closet. The sweep uncovered a memo written by GE to Apollo spacecraft director Joseph Shea, warning Shea of the danger of fire in the command module. Shea had passed the memo on to his safety and quality assurance people, who re­sponded that no significant dangers existed. GE, already in a sensitive situa­tion because MSC considered it to be spying for headquarters, did not push it any further.101

Webb reacted angrily to these revelations. He believed OMSF had been far too independent and secretive. Webb told Seamans, ‘‘You have to penetrate the [OMSF] system, don’t let Mueller get away with bullshit.’’ The problem, according to Webb, was a lack of supervision by NASA’s executive manage­ment. Mueller had ‘‘followed the policy in Houston of obtaining the very best men they could for the senior positions, and had, as a part of the process of obtaining them, given assurances that they would have almost complete free­dom in carrying out their responsibilities.’’102

After Seamans left NASA in late 1967, Webb expressed shock at the poor management system.103 Webb probably did not realize how decentralized NASA’s management really was. Executive managers routinely delegated most decisions to lower levels. In the wake of the fire, this did not seem wise.

When NAA refused to make swift and comprehensive changes — and even expected to be paid a fee for the burned-out spacecraft—Webb called Boe­ing to see if it would take the job. Boeing said that although it did not want to take over the job, if pressed it would do so. Webb returned to NAA, de­manding that it remove S&ID head Storms, further centralize Apollo project management, eliminate any fee for the failed spacecraft, and pay for improve­ments. NAA did not take the chance that he was bluffing. NAA was extremely unhappy with the entire situation because from its viewpoint, NASA was at fault. Shortly after contract award, over NAA’s objections, NASA had directed a change from a nitrogen-oxygen atmosphere to an all-oxygen atmosphere.104

One problem uncovered during the investigation was GE’s unwillingness to contest NASA over safety issues with a pure oxygen atmosphere. At the heart of the problem was industry’s reluctance to confront NASA when indus­try was dependent on government funding. Despite his substantial political acumen, Webb appeared not to comprehend this. He had hired GE and Bell – comm to strengthen headquarters’ ability to monitor the field centers in 1962; after the fire, Webb repeated his mistake by expanding Boeing’s role from integrator of the Saturn V to integrator of the entire Apollo-Saturn system to ‘‘penetrate the OMSF system.’’ Phillips, who understood the political prob­lems inherent in the GE and Boeing integration efforts, revised the Boeing contract to avoid the negative consequences of Webb’s misconception.105 In essence, Webb wanted to use GE, Bellcomm, and Boeing as an arm of NASA headquarters to control MSC, MSFC, and KSC. This could not work because these contractors could not challenge NASA field center personnel for fear of losing their contracts.

Boeing, as part of its contract, further integrated the management system. The ‘‘teleservices network’’ connected NASA project control rooms with hard copy data transmittal, computer data transmission, and the capability to hold a teleconference involving MSC, MSFC, KSC, Michoud (where the Saturn I was manufactured), and Boeing’s facility near Seattle. Boeing copied MSFC’s program control center design at each facility.106

After the fire, NASA placed even more emphasis on achieving high quality and safety through procedural means. In September 1967, NASA set up safety offices at each field center, along with the first project safety plan. The next month, MSC established a Spacecraft Incident Investigation and Reporting Panel to look into anomalies. A month later, NAA created a Problem Assess­ment Room to report and track problems.

Phillips ordered an astounding array of program reviews to prepare for Apollo’s upcoming missions. He wrote to field center managers to ensure that they used the upcoming Design Certification Reviews to evaluate all potential single-point failures.107 In January 1968, he ordered a complete system safety review, analyzing the interaction of the mission with the hardware, astronauts, ground systems, and personnel. Other reviews included those for quality and metrology, launch vehicle and spacecraft schedules, the communications net­work, flight readiness, mission planning, subcontractors, site selection, the Lunar Receiving Laboratory, flight evaluations, anomalies, crew safety, inter­face management, software, and lunar surface activities.108

NAA’s procedures exemplified the upgraded problem reporting system. Engineers reported failures on a Problem Action Record form. Reliability engineers sent failed components to the appropriate organization, which re­sponded by filling out a Failure Analysis Report describing the physical cause of the failure and the corrective actions taken or recommended. If the orga­nization determined that an engineering change was necessary, it submitted a change request to the change boards. The program control center tracked report status, and a centralized reliability ‘‘data bank’’ recorded the problem and its resolution. Follow-up failure reports and dispositions closed all failure reports.109

Another change in the aftermath of the fire was a further strengthening of configuration management, primarily through changing CCB operating pro­cedures. An October 1967 rule disallowed nonmandatory changes for the first command and lunar modules and required the MSC Senior Board to rule on any and all changes to these spacecraft. A February 1968 ruling required man­agers to consider software changes and their ramifications in CCBs. In May 1968, Apollo Spacecraft Manager George Low specified that the MSC CCB had authority over all design and manufacturing processes.110

By 1968, tough CCB rules slowed the program as trivial changes came to the attention of top managers. Eventually, even Phillips realized that central­ization through configuration management could go too far. In September, MSC managers classified changes into two categories: Class I changes, which MSC would pass judgment upon, and Class II changes, which could be ap­proved by the contractors. Classification did not by itself help much, so in October 1968 Phillips gave Level II CCBs more authority, while higher levels ruled on schedule changes.111

The Apollo project met its technical and schedule objectives, landing men on the Moon in July 1969 and returning them safely to Earth. Anchored by configuration management, Phillips’s system weathered the storm of prob­lems uncovered through testing and Apollo’s most severe crisis, the 1967 death of the three astronauts and the ensuing investigations. Despite strenuous ef­forts, congressional critics did not find many flaws with Phillips’s manage­ment scheme and concurred with NASA that the fire resulted from a tragic underestimation of the danger.

Configuration management was Phillips’s most powerful tool. Whenever problems occurred, his almost invariable response was to strengthen configu­ration management. Having found that his favorite method could be over­used, by the end of 1968, Phillips gave lower-level CCBs more authority. Con­figuration management formed the heart ofApollo’s system and has remained at the core of NASA’s organization ever since.

Social Groups, Values, and Authority

Alliances between scientists and military officers had grown during World War II on the Manhattan Project, in the Radiation Laboratory, and in opera­tions research groups. The Cold War furthered this military-scientific partner­ship. Appealing to the imminent Soviet threat, military officers like Bernard Schriever promoted the systems approach in weapons development to quickly design and manufacture novel weapons such as ballistic missiles. Working with his scientific allies, Schriever built an organization initially run by mili­tary officers and scientists. Similarly, Army Ordnance officers allied them­selves with JPL’s research engineers to develop the Corporal missile. Both Army Ordnance and Schriever’s ‘‘Inglewood complex’’ spent immense sums of money in concurrent development, designing, testing, and manufacturing missiles as rapidly as possible. The result for Atlas and for Corporal was the same: a radically new, expensive, and unreliable weapon.

Prematurely exploding missiles created a spectacle not easily ignored. JPL’s engineering managers resolved to improve on their ad hoc methods and em­ployed the systems approach their next missile, the Sergeant. The air force’s next-generation missile was the Minuteman, on which Col. Samuel Phillips developed the system of configuration control to better manage costs and schedules. Both second-generation weapons were far more reliable than their predecessors, partly because of the switch to solid-propelled engines, and partly because of changes in organizational practices.

Over time, social measures of success changed. Initially, simply getting a rocket off the ground was a major accomplishment. Eventually, however, Congress expected that its large appropriations would buy technologies that worked reliably. Soon thereafter, congressional leaders wanted accurate cost predictions so they could weigh alternative uses for that money. Cost over­runs came to be seen as failures. This was particularly true in Europe, where leaders promoted space programs mainly to spur economic development.

The Ranger program and its aftermath illustrated the power of Congress to change organizations. Under William Pickering’s guidance, JPL used a loose matrix structure where most authority resided with the technical divisions. When Ranger’s failures exposed JPL’s organizational flaws, Congress required JPL to strengthen project management and implement more stringent pro­cedures. Pickering and JPL’s engineers resisted these changes, but Ranger’s failures weakened their credibility. When National Aeronautics and Space Administration (NASA) Administrator James Webb threatened to remove all future programs from JPL, Pickering had little choice. He gave in. Similar pressures influenced the air force in the early 1960s and the European Space Research Organisation (ESRO) in the late 1960s. Systems management was the end result in each case. The European Space Vehicle Launcher Development Organisation’s (ELDO’s) attempts to strengthen project management did not succeed, because of weaknesses inherent in its authorizing Convention and the uncooperative attitude of its member states.

The first figure illustrates the relationships between the four social groups. In the early Cold War, military officer-entrepreneurs and scientists provided the authority and methods. I distinguish here between those military officers such as Bernard Schriever who promoted new systems, and others, such as Samuel Phillips, who brought them to fruition. Schriever acted in an entre­preneurial fashion and Phillips as a classical manager. In the air force, this period lasted from roughly 1953 until 1959, the heyday of the Atlas missile, before its many test failures led to change. Schriever acted as a visionary entre­preneur, albeit in an unconventional blue uniform. JPL’s period of military entrepreneur-scientific control came from 1944 to 1952, when JPL’s research engineers developed Private and converted Corporal from research to produc­tion. In both cases, expensive, unreliable weapons led to a concentration on cost and dependability for the next missiles, leading to approaches based on engineering and managerial values. Jack James at JPL and Phillips of Minute- man typified the no-nonsense managers that demanded dependability. Un­like engineers focused on research, such as Caltech’s von Karman and Malina, most engineers focused on the design and development of technological sys-

Cold War social groups and alliances. At JPL from 1944 to 1952, and in the air force be­tween 1953 and 1959, entrepreneurial military officers and scientists (along with research engineers) formed a social alliance to promote novel weapons. After these periods, man­agers in the military and industry formed an alliance with design engineers to control costs and build dependable systems.

tems. For them, creating a product that worked was more important than creating one that was new.

NASA’s history differed somewhat from that of the air force, because in the early years of NASA, the scientists and engineers controlled their own projects and methods. At JPL, the research-based methods prevailed in the labora­tory’s early years, and again later when Pickering shifted the laboratory into the space program, and satellite launches (and failures) were frequent as JPL raced with the clock and the Soviets.1 JPL’s new projects reverted to ad hoc committees to get fast results. Similarly, former National Advisory Commit­tee for Aeronautics researcher Robert Gilruth of the Space Task Group ran the early manned programs with little interference from NASA Administra­tor Keith Glennan. Engineers and scientists made decisions locally in a highly decentralized organization. After the Ranger fiasco at JPL, and after the ar­rival of George Mueller and Samuel Phillips in the manned programs, NASA’s high-level managers and engineers quickly centralized authority. A similar story was unfolding at ESRO, originally conceived of as an engineering ser­vice organization for scientists. By 1967, commercial interests predominated and European governments changed ESRO into an engineering organization run by managers to ensure cost control.

It is more difficult to determine who, if anyone, ran ELDO. With an am­bassador as secretary-general and economic nationalism the primary driving force, ELDO did not have a single dominant group, one could argue. Neither engineers nor scientists controlled the organization. Nor could managers fos­ter the communication necessary to break national and industrial barriers. If ever there existed a purely political organization for technology development, ELDO was it.

Each social group promoted its characteristic methods. Military officers and industrial managers promoted project management to centralize author­ity and used contractor penetration to closely monitor industry. Both groups also used competition to keep contractors honest and developed cost predic­tion and control methods such as configuration management and work pack­age management. Scientists promoted analytic techniques such as statistical analysis of reliability, network mathematics, and game theory. Engineers used

Authority changes at NASA and ESRO. In early NASA and ESRO, scientists and research engineers were allied with de­sign engineers to build new technologies. After several years, both organizations shifted to more predictable development, with managers and engineers controlling events.

Systems management methods classified by the social groups that promoted them.

a variety of testing methods, inspection and statistical methods for quality assurance, and design freeze to stabilize designs.

Ultimately, systems management is a stable system because its methods and processes maintain roles for each of its constituent social groups. For sys­tems management to remain stable over many years and projects, it had to have mechanisms for its constituent social groups to effectively interact. In the end, the primary mechanism became configuration management.