7

The Professionalization of Programming

In the development of professional standards, the computer field must be unrelenting in advocating stringent requirements for professional status, whether these include education, experience, examination, character tests, or what not.

—Charles M. Sidlo, “The Making of a Profession,” 1961

Too frequently these people [programmers], while exhibiting excellent technical skills, are non-professional in every other aspect of their work.

—Malcolm Gotterer, “The Impact of Professionalization Efforts on the Computer Manager,” 1971

The Certified Public Programmer

In 1962, the editors of the electronic data processing journal Datamation proposed what they believed would be the solution to the “many problems” that were “embarrassingly prominent” in the nascent commercial computing industry. The majority of these problems, they argued, were caused by the lack of “professional competency” among programming personnel. The recent explosive growth in commercial computing had brought with it a “mounting tide of inexperienced programmers, new-born consultants, and the untutored outer circle of controllers and accountants all assuming greater technical responsibility.” Few of these so-called computer experts were well qualified or experienced, and the result was the crisis of confidence that was plaguing the industry. The solution to this crisis, contended the Datamation editors, was the establishment of a new breed of technical professional: the certified public programmer.¹

By defining clear standards of professional competency, an industry-wide certification program would serve several important purposes for the programming profession. First, it would establish a shared body of abstract occupational knowledge—a “hard core of mutual understanding”—common across the entire professional community. Second, it would help elevate the public reputation of computer personnel from its current stature of “cautious bewilderment and misinterpretation,” to “at least, confused respect.” Finally, and perhaps most significantly, it would enable computer professionals to erect entry barriers to their increasingly contested occupational territory: the flood of amateur programmers—“the industry’s widely publicized upcoming incompetents”as the Datamation editorial dismissively referred to them—“would find their accession to financial stardom impeded by the need for specific qualification such as the passing of a reasonable test of competency.”² In fact, in 1963 the DPMA’s executive director Calvin Elliott named stamping out “bogus” data-processing schools as one of his organization’s primary objectives.³

The Datamation call for the professionalization of programming coincided neatly with the announcement by the National Machine Accountants Association (NMAA) of its new CDP examination. The NMAA, which would later that year rename itself the Data Processing Management Association (DPMA), represented almost sixteen thousand data processing workers in the United States and Canada.⁴ The NMAA had been working since 1960 to develop the CDP exam, which represented the first attempt by a professional association to establish rigorous standards of professional accomplishment in the data processing field. According to the NMAA’s 1962 press release, the exam was intended to “emphasize a broad educational background as well as knowledge of the field of data processing,” and represent “a standard of knowledge for organizing, analyzing and solving problems for which data processing equipment is especially suitable.” It was open to anyone, NMAA member or not, who had completed a prescribed course of academic study, had at least three years of direct work experience in punched card and/or computer installations, and had “high character qualifications.” The first year that the exam was offered, 1,048 applications took it—687 successfully.⁵

Despite being widely criticized for being superficial and irrelevant to real-world software development, the CDP clearly met a perceived need within the computing community. In 1965, 6,951 individuals took the CDP examination, and another 4,000 completed CDP refresher courses conducted by local DPMA chapters.⁶ A number of large employers, including State Farm Insurance, the Prudential Insurance Company of America, and the U.S. Army Corps of Engineers, extended official recognition to the CDP program, and the city of Milwaukee used the CDP as a means to assign pay grades to data processing personnel.⁷ By the end of 1975, 31,351 candidates had taken the CDP and 15,115 had been awarded the certificate.⁸ Although it is difficult to find accurate employment information for software workers in this period, estimates from the Bureau of Labor indicate these 15,115 CDP recipients constituted approximately 10 percent of the overall computing community.

The CDP examinations represented just one step in the DPMA’s ambitious “Six Measures of Professionalism Program,” which included not only the development of standards of competence and codes of ethics but also programs for public service, continuing education, and fundamental research. Of these six measures, only the CDP program achieved even moderate industry acceptance. Nevertheless, simply by articulating a clear professional agenda the DPMA claimed for itself a leadership role in the computing community. Given the general lack of agreement about what skills and educational background were appropriate for computing personnel, the CDP program promised to guarantee at least a basic level of competence. Employers viewed certification as a tool for screening potential employees, evaluating performance, and assuring uniform product and quality.⁹ Programmers saw it as an indication of professional status, a means of assuring job security and achieving promotions, and an aid to finding and obtaining new positions.¹⁰ The certification of practitioners was generally considered to be one of the characteristic functions of any legitimate profession, and the professionalization of programming was seen by many at this time as the solution to a growing sense of crisis within the computing community.¹¹ The “question of professionalism,” as it came to be known in the literature, would come to form the basis for explicit discussions of the software crisis in the late 1960s.

The growing discontent with a perceived lack of professionalism among computing personnel was in part a legacy of the massive expansion of the commercial computer industry over the course of the previous decade. As the Datamation editorial suggests, one response to the personnel crisis of the 1950s had been an influx of new programmer trainees and vocational school graduates into the software labor market. “The ranks of the computer world are being swelled by growing hordes of programmers, systems analysts and related personnel,” warned a report in 1968 by the SIGCPR, and as a result “educational, performance and professional standards are virtually nonexistent.”¹² And although computer specialists in general were appreciative of the short-term benefits of the ongoing personnel shortage in the computer industry—among them, above-average salaries and plentiful opportunities for occupational mobility—many believed that a continued crisis threatened the long-term stability and reputation of their industry and profession. “There is a tendency,” observed a report by the SIGCPR, “for programming to be a ‘dead-end’ profession for many individuals, who, no matter how good they are as programmers, will never make the transition into a supervisory slot. And, in too many instances this is the only road to advancement.”¹³ Many programmers worried about becoming obsolete and felt pressure to constantly upgrade their technical skills.¹⁴ Although starting salaries were high and individual programmers were able to move with relative ease horizontally throughout the industry, there were precious few opportunities for vertical advancement.¹⁵ Whereas technical specialists in traditional engineering disciplines were often able (and in fact expected) to climb the corporate ladder into management positions, the computer boys were usually denied this opportunity.¹⁶

Many of the job advertisements for programmers reflected these concerns about a lack of professional status and longevity. Employers promised new hires a potential career path that involved more than just mere technical labor: “Is your programming career in a closed loop? Create a loop exit for yourself at [the Bendix Corporation].”¹⁷ “Working your way toward obsolescence? At MITRE professional growth is limited only by your ability.”¹⁸ “At Xerox, we look at programmers . . . and see managers.”¹⁹ But as contemporary studies of such “dual ladder” programs for technical workers in the computer fields revealed, programmers rarely had many opportunities for professional development.²⁰ It was just not clear to many corporate employers how the skills—and personality types—possessed by programmers would map onto the skills required for management.

Figure 7.1

Bendix Corporation advertisement, 1962.

Given their growing uncertainty about the future of their occupation, it is not difficult to understand why programmers in the early 1960s were so concerned with establishing themselves as recognized professionals. Belonging to a profession provided an individual with a “monopoly of competence,” or the control over a valuable skill that was readily transferable from organization to organization.²¹ In more practical terms, professionalism offered a means of excluding undesirables and competitors from the labor market, thereby assuring at least basic standards of quality and reliability as well providing a certain degree of protection from the fluctuations of the labor market. Programmers in particular saw professionalism as means of distinguishing themselves from coders or other “mere technicians.” Professionalism offered increased social status, greater autonomy, improved opportunities for advancement, and better pay.²²

The professionalization efforts of programmers were generally encouraged by their corporate employers. An increasing number of corporate managers were beginning to blame their growing dissatisfaction with the rising costs of software development on the lack of professionalism on the part of programmers. Professionalism, or at least a certain form of corporate-friendly professionalism, was represented by managers as a means of reducing corporate dependence on the whims of individual programmers.²³ It was also thought that professionalism might solve a number of other pressing management problems: it might motivate staff members to improve their capabilities; it could bring about more commonality of approaches; it could be used for hiring, promotions, and raises; and it could help solve the perennial question, Who is qualified?²⁴ “The concept of professionalism,” argued one personnel research journal from the early 1970s, “affords a business-like answer to the existing and future computer skills market” by making computer personnel responsible for policing their own disciplinary identity.²⁵ Professionalism appeared to provide a familiar solution to the increasingly complex problems of managing the relationship between business and technological expertise.

In response to these various motivations to professionalize, programmers in the late 1950s and early 1960s worked to establish the institutional structures traditionally associated with the professions. These included the development of an academic infrastructure for supporting theoretical computer science research; support for industry-based certification and licensing programs; the establishment of professional societies and journals; the introduction of performance standards; and professional codes of ethics. Many of these institutional structures developed rapidly and were established on a provisional basis by the end of the 1950s.

But the existence of professional institutions did not necessarily translate readily into widely recognized professional status.²⁶ The early adoption of the structures of professionalism, however, obscured the deep intellectual and ideological schisms that existed within the programming community. Although many practitioners agreed on the need for a programming profession, they disagreed sharply about what such a profession should look like. What was the purpose of the profession? Who should be allowed to participate? Who would control entry into the profession, and how? What body of abstract knowledge would be used to support its claims to legitimacy? By the beginning of the 1960s, clearly discernible factions had emerged within the fledgling programming profession. Science- and engineering-oriented programmers worked to develop a theoretical basis for their discipline. They joined associations like the ACM that published academic-style journals, imposed strict educational requirements for membership, and resisted certification and licensing programs. Business data processing personnel, on the other hand, pursued a more practice-centered professional agenda. If they joined any professional associations at all, it was the DPMA. They read journals like Datamation, which emphasized plain speech and practical relevance over theoretical rigor. The tension that existed between these two groups of aspiring professionals—the academic computer scientists and the business data processors—greatly influenced the character and fortunes of the various professional institutions that each faction supported. Academic computer scientists struggled to establish a legitimate and autonomous intellectual discipline based on a sound body of theoretical research. Systems analysts and business programmers worked to improve their standing within the organizational hierarchy by distancing themselves from computer operators and other so-called technicians. Neither group was entirely successful.

This chapter will focus on the attempts of programmers to establish the institutional structures associated with professionalism, including professional societies, certification programs, educational standards, and codes of ethics. It argues that the professionalization of computer programming represented a potential solution to the looming software crisis that appealed to programmers and employers alike. But it also suggests that the controversy that surrounded the various professional institutions that were established in this period reveals the deep divisions that existed within the programming community about the nature of programming skill and the future of the programming professions. Many of the themes developed in previous chapters—the development of new programming technologies or more “efficient” management methodologies—are closely tied to questions of professional status. If skilled programmers could be replaced by automated development tools, for example, or by more “scientific” management methodologies, then they could hardly have much claim to professional legitimacy. The question of what programming was—as an intellectual and occupational activity—and where it fit into traditional social, academic and professional hierarchies, was actively negotiated during the decades of the 1950s and 1960s. Programmers were well aware of their tenuous professional position, and they struggled to prove that they possessed a unique set of skills and training that allowed them to lay claim to professional autonomy.

The Association for Computing Machinery

On January 10, 1947, at the Symposium on Large-Scale Digital Calculating Machinery at the Harvard Computation Laboratory, Professor Samuel Caldwell of MIT proposed to a crowd of more than three hundred the formation of a new association of those interested in computing machinery. His proposal obviously landed on fertile soil: within six months a “Notice on the Organization of an Eastern Association for Computing Machinery” was circulating within the computing community, and in September the first meeting of the Eastern Association for Computing Machinery was held at Columbia University. Seventy-eight individuals attended. Officers were elected, and the Executive Council was appointed. A second meeting, held in December at the Aberdeen Proving Grounds in Aberdeen, Maryland, attracted three hundred participants. The next year the organization dropped the word Eastern from its title, and was thereafter known simply as the Association for Computing Machinery (ACM).

During the 1950s the ACM grew steadily but not spectacularly. By 1951 there were 1,113 members, including 43 in other countries; in 1956 the total had risen to 2,305, and by 1959 it had reached 5,254. In the 1960s, the membership grew somewhat more slowly, and there were a few periods during which the total number of members actually decreased. Overall, though, the ACM continued to expand at a rate of about 16 percent annually. By the end of 1969 there were 22,761 regular members. Figure 7.2 shows the annual membership statistics for the years 1947 to 1972.

Figure 7.2

ACM members, 1947–1971.

From its inception, the ACM styled itself as an academically oriented organization. Many of the original members either were or had been associated with a major university computation project, and most were university educated, including a number at the graduate level. The focus of the organization’s early activities was a series of national conferences, the first of which was cosponsored by the Institute for Numerical Analysis at the University of California at Los Angeles. These meetings represented an outgrowth of an earlier series of university-sponsored conferences, and they retained an academic flavor. Many were low-budget affairs held at universities or research institutions, and they frequently made use of dormitory facilities. The papers presented were usually technical, and the proceedings were published. The ACM conferences never acquired the trade show atmosphere that characterized other national meetings. The National Computer Conference, which became almost entirely commercial, for instance, resembled a trade show much more than an academic conference. In fact, deliberate efforts were made to distance the ACM from the influence of the commercial vendors, particularly IBM. For many years the ACM resisted publishing its own journal, possibly because “some early ACM leaders saw the society as a declaration of independence from IBM, and, by extension, from all commercial considerations like the sale of publications and the solicitation of advertising.”²⁷ Until 1953, when it began publishing the Journal of the ACM, the ACM exclusively supported the National Research Council’s highly technical journal Mathematical Tables and Other Aids to Computation. Even then, the primary contents of the Journal of the ACM were theoretical papers, and the emphasis was on the dissemination of “information about computing machinery in the best scientific tradition.”²⁸ Articles were peer-reviewed, and every attempt was made to maintain rigorous academic standards.

Throughout the 1950s and 1960s the ACM continued to cultivate its relationship with the academic community. In 1954 it accepted an invitation to apply for membership in the American Association for the Advancement of Science. Since 1958 the ACM has been represented in the Mathematical Sciences Division of the National Academy of Sciences National Research Council. In 1962 it affiliated with the Conference Board of the Mathematical Sciences, which also consisted of the American Mathematical Society, the Mathematical Association of America, the Society for Industrial and Applied Mathematics, and the Institute of Mathematical Statistics. In 1966 the ACM established the prestigious Turing Award, the highest honor awarded in computer science. Almost half of the institutional members of the ACM were educational organizations, and after 1962 a thriving student membership program was developed.²⁹

The close association that the ACM maintained with the academic computer scientist proved a mixed blessing, however. Although the ACM was able to maintain a relatively high profile within scientific and mathematical circles, it was often castigated by the business community. Many business programmers looked on the ACM as “a sort of holier than thou academic intellectual sort of enterprise—not inclined to be messing around with the garbage that comptrollers worry about,” and the ACM leadership was characterized as “a bunch of guys with their heads in the clouds worrying about Tchebysheff polynomials and things like that.”³⁰ “These four-year computer science wonders are infinitely better equipped to design a new compiler than they are to manage a software development project. We don’t need new compilers. We need on-time, on-budget, software development.”³¹ A Datamation article from 1963 titled “The Cost of Professionalism” warned that the members of the ACM had to “decide whether it’s worth that much to belong to an organization which many feel has been dominated by—and catered pretty much to—Ph.D. mathematicians. . . . [T]he Association tends to look down its nose at business data processing types while claiming to represent the whole, wide wonderful world of computing.”³² A Diebold Group publication from 1966 characterized the ACM as a group “whose interests are primarily academic and which is helpful to those with scholastic backgrounds, theoreticians of methodology, scientific programmers and software people.” Although the ACM president immediately denied this depiction, calling it “too narrow,” the popular perception that the ACM catered solely to academics was difficult to counter.³³

The ACM leadership was not entirely unaware of or unsympathetic to the needs of the business programmers. In his unsuccessful bid in 1959 for the ACM presidency, Paul Armer urged the ACM membership to “THINK BIG,” to “visualize ACM as the professional society unifying all computer users.”³⁴ That same year, Herbert Grosch, an outspoken proponent of a strong, American Medical Association–style professional society (and later ACM president), roundly criticized the ACM for its academic parochialism: “Information processing is as broad as our culture and as deep as interplanetary space. To allow narrow interests, pioneering though they might have been, to preempt the name, to relegate ninety percent of the field to ‘an exercise left to the reader,’ would be disastrous to the underlying unity of the new information sciences.”³⁵ Several attempts were made during the next decade to make the ACM more relevant to the business community. In response to widespread criticism of the theoretical orientation of the Journal of the ACM, a new publication, Communications of the ACM, was introduced in 1958. The main contents of Communications were short articles, mostly unrefereed, on technical subjects such as applications, techniques, and standards.³⁶ In 1966 the Executive Council announced a $45,000 professional development program aimed at business data processing personnel. The program included short “skill upgrade” seminars offered at the national computer conferences, a traveling course series, and self-study materials.³⁷ There was even talk, in the mid-1960s, of a potential merger with the DPMA. In 1969, ACM president Bernard Galler announced a move toward “less formality, less science, and less academia.”³⁸

Despite these short-lived efforts to reconcile with the business community, however, the conservative ACM leadership continued to pursue a largely academic agenda. As early as 1959 it was suggested that the ACM should impose stringent academic standards on its members, and in 1965 a four-year degree became a prerequisite for receiving full membership. Frequent battles arose over repeated attempts to change the name of the association to something more broadly relevant. In 1965 a proposal to change it to the Association for Computing and Information Science was rejected; a decade later the same issue was still being debated.³⁹ When Louis Fein suggested in 1967 that the ACM faced a “crisis of identity,” ACM president Anthony Oettinger insisted vehemently that the ACM had no such crisis. In doing so, he reaffirmed the association’s commitment to a theoretical approach to computing: “Our science must, indeed, ‘maintain as its sole abstract purpose of advancing truth and knowledge.’”⁴⁰

This commitment to abstract science was further reinforced the following year when the ACM’s C³S announced its Curriculum ‘68 guidelines for university computer science programs. Curriculum ‘68 advocated a rigorously theoretically approach to computer science that included little of interest to business practitioners.⁴¹ Even when the ACM did recognize the growing importance of business data processing to the future of its discipline, the emphasis was always placed on research and education:

All of us, I am sure, have read non-ACM articles on business data processing and found them lacking. They suffer, I believe, from one basic fault: They fail to report fundamental research in the data processing field. The question of ‘fundamentalness’ is all-important. . . . In summary, this letter is intended to urge new emphasis on FUNDAMENTALISM in business data processing. This objective seems not only feasible but essential to me. It provides not only a technique for getting ACM into the business data processing business, but a technique (the same one) for getting the field of business data processing on a firm theoretical footing.⁴²

There is little question that throughout the 1960s, the ACM pursued a professionalization strategy that was heavily dependent on the authority and legitimacy of its academic accomplishments.

It was not until the 1970s that the ACM began to seriously reconsider its policy toward business-oriented practitioners. In 1974 the ACM Executive Council commissioned a series of studies on business programming as part of its long-range planning report. In doing so, the ACM was responding both to long-standing criticism and a recent spate of anti-ACM editorials that had appeared in the industry newsletter Computerworld. “ACM had become not so much an industry professional group,” declared one of these editorials, “as it was a home for members of educational institutions around the country to overwhelm us with their erudition on topics of vaguely moderate interest.”⁴³ The author noted that while most business data processing installations had standardized on the COBOL and FORTRAN programming languages, the ACM still supported ALGOL. He quoted ACM president Anthony Ralston to the effect that although only 25 percent of the ACM membership were academics, ten out of twenty-five council members were academics.⁴⁴

The long-range report noted that of the 320,000 software personnel then working in the United States, 85 percent dealt with business data processing. It admitted that while the ACM had a reputation for professionalism, “BDP [business data processing] people tend to be turned off by ACM’s academically oriented leadership. . . . BDP professionals feel that academics don’t understand what BDP needs, and they’re right.”⁴⁵ It concluded that any new ACM members were likely to come from business data processing, and recommended the development of a new publication aimed at that audience. The report signaled to many in the ACM that the organization needed to broaden its membership and become more accommodating. The next few years witnessed a bitterly contested presidential election (the cornerstone of which was a debate over business data processing), yet another attempt to change the name of the ACM to something more broadly relevant, and efforts to reconcile with its business-oriented competitor, the DPMA.

The Data Processing Management Association

The DPMA originated in 1949 as the NMAA. The NMAA was founded as an association of accountants and tabulating machine managers. It held its first convention in 1952, and grew rapidly over the next decade. By 1957 it represented more than ten thousand data processing workers in the United States and Canada, and by 1962 more than sixteen thousand.

In 1962 the NMAA changed its name to the DPMA. This was in part an attempt to expand its membership beyond finance and accounting professionals, and in part a reflection of the changing status of its discipline within the corporate hierarchy. As Thomas Haigh has suggested, punch card divisions at many large corporations had, by the beginning of the 1950s, acquired new status as the providers of strategic business information and other forms of valuable corporate data. The replacement of tabulating machine technology with electronic computers created a new role for data processors within the corporation; in fact, it was as part of a shift toward electronic data processing that most corporations invested in their first electronic computing equipment. From its inception, therefore, the DPMA represented the largest professional association of computing personnel.

The establishment of the CDP program later that year was part of a larger strategy of professional development. It was announced in conjunction with the DPMA’s “Six Measures of Professionalism” program, which included as the “marks of professionalism” self-education, standard measures of knowledge, continuing research, a code of ethics, and mechanisms for self-policing and disciplining practitioners.⁴⁶ The DPMA’s many national conferences, local chapter programs and seminars, and DPMA publications and home-study courses were all directed toward the self-education of individual members. The CDP program was obviously its provision for establishing a means of “measuring a minimum level of knowledge in the field.” DPMA graduate research grants encouraged contributions to the “knowledge of the field.” The DPMA code of ethics was part of its original charter, and was the first of such codes to be established for the computer-related professions. Finally, although the DPMA had no existing mechanisms for determining and punishing misconduct, it promised that the association would take a leading role in the development of an industry policing program. Although the DPMA’s original focus was on data processing supervisors, more than those of any other aspiring professional organization its programs were aimed at the broad computing community. Programmers and systems analysts were clearly part of its imagined community of practitioners.

Unlike the ACM, the DPMA made every effort to reach a broad spectrum of data processing personnel. Although originally open only to data processing supervisors, by 1964 the national leadership was making determined attempts to cultivate programmers within its membership.⁴⁷ The structure of the organization, which included strong regional chapters, allowed for diversity, local control, and rapid expansion. Each region had its own representative on the Executive Council who served with several executive officers and implemented policy decisions from the International Board of Directors. In addition, the DPMA’s official publication, the Data Management Journal, encouraged submissions on a much wider range of subjects than did the ACM’s Journal or Communications. The DPMA also maintained a close association with the editors of Datamation, another widely read industry journal that focused on issues of timely concern and practical relevance.

The DPMA’s inclusive approach to professional development brought it into conflict with competing societies, particularly the ACM. The differences between the two organizations mirrored the larger tensions that existed within the computing community: academic computer scientists versus the business data processors; theory versus practice. I have already shown how this tension affected the adoption of the DPMA’s CDP program: the ACM’s obvious lack of support helped to undermine the program’s legitimacy and prevented its widespread adoption. This opposition was based on both philosophical grounds—many in the ACM believed that the CDP examinations were superficial and irrelevant— and institutional ones, since control over an industry-wide certification program would have granted the DPMA considerable political authority.⁴⁸ The two group also sparred over trivial issues, such as unauthorized use of member-address databases.⁴⁹ Despite several halfhearted attempts to explore an ACM/DPMA merger, or at least to establish an interassociation liaison, the two groups rarely communicated.⁵⁰ When AFIPS was established in the early 1960s, the NMAA and other industry-oriented groups were treated with dismissive contempt, and the DPMA resisted AFIPS affiliation until the mid-1970s. At a meeting arranged by AFIPS officials, for example, DPMA representatives were kept waiting, without explanation or apology, for over an hour.⁵¹

In the year that it was introduced, the CDP examination attracted 1,048 applicants, 687 of whom passed successfully. The exam itself included 150 multiple-choice questions on programming, numerical analysis, Boolean algebra, applications, elementary cost accounting, English, and basic mathematics (not including calculus). In response to criticism from the many otherwise-qualified programmers who did not have formal mathematical training or college-level degrees, the educational requirements for the CDP were suspended until 1965. The other prerequisites—three years’ experience and “high character qualifications”—were so vague as to be almost meaningless, and appear to have been only selectively enforced.

By the end of 1965, almost seven thousand programmers and data processing supervisors had taken the exam. Figure 7.3 shows for the years between 1962 and 1973 the total number of candidates taking the exam, the total number of candidates who passed the exam, and the cumulative number of CDP holders.

Figure 7.3

CDP recipients, 1961–1973.

The data in figure 7.3 reveal the mixed fortunes and troubled history of the CDP examination. The striking early success of the program, which more than quintupled in size in its first three years, suggests that many data processing personnel saw certification as an attractive professional strategy. This corresponds well with evidence from industry journals and other documentary sources. A survey of the candidates in 1963 reveals a remarkable range of background, experience, and education.⁵² For the examination session in 1966, however, the education requirements outlined in the original program announcement from 1962 were finally put in place. These requirements included specific courses in math, English, managerial accounting, statistics, and data processing systems. Whereas participation in the exam in 1965 had jumped by more than 300 percent from the previous year (possibly in anticipation of the imposition of these requirements), applications for the session in 1966 dropped by almost 85 percent. Of the eighty-eight scheduled examination sites, twelve were dropped for lack of attendance. A major controversy erupted within the data processing community, particularly in DPMA-oriented publications such as Datamation and Computerworld.

Advocates of the academic requirements argued that such requirements not only elevated the status and legitimacy of the CDP but also were standard for most other professions, including law, medicine, engineering, and accounting. Opponents claimed that the specific course requirements were ambiguous, meaningless, and irrelevant. The DPMA Committee for Certification, which administered the CDP program, was flooded with letters from disgruntled applicants requesting special dispensation. Each case had to be individually evaluated.⁵³ In 1966 only 1,005 candidates were approved to sit for the exam. In 1967, this number dropped to 646. This posed not only financial difficulties for the DPMA but presented a grave threat to the perceived legitimacy of the entire CDP program as well. Faced with the imminent collapse of their membership support, the DPMA admitted that “the established eligibility requirements had unintentionally excluded some of the people for whom the CDP program was originally designed.”⁵⁴ The committee dropped the specific course requirements, providing a grandfather clause for those with three years’ experience prior to 1965, and requiring others to have only two years of postsecondary education. Applications for the exam session in 1968 jumped back to almost three thousand.

Over the next several years, the CDP program struggled to regain its initial momentum. Annual enrollments dropped again briefly in 1969, then leveled off for the next several years at about twenty-seven hundred. In an industry characterized by rapid expansion, this noticeable lack of growth represented a clear failure of the CDP program. With each year CDP holders came to represent a smaller and smaller percentage of the programming community. In 1970 the program faced yet another crisis: the announcement that a bachelor’s degree would be required of all CDP candidates, beginning with the examination in 1972. Once again a firestorm of debate broke out. The DPMA claimed that this new requirement merely reflected the changing realities of the labor market: since a college degree had already become a de facto requirement within the industry, requiring anything less for the CDP would severely undermine its legitimacy. The resulting controversy highlighted already-existing tensions within the data processing community, and further divided the already-fragmented DPMA Certification Council (many of whose own members could not satisfy the new degree requirement). Numerous observers called for the DPMA to relinquish control of the CDP examination to an independent certification authority. By the mid-1970s it became increasingly clear that the CDP program faced imminent dissolution.

In an attempt to restore momentum to their flagging certification initiative, the DPMA joined forces with seven other computing societies—the ACM, the IEEE (Institute of Electrical and Electronics Engineers) Computer Science Society, the Association for Computer Programmers and Analysts, the Association for Education Data Systems, the Automation One Association, the Canadian Information Processing Society, and the Society of Certified Data Processors (SCDP)—to form the Institute for Certification of Computer Professionals (ICCP). The DPMA had always been extremely possessive of its certification program, and its decision to relinquish control to an independent foundation reflects a growing sense of desperation about the future of the CDP.⁵⁵ The ICCP was charged with upgrading and expanding the CDP program, introducing new specialized examinations, and promoting professional development. In 1973 the ICCP took over responsibility for the CDP examinations. It also worked to develop a code of professional ethics to be adopted by its member organizations.

The ICCP failed to revive the CDP or institute a meaningful certification program of its own. Because it represented such a wide variety of constituents, the ICCP was hindered by the same internal divisions that plagued the larger programming community. Rivalries among the constituent member societies, many of whom were only superficially committed to the concept of certification, doomed the organization to internal conflict and inactivity.⁵⁶ The failure of the various competing professional associations to cooperate crippled the ability of the ICCP to develop meaningful certification standards. No single program was able to reflect the diverse needs of the collective software community. Furthermore, a series of highly critical assessments of the validity of the CDP examinations weakened popular and industry support.⁵⁷ The ICCP failed to present appealing alternative programs or examinations, and the organization languished during the 1970s.

In response to the inability of the professional associations to establish rigorous certification programs, the SCDP adopted an approach to professional standards that circumvented the ICCP altogether: state licensing of computer professionals. The SCDP was a grassroots organization of CDP holders dedicated to improving the status and legitimacy of the CDP program. Founded by the self-professed gadfly Kenniston W. Lord, the SCDP frequently challenged the wisdom and authority of associations such as the DPMA and the ICCP. For many years, Lord and his fellow SCDP member Alan Taylor carried out a vituperative verbal campaign against the DPMA (and later the ICCP) in the pages of the weekly newspaper Computerworld.⁵⁸ Taylor, a popular columnist for Computerworld, accused the DPMA of running the CDP examinations as a profit-making enterprise rather than an independent professional development program.⁵⁹ When the SCDP was denied formal representation in the ICCP in 1973, Lord proposed what was effectively a government takeover of responsibility for programmer certification. Unlike the certification programs voluntarily adopted by individuals and associations, however, government licensing would be mandatory. Since it is illegal to practice a licensed profession without the prior approval of the state, entry into that profession could be tightly controlled and monitored. Licensing would provide both control and protection as well as a certain degree of public recognition and legitimacy.

In 1974, the SCDP developed a model licensing bill and submitted it to a number of state legislatures. According to its model legislation, no person in a state that passes the SCDP bill could “practice, continue to practice, offer or attempt to practice data processing or any branch thereof” without either achieving a four-year degree in data processing and gaining three years of related experience, or successfully completing a certification examination and five years of experience. The bill also provided a five-year window in which those with twelve years of experience could be “grandfathered” into the profession. Practitioners were granted a twenty-four-month grace period in which to acquire the necessary qualifications. The legislation covered a wide variety of occupational activities and titles, including any that made use of the terms “data processing,” “data processing professional,” “computer professional,” or any of their derivatives. The state was given the power to revoke the certification of any registrant who committed fraud, was proved guilty of negligence, or who violated the professional code of ethics.⁶⁰

The proposed SCDP legislation is notable as the only concerted attempt in this period to encourage government involvement in the programming labor market. In fact, the specter of externally imposed state regulation had been raised as a primary justification for establishing certification programs in the first place: since self-regulation was considered to be one of the defining characteristics of a profession, surrendering control over this function to the state was essentially an admission of defeat. Observers warned that the lack of a solution from within the science would result in a solution imposed from without: “In several fields, the lack of professional and industrial standards has prompted the government to establish standards.”⁶¹ Ironically enough, even the defeat of the SCDP legislation proved humiliating to some practitioners; the state’s unwillingness to legislate data processing activities was perceived as a slight to the entire industry’s importance and reputation.⁶²

Although the model SCDP legislation was adopted by none of the states to which it was submitted, the fact that it was proposed at all reveals one of the primary shortcomings of voluntary certification programs such as the CDP: the lack of effective methods of enforcement. The inability, or unwillingness, of associations like the ACM and the DPMA to self-regulate was widely criticized by industry observers. Neither group had ever taken action against one of their members accused of fraud or negligence, and both had reputations for being unwilling to take strong positions on issues of public interest or safety. Indeed, the DPMA was unable even to enforce the proper use of the CDP trademark. Individuals and organizations that abused the CDP designation, either by claiming to have received a CDP when in reality they had not or instituting their own CDP programs, received only ineffective warning letters. No legal action appears to have been taken.⁶³ According to SCDP president Kenniston Lord, the inability of the profession to regulate its own activities justified drastic action in regard to state licensing: “One does not truly have a profession until one has the ability, legally, to challenge a practitioner and when proven guilty, to see that he is separated from the practice. . . . This is one problem that the SCDP bill will solve.”⁶⁴

The lack of ability and willingness of the DPMA to equip its certification program with teeth was not the only reason why the CDP failed to achieve widespread industry acceptance, however. The program had other shortcomings as well. From almost the beginning, the examinations had been tainted by accusations of fraud and incompetent administration. In 1966 several individuals reported receiving offers from an existing CDP holder to take their examinations for them for a fee.⁶⁵ A copy of the 1965 exam was stolen from a locked storage cabinet at California State College, and its disappearance was covered up by the DPMA Committee for Certification.⁶⁶ Complaints about testing conditions and locations were frequent and vociferous. For example, at one examination site at the University of Minnesota, the noise caused by a nearby drama club rehearsal of a sword fight scene “was so severe as to shower the room with particles of plaster.”⁶⁷ Other examinees suggested that poorly trained proctors (“the little old lady who passed out the papers”) were not only unable to answer even basic questions about content and procedure but also in some cases switched rooms without notice, started sessions early for personal convenience, and misplaced completed examination booklets.⁶⁸ Although such administrative snafus were hardly unique to the CDP program, they undermined public confidence in the ability of the DPMA to adequately represent the profession.

Another reason why the DPMA was unable to push through its certification initiative was a lack of support from other professional associations. An article in 1968 on certification and accreditation in the Communications of the ACM failed to mention the CDP program. This conspicuous neglect of the most successful certification program then available reflects a growing tension between the two competing professional associations. The ACM recognized that a successful certification program required a strong controlling organization. The organization that controlled certification would effectively control the profession. Indeed, the proposal that launched the CDP program in 1959 suggested that “the first association to undertake a Data Processor’s Certificate is going to be the leading association in the data processing field.”⁶⁹ Opposed to the idea that this controlling organization could be anything but the ACM, the Executive Council of the ACM worked to undermine the efforts of the DPMA at every occasion. In 1966 the council considered a resolution, clearly aimed at the CDP, to “warn employers against relying on examinations designed for subprofessionals or professionals as providing an index of professional competence.”⁷⁰ An early draft of this document referred specifically throughout to the “DPMA certification program.” Although the final published version referred only to unspecific “certification programs,” the target of its attacks was obviously the CDP. Later that year the Executive Council established a Committee to Investigate the Implications of the CDP. The first order of business for the committee was the drafting of a strongly worded objection to the use of the word professional in association with the DPMA exam, and the wording of subsequent exam and program literature eliminated all references to such language: CDP therefore came to stand for “Certified Data Processor,” rather than “Certified Data Professional.”⁷¹ Even this modest acronym was offensive to some professional groups. A member of a SHARE (an influential IBM users group) panel on certification was “disturbed to read [the] statement that many DPMA certificate holders are beginning to use the initials ‘CDP’ in their titles.” Such pretentious behavior, he suggested, “will quickly bring down upon DPMA the wrath of other professions. It is probably illegal in some states. I fail to see how it can conceivably benefit the cause of professionalism which DPMA and others of us are working toward.”⁷² Although the DPMA insisted that “many persons who use the CDP initials do so more to publicize the certification program” than to promote their own personal interests, pressure from competing associations forced it to abandon many of its more ambitious claims for the CDP program.⁷³ A statement in 1966 conceded that “it would be presumptuous at this early stage in the program to suggest that CDP represents the assurance of competence, or that the Certificate should be considered as a requirement for employment or promotion in the field.”⁷⁴ It is no wonder that so many employers and practitioners lost confidence in the ability of the DPMA to successfully administer an industry-wide certification program.

An even more troublesome problem for the DPMA was resistance from its primary constituency to its proposed educational requirements. The original CDP announcement included a list of specific academic prerequisites, including college-level courses in math, English, managerial accounting, statistics, and data processing systems as well as eight out of seventeen possible electives.⁷⁵ Many of the practicing EDP specialists who formed the core of the DPMA membership saw such requirements as being irrelevant, unattainable, or both. When the educational requirements were first enforced in 1966, applications dropped by more than 85 percent, never to recover.

The problem was not only that the new educational requirements were overly stringent for many aspiring EDP professionals; they were also entirely too specific. What exactly counted as a math, English, or managerial accounting course? Course titles and descriptions varied greatly by institution. Each application had to be evaluated individually to determine which courses legitimately counted toward the requirement. The Committee for Certification was immediately overwhelmed with paperwork: complaints, transcripts, notes from faculty, requests for exemptions, and so on. This was in addition to the massive efforts required to assure that each candidate had the requisite three years’ work experience and high character qualifications. It is unclear exactly what was meant by this requirement. It does appear that certain candidates were eliminated on the basis of having misrepresented their qualifications, or having committed fraud or other crimes, but no written standards for the high character qualification seem to have existed. The situation quickly turned into an administrative nightmare for DPMA officials. The specific course prerequisites were soon replaced with a more straightforward, although no less controversial, two-year college requirement. When this prerequisite was modified to a four-year degree in 1972, opposition became even more vociferous. The head of the West Tennessee chapter of the DPMA wrote to complain that he, along with about one-third of his chapter’s membership, had suddenly become ineligible to receive the CDP.⁷⁶ A Computerworld survey in 1970 indicated that many practitioners felt the new requirement “unduly harsh” and “ludicrous,” believing that it would decimate the data processing staffs of many smaller departments.⁷⁷ The always-outspoken Herbert Grosch (himself a PhD astronomer and president of the ACM from 1976 to 1978) declared that “this policy is very ill-advised. What the hell is so hot about college—it turns out a bunch of knuckleheads—and a knucklehead PhD is no better that a knucklehead CDP.”⁷⁸

Despite the strong negative reaction generated by these educational requirements, the DPMA leadership continued to insist on their necessity. Such requirements had always been considered an essential component of the DPMA’s professionalization program: only by defining a “standard of knowledge for organizing, analyzing, and solving problems for which data processing equipment is especially suitable” could programmers ever hope to distinguish themselves from mere technicians or other “sub-professionals.”⁷⁹ Like the academic computer scientists, business programmers recognized the need for a foundational body of abstract knowledge on which to construct their profession; they differed only about what that relevant foundation of knowledge should include. In insisting on strong educational standards, the DPMA was in complete accord with the conventional wisdom of the contemporary professionalization literature.⁸⁰ And by the end of the 1960s, it was true that many employers did prefer to hire college graduates—although not necessarily computer science or data processing graduates—for entry-level programming positions.⁸¹ According to a study published in September 1968 by the Office of Education, U.S. Department of Health, Education, and Welfare, 61 percent of the 353 business data processing managers surveyed preferred that programmers have a college degree. Over 60 percent indicated that educational background was a substantial factor in determining a programmer’s chances for promotion.⁸² As a recession hit the industry in the early part of the 1970s, this trend became even more pronounced.⁸³ An aspiring EDP school graduate, even with a CDP certificate, had little chance of breaking into data processing without a college degree. As one of these individuals lamented, “They told me 80% of all programmers don’t have a college degree. Now everywhere I go I’m told they’re sorry but they only want college people.”⁸⁴ Although the DPMA’s decision to raise the educational requirements for the CDP was highly controversial, it was also probably justified.

Ultimately, however, the DPMA never managed to convince employers and practitioners of the relevance of its educational standards, nor for that matter its certification exams. Neither group was convinced that a CDP meant much in terms of future performance. The DPMA Certification Council was not even able to pass a resolution requiring its own officials to possess the CDP.⁸⁵ In 1971, the Certification Council decided to drop the baccalaureate degree requirement. Although this decision was a response to pressure from within the data processing community, it was widely regarded as a sign of weakness rather than judicious concession.⁸⁶ As the director of the computing center at Virginia Tech wrote to the president of the local DPMA chapter, “The removal of the degree requirement has forced all of us to consider the attainment of the CDP not as an extension of our normal academic and work experience, but, as a matter of fact, something quite inferior to either one.”⁸⁷ His letter provides a stinging but accurate indictment of the failure of the CDP program to achieve widespread acceptance and legitimacy:

My experience indicates that people seek certification from their professional peer group for only two reasons. Either it is required by law or the individual feels that the mark of acceptance stamped upon him by his peer group is sufficiently important to be worthy of the extra effort to achieve that certification. Unfortunately, in the data processing profession, many, certainly most, of the people we recognize as outstanding professional achievers and accomplishers, do not hold the CDP.⁸⁸

One of the major criticisms leveled against the CDP examination by employers and data processing managers was that it tested “familiarity” rather than competence.⁸⁹ It was not clear to what skills and abilities the CDP was actually intended to certify: “The present DPMA examination measures breadth of data processing experience but does not measure depth. . . . It certainly does not measure or qualify programming ability. It makes no pretense of being any measure of management skills.”⁹⁰ The problem was a familiar one for the industry: although most employers in this period believed that only “competent” programmers could develop quality software, no one agreed on what knowledge and abilities constituted that “competence.”⁹¹ As Fred Gruenberger suggested at a RAND symposium in 1975 on certification issues, “I have the fear that someone who has passed the certifying exams has either been certified in the wrong things (wrong to me, to be sure) or he has been tuned to pass the diagnostics, and in either case I distrust the whole affair.”⁹² His attitude reflects the ambivalence that many observers in this period felt about contemporary data processing training and educational practices. If data processing was simply a “miscellaneous collection of techniques applied to business, technology and science,” rather than a unique discipline requiring special knowledge and experience, then no certification exam could possibly test for the broad range of skills associated with “general business knowledge.” “Given the choice between two people from the same school, one of whom has the CDP, but the other appears brighter,” Gruenberger argued, “I’ll take the brighter guy.”⁹³

Although the DPMA revised and updated its examinations annually, and eventually introduced a Registered Business Programmers exam intended specifically for programmers, it was never able to convince the industry of the relevance of its certification programs. One data processing manager suggested that the CDP was at best “a minor plus for the person who can measure up to other standards,” but that it would never be considered a “real” qualification for employment.⁹⁴ Another warned of a “lack of confidence” in the validity of the CDP exam: “I do not expect to apply for a CDP or to use the possession of a CDP as a criterion for employment.”⁹⁵ Still another resented a perceived attempt on the part of the DPMA to foster a “closed shop” mentality, promising to “continue to regard the CDP holder with suspicion as to motive and qualification, the level of suspicion being in inverse proportion to the date of the certificate.”⁹⁶ In the absence of a strong commitment to the CDP on the part of employers, many programmers saw little benefit in participating in the program. Those who did were increasingly self-selected from the lowest ranks of the labor pool—individuals for whom the CDP was a perceived substitute for experience and education.

Professional Societies or Technician Associations?

In spring 1975, on the eve of the annual National Computer Conference, a small group of the elite leaders of the computing community met in a nondescript conference room at a Quality Inn in Anaheim, California, to discuss the future of the computing profession. Similar meetings had been convened every year for the previous two decades, always with the intent to address the most pressing issues facing the computing community. Although the specific composition of the group changed from year to year, the attendees always represented the highest levels of leadership in the discipline: award-winning computer scientists, successful business entrepreneurs, association presidents, and prolific authors. The cumulative list of participants reads like a who’s who of the computing industry: Gene Amdahl, Paul Armer, Herbert Bright, Howard Bromberg, Richard Canning, Herbert Grosch, Fred Gruenberger, Richard Hamming, J.C.R. Licklider, Daniel McCracken, Anthony Oettinger, Seymour Papert, and Joseph Weizenbaum, among many others. This particular meeting included high-ranking representatives from all of the major professional societies: the ACM, the DPMA, the IEEE Computer Society, and the ICCP. These societies represented the largest and most influential constituent members of the umbrella organization, AFIPS. On the agenda was a discussion of the role of AFIPS in the professional development of the discipline.

AFIPS had been founded in 1961 as a society of societies. The immediate goal had been to provide a U.S. representative to the upcoming International Federation of Information Processing (IFIP) conference. IFIP had been established several years earlier under the aegis of the United Nations Educational, Scientific, and Cultural Organization (UNESCO). Beginning in 1959, IFIP hosted an annual international conference on computing. Each member nation was allowed to send representatives from a single organization. Since the United States had no single organization that spoke for its computing community, AFIPS was created to represent three of the largest computer-related societies: the ACM, the American Institute of Electrical Engineers (AIEE), and the Institute of Radio Engineers (IRE). (The AIEE and the IRE later merged into the IEEE.) It was hoped that AFIPS would eventually come to serve as the single national voice for computer interests in the United States.⁹⁷

From the start, AFIPS proved a disappointment. AFIPS did represent the United States at the annual IFIP meeting. It was given control over the lucrative Joint Computer Conferences, but beyond that, it proved incapable of serving as “the voice of the computing profession in America.”⁹⁸ It was crippled by a weak charter and a lack of tangible support from its founding societies. AFIPS was a society of societies, not a society of members, and it was therefore dependent on and subservient to the interests of its constituent societies, rather than to the larger computing community. In addition, several obvious candidates for membership, including the DPMA, had been conspicuously excluded from participation, and the AFIPS voting structure made it obvious that additional members would be unwelcome.⁹⁹ Even more limiting was a clause in the constitution, insisted on by the ACM as an essential precondition for its support, prohibiting AFIPS from placing itself “in direct competition with the activities of its member societies.”¹⁰⁰ Although the constitution was revised in 1969 to provide for stronger leadership and a more inclusive atmosphere, AFIPS continued to struggle for support and recognition. The DPMA did not join until 1974, for example, and even then without much enthusiasm. The gathering in 1975 of the computing elite at the Quality Inn in Anaheim represented one of the many attempts to reinvigorate interest in this ailing association. In 1989, just two years after celebrating its twenty-fifth anniversary, AFIPS voted itself out of existence. The loss of control over the lucrative National Computing Conferences left it financially unstable and without any clear means of support. Few in the community mourned its passing.

The transcripts of the meeting are revealing. The existence of a powerful professional association was obviously considered by the many influential members of the computing community to be the cornerstone of a strong professional identity. And yet rivalries between the member societies, particularly the ACM and the DPMA, proved to be an endemic and ultimately insoluble barrier to the establishment of this identity. Participants in the various associations disagreed over membership qualifications, dues, voting privileges, and certification and licensing proposals. More important, however, was the lack of widespread popular support for these associations. One Datamation study indicated that less than 40 percent of all programmers belonged to any professional association, and “probably less than 1% do anything in connection with an association that requires an extra effort on the individual’s part.”¹⁰¹ And even these low figures were probably inflated: a Wall Street Journal report from the same year revealed only that 13 percent of the data processing personnel surveyed belonged to any professional society.¹⁰² These numbers correspond well with the low level of interest in the CDP certification program. Although it is difficult to compile exact figures on association membership, it is clear that at best only a small percentage of the eligible population chose to participate in any professional society.

If strong professional associations were widely perceived to be an important element of professional identity, why did groups like the ACM, the DPMA, and AFIPS have such difficulty attracting and keeping members? AFIPS had some obvious structural problems that almost assured its ineffectiveness. Individuals could not directly join AFIPS; it was merely an umbrella organization for other associations, and possessed little real authority. But what about the ACM and the DPMA, the two largest relevant member societies? Both of these groups were established early, were relatively high profile, and published their own widely distributed journals. Both were frequently mentioned as candidates for the position of the professional computing association. Yet neither was able to consolidate its control over any significant portion of the discipline’s practitioners. The reasons behind their failure suggest the limitations of professional associations as an institutional solution to the software crisis.

The persistent conflict between the ACM and the DPMA revealed a much larger tension that existed within the computing community. As early as 1959, the outlines of a battle between academically oriented computer scientists and business programmers had taken shape around the issue of professionalism.¹⁰³ Although both groups agreed on the desirability of establishing institutional and occupational boundaries around the nascent computer-related professions, they disagreed sharply about what form these professional structures should take. Observers noted a deepening “programming schism” developing within the industry, a “growing breach between the scientific and engineering computation boys who talk ALGOL and FORTRAN . . . and the business data processing boys who talk English and write programs in COBOL.”¹⁰⁴ Individuals who believed that the key to professional status was the development of formal theories of computer science resisted subprofessional certification programs and tended to join the ACM; business data processors who were skeptical of “cute mathematical tricks,” either supported the DPMA or ignored the professional societies altogether.

It is obvious that the turf battles that raged between the ACM and the DPMA during the 1950s and 1960s helped undermine popular support for both organizations. In response to extensive Datamation coverage of a RAND symposium in 1959 on “the perennial professional society question,” one reader commented that he “hadn’t laughed so hard in a decade. Are these guys kidding? You won’t solve this problem by self-interested conversation about it, nor is it solved by founding another organization.”¹⁰⁵ In a retrospective in 1985 on the troubled history of AFIPS, Harry Tropp suggested that “the question of turf seems to have been there from the beginning. It shows up in the [1950s’] Rand Symposium. . . . There were the hardware and software types and then there were the users. We had the east coast/west coast turf problems. What I am hearing today is a whole new evolution of different turfs as this information processing society explodes.”¹⁰⁶ The fact that the DPMA refused affiliation with AFIPS until the mid-1970s—largely because of the perception that the latter organization was dominated by the ACM—was a major factor in its perpetual ineffectiveness and eventual dissolution (in 1987, just two years after it celebrated its twenty-fifth anniversary). Herbert Grosch in particular was dismayed by the pettiness of the ACM-DPMA debates, which he believed detracted from the overall goal of establishing a legitimate professional identity:

I couldn’t care less who publishes some abstract scientific paper! What I want to know is how do we pull together a hundred thousand warm bodies that are working on the outskirts of the computer business, give them a high-priced executive director, lots of advertising, a whole series of technical journals; in other words, organize a real rip-snorting profession? Whenever somebody starts worrying about which journal what paper should be published in, we get bogged down in an academic cross-fire we’ve been in for ten years.”¹⁰⁷

As damaging as these interassociation rivalries were to the influence and reputation of the ACM and the DPMA, what really hurt them was the lack of support that they received from industry practitioners. Neither organization was able to clearly establish its relevance to the needs of either workers or their managers. “Neither organization . . . has done much for the industry or for society as a whole,” argued one Datamation editorial from 1965. “We think the time is ripe to more clearly define larger, more important long-range goals which distinguish a professional society from a technician’s association.”¹⁰⁸ Employers looked to the professional associations to provide a supply of reliable, capable programmers. As was apparent from the impassioned debates about the structure and relevance of computer science curricula, however, it was far from obvious to many managers that formal educational programs contributed much to the production of professional programmers. The ACM’s continued devotion to theoretical computer science made it seem out of touch with the practical demands of business. The DPMA’s CDP program, although it was much more oriented to business data processing, failed to achieve widespread industry acceptance. As a result, it also was not able to guarantee the kind of standardized labor force in which corporations were interested. Employers saw little value in either organization.

The Limits of Professionalism

In his monograph on Office Automation in Social Perspective from 1968, Oxford sociologist Hans Rhee observed that “the computer elite are beginning to erect collective defenses against the lay world. They are beginning to develop a sense of professional identity and values.” But the process of establishing professional attitudes and controls, and a professional conscience and solidarity, Rhee suggested, had “not yet advanced very far.”¹⁰⁹ He could just have easily been describing the computing professions as they existed a decade earlier or a decade afterward. By 1968 computing had acquired many of the trappings of professionalism: academic computer science departments, certification programs, and professional associations. And yet most computing practitioners were not widely regarded as professionals, at least not in the eyes of the general public. In 1967, for example, the U.S. Civil Service Commission declared data processing personnel to be nonexempt employees, officially categorizing programmers as technicians rather than professionals. Although this decision did not affect the lives or practices of programmers, it represented a symbolic defeat for professional associations such as the ACM, which lobbied hard to have it overturned.¹¹⁰

The inability of programmers and other data processing personnel to successfully professionalize raises a perplexing question for the historian: Given the apparent interest in professionalization on the part of both employers and practitioners, why were these efforts so ineffective? As was described earlier, industrial employers in the 1960s complained not so much about technical incompetence as a general lack of professionalism among programmers. “It was his distressing lack of professional attributes that most often undermines his work and destroys his management’s confidence,” declared Malcolm Gotterer. “Too frequently these people, while exhibiting excellent technical skills, are non- professional in every other aspect of their work.”¹¹¹ Increased professionalism would presumably address the most frequent complaints leveled against data processing personnel: an overreliance on idiosyncratic craft techniques; an arrogant disregard for proper lines of authority; shoddy production quality; and a lack of commitment to the best interests of the organization. On the surface, the professionalization of programming appeared to be an ideal solution to many of the most deleterious symptoms of the burgeoning software crisis.

There are a number of explanations for the failure of most professionalization programs. Internal rivalries within the computing community undermined the effectiveness of groups such as the ACM and the DPMA. No single organization could meet the needs of a diverse community of computer people that included everyone from PhD mathematicians to high school dropout keypunch operators. As Louis Fein pointed out in his discussion of the ACM’s crisis of identity, “It is not clear . . . that an organization can play simultaneously the role of a profession, of an industry, and of a science. . . . I cannot see that ACM members, or IEEE Computer Group members, or DPMA members, or Simulations Councils, Inc. members, are members of a profession. They are practitioners or scientists or engineers or programmers—members of a technical society.”¹¹² The attempts of the computer scientists to rationalize the practice of programming and produce a body of generally applicable programming theory set them at odds with vocational programmers. The seemingly inconsistent and idiosyncratic practices of working programmers were used as foils for the elegant constructions of the academic computer scientists. The attempts of the vocational programmers to appeal to the language and ideals of science and engineering were ridiculed. When asked to explain the linguistic transition from coder to programmer, the prominent computer scientist John Backus dismissed it as purely rhetorical: “It’s the same reason that janitors are now called ‘custodians.’ ‘Programmer’ was considered a higher-class enterprise than ‘coder,’ and things have a tendency to move in that direction.”¹¹³ As the programming community broke down into competing factions—such as theoretical versus practical, certified versus uncertified, and the ACM versus the DPMA—its members lost the leverage necessary to push through any particular professionalization agenda.

In addition to internal rivalries, the aspiring computing professions also faced external opposition. For many corporate managers, professionalism was a potentially dangerous double-edged sword. On the one hand, “professionalism might motivate staff members to improve their capabilities, it could bring about more commonality of approaches, it could be used for hiring, promotions and raises, and it could help determine ‘who is qualified.’” On the other hand, “professionalism might well increase staff mobility and hence turnover, and it probably would lead to higher salaries for the ‘professionals.’”¹¹⁴ Computer personnel were seen as dangerously disruptive to the traditional corporate establishment. The last thing that traditional managers wanted was to provide data processing personnel with additional occupational authority. Professionalism was therefore encouraged only to the extent that it provided a standardized, tractable workforce; professionalization efforts that encouraged elitism, protectionism, or anything that smacked of unionism were seen as counterproductive.

Perhaps the most important reason that programmers and other data processing personnel failed to professionalize, however, was that the professional institutions that were set up in the 1950s and 1960s failed to convince employers of their relevance to the needs of business. A Computerworld survey in 1974 indicated that “no technical society has ever captured and held the attention of professionals in BDP.”¹¹⁵ Employers looked to professional institutions as a means of supplying their demand for competent, trustworthy employees. As we have seen, although computer science programs in the 1960s thrived in the universities, in the business world they were usually seen as overly theoretical and irrelevant. Likewise, the DPMA’s CDP program failed to establish itself as a reliable mechanism for predicting programmer performance or ability. Neither the ACM nor the DPMA offered much to employers in terms of improving the supply or quality of the programming workforce.

Given this lack of active support from employers, the professional associations had little to offer most data processing practitioners. Neither a computer science education nor professional certification could ensure employment or advancement. In response to a Computerworld article in 1974 titled “Why Business Users Are Turned Off by ACM,” AFIPS president George Glaser remarked that “the general lack of success of ACM in attracting business data processing professionals to its membership has relatively little to do with the nature and extent of the services it offers them. It is, rather, more attributable to a lack of interest on the part of these ‘professionals’ in any professional society.”¹¹⁶ Glaser’s comment can be read either as an indictment of the apathy of the average computing practitioner or the policies of the ACM; either way, it suggests the strained relationship that existed between the two communities. Many working programmers saw little value in belonging to either the ACM or the DPMA, and support for both organizations as well as professional institutions in general languished during the late 1960s and early 1970s.