Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
APPENDIX: INDUSTRIAL ATTITUDES TOWARD RESEE INTRODUCTION As part of its study, the panel undertook to gather certain basic statistical information concerning the flow of funds into industrial, academic, and other research and also concerning capital equipment available, sources on education of research personnel, and other relevant ~objective. parameters of research. It also sponsored a series of interviews with research managers in a representative variety of large and small industrial corporations. Interviews were conducted at six organizations. Because of the time constraints under which this project was carried out, we chose only six companies, all of which are located in the Northeast: · AAA Corporation. A major company in the computer marketplace, with a large and well-established research and development organization. · BBB Corporation. A quickly growing, medium-sized company with a history of innovative products. · CCC Computers, Inc. A small minicomputer systems company that has maintained one of the highest growth rates in the computer industry. · DDD Corporation. A small company, not normally associated with the computer business, that has made a significant investment in computer science research far in advance of serious commercial efforts in this area. · EKE Systems, Inc. A systems house that does the bulk of its business in contract research and development, but that is also involved in marketing software products. · FFF Technology Company. A small firm that markets a variety of high-technology computer-based products. The first version of this Appendix was prepared for the panel by S.R. Shiochet and F.J. Gilbert of MIT's Sloan School of Industrial Management and the Harvard Business School, to whom the panel would like to extend warm thanks. 75
76 Although these six companies cannot be considered a significant sample of American companies participating in the computer market, the wide variation among them has provided an interesting and useful range of response to our questions. The interviews conducted are reviewed in the remainder of this Appendix. However, it would be useful here to summarize the main conclusions that emerge from these interviews. The most striking observation is simply this: Little basic research is being carried on within the computer industry today, and there is even less interaction between basic research and applied research. We believe that the primary reason for this state of affairs is that, as noted earlier, the computer industry is not technologically mature. The raw mass of unexploited technological ideas and the rapid pace of technical developments based on existing or freely available science leave little opportunity for any but the very largest organizations to fund basic research in computer science. mese same factors, together with the inappropriateness of typical university structures for the management of large, practically oriented research programs, conspire to reduce business interest in most academic work in computer science, work that by comparison with such subjects as chemistry or biology might be expected to interest them profoundly. This point will be brought out in more detail in two of the discussions that follow, but the main facts appear clearly enough in the following summary table: Company AAA Corp. BBB Corp. CCC Computers, Inc. ODD Corp. EKE Systems, Inc. FFF Technology Co. Basic Research l Organized as research division None None Emphasized but not separately organized Sponsored research only None The level of research activity in the computer industry tends to fall off very rapidly as one goes from the largest industry leaders to mediumrsized and smaller corporations. Only AAA, 000, and EKE Systems, a firm that does research for the federal government, have any activities that qualify as basic research. THE INTERVIEWS Although the questionnaire used to guide the interviews covers a large amount of material, each of the interviews lasted only about one hour. Thus each interview ended up concentrating in those areas that were of greatest interest to the interviewee. In reporting these interviews, we have followed the emphasis of the individuals we interviewed.
77 AAA Corporation At AAA Corporation, our primary interview was held with Dr. Z. who is director of computer science research for AAA Corporation. He reports to Dr. R. who is the vice president for research. Dr. R. in turn, reports to the chairman of the corporation. In its present form, the Research Division at AAA Corporation is 20 years old. There are over 1000 people in the division, of whom two-thirds are technical staff. This number is growing by 2 percent per year. Research is 10 percent of the AAA Corporation's research and development budget and is almost exclusively conducted in the Research Division. m e division is supported via a bookkeeping tax on the other divisions. How Work Is Classified In the Computer Science section, Dr. Z sees the work as being divided into science and advanced development. One-third of his budget goes to science, the rest to advanced development. m e range of scientific research of potential interest to AAA Corporation is broad. Any project that can produce basic results in a field that could be useful to AAA is a candidate for funding. For example, because there is a general feeling that lasers will become quite useful to AAA, there are many laser projects, including some whose direct usefulness is not apparent Within the loose guidelines, projects and the people who work on them are evaluated purely in terms of scientific quality. Traditional evaluation measures such as the number of invited talks or papers published are used. Dr. Z emphasized that he takes a very hard line on science projects, insisting that they must remain pure science. The reason for this is that other criteria are applied to application projects, but in approving science, one is concerned to ensure that they are not disguised application efforts addressing vacuous applications. Dr. Z categorizes advanced development projects, the other two-thirds of the work done in the Computer Science section, into · Technology enhancement, i.e., the refinement of technologies that are in use by other divisions of the AAA Corporation. Usually, such work is done within the Research Division only if it is a longer range project than an operating division of AAA Corporation could be expected to support. · Alternative technology projects, which explore different ways of accomplishing functions that already are provided in existing AAA Corporation products or processes. For example, display technologies other than the cathode-ray tube are being explored. Dr. Z sees the challenge to the mainline AAA Corporation divisions implicit in these projects as being an important contribution of the research divisions; they either confirm the decisions of the operating divisions or emphasize alternatives that might not otherwise have been explored.
78 However, the Research Division does not undertake such projects unless they feel that they can develop a technology that is significantly better than a current field alternative that they might replace. · New business areas, which, as the phrase implies, explores opportunities of which AAA Corporation has not previously taken advantage. Work in robotics is an example of this. Initiation of Projects In spite of the large amount of activity that is involved, the current research culture in the AAA Corporation Research Division seems to center on researcher-initiated small projects. It is Dr. Z's estimate that more than 90 percent of all projects are initially proposed to him by the researchers who wish to work on them; most start as small-scale investigations of basic questions. Subsequently, the projects may remain small science projects or can grow to large-scale projects as they develop. For example, one basic device-technology project started in the 1960s as two people investigating a phenomenon; it has grown to 150 people. Work on an advanced data base query system has had a similar history of growth. In consonance with the aim of facilitating the start of small projects, an effort is made to minimize associated administrative tasks. Because only Dr. Z's approval and not a long bureaucratic decision process is needed to start a project, and because the early stages of most projects do not involve a great deal of money, the start of a new project is not a major event. Either a formal proposal for a project can be made to Dr. Z or an individual researcher can request 6 months or a year to look at a specific area before initiating a formal project. Bootleg projects are common; they are, in fact, encouraged by not charging for off-hour computer use. Dr. Z believes that all of this gives the Research Division the ability to react quickly to new ideas, without the burden that more stringent review processes would impose. He notes that about three-quarters of the ideas brought to him are rejected, but believes that this results largely from the large number of proposals generated by the informal nature of the proposal process. BBB Corporation We were able to interview two people at BBB Corporation--Dr. C, who is both the director for large business and integrated information systems and product line manager for office systems marketing in the Market Planning and Development Group, and Dr. A, the special assistant to the president.
79 Organization BBB Corporation sees itself as a systems house and attempts to organize itself in such a way as to facilitate the development of useful computer systems. In their eyes, this calls for efforts to break down hierarchical management structures, it calls for continual design review meetings among all the top people, and it justifies an emphasis on informality and blackboard sessions, which can involve the president of BBB Corporation along with other top people. This emphasis on frequent and informal meetings aims at improving the flow of technical communications within and among the system development groups who carry out the central technical activities of BBB Corporation. me company's complex and relatively flat management structure is also designed to help reduce organizational barriers to communication. Technical management at BBB is divided into two components. The first of these is the Market Planning and Development (MP&D) Group, which is organized around marketing areas. It has two directors, one for the small minicomputer system produced by BBB Corporation, the other in charge of its larger computers and integrated information systems. Within this part of the technical management structure there also function three product line managers, one each for office systems, telecommunications, and hardcopy devices. The second component of BBB technical management consists of a group of six vice presidents, all of whom report directly to the company president. Each of these vice presidents is responsible for product development activities in a specific technical area--large systems software development, telecom- munications software, hardware development, and internal applications (including diagnostic procedures, quality assurance, etc.~. With the exception of the hardware group, which is about 50 percent larger than the others, all of these technical groups are of approximately the same size. m e hardware and the software groups are interrelated through a matrix organization intended to promote the close cooperation required for computer systems development. Marketing planning and development {MP&D) is purposely kept separate from product development to prevent the planning function from getting bogged down in work on incremental product improvements. The MP&D people are responsible for providing long-range product improvements as well as new product development. Weekly design review meetings are held to provide overall coordination of BBB Corporation's marketing and technical plans. These meetings are regularly attended by the BOB president, the head of MP&D and his directors and managers, the engineering vice presidents, the head of manufacturing, etc. m e effect is to provide a fluid and informal technical management team with an emphasis on entrepreneurial activity by the product line managers. Budgets are used primarily to control expenditures relative to income by product line. Once again, this is a reflection of BBB's management style. In support of this informal, small-company style of management, BBB Corporation has built a new development center to bring all of their technical people into one place, further improving technical communication.
80 Attitudes to Research Despite its rapid growth, BBB Corporation does not perceive itself as being able to undertake much real research, which is regarded as a luxury. ffl ree years to market is about as far toward advanced develop- ment as they go--the market is Been as too dynamic to make longer range projects worthwhile. As an indicator of the pace of new developments, Dr. A pointed out that BBB introduces a new product about once every three business days. Activities having something of a research flavor center on the office of the special assistant to the president. The special assistant is charged with a variety of responsibilities, ranging from being an Industry snooper. to running the corporate library and overseeing BBB's patent operation. Technical observations and ideas for new products are circulated among the group of engineering vice presidents and MP&D in memos and informal reports prepared under his supervision. Interest- ing suggestions are followed up by the advanced software development group. Dr. A sees these groups as the technical driving force in the company. He stated that approximately 20 percent of their work falls into the category of advanced product studies of this type. In addition to his activities as an information gatekeeper, Dr. A administers a variety of more aggressive activities for bringing outside ideas inside. mese include the following: · Creation of special development groups to do in-house what someone outside has been seen to do. For example, a daisy wheel printer was developed in this fashion. · Joint ventures with small companies that have demonstrated special expertise. BBB participated in the development, and was the first user, of both the Intel 8080 microprocessor and the Shugart double sided, double-density floppy disc drive. · Acquisition of small companies for their technology. The technology for BBB's new image printer was purchased and YYY Incorporated was acquired for their photocomposing and digital voice expertise. Word processing systems were offered as an example of the develop- ment of an innovative product. m e BBB Corporation's approach to this field was to diversify their prior product line, moving out from the existing base, which had emphasized calculators and typing systems. They waited 2 years for the technology to become economical for the application and then entered the word processing market as swiftly as they could. Personnel Issues In hiring new personnel the BBB Corporation tries for a mix of new graduates and seasoned people, but seldom looks for very specific skills or technology-transfer opportunities. This goes back to their
81 feeling for how dynamic the marketplace is. Specialized skills are not that valuable: solid basic skills are what they look for. They do not like to employ consultants except to gain perspectives on market trends or to assess new markets. Nothing special is done in the area of rewarding development personnel beyond such normal business practices as a profit-sharing program. Hard work is what people get paid for at BBB. m e major personnel problem reported at BBB is finding enough well-qualified technical people. As it was put to us, the availability of people and the general quality of computer science education "stinks." BBB sees this problem as being so acute that they have sponsored the formation of an Educational Institute, a new school aimed specifically at training high-quality computer science people. The central problem that BBB hopes to overcome by the creation of this institute is the lack of attention to practical applications skills in conventional academic computer science education. CCC Computers, Inc. At CCC we interviwed Dr. M, Manager of Research and Development. As a prelude to this interview, Dr. M thought it important to sketch CCC's history to provide a context within which the role of R&D in the company could be understood. CCC Computers, Inc., has grown by an incremental process based on evolution of existing products, rather than by the introduction of completely new products. Originally a spinoff from HHH Computers, CCC produced as their first product a copy of a minicomputer that HHH had decided to discontinue. Much of CCC's early software was acquired via third-party agreements, capitalizing on the work of others. Even such a recent and innovative product as their communications network is based on research work done at the University of California at Irvine. AS a result of this strategy, CCC has been able to sustain an extremely high growth rate. Sales doubled each year until an intentional slowdown about a year and a half ago was put into effect to allow them to catch up with the organizational problems attendant on such rapid growth. Now in its eighth year of operation, CCC expects their sales to exceed $200 million. Organization of Technical Activities The bulk of technical activities at CCC are managed by the vice president for research, development and engineering (RD&E). Reporting to the vice president for RD&E are several product line directors and a director for the Research Division. In turn, the director for research has reporting to him both a research and development group (headed by Dr. M) and an advanced development group. The advanced development group is responsible for such activities as investigating new chips to see what can be done with them, maintaining contact with component vendors, coordinating company-wide projects, and so on. The research
82 and development group is small, only 11 people of a total of 300 or so in RD&E. The present activities of this small R&D group are concen- trated in three areas, one person is assessing the 'potential of functional programming, a few are involved in the development of future computer architectures, and the rest are working to define CCC's approach to office systems. The budgeting process at CCC is informal and reflects the loose organization and entrepreneurial atmosphere entailed by explosive growth. Although a corporate planning organization exists, it seems to have little real influence on the budgeting process. Instead, the director of research shapes his own ad hoc budgets, which are almost always approved by the vice president for RD&E. Dr. M sees this fluid, growth-oriented environment as a very challenging one for the management of research and development. He feels that handling this sort of challenge is a'real specialty within the field of R&D management--one that is common to the many smaller companies in the computer business. Research and Development Strategy There is not yet much of a basis for assessing the success of research projects within CCC since nothing that has been worked on in the R&D group has yet moved into the marketplace. Given the current state of the economy, Dr. M sees his job not so much as that of developing new products, but more as that of exerting a positive influence on the product line development groups by keeping them abreast of technical developments outside CCC. He noted that the product development groups get most of their input from marketing and tend not be very enlightened in their technical perspectives. m e R&D group also does some internal consulting for the development groups on particularly complex problems. Dr. M hopes to see the role of the R&D group evolve toward technology demonstration, but this can only happen as the financial and organi- zational stability of the company grows. To date, research has fielded only small project teams owing to the lack of manpower. However, the R&D group is now moving toward critical mass in a few areas. Almost all of the ideas that they pursue are internally generated and are not suggested by marketing or top management. Unofficial projects are allowed--this is seen as the best 'way of getting new things started in the organization and letting them live long enough for people to form opinions about their relative merits. Internal communications in the research group are handled on an informal basis and simply rely on the small size of the group. Communication between the research and product groups takes place in periodic top-level meetings to discuss the implications of new technologies and research trends, and in the internal technical consulting performed by the Research Division. However, Dr. M feels that there is a real problem in getting the development people to pay attention to what the research people have to say because they are so preoccupied with the next incremental product improvement.
83 Communications with the academic world make use of the usual mechanisms, e.g., CCC maintains a technical library and sends people to industry conferences. Dr. M would like to see the communication improve. When asked to comment on the quality of computer science research in general, Dr. M found himself uneasy without being able to say exactly why. He believes that the major universities {which he believed to be three in number--MIT, Stanford, and Carnegie-Mellon) appear to be having problems defining their directions. He sees them as following industry trends, not leading the development of new technical directions. In his view, graduates from the lesser schools seem to be fairly well trained in basic skills such as programming, but tend not to have any useful notion of what doing research is all about. He expressed the view that industry interest in research is cyclic; periods of interest in long-range developments alternate with periods of intense exploita- tion of new developments. On the whole, however, he thought that the industry has not matured enough to feel a need for long-term develop- ment projects. For all his misgivings concerning the quality of the work done in universities, Dr. M said that his group pays more attention to the ideas coming out of academia than to those coming out of industry. The major reason is that when industrial results become public, it is because a competitor has just introduced them to the market. Academic people are taken into confidence by everyone and hence enjoy a special position as information carriers, which he believes is an important part of their technical role. Personnel Issues The major personnel issue that has had to be faced in the Research Division is the difficulty of recruiting good people. Dr. M said that this problem has been particularly acute during the past few years, during which CCC Computers was relatively unknown. m e people to whom they made offers kept turning them down in favor of places like Xerox, Texas Instruments, or IBM Research Centers. Recently, this situation has improved because CCC has become better known and also because the larger research centers have stopped hiring so many people. At present, the CCC research organization is not having too much 'trouble finding people, although all the large technical organizations within CCC are still having problems because of their relatively greater need for people. In hiring new people, they are not usually looking for anything beyond a general competence in computer science. They stay away from people with too narrow a specialization and have not had much luck finding people when they needed someone with a very specific background. CCC makes little use of consultants except occasionally for market trend analysis and formulating business strategy. Even in those few cases in which consultants are employed, Dr. M believes it to be mainly for internal political reasons, not because they can provide technical expertise unavailable within the company.
84 Protection of Intellectual Property Intellectual property laws as yet have little influence on CCC strategy. CCC holds very few patents, although all of their software is copyrighted. They do not believe that legal protection makes that much of a difference in the computer business. Weir view is that software protects itself through the difficulty of integrating new models into large and complex systems and because of the difficulty of transporting software systems between different architectures. DDD Corporation The information in this section was gathered during an interview held with Dr. W. vice president for research, development, and engineering, and supplemented by information gleaned from a lecture on The Changing Role of Research at 000" by Dr. C, vice president for corporate research. Organization m e president of DDD Corporation has two staff groups that report to himr-corporate planning and the office of the chief scientist--and he is also assisted by a number of vice presidents for different functional areas, including research, development, and engineering (headed by Dr. W) and corporate research (headed by Dr. C). On the development side, groups are organized by product lines, such as copiers and office systems. m e research activities of DDD are centered in three laboratories that are specialized by technical areas, the A Research Center, which focuses on marking and imaging technologies, the B Research Centre Canada, which is concerned with materials processing research, and the C Research Center, whose focus is on digital systems research. Classification of Technical Activities DDD classifies their technical activities in four complementary ways: · Time phase. Projects are considered to move through four distinct phases; research, technology demonstration, development engineering, and design engineering. · Geography. Projects are distributed among research and development facilities in Japan, Canada, England, and the United States. · Technical specialty. Development efforts are planned to relate to existing product lines, such as copiers or office systems. · Expected time-to-market. They recognize a difference in expected time-to~market between research that is unique to DDD and research that is consonant with developments in the rest of the industry. Dr. W said that, in his experience, it has taken 10 to 15
85 years to successfully commercialize developments that were unique to the B Research Centre, while commercialization of projects similar to those under active development by others can occur quite quickly, sometimes in a matter of months. They try to stay away from defining any notion of "success. for individual research projects. Instead, their research efforts are oriented toward building corporate expertise in areas believed to be important to 000's long-range business interests. ffl ey find that market evaluation is too long in coming to be of much use. By the time a commercial ~judgment. is in, both market and technical conditions have changed so much that it is hard to draw any useful lessons. For example, a laser printer project was started in 1964 and has now achieved a good market position. However, the primary application area that is now envisioned for it (office systems) is not the one for which it was originally intended (high-volume computer printout) and in which most of its sales have been to date. Research Strategy Computer science research has a fundamental place in 000's long-range strategy. They see their present products as assisting only a very small part of the overall flow of paperwork in an office. Referring to Theodore Levitt's discussion of "marketing myopias Dr. W emphasized that DDD sees itself in the office communications business, not in the business of supplying any narrower group of products. They have seen the "electronic office," involving the rationalization and mechaniza- tion of all common office tasks, coming for some time and feared that if a competitor was first to incorporate copying into an overall office system they might find themselves locked out of a significant share of the market. They therefore consider some of their leading-edge technological products, e.g., their present word processing products, to be of great strategic importance because of the market entry that they provide, even though they currently account for only a very small portion of the corporation's revenue. 000's basic strategy for entering the office systems market was to develop a corporate base of expertise in digital systems far in advance of product requirements. Their feeling was that it was crucial to invest the time to develop and experiment with their own approach to the design of electronic office systems, on the theory that this would then allow them to formulate intelligent product development plans and to implement them quickly and successfully when the right time came. TO implement this strategy, DDD brought in Dr. C 10 years ago to lead the formation of the C Research Center. in explaining his charter for this activity, Dr. C pointed out that the role of the research laboratories within DDD is defined as strictly research, not research and development. Their basic goal is to demonstrate the technological feasibility of new ideas, ~hardening" concepts that circulate at large in the technical community and developing an internal capability to deal with them. However, Dr. C said about this arrangement that
86 although separation of research from the demands of product development is important, many problems within DDD derive from this too-strict separation of the research facilities from the business arm of 000, and the accompanying lack of adequate transfer and development mechanisms. When asked whether there was a definable DDD style of research, Dr. W said that he did not think that there was enough of a tradition to give a good answer. At the A Research Center, they do mainly basic science, and the work there is normally carried out in separate small groups. At the C Research Center, they are trying to do systems synthesis and have made a large investment in the facilities needed to coordinate their approach to digital systems development, but most work is still done by individual investigators. The comment about their C Research Center summarizes what Dr. W believes differentiates computer science research from the other types of research done by CCC Corporation--the need for expensive equipment and the effort that must be invested in coordinating separate develop- ments so that the work results in integrated systems rather than disjoint "packages. n Controlling the Research Process The overall planning and budgeting process at DDD revolves around a series of annual reviews. A corporate-wide 5-year strategic research plan is revised annually. Later each year, they revise and extend a more detailed 2-year operating budget that focuses on head counts, capital requirements, facilities, and so on. The long-range planning process is structured along lines defined by 000's classification of technical activities. An effort is made to assess their capabilities in each of the technical specialties they see as important to their business goals, to categorize (by time phase) the state of each project in progress, and to evaluate the performance of each product line and to determine whether it is receiving the technical support it needs. Both the long- and the short-range planning processes involved four interacting agencies that are seen as representing two sets of counter- balanced interests. On the technical side, there is the corporate research organization (headed by Dr. C) and the engineering RD&E group (headed by Dr. W). On the business side, there is the set of functional business organizations, each of which develops and proposes its own plans, and the corporate planning group, which is responsible for integrating these separate plans. Together with the chairman, president, and executive vice presidents, these four agencies form the Corporate Policy Committee. m e corporate planning group and the engineering RD&E staff track R&D projects, corporation-wide, in weekly review meetings. Although not too much pressure is applied, the research groups are not left to themselves for long periods of time. Dr. C commented that managing the research function raised the problem of steering a steady technical course through business planning storms. In fact, a debate concerning the future direction of the C
87 Research Center is currently raging. This laboratory was originally set up to develop a corporate capability in the general area of digital systems. Now that DDD has reached the point of moving into computing as an active business area, the question arises whether the C Research Center should still aim to stay 8-10 years ahead of the market or should begin to be more product and business oriented. m e ability to continue working on the feasibility of good concepts requires a careful and well-justified project selection process. Dr. C outlined a rather formal set of criteria that he uses for reviewing individual technical projects. The effort expended in these project reviews reflects the strategic importance of computer systems to the DDD company. The criteria applied are as follows: · Estimated relevance to projected DDD business areas. This implies a need to project business plans out for about 8 years. · Ripeness of the field for research exploitation: Are there good ideas to pursue? How many years will it take to achieve useful results? How many prior successes and failures by others have there already been? · Is the target technology or capability better obtained through purchase from vendors or by acquisition? · Is the magnitude of the necessary investment within 000's resources? · What is the opportunity cost of displacing an existing research program? · Is there an adequate prospect that a successful result could be transferred downstream? Dr. C noted that intellectual property laws do not play a role in the project selection process since they believe that there is no well-tested patent or copyright protection available in this area. The one major computer-oriented product that they have patented (a type of computer network) is being promoted as an office computer communications standard. For this to succeed, the DDD technology must become widely accepted and must therefore be available to imitation, making patent protection largely irrelevant. In this connection, Dr. W commented that software is easy to copy in isolation; fielding modular, well-supported systems is very difficult. EKE Systems, Incorporated This interview was given by Dr. R. the executive vice president responsible for all research and development activities at EKE Systems. Classification of Technical Activities The contractual basis of most of the work carried out at EKE Systems establishes an easy and clear distinction between research and development activities. Each of their contracts is awarded either for
88 the investigation of specific technical questions or to develop and deliver well-specified systems. In a research project, Success is defined by the granting agency. Typically, success depends upon three factors: 1. New scientific content, usually the answers to specific technical questions. 2. A demonstration of practical techniques. 3. Generation of interest elsewhere within the government or in academic circles. Although Dr. R agreed that there is such a thing as ~basic" research in computer science--he mentioned mathematically oriented work on the theory of computation--he said that there is a very fine line between most other types of research and advanced development. EKE gets involved with basic theory only in the context of specific projects, although they intend, when they can afford it, to develop an ongoing program of their own in theory. Organization of Research m e marketing arm of the R&D division acts as a liaison with the various federal agencies that award research or development contracts. Project concepts arise in roughly equal measure from within the R&D divisions (where prospects are then promoted to the agencies by marketing). The company's commercial products reflect, and for the most part stem, from its contract R&D projects. Budgets for the two research divisions have the nature of business plans rather than forming part of a corporate product planning strategy. In the product divisions the development groups propose their own budgets, which are then reviewed by the division directors and finally by Dr. R. The division directors have profit and cost goals that keep these budgets down and long-range product goals that keep them from approaching zero. This type of management relies heavily on the fact that EKE Systems is still a rather small organization involving only about 150 people altogether. Product planning and marketing efforts are guided by a 5-year plan that is revised as needed by occasional {about once a year on average) joint top management committees. In terms of research contracts, EKE takes just about anything that they can get. However, they are starting to emphasize development projects over research projects because of their larger size. EKE leans to the use of small project teams that take a single project through to completion. Research style is not usually specified in their contracts. The research divisions generally condone unofficial projects. Both of EKE's major commercial products (a data base management system and an electronic mail system) started in this way. Typically, an oppor- tunity is perceived as obvious by the research personnel and someone starts to tinker with the initial design of a system. At some point
89 the project comes to management's attention and is reviewed for official support. If a go-ahead is given, the project is handled as a normal contract project would be. Some of the key design personnel for a project that has germinated in this informal way may be moved into a product division (possibly a newly formed one) if the project reaches the commercialization stage. The progression of the electronic mail system design project illus- trates this pattern. Approximately six months of informal tinkering took place before the idea was officially adopted. After a financial partner and first user {YYY Computer Corporation) was found, a project team was organized and spent about one year developing a prototype version of the system. Another year was devoted to marketing develop- ment (opening sales offices, etc.) for the new product before a more aggressive market-push strategy was adopted. Overall, this product took 2 1/2 years to commercialize. Intellectual property laws (in particular, the lack of appropriate ones) are beginning to play a role in directing the course of new product planning at EKE. The only protection that they believe is workable for software products is the trade secret. Thus they are starting to steer away from projects that cannot easily be protected in this manner. For example, a great deal of effort was absorbed in the design of a good user interface for the electronic mail system -a mail system is the type of program in which, to a large extent, the user interface is the system. But it has now been copied and is unprotectable--the design of a user interface is not copyrightable and clearly cannot be kept secret. - Technical Communications Communications between the research divisions of EKE and marketing rely on the small size of the company, although there are also biannual meetings of the top management to review what is going on. Marketing has exerted an influence on product strategy, but has not been a major source of ideas. Communications among EKE's technical people also rely on the small size of the firm. Most R&D personnel are quartered in the same office complex, and it has an ongoing program of research seminars (by both insiders and outsiders) and a technical library and has a cooperative arrangement with the MIT library, which is near EKE's main location. Communications with the academic world involve the use of faculty members as consultants and exchanges of seminar speakers with univer- sities. EKE emphasizes the recruitment of academics on sabbatical to work as regular project team members. EKE makes frequent use of outside consultants, generally university faculty, and tries to maintain long- term relationships with them by setting up one day a week consulting arrangements. These people usually participate as regular project team members, not as occasional advice-givers. They find this a useful way to get around the difficulty of finding good full-time people.
so Personnel Issues Dr. R is not dissatisfied with the composition of his research staff, although he regards it as being too small. His major problem is an inability to find enough qualified people. The R&D groups at EKE are chronically undermanned--there is always more funding potentially available than they can find available people to spend it on. People, primarily technical people, are their limiting resource, and marketing people are also difficult to come by. In hiring new technical personnel, they usually look for expertise in the central areas of their current projects. Other than that, they do not care what else the person has done as long as he has a solid background in the fundamentals of computer science. Although they hire almost all of their new graduates from a limited set of schools (MIT, Stanford, Carnegie-Mellon), they do so because of the perceived excellence of the training and not because of any particular wish to achieve technology transfer. In the R&D divisions, research people are rewarded by a 5 percent profit distribution awarded for contributions, which are assessed on qualitative basis. m e product divisions make a similar arrangement for project managers. The State of Computer Research Development Dr. R perceives the general quality of computer science research as being quite high, but believes that almost all of the good work is product oriented and is being done in industrial laboratories. He sees this as one result of the dynamic nature of computer science: it is becoming an experimental science that requires large projects in order to make interesting advances. He contends that this is also the reason for what he perceives as the generally poor quality of academic work. Since few universities can provide the required resources for signifi- cant research, he strongly favors the idea of Centers of excellence.. He believes these centers could also provide high-quality training. All the new graduates EKE hires come from institutions regarded as centers of excellence, and currently they are happy with the quality of these people. When asked about innovations with real impact on the industry, Dr. R characterized LSI, personal computers, time sharing, data base management systems, and high-level programming languages as being the most important. With the exception of MIT's project MAC, he could not think of any academic work having substantial impact on these areas. In his view, the real innovation has all been industrial. FFF Technology Company This short section is based on a brief interview with Dr. H. the director of research and development at FFF Technology, and is included because FFF, a small firm that manufactures a variety of high- technology products, is a type of company that, though different from
91 any of the others covered in this report, is rather common in the computer industry. Dr. H classified the bulk of the work he is responsible for as product development activity. FFF Technology is in a position in which research or even advanced development is an unaffordable luxury. They view their aim as that of quickly taking advantage of price changes and new technology developments to build what they know the market wants next. Most of their major product concepts come from extrapolation of market trends or from working to solve specific customer problems. They rely heavily on the academic and industrial research that is available through technical magazines, conference proceedings, and trade shows, and try to pick up ideas on how to exploit existing tech- nologies in novel ways or in new areas. They watch market and technology/price trends carefully. They also work actively to meet to exchange ideas with academic researchers. But they perform no research. A REFLECTION ON "E INDUST~= I=E=~ - S AD Q=STIO~IRE In our industrial interviews, we found universal agreement that almost all of the major innovations in the computer industry have stemmed from industrial research and development groups focusing on specific products and applications. Typical of the advances cited as important innovations were large-scale integration, minicomputers, time sharing, data base management systems, and personal computers. Those few innovations that grew out of basic research, such as the ARPANET or the early work on satellite communications, were seen as large projects whose magnitude demanded federal funding and were carried out for the most part by contract research and development companies. Our general picture of the distribution of research activities in the computer industry was well summarized by one of our interviewees, Dr. C of the BBB Corporation, who offered us a perspective on his own organization's position in the computer industry by classifying computer corporations into three categories based on their size: 1. Small entrepreneurial ventures that are struggling just to manage production. 2. Intermediate companies (like BBB) that do a great deal of advanced product development, but very little research. 3. Giant organizations such as IBM, Texas Instruments, and Xerox that maintain corporate research laboratories and have the resources to pursue basic research. University work was characterized by Mr. F at BBB as ~tangential" to the needs of industry, due to the universities' lack of contact with product concerns. Dr. M at CCC Computers saw universities as tech- nological followers rather than developers of new concepts, but not as really capable of carrying out the types of large development projects that seem to generate interesting innovations. The major users of academic research in computer science seem to be the smaller firms like EKE Technology who are not leaders in innovation but who rely on university work to provide them with a competitive edge in exploiting existing technologies.
