In recent years, it has become apparent that software is the driving technology in the computer industry. From the evolution of the UNIX and Windows platforms, we have seen that applications and markets are defined by software, rather than by the hardware in place, as had been the case with earlier platforms like the mainframe, VAX and AS400. Moreover, we have been told in our interviews that as much as 70% of the cost of developing a new hardware system is actually spent on software development. Our research also confirms that 80% of the lifetime cost of purchasing and upgrading a computer system is actually spent on software purchase, license, development (in-house or outsourced) and maintenance. Even microprocessor design, once a CAD-based activity, is now largely a coding and debugging process using a special-purpose software language called a hardware definition language.
As processor and peripheral manufacturing move to the Third World, we recognize that the future of the US computer industry lies in design, networking, telecommunications and interactive content, all based on sets of complex software technologies. On the downside, we also see that major computer industry problems such as interoperability, systems errors and project failures are in large part software problems. Thus, our examination of the computer industry must include a careful measurement and analysis of the software industry. Through interviews, student research and small-scale surveys, we are well along in the process of identifying the key issues - business, public policy, and educational issues, that have shaped the software industry. As we continue to investigate how these issues will influence the future development of the industry, we have also established an extensive, multi-faceted dialogue with software industry leaders and other interested groups in the US and abroad to augment our research activities as well as to disseminate our research findings most effectively.
The software team's first area of investigation into the software industry occurred in 1993, when Professor Edward Feigenbaum and a dozen graduate students spent several months studying the troubled Japanese software industry. They identified several key factors that appeared to account in large part for the problems the Japanese were experiencing in fostering a domestic software products industry. Among these key factors were:
The team went on to examine the same key factors in the US to see whether and how they differed, and whether these differences accounted for the relative success of the US software industry. As reported in more detail in earlier papers and conferences, as well as in the previous Sloan Progress Report, we found a substantially different business and cultural environment in the US which appeared to nurture software entrepreneurship, and to create a favorable climate for successful long-term growth of the US software industry.
Under the auspices of several graduate seminars and classes, students from both business and engineering disciplines studied various aspects of the US software industry. These studies included literature searches, small-scale surveys, and over 80 interviews with industry insiders.
One of the more significant and consistent results that we found in the course of our interviews was that software development is still considered a craft; that developers cannot control or manage the process effectively, and too often cannot balance the demands of innovation, time to market, quality and cost. Even leaders in software product development, such as Microsoft, still cannot accurately predict the time and cost to produce a major new product.
As a result of the prevalence of these issues in the minds of software executives and other software industry insiders, we decided to undertake a study of product development practices among software publishers. Over the summer quarter of 1995, software team researchers and a small group of graduate students developed a survey methodology and appropriate survey instruments. Our goal was to establish benchmarks, uncover best practices, and identify the key issues in software product management. An overview discussion of the pilot survey with its initial findings is available.
We continue to expand and enhance research reported in previous periods. For example, while the central focus of our software management practices survey was software development, we also used it as a tool to quantify some of our earlier anecdotal evidence about the industry's key issues and concerns. In addition, we continue to interview software industry participants, but have extended our discussions to include more specialists, more middle managers, and more representatives from foreign governments and companies. Some of these findings are described briefly below.
Patents. A substantial number of American software industry executives pointed to the US patent system as a potential source of problems for the future of the software industry. Interviewees were divided about whether patents should be granted on software, whether copyright was adequate, or whether new solutions for protecting software innovations should be devised, because of software's special status as half invention, and half creative effort. However, all expressed concern with the inability of the Patent Office to recognize prior art, to grant only appropriate patents, and to process applications in a reasonably timely manner. Many feared the existence of landmine patents (unknown to later developers) and patent applications still in process that would force their products off the market when they were later brought to light. Others were concerned that new software development would soon take a back seat to instigating and defending patent infringement suits.
Our findings confirm that many problems exist in the patent system, and that the Patent Office and the courts cannot keep up with the changing technology. We believe, however, that American pragmatism and an ever-growing market will keep problems under control through the growth of partnerships and licensing arrangements, and the still prevalent practice of many software companies to avoid seeking patents at all. These companies exploit the rapidly-changing market without benefit of protection, and have already moved on to newer products before intellectual property questions arise.
Two notes: (1) In contrast to the fears expressed by software company senior management, engineering and marketing managers express far less concern with patent-related problems. In last summer's pilot survey, only two software development managers said that patent considerations motivated any of their actions. In one of those situations, the company licensed a patented technology from another vendor; in the other, the company decided to drop a feature from their product. (2) While many software products companies still eschew the patent application process, 100% of companies included in our pilot survey applied for patents on their products, and 91% also copyrighted their works.
The Network is the Platform. The network is the new platform, replacing the standalone PC in the last half of this decade, just as the PC replaced timesharing on the mainframe as the platform of choice in the 80's. The success of the PC was driven by what we called a "killer app" for the platform -- the application that made large numbers of people decide they needed a PC. This application was the spreadsheet, which gave business people a powerful new functionality that they could understand and use immediately. The networked computer still lacks the powerful new application that will make the new platform a business or consumer necessity. Perhaps the Web browser will prove to be the killer app the medium needs; however, the supporting technologies that this application requires to be truly a driving force are still in short supply: most significant among these supporting technologies is adequate bandwidth into the office and home.
Foreign Competition. Foreign competition in the US software industry is small but growing. We continue to monitor the handful of foreign firms participating successfully in the US, but these companies still dominate few market segments. Successful entrants in the US include Nintendo and Sega. While foreign owned, these two companies make extensive use of third-party developers in the US, and dominate a segment that is characterized by a proprietary platform. SAP of Germany and Baan of the Netherlands are also gaining a rapid foothold in the client/server systems marketplace for customizable, off-the-shelf solutions. Overall, however, European and Asian software companies in the US remain conspicuous by their absence. Many factors, including longer development cycles and less available capital for new ventures, continue to dampen competitive challenges to US dominance.
Contract programming is one area that continues to experience strong growth in a number of Third World countries. Tata Consultancy and several other Indian software companies have together grown to a half-billion dollar offshore programming business, doing custom programming - especially for maintenance and for client/server systems transitions, for large corporate MIS operations. In contrast to corporate software development activities, software publishers appear to be much more reluctant to use offshore software development for their cutting-edge products. In the pilot survey we found that, of the 18% of companies that outsourced any portion of their new product development, none utilized a foreign contract programming firm.
Packaged Software Sales Channel. The consolidation of software vendors which is ever-present in the news is less of a problem among publishers than some believe. While some business observers fear that consolidation in mature segments will lead to higher prices or fewer choices, we see that new platforms, the growth of networks and the introduction of new processor and multimedia technologies continue to enable new software and systems solutions at an increasingly rapid rate. We do not expect the pace of innovation or new product introduction to slow down in the coming decade.
Training of Software Engineers. In several student studies, and reinforced in interviews and a small-scale survey, we found that many people engaged in the development of software were not trained software professionals. For example, those developing software embedded in cars, cellular phones, aircraft and consumer electronics were far more likely to be mechanical engineers or electrical engineers with minimal training in software development methodology. As embedded software becomes more complex and more networked, and as it must be maintained and modified over multiple releases, we believe that significant failures and problems will become more commonplace, partly because of the lack of software development training of those responsible for this software.
One particularly interesting student study found that microprocessor designers no longer designed chips in CAD environments; rather these electrical engineers develop chips by programming their specifications in a hardware design language (HDL) and then running simulations to test their functionality. More than half of those designers questioned were willing to acknowledge that they were indeed programming rather than designing in the traditional sense. We are working to communicate a broader view of what constitutes the full range of software and software development activities, to encourage more educational disciplines to recognize the need for software training as part of their curricula.
Quality. While a few segments of the software industry, such as aerospace and telecommunications, have long emphasized quality in software development, the US business and consumer software products market has enjoyed rapid growth and acceptance in spite of a persistent quality problem. European and Japanese software publishers appear to maintain higher standards for software quality. To date, however, US software product sales have been driven by time-to-market and features/functionality rather than product quality.
Our pilot survey results confirm quality's relatively lower priority: Eighty-one percent of respondents cited "technology leadership" as the best characterization of their products' positioning, while only 9% said that they positioned their product primarily on the basis of quality.
We expect this situation to change in certain segments in the next few years. Embedded software is likely to be one of the first beneficiaries of a new thrust in software quality. If this prediction becomes reality, Japanese and European publishers will have new opportunities to play in the US market. The SEI and other standards organizations are working to improve the quality orientation of software development in the US, but such changes will be only slowly accepted among US developers and publishers, and that acceptance will be driven by consumer demand. As the CEO of Autodesk, Carol Bartz, eloquently pointed out in a recent meeting of the Software Forum: "The industry doesn't know how to produce quality software. But we know we have only a few years to figure it out. They are starting to expect this stuff to work!"
In a September 27, 1994 article in the San Jose Mercury News, called Culture Played Role in Success of Software, columnist James J. Mitchell discussed some of our research findings on the Japanese Software Industry. The article is based on a talk which the founder of the Software Research Team, Edward A. Feigenbaum, gave to the Computer Industry Forum of the Stanford Computer Industry Project on September 20, 1994.
A draft of a short paper is available which describes some of our initial findings in more depth. Another working paper draft presenting a perspective on the problems of regulating the NII is also available. Both papers can be found in the Stanford Computer Industry Project Research Highlights section. A slide presentation of a recent conference talk on US Initiatives on the NII is also offered.
As one effort to further progress in the area of better data collection, the Software Study Team has submitted a recommendation to the Economic Classification Policy Committee of the US Office of Management and Budget to revise the Standard Industry Classification system as it relates to the Software Industry.
Please click here if you wish to be added to our Events Announcements mailing list, or to send any comments or questions.