For Computing Research, The Past Is Prologue, A CSTB Report

By Jerry Sheehan

The computing research community has an enviable track record of contributing to innovation. From graphical user interfaces to relational databases to speech recognition systems, many of today's desktop computing technologies trace their roots to university and industry laboratories. Popular accounts of such technologies tend to properly credit researchers for their pioneering efforts, but often overlook the contributions of the federal government to the innovation process.

A new report from the Computer Science and Telecommunications Board (CSTB), "Funding A Revolution: Government Support for Computing Research," remedies this imbalance. Authored by a committee of computing researchers, historians of technology, economists, and former government officials, the report reviews developments in computing between 1945 and 1995 to highlight the role of government in supporting research. The report examines specific developments in the areas of relational databases, the Internet, theoretical computer science, artificial intelligence and virtual reality, gleaning lessons that can guide future government efforts to support computing.

Tracing Government Contributions

The report finds that the government exerted tremendous influence on the development of computing technology. The federal government launched the computing revolution immediately after World War II and has continued to sustain it, even as the industry itself has evolved. It did so by contributing funds for research, supporting the education and training of students, and providing research equipment for university laboratories. As such, government became an essential element of the nation's research infrastructure in computing, complementing the roles of industry and universities.

Federal research funding is perhaps the most visible government contribution to computing. From less than $10 million in 1960, federal support for computer science research climbed to almost $1 billion in 1995. Though declining as a fraction of the nation's total computing research budget (until industry cut back its funding significantly in the early 1990s), federal funding supported a range of innovative technologies, including timeshared computing, computer graphics, packet-switched networks, and intelligent systems. In fact, over half the research papers cited in U.S. computing-related patents filed between 1993 and 1994 acknowledge public funding as a source of support.

Such funding had a profound affect on universities. Nationwide, federal funding comprised roughly 70 percent of university research funds in computer science between 1976 and 1995. Much of this funding was used to support graduate research assistants. A growing percentage of graduate students in computer science have received federal support for their studies, even as enrollments have increased. In computer science departments at Carnegie Mellon, MIT, Stanford, and the University of California at Berkeley, over half of all graduate students (master's and Ph.D.) received financial support from the government between 1985 and 1995. In addition, most of the funding used by academic computer science departments to purchase research equipment comes from federal agencies.

The results of such investments have been far-reaching. They have enabled federal agencies to perform their public missions and also create the technology base and human resources upon which the computing industry has been built. Companies such as Sun Microsystems, Silicon Graphics, Informix, and Netscape Communications were founded by researchers who developed technology with government support. Others, like IBM and Microsoft, have continued to hire trained graduates and have incorporated research results into their product lines.

Supported Research

Government funding has been effective in stimulating innovation largely because it supported work outside the scope of most industrial research investments. A significant fraction of federal monies has been aimed at long-term, fundamental research. Such research can produce large payoffs, but is inherently risky and often takes decades to produce tangible results. Further, the results can be so basic that it is hard for any individual company to keep its competitors from adopting them. Hence, industry can only support limited amounts of such research.

Federal efforts aimed at building large systems have also produced significant results. Consider the Air Force's Semi-Automatic Ground Environment (SAGE) project of the 1950s. Intended to develop a command and control system for responding to attacks by Soviet bombers, SAGE involved researchers from MIT, IBM, and other laboratories. The system served as a testbed that sparked innovations ranging from real-time computing to core memories to computer graphics. Many of the pioneers of computing learned their craft through hands-on experimentation with SAGE and subsequently staffed the companies and laboratories of the nascent computing industry.

The Internet is a more modern example of this phenomenon. Federal support created the community of researchers who developed and promulgated networks based on the Internet Protocol and invented a range of related applications, such as e-mail.

Federally sponsored research has also built on pioneering work in industry. For example, both relational databases and reduced instruction set computing (RISC) technology were first proposed by researchers at IBM. The company did not immediately commercialize either technology. Rather, commercialization was hastened by federal support for university research that further refined the technology and built a pool of expertise. Many researchers involved in these projects assisted in the subsequent commercialization of the technology, working with established companies or starting new ones.

Sponsored research through NSF tends to support small research programs undertaken by individual investigators. It has funded numerous small-scale efforts that planted the seeds for larger programs. NSF's support for MIT's Compatible Time Sharing System in the early 1960s became the basis for DARPA's much larger Project MAC. This project explored a broader range of technologies for human-computer interaction, of which computer timesharing is only one example.

Structuring Federal Support

Why has federal support been effective in stimulating innovation in computing? The report offers several suggestions.

First, computing research has benefited from multiple sources of federal support. The most obvious benefit of diverse funding agencies is the opportunity for researchers to seek project support from multiple sponsors, widening the range of applications developed and approaches taken. Diversity helps to ensure continuous support for research areas as they continue to mature.

Second, agencies were able to recruit strong program managers. Visionary leaders, such as J.C.R. Licklider, Ivan Sutherland, and Robert Taylor, who were drawn from the research community for short tours of duty, were instrumental in identifying new research areas and ensuring sustained support.

Third, federal agencies have continued to adapt to changes in the political and technological environments.

Looking to the Future

History cannot predict the path computing technology will follow in the future. Research is inherently uncertain. Few could have predicted the success of early research programs on packet switching, fewer still the subsequent growth of the Internet. History cannot determine which policies will be most effective in stimulating research in the face of future challenges. But, as the latest CSTB report demonstrates, history can guide efforts to rethink federal support for computing research, even as it enters a new millennium.

"Funding A Revolution: Government Support For Computing Research" is available from National Academy Press, Tel. 1-800-624-6242. Currently available in prepublication format, the final book will be available in January. It is also on the web at http://www.nap.edu/readingroom/enter2.cgi?0309062780.html. Jerry Sheehan is a Senior Program Officer at CSTB. He can be reached at JSheehan@nas.edu for comments or questions.