RESEARCH INFRASTRUCTURE

To understand the state of HCI research in Japan and prospects for the future, it is necessary to understand the Japanese style of conducting research, which is considerably different from the American style. The essential difference is that research in Japan is centered in large corporate research labs, not in universities. A number of causes and effects surround this phenomenon; it is not always clear which are the causes and which are the effects. The following endeavors to explain some of them.

Most researchers in Japan are hired by corporate research labs after graduating with MS degrees. Ninety percent of corporate researchers are hired after they earn their MS degrees; the other 10% are more or less evenly split between having BS and PhD degrees. At some companies, a slight increase in PhD hiring is occurring. The companies provide further research training as part of the until-now traditional life-long employment in large Japanese corporations. Researchers typically remain in the lab for 5 to 15 years and then move on to development groups. In some cases, a researcher will go to a development lab for 2 or 3 years, then return to research for a time before making a permanent return to development.

This overall pattern is a very effective technology transfer strategy for two reasons: First, people transfer technology much more effectively than do papers and reports. Second, relationships the transferred researchers have developed with younger researchers still in the lab ensure good continuing communication between research and product development.

Japanese companies look to universities more for graduates than for ideas. In general, corporations do not fund university research. Large companies fund some universities, generally at low levels, mostly to maintain relationships with well-known professors. Several of the JTEC team's contacts indicated that this is fairly common. Also, panelists were told that tax laws do not permit deducting university research support that has specific deliverables to the company; so funding that is provided is for general support of programs. There appear to be exceptions, such as the TRON project at the University of Tokyo, which has had considerable corporate support (see TRON site report in Appendix C).

Professors at Japanese publicly-supported schools are not allowed to accept salary or consulting fees from companies. As full-time public servants, faculty are paid for twelve months, so other support for summer salary is not needed. Unlike the situation for faculty at U.S. schools, being paid for consulting is also forbidden. Eight of the ten top schools in Japan are publicly supported, so this policy greatly diminishes the financial incentive for faculty to seek corporate support, and it decreases the opportunity for companies to discover the professors' innovative research ideas.

In contrast to the situation at public schools, professors at private schools (such as Keio and Waseda Universities, which are the strongest and in the top ten), do work with companies, and indeed are expected to seek outside support. At least one senior faculty member from one of these schools is also head of the computer science lab of a major corporation and has successfully transferred ideas from the university to the company.

Professors at publicly supported schools in Japan do not have sabbatical leaves. This removes one means for professors to become more familiar with industrial needs: it is quite common for a U.S. professor to spend all or part of a sabbatical, or a summer, in a corporate research lab or development group.

Direct Japanese government funding of university research is modest. There is no Japanese equivalent to the combination of the U.S. Air Force Office of Scientific Research (AFOSR), the Army Research Office (ARO), ARPA, the Department of Energy (DOE), the National Aeronautics and Space Administration (NASA), NSF, and ONR to invest billions of dollars in university research. Japan's Ministry of Education does provide some research funding to a limited number of prominent universities. Japan's Ministry of International Trade and Industry (MITI) research funding goes to corporate coalitions, and the bulk of the funding goes to corporate labs or to the Advanced Telecommunications Research Institute (ATR) (see ATR site report in Appendix C). The coalitions do sometimes pass some funding on to universities.

On the other hand, a recent Japanese government program is injecting major research funds into universities, apparently as a way to help the weak economy. For instance, Hokkaido University is receiving funds to build a large HCI research center.

The consequence of this research environment is, on the average, a focus in Japan on the relatively short-term and on incremental improvements. Product evolution is often valued over product revolution. Research is conducted in a structured, corporate-driven environment. Contrast this to the longer-term, more adventuresome, less inhibited, perhaps more creative style of research found in U.S. university labs, where groups of graduate students have the freedom to pursue ideas and take pride in being different. Industry-founding ideas have emerged from such environments. SGI and Sun are prominent examples.

While the above is true on average, the JTEC panel did see some longer-term, far-reaching projects and some impressive creativity, as exemplified by the Phin project at Fujitsu (see Chapter 4) and some of the work at Sony's Computer Science Lab. Perhaps as a consequence of Japan's corporate-centered research, the percentage of annual revenues spent on research and development tends to be higher than for many U.S. companies. JTEC panelists were consistently shown graphs indicating that 7-8% of gross revenues are being reinvested in R&D in Japanese companies. This contrasts with the following recently published figures for U.S. companies' R&D investment percentages (Business Week 1995):

Comparisons of this sort must be viewed with great caution: the methods by which these figures are derived in Japan and the United States may differ greatly. Since corporate cost allocation methodologies probably differ in the two countries, it would not be appropriate to make a big issue of this observation.

Comparing government R&D funding in Japan and the United States is even more difficult. For example, government R&D funding figures for Japan often do not include salaries, which are paid separately by either companies or the Ministry of Education. Conversely, U.S. government R&D funding projects often do include salaries as 50% or more of total budgets. Furthermore, official government funding levels in Japan are supplemented through semiofficial channels (e.g., the Bicycle Racing Association of Japan, which has been known to contribute proceeds from racing events to government-sponsored R&D projects!).

The Japanese cultural emphasis on consensus and uniformity and seniority limits the adventurous and creative thinking that is part of research. The JTEC panelists were told of the Japanese saying, "The nail which stands out is pounded down." Risk-taking in Japan tends to be discouraged. As well, the emphasis on seniority in the research labs, recently changing in at least one lab, the Sony Computer Science Lab (see Sony site report in Appendix C), makes it hard for bright young researchers to challenge the "conventional wisdom" of their elders and introduce radical new ideas.

Not taking risks has another effect. In the United States, numerous start-up companies have been established by recent university graduates, sometimes with their professors, to commercialize promising innovations. Venture capital firms are eager to help launch promising companies. In Japan, there are neither the venture capital firms nor the desire to take the risk of starting a company. Again, there are exceptions. JTEC panelists met with Kazuo Nishi, President of the ASCII Corporation, a company that Nishi started ten to fifteen years ago that today has sales of about $400 million.

Finally, since most new researchers in Japan have experienced neither a Japanese university research lab (recall that most researchers go to work after earning their MS degrees) nor a U.S. lab, they do not have the frame of reference and the set of expectations that graduate students in the United States develop.

Readers are cautioned not to take any of these statements as absolutes. There are always exceptions. Also, readers should recognize that the JTEC study team spent only 9 days in Japan, so that the collage presented here is undoubtedly incomplete. However, more knowledgeable individuals who have reviewed this material state that it is generally accurate. More detail on many of these issues can be found in an informative article by Notkin and Schlichting (1993).


Published: March 1996; WTEC Hyper-Librarian