What Should Industry Expect from Academic Engineering Research?
John A. Armstrong
Before turning to what I feel are the central issues confronting academic engineering research today, I want to give a list of topics that are not, in my view, the most important issues. Matters of indirect cost recovery, of getting sponsors to pay the full costs of research, matters of fraying university research infrastructure are all important, but they concern a large population of university departments, so I do not think they are peculiar to engineering research. Nor do I think it is very important to concern ourselves about undue influence by industry on the academic research agenda. There is no way to have a determining effect on that agenda without putting up substantial amounts of money, and I do not think that industry is likely to do so in most fields. On the other hand, mutually beneficial relationships between engineering research departments and their related industrial sectors are matters of concern, and I will have something to say on that topic later in this paper.
In addition, although the changes in corporate R&D investments are clearly relevant to the future of academic engineering research, I believe it is misleading to describe these changes as the ''apparent abandonment of in-house capacity for basic industrial research." There is more to it than that, and I will have more to say about that later in this paper as well.
Finally, I do not think that we should be talking about "the cures for national competitiveness concerns being found in academic engineering departments" if that is taken to imply the simplistic notion that competitiveness problems stem from lack of technology transfer of new results from academia to the private sector. The competitiveness problems of American
industry have almost nothing to do with lack of technology transfer from universities, or national labs for that matter. The responsibility for deficiencies in our industrial performance rests largely with failures in the private sector, failures of strategy, investment, and training—failures, in short, of management. There have also been failures of government policies in the areas of macroeconomics, access to overseas markets, and the balance between consumption and savings. Neither the private nor the government failings will be cured, or even helped, by more engineering research. Trying to cure poor industrial performance in the short term by more university research is like asking for helpers when pushing on a rope.
Having said that technology transfer is not one of the real issues, I hasten to say that I do believe that academic engineering research can make a major contribution to improving the ability of our nation to realize the fruits of its investments in basic and applied research. But that contribution will have more to do with the nature of the advanced training given to Ph.D.'s than with the specific research results produced as part of that training.
I want now to summarize what I think are a few of the real issues confronting academic engineering research today, as seen from the industrial perspective.
Recently I gave a Compton Lecture at MIT titled "Is Basic Research a Luxury Our Society Can No Longer Afford?" My answer was "No, it is not a luxury, it is a source of national competitive advantage." But that answer was strongly qualified. I argued that national leadership in basic research was neither necessary nor sufficient for society to achieve its economic and environmental goals. The reason is that successful R&D represents less than 5 percent of the process by which wealth and jobs are created. Countries that do much or all of the other 95 percent in a world-class way can be successful in reaching their goals without being world leaders in research. But I went on to assert that a nation that does the 95 percent competitively, and leads in basic research, may expect to have a comparative advantage. And since the United States currently enjoys world leadership in many areas of research, we ought to be careful to preserve that advantage while, in parallel, we address deficiencies in our national performance in the 95 percent of wealth creation that is not R&D.
It is in thinking about the larger role that scientists and engineers can play in some of these "downstream" activities that I have been led to ask the question "What is an Engineering Science Ph.D. For?" In what ways can academic engineering research contribute more effectively to the rest of the processes by which new knowledge is turned into societal value, processes that for the most part lie outside traditional science and engineering, but to which, experience has shown, engineers and scientists can make major contributions? In my view, this is the overriding issue confronting academic engineering research today.
But before expanding at length on this question, it is important to keep it in perspective with a set of other important considerations. The first of these additional considerations is that the possible role of the government vis-à-vis research in support of civilian technology is very different from the government role (which has been eminently successful) in supporting military technology. The government was the customer for military systems, and as such could guide and control its support of research by whether its clearly understood needs were being met, or were likely to be met. But the government is not the definitive (nor even a very important customer) for most civilian technologies. (Indeed the government is a customer that drives most companies "up the wall" because of its cumbersome, often wrong-headed, and counterproductive procurement practices.) It is not expert on the results to be achieved with civilian technology nor on how to measure success. Therefore it should exercise a healthy degree of caution in creating policies and programs to direct research funds toward nonmilitary national goals and in pushing for technology transfer from the national labs to the civilian sector. I do not see sufficient appreciation of these dangers in the public discussion, although the plan to channel funds to universities through the industrial members of industry-university consortia is clearly an attempt to deal with this issue. Whether that will turn out to be a good idea or not, time will tell.
The next consideration to be kept in mind relates to changing industrial R&D portfolios. While many industries have been reassessing what types of applied and engineering research are likely to be of significant help to them in achieving comparative advantage in the coming decade, the engineering research portfolios of universities will probably change much too slowly to be in step with the needs of many sectors. This lack of being in step is more a matter of balance than a lack of appropriate investments altogether. Said another way, academic engineering research is well adapted to creating new programs (witness the rapid emergence of programs relating to manufacturing). But academic engineering research (and all other forms of academic research as well) are just about incapable of stopping programs, of scaling back investments, and of redirecting the work of faculty. This guarantees a number of glaring mismatches between academic engineering research and industry in the coming decade. Clearly I have a different view than many of my academic colleagues about the problems of undue industrial influence!
So, before we conclude that downsizing and other changes in corporate R&D portfolios are all regrettable and shortsighted, we should ask whether particular R&D areas are still such good investments as they may once have been. Not only does the leverage of a particular field of research in industry change with time, but there can also be a change in the relative importance of the university engineering research contribution to such a field.
Of course, this balance has shifted more in some fields than in others. The newer the field, and the less the aggregated industrial R&D investment, the greater the significance of university-derived results.
Whereas today's best example is the whole set of technologies spawned by progress in understanding the molecular basis of life, in the 1960s and 1970s electronics and computer science were the leading examples. Then, the relative contributions of university-based research results in electronics and computer science were much greater than they are today. But now the resources devoted to R&D in industry, and the accumulated knowledge, know-how, and investment, make industry less dependent than was the case twenty years ago.
Ironically, during the 1980s there was substantial attention to building silicon-technology-based research facilities in universities just before it became clear to the electronics and computer industries in the 1990s that the value of hardware technology and expertise was declining in relation to that of software, systems, and applications expertise. Industry's response in some cases was to leave hardware technology development altogether, and in other cases to scale back the investments being made, often by forming consortia to share risks and investments. What academia's response to these developments will be is far from clear.
How might university engineering research have avoided "zigging" just as the electronics industry "zagged?" Better relations between industry and academia, focused more strongly on their principal enduring common interest, might have helped. That "principle enduring common interest'' is, of course, highly trained students.
It follows, I believe, from all of the considerations I have listed that academic engineering research may need to rethink the importance it attaches to research results per se in relation to the value of the Ph.D. training through which those results are obtained. In a word, for the next decade or so, the training will be more important than the research results, at least in many fields. This is an institutional issue as well as an issue to be faced by university faculty as individuals. Clearly it is an issue for the mission agencies that support academic engineering research as well. They need to reassert that they have an explicit mission to foster graduate technical education as well as to support research whose results are useful to them.
In any future rethinking of engineering science Ph.D. programs, one should examine not only the appropriate portfolio of technical areas and programs maintained, but also ask, How can academic engineering research be more effective in helping the nation achieve its goals for more and better jobs, a rising standard of living, and a more sustainable relationship with the environment?
In short, one should ask the question, "What is an engineering research Ph.D. for?" Although I am only a visitor to academia, I propose to devote the balance of my remarks to addressing this mildly provocative question.
Many engineering schools have recently reassessed their master's degree programs, and many schools have made significant changes in those programs. It is also true that many professional schools have thoroughly revamped their curricula, expectations, and culture. A good example of this is the New Pathway program at the Harvard Medical School. But there has been little serious reassessment so far of the underlying assumptions, expectations, and requirements of Ph.D. programs in science and in areas of engineering closely allied to science, the areas we have been calling engineering research. In my view it may be time for such a reassessment.
Of course, in many respects the Ph.D. programs in science and engineering are in good shape. The technical sophistication of new graduates in their specialties is often breathtaking. New Ph.D. graduates are still the best "vehicles" in the world for transfer of new insights and new ways of doing things.
And yet . . . there are serious problems as well, problems I came to see over many years of hiring and managing new Ph.D.'s. In brief, it is my view that the training of new Ph.D.'s is too narrow intellectually, too campus-centered, and too long. Furthermore in my experience, many new Ph.D.'s have much too narrow a set of personal and career expectations. Most do not know what it is they know. They think that what they know is how to solve certain highly technical and specialized problems, like designing microprocessors or writing high-speed networking protocols.
Of course, what they actually know that is of lasting value is how to approach and solve problems starting from powerful and fundamental points of view. But to my surprise, most do not understand that that is what gives them any edge they may have over young people of their own age who are already out in the workplace without a Ph.D. but with a six-to eight-year head start in experience.
This is all part of what one might call the Ph.D. paradox. The phrase is simply a way of drawing special attention to what we all know, but which is not, I think, sufficiently taken into account in the design of Ph.D. programs. To earn a Ph.D. in engineering research, a young person is expected to make an original contribution to fundamental engineering science. To get to the frontier, it is expected that one will ask a narrowly defined set of questions, and in that narrow region, think or experiment deeply. In the course of this deep but narrow exploration, the graduate student acquires a powerful methodology for formulating and solving technical problems, starting with an understanding of the fundamentals of the subject. He or she learns how to pose a problem, decide what data or experiments are required to solve it, obtain that data, analyze it critically, and then defend the conclusions vigorously. He or she has learned how to acquire new skills, including the ability to understand and use just about any form of applied mathematics. The Ph.D. candidate has, in a word, learned how to learn at a very sophisticated level.
The ''paradox," of course, is that in the course of deep, specialized inquiry, one acquires an intellectual armamentarium and outlook of great general utility. The training of the scientific or engineering specialist in fact provides much of what might be termed training for the advanced technical generalist. It is a further part of the paradox that many new graduates do not seem to value this powerful generalist capability—perhaps because their professors do not value it either.
This overspecialization often has unfortunate consequences for new engineering scientists. Overspecialization can result in a lack both of perspective and of self-confidence; new Ph.D.'s often believe themselves ill prepared to venture outside their specialty to use their powerful training in jobs in development, manufacturing, and technical management, let alone in tasks even farther afield from their specific training. The burden of overspecialization is compounded by their often total lack of work experience outside the university and by a culture that often suggests to them in not so subtle ways that becoming like their professor should be their goal and mark of success.
This paradoxical situation is due in part to the lack of serious requirements for scientific and technical breadth in the typical graduate curriculum, as well as to the fact that there is little or no encouragement, and a lot of implicit discouragement, for the young person who wants to spend time during graduate school off campus in a setting where technical knowledge is actually used. There is, in short, almost no value assigned to technical breadth or to real-world experience as an essential part of Ph.D. training.
You may recall that I asserted that the typical Ph.D. degree takes too long to acquire. I firmly believe that to be the case, and I see no contradiction between shortening the time to obtain a Ph.D. and my just expressed desire to see young people spend more of their time away from campus as part of their training.
I hold these seemingly contradictory views because of a hypothesis I have about why the typical Ph.D. takes so long. It is only in part because of course requirements and faculty pressure to get more research results for a thesis. It is due in large part both to the students' comfort with graduate student life and to their anxiety about what it will be like in the outside world when they leave the university. This is all possible, of course, because universities and funding agencies permit and support such long stays.
If I were a sociologist I would test the following hypothesis, both retrospectively and prospectively. What is the average length of time to the Ph.D. of young men who are married and have small children while they are graduate students? The answer I expect is that it is up to two years shorter than the average. (That the opposite result will obtain for young married women graduate students is altogether a different problem!)
Just as experience of family responsibility tends to shorten one's tolerance for the life of a graduate student, so, I believe, will experience out in
the world of technical work tend to lower the typical graduate student's anxiety about finding a job and starting a career.
Now we cannot require graduate students to get married and start families, but we could exert serious pressure to bring the average duration down by a year or 18 months. Shortening the average duration of graduate study will lower the cost to the nation for training a given number of young scientists and engineers, and it will saddle the graduates with less of a disadvantage with respect to their contemporaries who are years ahead in gaining experience and seniority in the workplace.
What can industry do to help? It can and should be responsive to setting up cooperative arrangements with engineering research departments. I believe that small firms and start-ups have the most to gain by such arrangements, and also the most to give students in the way of broad perspective. Many of our best graduate schools are surrounded by such small companies, many of which have been started from university science and engineering programs. However, except for the students of faculty members connected with these spin-offs, these exciting firms are invisible to the majority of graduate students.
I know well, by the way, that it is hard, time-consuming work for faculty members (as well as for their industrial colleagues) to set up mutually advantageous joint projects involving graduate students. But because they will, on average, contribute so much to the improvement of the student's education, both faculty and industry managers should make the time.
There is more that industry can and should do. Companies should be more willing than they now are to have key technical people spend time in universities as adjunct faculty. The improved perspective they will bring to faculty and to graduate students will be more than enough to offset the substantial effort it will be to initiate such arrangements. Similarly, there is far from sufficient value placed on faculty members having professional experience in the outside world. And by that I mean more than the casual knowledge that consultants obtain of the culture, the problems, and the Intellectual value that exists in off-campus engineering research.
It is true that, both as individuals and as members of their discipline, professors take pride in the fact that many of their students turn out to have highly successful careers in industrial management, or in government service, or in the business world generally, or as teachers and professors in nonresearch institutions. But this is all thought to be irrelevant to the graduate curriculum. The curriculum is still characterized overwhelmingly by what is necessary for the training of future research faculty members.
Although these nontraditional uses of the Ph.D. have been around for a long time, their importance both to society and to society's support for the scientific research enterprise requires that they be taken into account in new ways. The reason is that, as described in my first Compton Lecture, getting
the R&D done "right" is less than 5 percent of the job of turning new knowledge into the social and economic utility for which society supports scientific research in the first place (including basic research). The other 95 percent of the job has to be done in a world-class, competitive way if the society that pays for the research is to be the society that gets a fair return on its investment.
In the doing of the other 95 percent of the job, many people with skills outside of science are needed, to be sure. But even so, there is much of the 95 percent of the job that is not R&D that can and should be done by people with scientific training. I say "can" because much of this work is done best by those with technical background and understanding; and I say "should" because societies that do bring technical generalists to bear on this work will have an advantage in world competition and will get more for their investment, sooner, than countries whose scientists and engineers play less frequent and prominent roles beyond the laboratory.
The presumption seems to have been that the apprenticeship process designed for the traditional science Ph.D. degree would do as well as needs to be done in fitting graduates for employment as science and engineering Ph.D.'s in what I have called the nontraditional roles. Certainly the traditional Ph.D. training is not bad preparation for nontraditional roles, but it is hard to believe it cannot be done better. Society's support for academic research may even depend on its being done better.
Indeed, I believe society is poised at this moment between developing an enlarged expectation of what scientists and engineers can do, on the one hand, and concluding that we have been largely overrated in our contributions to society, on the other hand. If this perception is correct, it behooves us all to take the improvement of graduate science and engineering education very seriously.
I have already said that industry can and should do more to help enlarge and augment graduate technical education. But it must also be said that those of us outside the university cannot possibly be major actors in this reassessment and revamping. The most that we in industry can do is offer to help where appropriate and to transmit our sense of the urgency of the task of rethinking Ph.D. training. We feel this urgency because the students at issue are of enormous importance to our own future and because we believe that society's continued support of the university engineering research establishment depends in no small part on that establishment's doing, and being seen to do, a better job at fitting technically trained citizens to play their full role in achieving the goals of society to which science and technology can contribute.