John A. Armstrong, "What is a PhD For in Science?" Compton Lecture (3 of 3)

Search transcript...


MODERATOR: This year's Compton lecturer is John Armstrong. He was formerly Vice President of Science and Technology at IBM. He is now beginning this series involving various issues quite relevant in these times of great uncertainty in terms of the direction in basic research. I think many of you have heard much of his biographical information in previous talks, so I will forego that and just take the opportunity actually to thank him for his efforts here at MIT in the last several months.

John has come here with some perhaps delusions of taking it a little bit easy for a few months. And instead, has done actually a terrific job intersecting with the graduate students in particular at the Institute, and certainly with proper physics.

He has taken the initiative in holding free lunchtime discussions with the order of 50 or so grad students discussing all kinds of options which students might pursue in their careers with degrees in physics and in other sciences. Students certainly have responded with great enthusiasm.

And John did, in fact, foolishly got carried away and offered to be a personal resume reviewer for each and every student who comes to his office. So the invitation is now extended to all of you.


And absolutely. Well, in fact, I'm first in line, Tom.


So John actually has had a wonderful impact in terms of interaction with our students. And I think that's all I will say, and turn it over to him. His talk today is, in fact, on, what is a PhD for in science?


ARMSTRONG: Thank you very much, Ernie. And I did, in fact, make that offer. I critique, not write, resumes.


I want to begin by saying again how pleased I am to be the Karl Compton lecturer, and how grateful I am to many here in the audience and to many at the Institute for your hospitality and for very stimulating discussions and interactions.

Now my first lecture some weeks ago was entitled, Is Basic Research a Luxury Our Society Can No Longer Afford? Some of you may have heard that. The answer that I gave at that time was no, basic research is not a luxury. It is a source of competitive advantage. But that answer, I took 45 minutes to give, and was quite strongly qualified, as some of you may recall.

I argued that leadership in basic research was neither necessary nor sufficient for a nation to achieve its economic and other goals, the reason for that being simply that successful R&D is only about 5% of the job of turning new knowledge into economic value and societal value. And then countries that do much or all of that other 95% in a world class way can achieve their goals without being leaders in scientific research.

But I went on to assert-- and I firmly believe-- that a nation which does that 95%, the other 95%, so to speak, of the job in a competitive way, and is a leader in basic research, such a nation will enjoy a competitive advantage in the economic struggles that we are all going to be part of over the next decades.

And since the United States currently enjoys leadership in basic research, I argued it would be foolish for us to do anything to forgo that advantage while we are, at the same time, addressing the deficiencies in our national performance on that other 95% of wealth creation, which is not research and development.

And it is in thinking about that other 95% of the job and the roles that scientists might play that I'm led to ask the question of today's lecture. What is a science PhD for?

Other than my early career as a working scientist, my qualifications for asking this question are only those of an employer and manager of science for many years, and an observer of both of the great strengths and some of the weaknesses that scientists bring with them when they venture away from the university. But as you know, I have no experience as a university professor or administrator, and so I'm sure there are many aspects of this problem of which I'm ignorant, and which I will miss here today.

I hope to convince you and others outside the university that there is a need to assess the strengths and weaknesses of the current apprenticeship system of PhD training in science, and to look at the uses both tradition and non-traditional to which, if I may put it that way, a science PhD is put. We need to look at those uses when we're thinking about how the system might be improved.

My hope is really only that these lectures will stimulate discussion, will help to stimulate discussion among all the stakeholders, including academics, scientists in private institutions that employ scientists, and in the funding agencies. That discussion needs to go beyond the difficult and immediate and pressing issues of allowable overhead costs, the size of the National Science Foundation budget, the budgets of some of the-- the science budgets of the mission agencies.

Those are urgent matters. But we need, as I say also, to think about the role which scientists will need to play in the coming decades, and what part both the funding agencies and future employers should play in that process. Of course, I'm well aware that most of the analysis and the suggestions for change, if any, must come from those most closely involved, mainly in the university.

It's widely acknowledged that with the end of the Cold War and the escalation of international economic competition, the rationale for funding non-health-related science in the universities has changed. And it's beginning to be understood, I think that this implies increased attention in the university to basic and applied research in the so-called strategic areas of science areas, which are a good bet to be helpful in achieving national goals.

Part of the difficulty of this whole subject-- and we discussed it at the first lecture-- is that there is widespread and I think justifiable nervousness about how to balance the vigorous pursuit of strategic basic and applied research with the appropriately vigorous support of research in areas that I call discipline-only driven.

In a lecture at UC Santa Barbara about two years ago, which was entitled Research and Competitiveness: Problems of a New Rationale, I pointed to many dangers for research universities if they jumped without careful thought onto the research for competitiveness bandwagon. I failed to discuss one major danger in that talk, namely the danger that science faculties would try to adjust to the new environment without asking whether change is needed in the nature of PhD training for some or all the sciences. So that is the main topic of my remarks today.

Many of you know, I'm sure, that other branches of graduate education are re-examining the training that they provide. For example, medical education is being rethought from the ground up. An example of that is the new Pathway program in the Harvard Medical School.

As I've already suggested, I don't know how much of the topics I'm going to discuss today are relevant to PhD training in the biological sciences. The rationale for support and research in the biological sciences has not changed in the way that it has changed for physics, material science and mathematics, and some parts of chemistry. I hope, however, that the leaders in those fields are alert to the possibility that changes in our areas of science, so to speak, in science policy may spill over, in effect.

Now I'm going to give a brief characterization of PhD training in science. Some of you may feel a caricaturization of it. I'm going to tell you what you all know already, and what many of you, as I look around, are right in the middle of. But a description of the present system is necessary for analysis, and I hope to reach people with this discussion beyond the university.

PhD training in the sciences, as you well know, is best described as apprenticeship. Graduate students attach themselves early and tightly to individual professors. Indeed, the attachment is in many cases that the students identify more strongly with the research group of their professor than they do with the department or even the university of which they're a member.

Despite mechanisms for ensuring common standards, I think it's fair to say that the nature of that apprenticeship depends strongly on the personality and the intellectual style of individual professors. PhD candidates in different groups often have very different overall experiences, and have to be de facto different standards.

But this is after all what one would expect of apprenticeship as opposed to qualification by standardized testing or certification. And I believe the system has served us well, in part because of the flexibility that I've just described. Indeed, some of the suggestions I have would increase the flexibility from the student's perspective.

In addition, many aspects of the present system of training in the sciences are strongly influenced in ways that you understand better than I do. By the way money flows from funding sources to participants, to the students, to the faculty, and to the university as a whole. But despite the continued existence of some graduate fellowships, for example, in which money flows directly to the student, most apprenticized scientists are funded, as you well know, under research contracts and grants to their professor.

This system evolved to its present form during the Cold War in an atmosphere in which the long-term trend was ever-increasing budgets and ever-increasing numbers of positions. I repeat, I think the system has served us well, but I believe it's time to take a fresh look at it.

Now I'd like to mention just briefly what I believe are some of the principal strengths of training for science at the PhD level. Now what I'm about to say will be obvious to those of you who are faculty members, but I know from recent talks with graduate students that what I'm about to say is not well understood by some of them. Perhaps at that stage, cannot be.

But it is my firm belief that in addition to and more important than the highly specialized scientific knowledge that they obtain, PhD graduates acquire a powerful methodology for tackling technical problems, starting from an understanding a drive to understand the fundamentals of the subject.

They learn how to pose a problem, how to decide what data or experiments are required to solve and obtain that data, analyze it critically, and then dispend their conclusions vigorously. They learn how to learn new skills, including the ability to understand and use just about any form of applied mathematics.

To be sure, the PhD student has the experience in at least one area of going to the research frontier and coming back with some new insight or new knowledge. But the scientific results thesis work, like a large fraction of all published research, is of trends and is often of transient value. The true and lasting value of PhD training is to have learned not an answer, but a process that reliably leads to answers.

These are the main strengths of scientific training at the graduate level, and they are strengths whether one uses them in a lifetime of further scientific research or in a lifetime of diverse work in service in any of a host of non-research, science-related occupations.

Incidentally, it would be foolish to imply that it's only PhDs in science whose training provides them with this intellectual armamentarium. Many forms of engineering pursued to the PhD level confer similar benefits. But this is true principally I think because in recent decades, the engineering curriculum has moved strongly towards engineering science as its basis.

That is certainly not a novel remark, but what has, I believe, both some novelty and urgency is asking whether the training and the culture-- and I use the term culture advisedly-- acquired by PhDs in science can be improved to deal with the challenges and opportunities that we as scientists and we as citizens will face in the coming decades.

My belief that it is appropriate to do this reassessment may be strengthened or at least it is strengthened in my mind by taking a look not only at the strengths which we just discussed, but at some of the weaknesses of the present system of PhD training. And weaknesses which sometimes initiate the strengths we've just described. Now I have to say here that, of course, I have only a limited perspective on these matters. But I will tell you a number of things which come out of that respect.

It's no novelty to say, to observe that new PhDs are typically very narrowly specialized in the area of their thesis research. Indeed, how could it be otherwise, given the current state of science and the way that funds are awarded?

However understandable this specialization may be, it has unfortunate consequences in my experience for the new scientist's view of himself or herself. Sometimes this overspecialization results both in a lack of perspective, which is understandable, and a lack of self-confidence as a scientist. New graduates often believe themselves ill equipped to venture outside their specialty area, to use their wonderful and powerful training in jobs, for example, in R&D or manufacturing or technical management.

This burden of overspecialization is compounded by what is often a total lack of work experience during the graduate student years outside the university, and why a culture which often suggests to them in not-so-subtle ways that becoming like their professor should be the main goal and mark of success. I should hasten to add that some of my best friends are professors.


Worse, I would say that culture of graduate training in science or at least in some of the sciences often suddenly imparts an intellectual pecking order to the student's view. It is a view to the effect that physics is better than chemistry. Pure is better than applied. That's for sure. Physics is better than chemistry. Both are better than engineering. And the discipline and the intellectual content of manufacturing, for example, are never considered at all.

We need to try to rid Western civilization once and for all of that cultural bias. If-- well, for two reasons. First of all, it has nothing to do with nature. Nature doesn't understand those distinctions, nor respect them. Secondly, those distinctions and the attitudes that come with it often interfere mightily with young scientists having as successful and exciting careers as they might otherwise have.

Now I'm aware that many of these difficulties are exacerbated for universities in what is an unintended consequence of perhaps the dark side, if you will, of the peer review system, which is so important for ensuring the quality of scientific funding distribution. The dark side of that system is the pressure to go ever more deeply and expertly into a narrow field where research results can be, quote, "guaranteed." That atmosphere, of course, stifles a real sense of intellectual adventure.

The notion of guaranteed results from research is obviously an oxymoron. But there is a great tendency of peer review panels to travel in packs, and that behavior stifles risks taken by faculty, and it is a lesson that is not lost on graduate students.

Now having laid out so briefly the strengths, which I believe are really formidable, and the weaknesses, which I believe can be dealt with of this apprenticeship system, the next part of the discussion is really to talk a little bit about the uses, the traditional and the nontraditional uses to which science PhDs are equipped.

As this audience certainly knows, the PhD degree is necessary but not sufficient for becoming a member of the science faculty at a research university. It's necessary for becoming a professor in almost all non-research universities and in many four-year colleges. And it's close to becoming a necessary qualification for membership on the research and development staff of many governmental and private collaborators. Now what are the-- and it is those uses, of course, for which the apprenticeship program has been designed and evolved over the years.

Now what are the nontraditional uses through which PhDs and science equip? The list will be familiar to you as well, but it is central to my argument that these nontraditional uses will be increasingly important in ensuring that society gets a timely and satisfactory return on its investment in scientific research.

In many fields, as you know, a large fraction of the PhDs do not go on to a career either in research or in academia or in government efforts. I think the PhD mathematicians, the number is almost 90% do not go on to do research.

Indeed, many go into technical work in industry outside their area of expertise and training. And so it is not common-- and some of us a week or so ago met young condensed matter experimentalists who were working in a medical research firm. You will have colleagues with similar cases. Many PhDs in science will, of course, go on and become involved in the management of applied research, engineering, and even manufacturing programs and industry.

Where as I have repeatedly seen-- this is something you need to understand-- PhDs in science can often compete successfully with engineers, including those with engineering science PhDs. And in observing that, I'm not simply calling for scientists to be trained like engineers. I'm just pointing out the breadth and the power of the science PhD training.

In these positions, PhD scientists are often very well placed to help improve the efficiency of the processes by which new technical knowledge is turned into economic or other value to society, the society which funds graduate training and scientific research, and which expects those returns as its primary result from the social compact of supporting science.

Now in addition, the holders of science PhDs participate strongly in the management, federal, and state agencies, which fund science, serve as technical staff in Congress, in other governmental organizations, and in many other places. And in addition to those who teach at the four-year college level, there are many-- I think increasingly many who will teach science below the college level, or otherwise be involved in the interpretation of science to the public.

Finally, science PhDs increasingly use their quantitative modeling and analytical skills outside science all together. As The Economist magazine pointed out a few weeks ago, there is a regular stampede of physicists and others to Wall Street. Now what that says about your or my pension fund is another matter all together.


In short-- and here's the point. In short, a PhD in science should, for many purposes, be thought of as a technical generalist degree. Although it's nontraditional uses have been clear for a long time, their importance to society and society's support for scientific research requires that they be taken into account in new ways. And as I said, the reason is that 95% of the job which is not R&D, the job by which new technical knowledge gets turned into societal and economic benefit.

The other 95% of the job has to be done in a world class way. And the competitive-- excuse me. In a world class competitive way, if the society that pays for the research is to be the society that gets a fair return.

Now it's clear that that 95% of which is not R&D involves many people with skills outside of science, to be sure. But even so, there is much of that 95% of a job that can and should be done by people with scientific training.

I say can be done, because much of that work requires technical understanding. And I say should be done, because societies which do bring highly trained technical generalists to bear on this work may well have a comparative advantage over, in the economic competition, over countries whose scientists play less prominent roles outside the laboratory.

Now-- and here's an important point of the argument. The presumption, I think, has been until now that an educational process, the apprenticeship which was designed and works well for the traditional uses of the science PhD, would do well-- would do as well as needs to be done, and perhaps does as well as could be done in fitting graduates for employment as science PhDs in what I've called the nontraditional roles.

Now I've already stated, and I firmly believe, that the traditional training is not bad preparation. In fact, it is quite good preparation in some cases for these nontraditional roles. The question is, can it be done better, and are other societies doing it better, or going to learn to do it better?

Indeed, I believe that society may well be poised at a moment between developing an enlarged expectation of what scientists can do, on the one hand, and concluding that we have been largely overrated in our contributions to society on the other. If this perception is correct, it behooves us to take the improvement of graduate science education very seriously indeed.

Moreover, if I may say so as a non-academic and as a visitor to this but also to other universities, students deserve better than the present system tends to give them. They are over 30 years old in many cases before receiving the PhD or completing the increasingly common two to four years post-doctorate positions. Only then in most cases do they get to try and move up in the world outside academia, a world of which many are ignorant in large measure because they have never been-- neither being encouraged or allowed to spend time outside during their studying.

The human cost, the inefficiency, the disappointment which often occur, it seems to me, are no longer defensible. Nor is it any longer acceptable that graduating students in the sciences have often so narrow a view of their own scope for successful careers.

Well, one of the most important lessons I learned early on was that it's easier to ask questions than to answer them. But I will share with you some thoughts about where improvements might be made in the overall effectiveness of science PhD training.

My first comment is that radical change is not required. Training by apprenticeship, training by doing under the direction of a more senior and experienced expert really does work. If the method were not available to us, we would hail this invention as a major breakthrough.

The changes that may need to be made, well, the system's still perfectly capable of producing new generations of researchers and new hierarchies of professors. To the extent that proposed changes are perceived as drawbacks for faculty or for funding agencies, I would argue that the overall gain to students and to society might compensate for those.

So I have five suggestions. And I really offer these in the full knowledge that I am not really an expert in this topic. My first suggestion is that this whole area, this whole topic of reassessment of graduate training be discussed and debated widely. There is already widespread discussion of the rationale for, the goals of, and the means by which one can ensure a return on the national investment in science.

An examination of the training of science PhDs ought to be part of that discussion. And that should take place in departmental and university-wide groups, in professional societies, in groups at the National Research Council, the National Academies, and in the funding agencies and in the R&D enterprises of the private entity.

My second suggestion is that one welcome this heightened realization that a science PhD is, in fact, a technical generalist degree in addition to being the threshold for research and university teaching. Faculties should then begin to work out the consequences of such an explicitly enhanced view of the PhD.

My third suggestion is that after discussion with the funding agencies, we should try to ensure-- and I believe ensure in the national interest that the proportion of students supported by fellowships be increased. The aim should be that a progressively larger and larger fraction of them are supported in their first one or two years by direct fellowships rather than by support for contracts attached to individual research professors.

Departmental internships of the kind that are funded by NIH would also help here. By the way, in making this suggestion, I'm fully aware of the fact that it goes counter to the trends of the last 20 years. However, in my view, those trends are no longer in the national interest.

Now if fellowships were more widely the source of graduate student support, it would begin to introduce a degree of flexibility, which would have several beneficial effects. First, it would partly mitigate the pressure to join a research group in the first year of graduate school, and thus start specializing before one has really had a chance to do the right amount of work and earn it.

Fellowship support for the first two years would also make it possible for students to spend at least one, perhaps two summers away from the university in work settings which could help enormously to give them the needed perspective, experience, self-confidence, and contacts.

I trust students to have a reasonable sense of what is in their own interest, and I would give them more flexibility than the present system does. I believe most would use it in ways that would help to mitigate the effects of overspecialization.

Moreover, fellowship support-- that is, increasing fellowship support might introduce new forms of competition between the science departments of research universities. just as it is now the case that word gets around among young women as to which schools and departments have made the most effective efforts to deal with the special problems of women in science, with the result that some departments are much more successful than others in recruiting women as science graduate students, so word would get around as to which math or physics or chemistry or material science departments are best informed about and most usefully coupled to the non-academic world of scientific employment.

To go with this increased flexibility that would come with more fellowship support, it might be useful for gradual curriculum to include possible suggested tailorable tracks, tailorable to the needs of students who envision careers outside the research university. It might offer a way to broaden one's preparation, for example, for college teaching, or for a career in government service, or a career in a technical sector of industry. Some universities at the undergraduate and master's level are experimenting with programs in interdisciplinary areas, such as environment and designing degree programs which consciously try to provide unusual breadth without undue loss of depth.

Now it may be objected that such possibilities are open to PhD students now. My reply is yes they are, but there isn't a lot of encouragement to take them in most universities. And by the way, I am not-- I beg you to believe I am not lecturing at MIT. I am talking about, as well as I understand it, graduate training in general. And I have some familiarity with a number of other research universities.

In my view-- well, excuse me. It may also be objected that such accommodation of different tracks is not required in the best research universities, since their graduates are all going to be real scientists.


These other options and possibilities are appropriate perhaps for less prominent schools and departments. Well, in my view, the leading science departments should aspire to train not only more than their fair share of the best researchers, but also more than their fair share of the leaders in science and science-related fields.

My fourth suggestion is that we should explicitly and actively encourage PhD students to spend time in user environments outside the university as part of their apprenticeship. Now it may be objected that it already takes at least six years to get a PhD, and that time away will only prolong what is already too long a process. I think the truth might be otherwise. In the long run, by which I mean by the time a student is 35--


--the benefits to him or her and their employer will far outweigh the cost, a cost which I suppose is mainly borne by the faculty advisor at the university and the funding agencies. But I'm trying to put the students and society first here.

One way that time away might be organized would be in programs analogous to the co-op programs now used mainly for undergraduate and master's students. The MIT co-op program is a terrific program, and it has a terrific reputation in the industry. It's no longer the case, if it ever was, that there are not scientists and others outside the university who are capable of helping in some way to supervise PhD theses.

Moreover, the deficiencies of such shared supervision will almost certainly be more than offset by the benefits to all concerned. Students, faculty, industrial colleagues, and others. Now of course, not all internships-- time spent away, a summer or a semester or whatever-- not all time spent away by a long shot would be suitable for thesis work, but some would.

In thinking about this, it has occurred to me that substantial intern-type experience during graduate school may be one of the best-- indeed, one of the very few ways in which to effectively counter the ills of overspecialization that are forced on us by the progress in science and by the competitive culture of research and the practices of peer review and the funding process.

I think one should plan-- if one took this at all seriously, one should plan to begin offering this option of external internships in a small way first. It takes time to build a cadre of outside organizations and individuals who will be able and willing to provide meaningful internships. One should plan to involve people from those organizations in the creation of such programs in the first place, and make sure that some of them have a sense of ownership and personal commitment to making it work.

Local and/or small businesses are natural places to start, since the perception of mutual benefit is likely to be largest on the part of small organizations. For example, because they will be very sensitive to the cost effectiveness of getting technical work done by non-permanent but highly skilled people.

You would also have to expect that the willingness of outside organizations to take on graduate students as generalists rather than as specialists will depend on many factors that vary by company, by industry, and with the economic climate. If an organization is expanding, it will be easy [INAUDIBLE] students with the generalist [INAUDIBLE]. If the organization is declining, it will be much harder. But in general, I think small organizations, startups, and local companies are easier to crack in this regard.

Now I also think that it's worth working on both from the university side and national lab or private industry side, work on minimizing the financial and bureaucratic barriers for family members to enter into two partnerships with scientific and ethical [INAUDIBLE], since these genuine partnerships are often excellent vehicles for providing non-university experience for graduate students. And indeed, there are many good examples of such partnerships that I know about here in science at MIT.

My fifth and final suggestion has to do with opportunities for science faculty members themselves. We've so far been discussing the options for broadening the perspective and experience of students.

But there is value as well to providing opportunities for faculty to broaden their knowledge of the ways in which science is put to work in the world outside of university. So I believe university, funding agencies, and private industry and national labs should all consider helping to provide fellowships, consulting sabbaticals, and other incentives and means to enable professors, if they choose, to spend time away from university in non-academic settings where science is used.

By the way, the community does one aspect of that very well. University professors well understand their obligation to take a term, for example, as rotators at the National Science Foundation.

Well, to summarize my recommendations and then leave some time for discussion, I suggested first that re-examination of PhD training is science be part of the discussion agenda of the many groups who have a stake in science PhDs. Second, the fellowship support be used increasingly to help mitigate some of the problems of too early specialization.

Third, that serious thought should be given to have a broadened perspective, outlook, and experience of graduate students through optional but strongly encouraged technical internships outside academia. Fourth, that analogous opportunities be created and be considered normal for faculty. And fourth-- and I haven't laid much groundwork for this. And finally, rather, that the foundations and other non-governmental sources of funds look hard once again at whether they cannot, at the margins at least, lift some of the enormous pressure for scientific conformity.

Now let me repeat that I do not expect to have come up with the best ideas or necessarily a lot of good ideas for improving graduate training. That can only being done by those of you who are more deeply involved.

And finally, I want to observe that these changes by themselves will not relieve many of the other major stresses on the research university system. And they're not suggested with that in mind.

They are suggested simply with the idea that they might improve the careers and the social utility of future science PhDs. And by so doing, strengthen the research university and our national commitment to scientific excellence. Thank you very much for your attention. I would be happy to entertain questions and comments.


At some point, Ernie, you may have to moderate. As long as the questioning is friendly, I'll do it.


AUDIENCE: [INAUDIBLE]. For those of you who don't know me, my name's [INAUDIBLE]. I founded what's called the Young Scientist Network. And I agree with everything you said here today. I have one comment that might make things better.

Prospective employers need to make things better for young people trying to find jobs outside. And that would be to advertise in places like Physics Today or Science magazine. People looking for jobs are looking in those magazines, and don't really know where else to look for them.

ARMSTRONG: Well, the problem which Mr. Ellsworth just raised, how is that one convinces in the first place-- that is, not everyone in the industry is as convinced as I am about the value of the PhD in physics.

AUDIENCE: We're working on it.

ARMSTRONG: Indeed, many of my colleagues at IBM after getting to know me may be less convinced of that now.


But there is a serious dilemma here, and the students I met with over the last few weeks and discussed that at some length. There is a tendency to look for people based on their specialties. And if you don't happen to be looking for the specialty or if it doesn't happen to be in demand this year, that really complicates life.

I think there are-- that part of the job mainly falls on the outside organizations. And it's one of the reasons why I'm addressing these remarks and subsequent [INAUDIBLE] to them as well. I think it is, in fact, possible to make progress there. And as I said, it's easier when firms are growing. It's different for different firms. Mark?

AUDIENCE: I like many of your suggestions. And in particular, I like some aspects of the idea that graduate students be supported by fellowships. But there are some problems. You addressed the problem of prolonging the PhD. With regard to the external internship, you can argue that it's possible that that might be part of the PhD thesis process.

I think our experience at MIT is that by getting students into research at an early stage in their graduate career shortens the average time to a PhD compared by other universities. And if all the students were supported as fellows, I think what you would see is a lot of students would add those years to the amount of the time it takes to get a PhD.

ARMSTRONG: Well, I think that is a danger. On the other hand, as I said, I have a lot of confidence that people will act in their own best interest, especially if they're given some guidance. That's why I had this suggestion about tracks.

I spent a year away between college and graduate school. I spent a year traveling in Europe. It was easily the best investment of time I ever made. And I spent six summers in a non-university research lab up to that-- or in that time as well.

So there is that danger. And what I propose really is that this be considered an option, a real option. That it be required is beyond my fantasy. Or beyond the opportunities, really. I mean, it would just not be possible to tell people that they had to do this requirement. But I think it's possible.

AUDIENCE: I just wanted to add [INAUDIBLE].


ARMSTRONG: [INAUDIBLE]. There are [INAUDIBLE] that I'm sure I've missed. Yes, please.

AUDIENCE: [INAUDIBLE] graduate students [INAUDIBLE] come back to work for [INAUDIBLE].

ARMSTRONG: Oh, yes. [INAUDIBLE], right.

AUDIENCE: Second question. If you model it on the NIH apprenticeship system, would it be restricted to permanent residents [INAUDIBLE]?

ARMSTRONG: Well, that's a set of questions that I'm not really confident to answer. I don't know why-- well, I just don't know. They should not be, in my opinion. First of all, there is no thought in my mind that people would be required to go back and work where they intern.

But of course, they would have contacts that would make it easier for them to do so or to [INAUDIBLE]. The question of citizenship is one that I'm sure Bob or Dave or Ernie would understand better.


[INAUDIBLE], as you know, right? [INAUDIBLE]? Yes? [INAUDIBLE].

AUDIENCE: John, there are a lot-- and especially at smaller universities of stresses that are going to cause the individual professors to coalesce their groups into sort of professor group research. I wonder if from the perspective of your talk you felt that a single one professor, one group, or both professors [INAUDIBLE] would be better in the future.

ARMSTRONG: Well, that's a really interesting-- I hadn't thought about that pressure [INAUDIBLE]. Depending, again, on the members of the group, the professorial members of the group, it could have some very beneficial consequences, I suppose, given that there would be a wider range of experience and background and contacts to draw on. But I don't think that would be a fundamental or necessarily detrimental change to the system. Yeah? Oh, Carl.

AUDIENCE: You specifically addressed your remarks to PhD and PhD in science, although you made allusion [INAUDIBLE]. Would you change any of these recommendations or statements [INAUDIBLE] directly address the issue of a PhD in engineering?

ARMSTRONG: Can you hear the question? It was, although I alluded to PhDs in engineering, would I change my analysis if I were talking to primarily engineers? And of course, there are some in the audience.

I think the only-- I have talked about some of these ideas with the Engineering Council and with the [INAUDIBLE] School. If I may say so, I think the engineering faculty are-- and again, in general. I don't want to say specifically about MIT or Harvard, or any other place. I think generally, the engineering faculties are further down this road than science faculties, which is why I presumed to give this talk the title that I did. Yeah?

AUDIENCE: I agree with your point about putting students first and encouraging them [INAUDIBLE].


But one often hears these horrible anecdotes about the professor who blanches at the thought of his student taking foreign language or engaging in business study courses, for example. And how would your proposals reconcile that tension between the students' long-term interests and professors' vested interest in keeping the students [INAUDIBLE]?

ARMSTRONG: Well, I addressed that, I thought. My first recommendation, remember, was that this whole topic needs to be vigorously and openly discussed. And as an outsider, I can't do more than say that if it were, and if it were also discussed at various national levels, I believe ways would be found to harmonize these interests in a better-- I don't think there's--

If I came across in this lecture as tending to demonize the process, I did not mean to. I'm just thinking maybe it could be somewhat improved. Yes, Paul?

AUDIENCE: John, I found your talk very provocative, and I thank you very much for that. You seem to have a very interesting thesis, which is that a PhD-- and I would say in engineering as well, [INAUDIBLE] science-- can serve as a basis for a career outside of science or engineering in modern society, which is so technologically based, and it gives the person a lot of skills. And it seems to me that that aspect might be attainable for someone who has only a bachelor's in science or engineering.

Another question is, what is the added value of the PhD? And I heard you say the ability to formulate problems, to attack them in a methodological way, and so on.

ARMSTRONG: Is it all that great an added value?

AUDIENCE: Well, my question is, is that added value worth it in some sense for someone who does not aspire to work within the system of science or technology?

ARMSTRONG: Well, I would observe-- I mean, I've thought a lot about those questions, and I did not deal with them as well as I might have. I think part of it, for lack of time.

First of all, I had many colleagues in senior management at IBM who were very effective technical managers whose only formal training was BS degrees. So I did not mean to suggest that one can only get this power, this approach to problems by the science PhD. However, I believe that the experience of doing that apprenticeship does, in fact, confer extra skills or it adds to the weight of the artillery that can be brought to bear.

I also think that the fact-- I was arguing also to sort of face what I think is the reality, that a very large fraction of people who start-- who genuinely enjoy research and who start a research career will end up doing other things. I mean, I did not go through graduate school with the notion of ending up managing the [INAUDIBLE] or the [INAUDIBLE].

But the various aspects of my own case in which I was fortunate to have this wide range of experience, and what led me to believe that the very large fraction of research scientists who will end up doing other things sooner or later in their careers, and importantly would be better at those jobs, and would, for example, find their way to them more easily and more quickly if they had the wide range of experience I've talked about. But it's a subtle argument, and it's clearly debated.

On the other hand, I think the view that PhD training is just for people who are going to spend a life in research, I really believe in today's world that's a non-starter. At least unless one is willing to see the whole enterprise shrink by a factor of three or so. Yeah?

AUDIENCE: You suggested that there would be a project that might be done in a user environment during one's PhD training. I guess my concern is, how would you structure that project so that it would be a valuable experience?

As it is, it seems that it's somewhat hit or miss, finding a good project that's the right scale for a PhD. And certainly, my experience as doing an undergraduate project which was six months or eight months is that it's much more a hit or miss whether you can do something that you can learn from and still make a contribution to. How would that work?

ARMSTRONG: Well, I think first of all, that's why I suggested we'd have to start small. A second way, it's something that a lot of effort has to go into. But I know Professor [INAUDIBLE] cases where that has worked out very well, and others. So I think it's a question of raising the sense of priority and attention that we devote to that.

It's not going to be easy, but then change is hard. And you have to think about how to make it really a mutually perceived benefit on all sides. That's a very important part of this as well. And the hook is about to come out.


MODERATOR: It's self-defense. I'll take over the moderation. Perhaps two more comments. [INAUDIBLE].

AUDIENCE: I just wanted to ask a question. [INAUDIBLE]. And the thing is, I would imagine that the problems that are associated with having a PhD thesis [INAUDIBLE] partly the university [INAUDIBLE]. And that's something I wish [INAUDIBLE] you would tell me more about, because I know it has [INAUDIBLE]. Not necessarily [INAUDIBLE]. And how did that go at IBM?

ARMSTRONG: Well, by the way, I think first of all, this is a very common concern. My own experience, it is-- and [INAUDIBLE] in my own experience that IBM may be peculiar in this regard. It was the easiest difficulty to be overcome. For example, if something is patentable-- that is, if a patent comes out of the work-- well, all right. Somebody better get on [INAUDIBLE] and file a patent before the thesis is due, or before the publication is to be published.

AUDIENCE: But you said something very important [INAUDIBLE]. In your experience, it would be helpful.

ARMSTRONG: Well, last time at the first lecture, remember, I said that [INAUDIBLE] the R&D world is really 10 or 20 times bigger than the part that is sort of visible to the world of science and the university. So there is a small part of work in industry that would lend itself to thesis work. But still, it's a substantial part. In absolute terms, it's substantial.

Now by the way, I think I did not make clear, and I do want to make clear that I don't think that these internships have to be thesis-related. I think that's asking too much. There aren't that many opportunities. Many of the benefits of seeing how science has done or technology has done out there, many benefits, the context, the perspective would come from doing work which was full of content, but which was not necessarily published.

Now that will interrupt-- if one did it after year 4, obviously, that would be a serious problem. If you do it after year 2 or 1, it doesn't strike me as a serious problem. But I don't think it's possible [INAUDIBLE] let's say, as a customer in Japan to have 60% of all PhDs for work done in industry. I'm not suggesting that [INAUDIBLE].

MODERATOR: So one last question.

AUDIENCE: You have [INAUDIBLE] suggest to reduce the total number of PhDs. [INAUDIBLE] the nontraditional career path, is there any [INAUDIBLE] for PhDs?

ARMSTRONG: I'm not sure I understand the question. I did not explore the question of how many PhDs the world ought to have. And there are some people in the audience who have been thinking about that, and I have some thoughts. And that would be another lecture.

But the fact is that as I said, it's just plain fact that many, many PhDs end up working outside of research. Moreover, in a number of fields that I know well-- for example, computer science in industry-- the willingness to-- or the understanding of the value of PhDs is increasing. So is new development.

When I started at IBM 30 years ago, it was almost unheard of that any of our 40 development labs hired PhDs. Now it's routine. I mean, the world does change. People do.

And so I think the answer is that we can do better with sort of the size of the enterprise it is now. Whether that enterprise will or should shrink, as I say, is a subject for another lecture. Thank you very much.