About Joseph Giordmaine
Joseph Giordmaine retired as vice president of physical science research at the NEC laboratory in Princeton, New Jersey. He received his bachelor's degree from the University of Toronto in 1955 and did his Ph.D. work at Columbia University on the use of the maser amplifier in planetary astronomy, working with Charles Townes. After leaving Columbia, Giordmaine went to work at Bell Labs in 1961, working with ruby lasers and harmonic generation, eventually moving into nonlinear optics. In 1965 Giordmaine and Bob Millar worked together to develop an optical parametric oscillator. In the 1970s Giordmaine was promoted to laboratory director, and from the 1970s through the early 1980s held a variety of management position at Bell Labs in the research area and, eventually, in the company's development area. In the late 1980s Giordmaine left Bell Labs to help set up a Western-style basic research laboratory for the Japanese company NEC, where as vice president of physical science research Giordmaine also sits on the Board of Fellows.
The interview begins with Giordmaine's education at the University of Toronto and his work with maser amplifiers with Townes at Columbia, then moves to his experiences with Alley Trevon and his eventual move to Bell Labs in 1961 as a research scientist. Giordmaine discusses his early experiments at Bell Labs and his work in the field of nonlinear optics. He describes Bell Labs' openness to long-term basic research principles during the 1960s, briefly discussing how optical fibers "won" over pipe and microwave technology in telecommunications development. After discussing his optical experiments, Giordmaine describes his experiences as a manager at Bell Labs and the complex managerial problems caused by separate research and development laboratories. Describing his attempts to bring the two types of laboratories and scientists in contact with each other, Giordmaine suggests that certain research environments are more conducive than others for basic research and applications research. Giordmaine discusses his move to NEC and the ways in which he brought his research management experience to the Western-style basic research laboratory established by NEC at Princeton. The interview concludes with a discussion of Japanese research attitudes and the advantages and disadvantages of working within a foreign-owned company. Giordmaine asserts that the NEC Princeton laboratory proves that the Japanese can and are willing to do basic research and should not be seen as merely building on basic research done in other countries.
About the Interview
Joseph Giordmaine: An Interview Conducted by William Aspray, Center for the History of Electrical Engineering, April 18th, 1995
Interview #251 for the Center for the History of Electrical Engineering, The Institute of Electrical and Electronic Engineers, Inc.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, 39 Union Street, New Brunswick, NJ 08901-8538 USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
Joseph Giormaine, an oral history conducted in 1995 by William Aspray, IEEE History Center, New Brunswick, NJ, USA.
INTERVIEW: Joseph Giordmaine
INTERVIEWER: William Aspray
DATE: April 18, 1995
PLACE: Princeton, New Jersey
Childhood and Education
This is an interview of the eighteenth of April, 1995, in the NEC offices in Princeton, with Joseph Giordmaine. Now, could you start please by telling me something about what your parents did and your early life, where you were born and when?
All right. I grew up in Toronto, Canada. I was born in 1933. My father immigrated to Canada from Malta. In Canada, he performed in various amateur musical groups and also became a very well known professional magician in the Toronto area, even appearing eventually on several American television shows. He also had a technical background, and worked as an electrical technician. When I was in grade school, he often showed me electrical experiments; I think that got me interested in science, originally. And one thing led to another. I became very interested in radio experimentation and so on, in high school, and decided very early on that I wanted to be in engineering or science. I studied at the University of Toronto and received my bachelor's degree there in 1955.
Graduate Work on Maser & Astrophysics
When I was approaching graduation, Charles Townes, who at that time was at Columbia University, came to visit Toronto. I was in my senior year and had been accepted at another graduate school, for graduate work in physics, but he came and gave a talk on the maser. At that time the maser had just been invented, had just been demonstrated. This talk was in late 1954, and I think the demonstration had just been done. I had never heard of it before. I was not quite sure I understood the maser completely from his talk, which was my first exposure to it, but at that time it was quite clear that here was something completely new and different that I had never heard of before. This somehow brought together some interests that I had in microwave spectroscopy, which I had been reading up on at that time, and in electrical experiments and microwave technology. So, this seemed like an ideal field, and it was clear that this was a radically new direction in that field, so I applied on the spot, and got some of the professors there to put in a good word for me, and I soon was accepted to Columbia.
I went to Columbia in 1955. My graduate work there was in a field related to microwave spectroscopy, but which led in somewhat different directions. At Columbia I had the opportunity to work on the development of a maser amplifier for radio astronomy. Although he's best known perhaps for work on the laser and maser, Townes' enduring interest over the years, especially in recent years, has been in astronomy: radio astronomy and infrared astronomy. This was an opportunity for him to combine the maser work with work in astrophysics and astronomy. I think his astronomy interests probably complemented the work on the maser. He started the maser with the idea of having a low noise amplifier for use in radio astronomy which would allow work in the microwave region where he had a great deal of technical expertise, and which would allow work with lower noise levels in the microwave region. That drove the work on the maser amplifier, which I was fortunate enough to be involved in as a graduate student. So we built the first maser amplifier used for radio astronomy.
We got a decrease in noise level of a factor of about ten, and a noise temperature between eighty and a hundred degrees kelvin. This was a factor of about ten in noise level, in noise temperature, relative to what had been previously done, and that translated into, a factor gain of about three to four, which was really pretty significant. We made observations of planetary astronomy at the naval research laboratory. We mounted the maser amplifier at the focus of an antenna down there. This was a large vessel containing liquid helium, and mounting this at the focus of their large fifty-foot antenna was really very difficult. But we solved these difficulties, and we made a number of observations of planetary radiation in the x-band region, primarily from Jupiter and Venus, so we were able to improve the precision of the temperature measurements on Venus and come up with some atmospheric models that explained the variation in the limited data that was available at that time: the variation in the black body temperature, the surface radiation temperature of the planet, as a function of frequency. So that was my thesis.
And did this get picked up by the astrophysicists who were interested in this line of research?
Yes, it did, but it was a difficult tool to use. Two dedicated graduate students were required to keep it going. A colleague of mine at Columbia, Lee Alsop, did thesis research with Charlie Townes on the same subject. We were jointly involved in this work. Alsop did observations, on a different astrophysical object, not the planets, but we worked together — this was the kind of effort that really required graduate students. Developing the maser amplifier into a practical tool would really have required a great deal of work. And, at that time there was competition, because the parametric amplifiers had been developed at a number of laboratories, including Bell Labs, for the Telstar amplifier experiments, which were being done about the same time at Bell Labs. Ours was the first to be used for radio astronomy purposes. But Bell Labs had a far more elaborately engineered device designed for the Telstar program, and it worked at the same frequency. They were developing both maser and parametric amplifiers.
At that time the parametric amplifier, though it didn't get quite as low a noise temperature as the maser, was easier to operate, and didn't require a liquid helium temperature. I think it worked at nitrogen temperature, or perhaps at room temperature. Because it didn't require a liquid helium temperature, it was sufficiently easy to operate that it turned out to be the more practical device for the Telstar experiments, though there were some observations using the maser device with Telstar. So the master device's uses were limited. Because it required helium temperature, and the helium temperature had to be at the focus of the antenna, the master had limited practicality. I think that later, Arno Penzias, at Bell Labs used maser amplifiers, but because of the difficulties they were used only in a very small number of locations.
Another question to fill in some background: with your undergraduate and graduate education, can you talk about your interest and your training in mathematics and physics and engineering fields? Do you see yourself as a mathematician or physicist or engineer, a theoretician or experimentalist, and just how did your training look?
Well, I'm an experimentalist. When I started at the University of Toronto, there was a combined program in mathematics, physics and chemistry: students interested in any of those fields went through the same first-year program. I began my first year knowing that I was definitely interested in mathematics, but in my first year I was exposed to a very impressive chemistry teacher, Frank Wetmore, who's since passed away. He dramatized chemistry so much that I decided to concentrate in chemistry, and spent the next two or three years' chemistry option. But then my senior year I moved into the physics and chemistry option, because that seemed more exciting at that time. So in that way I got into physics, and that led to my going to Columbia.
Ali Javan and Bell Labs
So after you'd done your dissertation work and you're thinking about your career. What were your opportunities and how did you choose what to do?
Well, about the time I was getting my degree at Columbia, the idea of generalizing the maser to laser was really in the air. Charlie Townes had thought a great deal about it. He discussed it with graduate students very widely at Columbia, and we had had a number of informal meetings to discuss ways of going about it. I worked on one approach, although my main interest was still in the microwave area. Just before I left Columbia the first laser was built by Theodore Maiman, and then shortly thereafter, while I was still at Columbia, Ali Javan built the first helium neon laser at Bell Labs. I had a chance to go out and visit Bell Labs at that time to meet Ali Javan. Actually, Ali Javan had been at Columbia when I arrived there. I was his assistant in fact. That was an extremely valuable experience, because he was a master experimentalist, and working with him was a very valuable part of my training, especially the work on microwave circuitry.
After I'd been at Columbia for three years, Javan went to Bell Labs and began to work on helium neon laser. At Columbia, he has begun to develop his ideas of how a helium neon laser should work. When he arrived at Bell, he had a very clear idea of what he wanted to do. It took him over two years to get it to work. I stayed at Columbia for a year after I got my Ph.D., working as an instructor. I soon decided that really I wanted to go somewhere else. When I began to think about where I wanted to go after Columbia, I visited Ali Javan.
Was that because you had decided that you didn't want an academic career, or because you wanted to get broader experience?
In an industrial lab at that time, there were so many research positions available, that you could feasibly consider that to be a career direction. Very few people have that kind of option in industrial labs today. By career option in industrial labs, I mean the kind of position where I could go to an industrial lab and be offered an empty lab, and told to go off and do some research, without being given a specific assignment. At Bell labs a number of people were being hired then with that sort of charter. That appealed to me a great deal, because I wouldn't have to be teaching. I did teach at Columbia and I found it very time consuming. Lecture preparation took a long time; in an industrial lab at Bell, I'd have a hundred percent of the time to do research. That was very attractive. And it was clear that the laser would be a revolutionary thing, though I didn't quite realize then how. I decided at that point that I really wanted to do research on lasers.
Lasers, Harmonics & Nonlinear Optics
So, I applied to Bell Labs, I got interviewed at Bell Labs. When they asked me what I wanted to do, I told them that I wanted to do work on [word] spectroscopy on lasers, utilizing the very high resolution obviously was going to be available from lasers. So I came to Bell Labs June 1961. When I got there, ruby lasers as well as helium neon lasers had made their appearance. The labs down the hall were very heavily involved in ruby laser research, although the first laser had operated elsewhere. There was — Art Schawlow, Bob Collins, Don Nelson, Wolfgang Kaiser, Geoffrey Garrett had collaborated on what turned out to be the second paper of the ruby laser. This paper followed the report of the discovery of the ruby maser at Hughes. They got together and very quickly recreated what had been done at Hughes, but carried it a lot further. They demonstrated the coherence of the laser light, the real laser properties, which had been only partially done in Maiman's first work. So they had demonstrated a solid-state laser for the first time, in a way that revealed all of its properties and all of its potential. There were ruby rods around, flash lamp pumps, and so within a few weeks, having borrowed things, I set up a lab and had a laser flashing in my lab. I didn't know what to do with it at that time. It wasn't the kind of laser that was suitable for what I had originally in mind, namely [word] spectroscopy, which would have required a gas laser. At that time, setting up a gas laser was a major enterprise; it would have taken months of planning, which I really hadn't yet begun to do. It was fascinating to set up this ruby laser. I could borrow the parts and set it up.
I set up a laser in my lab, and I started flashing it at crystals. I had a spectrometer ordered at that time, and one of my objectives was to look at the possible changes in the spectra of the light going through crystal. I was looking at the diffraction effects, which at that time were all very strange and foreign. I was really just getting warmed up and trying to understand what a laser was. At that time, just when I was engaged in the preliminary activity, the first report came from Peter Franklin and his group at the University of Michigan treating their first observation of second harmonic generation. What they had done was to shine a similar laser at a crystal that had no center of symmetry — in that case, quartz — and reported a very, very small amount of second harmonic light. That paper came out in Physical Review Letters, and I was kicking myself because I had been shining the laser at various crystals just to get a feeling of what was involved in the experiments, but they had had the really brilliant idea that in order to see second harmonic generation, one would need a crystal that had no center of symmetry. They recognized that. I immediately reproduced their experiment. Because of all the work on piezoelectric generators during the war, Bell Labs had an enormous supply of crystals like quartz, and KDP [Potassium Dihydrogen Phosphate], and other ferroelectric crystals, noncentrosymmetric crystals, and so forth. So I was able to round up several of these crystals, and shine light on them, and after a couple of weeks I was able to reproduce the harmonic generation experiments.
Now at that time, Bell Labs had all sorts of resources. One of their great human resources was physicist Walter Bond. He had been at Bell throughout the war and had made very important practical contributions to the understanding of the use of piezoelectric crystals in sonar detectors and acoustic detectors. Bond had a practical knowledge of crystallography, and of crystal cutting — practical things that one needs to know to make a piezoelectric detector. I immediately began to try to understand what the issues were in second harmonic generation. So I did careful experiments with KDP, which turned out to be a little better than quartz. I started rotating the KDP, and making systematic observations with the spectrometer, and I soon found that as I rotated the crystal I saw very systematic increases in the intensity of the light. As the crystal was rotated, the intensity of the light got very strong, and then receded again as I went past a certain angle. So I went systematically into this, and dug out the information on the anisotropic refractive index of these materials, and started to theorize about how the light was generated in such a crystal. I cut crystals in different directions, and found that as I approached a certain direction I saw an extremely large generation of second harmonics, thousands of times more than from putting the crystal in the beam. I worked through what was going on, and it turned out that I was seeing a phase matching effect in which the light coming into the crystal was coming in one polarization, and the light coming out of the crystal was in the opposite polarization.
The second harmonic generation was very weak, because the fundamental light and the harmonic light go out of phase with each other because of the difference in the refractive index. But, because of the birefringence I could make up with the birefringence what was lost with the dispersion. The dispersion tended to slow down the harmonic relative to the fundamental. But I could arrange it so that, by choosing the angle, the birefringence speeded up the harmonic relative to the fundamental, and if I turned it to the right angle I could get them to match. That turned out to have an extremely big effect — literally thousands of times more intense. When I discovered that, I was able to put out the second publication in the field of nonlinear optics. That really launched me on my way, and I quickly followed this up with other things. I soon forgot about my intentions of doing high-resolution spectroscopy. The whole area of nonlinear optics developed into a very fruitful field.
I worked on various aspects of nonlinear optics for the next few years. This culminated, for me, in 1965. I was collaborating with Bob Millar, an expert in ferroelectricity. We thought of taking advantage of the nonlinearity to build a parametric oscillator. The idea of a parametric oscillator had been predicted in the Russian literature of parametric oscillators, which we weren't aware of at the time. We first had to learn how parametric devices work. There were experts at Bell Labs in that area, one of who had written a very readable book on the subject. We easily learned the background of parametric devices, and the theoretical techniques for doing the calculations. The book had been written from the point of view of microwave parametric oscillators. We adapted that theory to an optical parametric oscillator device, and we calculated there should be enough gain, using kdp as a non-linear optical crystal.
So we set up an infrared laser — a neodinium calcium tungstate laser, since there was no neodinium yag, at that time — and set, calculated what the angle should be to get the phase matching, and the way that the parametric generation should change as a function of angle, because we were generating two frequencies of light out, out of the incident frequency, that added up to the input frequency. The experiment worked this way: we had an infrared source of one micron, which was the neodinium calcium tungstate source — that we harmonically generated to fifty three hundred angstroms, roughly, and that was the pump for the optical parametric oscillator, two frequencies that added up to the frequency of the pump, and sure enough, one night, this thing worked. It had a sharp threshold, and an enormous efficiency coming out of this thing. It was very exciting time. Since that time the optical parametric devices have gone through a lot of phases — initially they were used as a tunable source; then the dilasers really turned out to be a much more practical tunable source of light. More recently they've come back, because of scientific interest, because the photon correlations between the two beams that come out of a parametric oscillator are correlated in ways which make them a very good experimental source for squeezed light and other quantum optics experiments. By now — the technology has improved so much that they're now also a valuable tunable source of laser light. I did a number of other experiments in the area of nonlinear optics at that time, in the early seventies.
Bell Lab's Interest in Optical Technology
Before you go on, tell me: why was the Bell Labs management interested in this line of research? How did it fit into the greater objectives of the laboratory?
Bell Labs had decided, early on, that optical technology would be important for communication. Now that required a certain degree of vision, because in the early sixties the common view at Bell Labs was that this communication would take place only for long haul applications, and that the communication would be achieved through pipes. By controlling the refractive index of gasses in the pipe, or through some other method, light would go through these pipes. This would be a very substantial transmission medium, and a very expensive one — much more costly and much more complex than what actually developed as a result of the invention and recognition of fibers. But the Bell Labs had recognized from the very beginning that the ability to control coherent light would be important for communications. So, Bell Labs enthusiastically supported work of any kind that would advance the scientific understanding of handling light, of studying light's properties. This clearly included processing light through non-linear optics. They were very supportive of anything that I wanted to do, or that others wanted to do, in the area of nonlinear optics.
They initially sponsored and encouraged Ali Javan's work. Javan's work was very interesting, because he was very demanding — his work was much more expensive than work done by the average research scientist at Bell Labs. It required more space; it required more equipment. In retrospect, now that we know how to do it, we know it wouldn't have required very much. But back then to attack the problem in a grand way, and to optimize the chances of success in what already was becoming a competitive kind of field, Javan had the right idea. You had to do things right, you had to take several approaches, and that meant a much more costly approach. Management at that time — Ted Gibralt, and others, in the physics research division management — recognized that this was a good bet, not just scientifically but also in terms of telecommunications technology, that it made sense for a basic research laboratory to do this. This way probably an ideal situation for a lab to foster basic research which would have some impact. Really good basic research is open-ended. You don't know exactly where it's going. If you knew where it was going, it wouldn't be basic research. You put things in the hands of a researcher in whom you have confidence, and you give that researcher real support. You don't tell him or her what to do. You give them a free hand. And in Javan's case that worked out. Their bet, which was a very substantial investment, obviously paid off.
This wasn't applied-development kind of research, because the researcher didn't know exactly how he would approach all the problems, but the research had focus, it had long-term direction. It was a little like the early work on the transistor. Management had a general goal; they believed that understanding of solid-state physics, the physics of transport, and particularly semiconductors, would achieve some goals in ways which would improve on vacuum tube technology. But management didn't really know how to go about the research. They brought in the right scientists, and gave them the opportunity to do work that management hoped would lead to results that would be useful for telecommunications. The work being done on the laser in the late fifties really paralleled the work that had been done ten or twelve years before on the transistor. And there was a very parallel kind of management spirit. Each company believed that the work was in an important area without fully understanding how the research would be used for the company's goals. Laser development at Bell Labs and transistor development were quite parallel in many ways.
When the management recognized promising research results, was there a transfer mechanism for moving this into an applied area? How did that occur?
That took a longer time. As soon as the laser came along after work at Murray Hill, Bell Labs started work at Holmdel to study the transmission effects, and to develop it. The main effort was on transmission and the understanding of propagation effects of light in the atmosphere, propagation effects of light through pipes. At Bell Labs at that time there was a very sophisticated and extremely elaborate technology in optical waveguides, waveguide propagation in a highly transmitting mode. Stuart Miller was in charge of this technology. For many years Bell labs had done research on waveguide propagation with the idea that, over the long haul, this would become a viable means of low loss transmission. Because they were oriented towards pipes, Bell Labs naturally thought transmitting laser beams through pipes, with the beam isolated from the wall of the pipes, perhaps through some gas flow, or through some other means of having a small refractive index, a gradient in the pipes somehow that would isolate it from the wall. In that time, the early sixties, however Charles Kao's discovery of the optical fiber became known, and so optical fiber work began at Bell Labs. That work was done largely at Murray Hill. Immediately there was much development on the fabrication of fibers, and this led, very systematically, into development at Bell Labs, ultimately leading to their very major manufacturing facilities in Georgia, for fiber propagation.
Transition to Optical Fibers
This presents an interesting kind of management question. The company had been fully supporting this alternative approach, before the fibers became available, and there were researchers who were geared up, and had been strongly supported, and then the company went in a different direction. How did the management deal with the problem of workers who wanted to continue in a certain line, or whose investments had been put in this direction? There was a management issue here.
Yes, there was. Because I wasn't directly involved in that decision-making, I don't have a very good answer. My impression though, was that it was handled very smoothly. As soon as the high transmission of the fiber was understood, there seemed to be a very universal recognition that the fiber was going to be the winner. Because, in fact, propagation through pipes had really not gone very far. It had gone far in the microwave region, and the decision to drop the microwave approach was a far more difficult decision, because there was a great deal more technology in that area — a great deal more experience, a great deal of practical understanding. Even as the fiber research continued the microwave work continued too, because it wasn't completely clear then just how successful the fiber optics would be, whereas the microwave technology and the pipe technology, were both very well understood — demanding, but very well understood. So that research continued for a few years, after the fiber work began. That was a much more difficult decision, because there was a much greater investment in that technology than there had been in the laser pipes. Laser tube propagation had been just at the research stage. There was confidence that somehow it would work, but that was a relatively minor investment, so I don't think that was a difficult transition. I think the really difficult transition came with the decision to drop the microwave transmission technology and accept the fact that the future lay in optical fiber.
Managing Research at Bell Labs
I see. I disrupted your flow earlier. You were starting to tell me about other work you were doing in this period.
Yes, in optics. I did a number of experiments in collaboration with many others at Bell Labs, including Wofgang Kaiser, Sergio Porto, Bob Millar, Don Nelson, and others, on various aspects of nonlinear optics. But in the late sixties I was fortunate enough to be promoted to be department head, which slowed me down somewhat. Then in the seventies, I was promoted to laboratory director. At that point, my research output dwindled very quickly, and I found it very difficult to carry on experimental work while holding this responsibility. I think others found ways to do it, but I found it very difficult. From the mid-seventies, my main activities at Bell Labs were in management. I was director of various laboratories: first, the solid-state electronics laboratory, which was the laboratory in which I had done all my work. This brought me in contact with a wider range of interests and activities at Bell Labs, including III V semiconductors, which was a very big part of the work in that laboratory. During that time there was early work on light-emitting diode materials. Some of the problems were that it prevented efficient light-emitting diode materials in galling phosphide, particularly, were solved by people like Ralph Logan, at that time. Also at that time — and even before I headed that laboratory — the first continuous lasers, the CW room temperature lasers, were being built in that laboratory by Mort Panish and Isuelt Hiachi. That was revolutionary, because their work indicated that the solid-state lasers — the ideal match for the fiber technology — were going to be practical devices. I was also involved in the management of laboratories that carried gallium aluminum arsenide laser technology to practical devices. I held management positions, then, in the research laboratories of Bell Labs, and then, in the early eighties, also in the development part of Bell Labs.
Before we go on, take a moment to tell me a bit about your experiences as a manager. How was the management structured, and what were the duties and challenges, both at the department head level and at the director level? What did you do, what were your greatest frustrations, what were your greatest pleasures? What was the job like, and how did it differ from what you saw at other places?
It was different because Bell Labs had a tradition, maintained throughout that period — the late seventies and early eighties — of having an independent research laboratory in one part of the organization and a development laboratory in the other part of the organization. The independent research laboratory had scientists who had been hired to do independent research. They expected to do the work that they wanted to do, that work which they chose to do. This work might or might not be related to the interests of the company. Whereas scientists in the development laboratory were involved solely in work where a product was clearly in line. Western Electric, the part of AT&T responsible for producing the products actually sponsored and paid for the development. So there was a clear division: there was a development laboratory and there was a research laboratory.
The real issue in the success, or lack of success at Bell Labs, was the managing of the interface between the research laboratory and the development laboratory. The research laboratory consisted largely of people who, and I don't use the term disparagingly, but who really were prima donnas. Their success in their own line of research had been determined by their independence, and their vision of some particular result, or excitement over research in some specific area. To the extent that they were successful scientists, their goals often were determined by scientific goals rather than the success of the company. As in any company success in taking advantage of the basic research comes through finding people who would be scientifically creative, and productive. But, if you're successful in recruiting those people, those people are not driven by the company objectives. Their directions, and their sense of who their peers are, in universities and other companies, the people they meet at scientific society meetings — that's the audience that they live in. These people, and I'm talking about the best people, these people have scientific objectives, and although the times are changing very fast today, nevertheless many of those people really felt that, then, Bell Labs was their patron. Their work if it was in the right areas would have value for Bell Labs, but overall they were concerned primarily with scientific creativity. Many of them felt themselves to be part of a community that consisted of the other scientists working in their field, rather than part of the AT&T community.
In retrospect, I'm particularly sensitive to that now, as a member of a Japanese company. In Japan that scientist working in an industrial laboratory feels a part of is the company, not the scientific community. That's an oversimplification, perhaps, but a scientist working in a company in Japan is working for the company, not for the science. So at Bell Labs, and at IBM and Xerox and other places, you had a group of people in research laboratories whose motivations and whole directions were directed towards the scientific community. Now, on the other hand, they're supposed to be interfacing, if their research is going to have an impact on the company, with this development department. And the development department also consisted of first-rate people, at Bell Labs, whose loyalties and whose responsibilities were really quite different. And yet these people had been trained in the same school, in the same kind of tradition.
So how do you manage the interface between these two groups? It's a very difficult management challenge, and it's one that I don't think has been very successfully handled in the United States or anywhere else. It's had its successes, first of all the transistor and the laser, that came out of Bell Labs, and a number of other advances. But it's very difficult interface, and dealing with that interface, although it may not have been the most time-consuming part of the work, was a very important part of my responsibilities, as I saw them, and also [those of] my opposite numbers in the development area. There would often be — and this wasn't unique to Bell Labs — frictions, because one side would look at the other and think it saw a certain narrowness, unwillingness to accept certain ideas; one side would look over to the other side and see people whose loyalties perhaps were elsewhere and not with the company, a very complex interface. Dealing with that was a very important part of my responsibility.
I also helped people in the research area, the area which was my responsibility, to get exposed to the excitement of the challenge that was on the development side. I tried to expose people to that challenge, to stimulate that challenge, without having people feel that they were being redirected against their will. I tried to encourage work that was scientifically exciting but that had some opportunity for application without issuing any orders. At Bell Labs, research management never issued any orders to do anything. That wasn't the way — that isn't the way — that a research lab works. On the other hand, I wanted to provide opportunities, to work with people, to encourage people, where people are doing some work that maybe had limited scientific interest, but that they were very excited about. Bell Labs would always support work that had first-rate scientific significance, even though its applicability to Bell Labs technology or to AT&T technology might be very problematical or not very clear. If it had been first-rate work, AT&T would support it. But what AT&T did not want to do was to support work that would generate a lot of papers, without having any impact on the technology. I mean, that's death!
So, what do you do when somebody is pursuing that kind of work? You really have to make suggestions of other options to try, make other contacts available, and try to change that situation, so that our work improved our scientific output, or improved our technological output. The thing about Bell Lab was that they understood that the time scale of basic research was not short. Now that's changed of course in recent years, but at that time, it was understood that five years, ten years, was a horizon that was not unreasonable for basic research. We're far less patient these days. In this institution, actually, we operate a little bit like the old Bell Laboratories in many ways: there's a respect for the value of long-term work. Here that's probably on a smaller scale relative to the company's size. At that time, however, it was clearly understood that some of the best scientific work was inherently long term, and that it had to be supported.
What sorts of personal and technical skills were required of the manager? You've suggested to me that the research manager had to have a breadth of understanding, and a depth of understanding, so that he could evaluate the research that was being done by the research scientists in their organization, and then help them to choose paths; there must have been personnel skills — but I'd rather have you talk to this point. What do you think were the traits that made a good research director within Bell Labs?
Well, I'm not sure that I was a good research director, first of all! I'm not sure I can answer the question directly, but I think it was clear that the scientists, if they were going to have a good relationship with the management, the scientists would have to feel that the management was interested in their work, had some degree of understanding of their work. I think that most researchers would like to have as little to do with the management as possible except to be sure that the support was there when it was needed. But people realize, of course, that their own career directions depend on the management's perception of them, and so I think the key thing is that the scientist really has to perceive that the manager understands the work. Maybe not so deeply that he can actually be considered a researcher in that field himself, but enough that he understands it well enough, that the scientist feels that his or her interests are being defended in the organization at the time of the performance reviews, and on other occasions. A person has to have confidence that the management is supportive of the work and is going to represent the work well. And I think the researcher has to feel comfortable. If the researcher feels that the manager is always pressuring him, or changing the direction of the work, and pushing and pushing, in ways that are not congenial, or that are not comfortable.
Ideally speaking, the manager has to promote a sense of community. All of this takes time. Maybe it took me more time than it takes others, but it all takes time. But I think that a feeling of community and trust and also a feeling that you know something about what the person down the hall is doing are important. So, there has to be a sense of togetherness, there have to be ways of mixing people, and broadcasting the work internally, so there's a community, not just a group of people who are getting their research support paid for by Bell Labs and who are always on the phone to university xyz and IBM and their own community. That way the company doesn't take advantage of those people quite so much. You want them to be part of that community, although some may not feel that, because they gain from being part of that community, and the community obviously gains from their work.
Exposing Researchers to Technological Side
An important aspect of the relationship between research and development is the question of technology transfer: how do you make that transfer from research to development? It's pretty obvious what you need to do, or attempt to do in that situation. But there's another aspect to that isn't often emphasized, and that is the reverse question of getting the researcher exposed to the technology. In retrospect, I saw that in my graduate work on the invention of the maser. The maser was invented where it did and by the person who invented it was that that person had not only the physics knowledge — the microwave spectroscopy, the detailed knowledge of molecular spectroscopy, which hundreds of physicists had — but also was able to combine the physics, the molecular spectroscopy in that case, with the technology. Townes had worked at Bell Labs, in the early years, had worked there during the war, had gotten familiar with all the microwave engineering, so that he could bring to bear, in the work that led up to the maser, the microwave, combination of microwave engineering and molecular spectroscopy. That was just really an explosive mixture, because it required both, an intuitive understanding of the electrical engineering issues of the microwave resonators and the wave guides — one just had to have an intuitive understanding of those issues to think of what amplification meant, in the context of molecular spectroscopy. So there was a case where the basic research was aided by being close to the technology.
There's also the case of the integrated circuit, for example. It's very significant that the integrated circuit came out of Fairchild, and TI, and not out of Bell Labs. I think one reason for that is that this interface between the fundamental research and the applied research, in the area of silicon technology, for better or worse, was designed in a special way at Bell Labs. Shortly after the transistor was invented, the management responsible for the silicon device physics, and the device technology, made an early decision to have that research done in the development area. It was really applied research; it wasn't development, it was really research, because of that decision, that whole effort then became a little isolated from the basic research organization. They were doing basic research in silicon also, first class research, but it was separated from the main part of the basic research organization. In the research area you didn't have to do serious experiments, you could be guided by your intuition, and you weren't expected to have some important results to report.
There was a lot of room for work that was really under the table kind of work, work you didn't tell your management about, because if it failed it had no significance at all, and in a sense you had wasted your time. Many people handled that kind of problem by not informing their manager; you just did it. If it worked you had something very important, but if it failed, you didn't have pie all over your face. That was an accepted way of doing business in research. As a manager you didn't probe too much into what people didn't want to tell you. But in development, you can't really operate in that kind of way, quite so readily. I think the fact that research and development had such limited contact may have been why it was difficult to do the kind of playful work that would have been required to lead to the integrated circuit. Because in the early work on the integrated circuit, you're connecting two things, multiple copies of things together, and there's no science in that. There's no quantum mechanics, there's no depth. On the other hand what's needed is not depth perhaps, but a vision of what this thing could mean as a device. Now there was in fact some early work on, that was sort of suggestive of future integrated circuits, at Bell Labs. Ian Ross, for example, did some experiments connecting various circuit elements together, including transistors, but the real invention came elsewhere. And I think in retrospect I think that if the organization had been a little different, had had more emphasis on bringing the research climate and the development climate together, and pushing them together — because they don't attract each other, they have to be pushed together! I think that that might have led to a different result. But it is significant that Bell Labs climate was such that it could produce the transistor, with the depth of scientific understanding that was needed, but was not able to produce the integrated circuit, which required a different kind — not quite so much the scientific depth, but a kind of combination of development and research views that come when you have a successful interface between those two areas.
NEC Research Laboratory
I'm mindful of the time. Maybe we should move onto the next stage in your career. What happened?
Well, I had various positions in development and research, and fully expected to have my whole career at Bell Labs, but in the late 1980s I was approached by NEC with the news that they were planning to start a basic research laboratory in the United States. This was something like four years after the divestiture in AT&T, and it was beginning to be clear that AT&T, because of the divestiture, could not, in the long term, support the level of basic research that it had in the past, because it had become a smaller company. This was not because it didn't understand the value of basic research; however, the level of competition in the electronics area was changing, and because AT&T no longer held a monopoly, AT&T could no longer charge the expenses of basic research in the same way that it had when it had a monopoly. Because it thought at that time of course that basic research was an expense of doing business. It was defined as —
Part of the rate structure.
It was part of the ratestructure. It was part of the way of doing business. But when the whole system became competitive, it was clear that research was going to have to justify itself on an economic basis, and that AT&T was going to have to compete with companies that were getting along very well without doing any research. It was clear that times were changing. Now at the same time NEC and other Japanese companies — especially in the 1980s when there was the flexibility to do things — NEC at that time of course had a much smaller basic research organization than AT&T had. NEC in fact had been founded, actually in 1899, as a joint venture with AT&T. So NEC has had over the years various close connections with AT&T, and a number of the people in the senior management of NEC have had — been employees of AT&T at one time or another. So they had a very good feel of what AT&T basic research was like; and at that point they concluded that they should really be doing more basic research. And they decided around 1985-87 that they should do that was by enhancing their basic research in Japan, without doing all the research in Japan; they would do some of the research in the West. And I think that decision reflected their understanding of this whole atmosphere that I've described at Bell Labs — I mean the awareness that the best scientists really related not to the company but to the scientific community.
That kind of situation is really hard to think about in Japan. Scientists are as good in Japan, if not better — they're certainly as good as the scientists in the United States, but they operate in a completely different cultural milieu. The idea of having people operating within a company as scientific entrepreneurs is almost unheard of in Japan. NEC recognized that having scientists working for science's sake within a company gave the west a certain advantage in doing basic research. You get people willing to go out and do crazy things, take great risks of failing, and in the Japanese environment it's a little more difficult to take risks. It's not considered in quite as friendly of a way as it certainly was in companies like Bell Labs and IBM and Xerox. In those companies, failure was understood; it was understood that to do good research you had to start projects that had a high risk of failure. To have a long-standing environment of that kind, you also have to have a management structure that's tolerant of failure, because if somebody fails and his desk is found out in the hall the next morning, nobody else is going to undertake that kind of research. So there has to be an obvious acceptance, within the institution, of failure.
So NEC recognized that there were certain culturally different ways of doing basic research in the west, although the scientific caliber of the people was the same on both sides. NEC felt that it would be an important step to support a research lab in the west. So they started it up, and the first president of this laboratory, Duang Kong, is a long standing employee of Bell Labs, who had taken early retirement just at the time he was picked by NEC. In NEC, it's very important that the top management really be known to the company. Duang Kong had a long-standing relationship with Mikiuki Anihara, who was at that time and is still the chairman of NEC's board of directors. So he was very well known in Japan — although he's Korean, and that was an unusual situation, having a Korean be the president of a Japanese company located in the United States. When the company got started, the funding was intended to allow us to build up to population of about fifty principal investigators, fifty scientists, with a total population maybe of about a hundred, hundred and thirty, hundred and forty. Right now we're at forty scientists and a total population of about one hundred.
We're well along the way to our planned level, but because the years have been very tough economically, over the last few years our growth has been slower than was anticipated, so we're not there yet. We're growing slowly. But it's a measure of the long-term viewpoint of a Japanese company that in these very difficult times, after losing money after about two years, the first time that had happened since the 1970s oil crisis, the company had to freeze or cut back its overall support of R&D, but nevertheless allowed us to continue growing, during that phase. Their long-term point of view has been an asset for us; I think an American company in these tight times would not have been as patient, and as supportive, and supportive of a long-term point of view as NEC has been. So they've encouraged us to grow, they've allowed us to make all the decisions on hiring, and on the directions that we want to emphasize. They've maintained the view that if we're successful in doing first-rate science — in areas that obviously have some connection with the company's business, where success would effect the company's future business, particularly in the computer area — that if we do good enough science, it will all pay off for NEC in the long run. So we've tried very hard to promote connections within the institution. Our work here is half physical science and half computer science.
And I've tried very hard to promote the connections between basic research in computer science and the physical sciences, and I've discovered that this is very difficult. The modes of thinking, the whole perspective of basic researchers in those two areas is really very, very different. And the language is different. We've tried quite novel ways, I think, of trying to promote that interface. Rather than having one corner of the building be used for computer science, and one corner of the building be used for physical science, which is the case with all other basic research laboratories that I know about, physical science and computer science are in adjoining rooms. There are a lot of casual conversations, people go to the canteens, they talk to each other in the halls, and socialize — that's helped lead to some interesting connections. So we have, for example, chemists whose interest is in the growth of open framework materials, so-called open framework structures that we hope can be built, with metal atoms, for example, to make things like quantum wires, quantum dots, allowing nature to do the photolithography, in a sense, and all at the microscopic, quasiatomic kind of level.
We have a chemist who's concerned with those kinds of problems, and the synthesis of inorganic materials, and next door to him is a computer scientist who understands graph theory and theoretical computer science and algorithms, and these people get together, for example, and come up with schemes for defining for a whole class of materials, what the mathematical limitations are on the structures that can be grown. This can lead to results of good chemical interest — made possible by someone with computer science interests, but in an area where you wouldn't have the connection unless you did this kind of collaboration. So we're trying to promote those kinds of connections. We have people interested in optical interconnections in computers, and physicists, electrical engineers, who are talking with computer scientists, who understand the architectural issues far better. Those kinds of interactions are developing very well. At the same time, we have a physicist here who's become interested in fuzzy logic, and has found ways of fundamentally enlarging the applicability of fuzzy logic; by looking at the definitions, at the operations in fuzzy logic — has found a way to really bring inference into fuzzy logic. There's been very little use of inference in fuzzy logic in the past, and a physicist here was able to do that, in part by collaborating with a scientist in the NEC laboratories in Japan, who's a computer scientist. We've had a number of these kinds of interactions.
Right now we're seeing it, for example, in biophysics. We have biologists, and we have a small biophysics group. There's a lot of interest in DNA computing, and a scientist here in the computer science division has shown that these computers which have been demonstrated to be able to do Monte Carlo calculations, can also be Turing machines. So it looks as though these computers have broad capabilities. But we've been able to do that work because the computer scientists, who brought to it all the understanding and feeling that you'd need to reach a conclusion like that, have down the hall biophysicists who understand the DNA synthesis and DNA programming and DNA replication and so on, and the two can talk to one another and really exchange information. And so we know we can work to move quickly into a new field that requires those people to be talking to each other. That's very important. I've found that some of the experience that I've had at Bell Labs in helping people communicate with each other has been very important here.
Culture of and Recruitment to Lab
I'm watching the clock. I want to ask two more questions; although I have many more I could ask. I'm curious about this organization structure. The two questions are: as the Princeton lab has actually evolved, has it become more like a traditional American research lab, or like a Japanese research lab, or some hybrid of the two? The other is, as you've tried to attract talent into the laboratory, has it made a difference that this is a Japanese company rather than an American company?
Well on the first question, we are not at all like a Japanese laboratory. We are relatively isolated, obviously, by distance and by language, from the Japanese lab. This is a great difficulty for us. As technology develops here, issues of technology transfer arise. And they have arisen in a few areas — although not in the major areas, yet. I think there'll be difficulties to be overcome. But we also suffer, I think, from the fact that this reverse direction that I was talking about — the exposure to the high technology that exists in the production divisions in Japan, and the R&D group in Japan — we don't see that in the intimate way that someone working over there would. We see it when we visit, when we collaborate with them, but it's far less of an intimate contact. So we don't have the advantages of having exposure to the development laboratory.
On the other hand, we were founded to be an American-style laboratory. In fact we only have one senior Japanese scientist. We have one visitor, and there's one Japanese scientist who's a senior management person; he's the liaison person. He's an experienced scientific manager, but he's the only permanent person from Japan we have here. So we're American, our style is very American. In fact, we have management tools here, and techniques here, that are nothing like Bell Labs' tools and techniques. We're really more American than Bell Labs in that we have a management structure that consists of a Board of Fellows — there isn't time to go into details, but the board of fellows is really a collegial arrangement, it's much like a university faculty arrangement. The decisions on hiring, the promotion decisions, are all handled by a board of fellows. I mean, as vice president of physical science research, all the scientists nominally in physical science report to me. But in fact, I'm simply a member of the board of fellows. I handle the administrative work, but I'm a member of the board of fellows, and the board of fellows makes the key decisions. And the board is chaired not by a manager, but by one of the senior scientists. So, in fact, we have a more academic style than perhaps even the old labs at Xerox or IBM had. So we're not at all similar to a Japanese structure. In fact, we're on the far side of an American structure in many ways. Now let's see, the second question was —
About attracting talent.
Yes. Interestingly enough, I do think that has been an issue, but with a minority. With this minority, I would say that maybe about five or ten percent really have questions about what it means to work for a foreign-owned company. The reason that more haven't had that view is because we assure everyone that everything that we do here is published in the open literature. We simply ask that prior to publication, people publish an internal memorandum which is circulated internally here and circulated in the R&D organization in Japan. And if there's patentable information, that we have a chance to file. We do ask that. But after the thirty days — and it's only thirty days, we have an absolute maximum of six months, which has never been invoked — in fact I think we've just had one or two cases which have gone over thirty days. But thirty days, we assure people, in general, after they've written up a manuscript, it will be published. Everything. So that everything we do here goes into the scientific literature. We're making jobs for scientists in the United States.
Oh, okay. All right, thank you. We're making jobs for scientists and other technical support people, and administrative staff, at a time when the scientific support in this country is going down. So the fact that we're contributing to the scientific community here, that we're bringing money into the scientific community here, has made people more comfortable. I think that if we were doing development work, which would be completely confidential, and our role was going to the universities, seeing what was going on, and taking technology, patenting it, not publishing, sending it to Japan, I think there would be far more sensitivity. But I think the reason we've been successful in attracting the people that we have attracted is that we can argue that we're not getting one dollar of scientific support here, and yet we're contributing to the scientific community in very significant ways. I think people can come here and really feel that they're a part of a larger process by which the Japanese are contributing to our scientific environment. Obviously, it's not altruistic on their part, but it's reversing a trend. Before these kinds of laboratories existed — and there still aren't very many of these — it used to be that people would complain that the Japanese had very outstanding development organizations, that they were competing with us, that they had all the memory business, they had all the CD business, and we make the inventions in the United States, we do the research, they acquire the results and they run with it, and then they're our toughest competitors. We do the basic research here, they don't do the basic research, and yet they benefit from it. This laboratory is part of the answer to that objection.
We can say that they are putting it back, and that they are contributing to the basic research community in this country. And this also makes people — it made me — feel very much more comfortable. So it seems to be working. Now to see whether it's really worked, in the long run, I think it's going to take about five more years, because in ten years you should be able to see progress. We look forward to '99, which is actually the one-hundredth year anniversary of the company. We're doing important work now, of course, but we hope to have something by then which really has a big impact on the company by ten years. And they have a right to expect that, I think.
Very good. Thank you very much. This has been most interesting to me.