Michael Godfrey, Information Sciences Pioneer

The advent of modern computer systems and the expansion of information sciences provides a fantastically rich historical narrative. In the discussion below, readers are invited to learn from the perspective of Michael Godfrey, who was a witness to, and central participant in, many of the most important advances that have made possible our ever-increasing technological world.

From childhood in Oregon and then Switzerland, Godfrey always pursued his interests in science and engineering, and after completing his education at Deerfield Academy, he arrived as an undergraduate at Caltech. His campus experience was rich both intellectually and socially, and as he reflects, it provided him with solid footing to pursue new courses of study at the London School of Economics. From Princeton and then at Bell Labs, Godfrey was well-positioned to apply his dual expertise in economics/statistics, and electrical engineering. He saw Bell Labs in its heyday as a vibrant environment that undertook both fundamental research and problem solving as the country's communications systems became massively more complex.

After the divestiture of Bell Labs, Godfrey rounded out his professional affiliation at Stanford University, where he ran a research lab in the Information Systems Group, and between the chronology of his appointment and recent advances in computation, he experienced first hand the many ways that computers were revolutionizing all aspects of society. It was also at Stanford that Godfrey became more interested in computer history (broadly defined). At the end of the discussion, and reflecting on his long career, Godfrey emphasizes the importance of developing a more complete understanding of the information content of physical systems, which is an endeavor that is certain to keep busy future generations of information systems experts.

DAVID ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Thursday, March 30th, 2023. I'm delighted to be here with Michael D. Godfrey. Michael, it is great to be with you. Thank you so much for joining me today.

DR. MICHAEL GODFREY: Thank you!

ZIERLER: Michael, to start, would you please tell me if you have any current or most recent institutional affiliations, and your most recent title?

GODFREY: Not anything that is all that official. My last title was Visiting Professor at Stanford. A title before that was the Schlumberger Professor at Imperial College in London, which was an endowed professorship, which didn't last very long, due to Margaret Thatcher and various forces of evil. That's when I went to Stanford and stayed there for quite a long time. I ran a research lab in the information systems group (ISL) at Stanford.

ZIERLER: Tell me about some of the most important research topics that you focused on in your career.

GODFREY: Let's think about that. My early work with John Tukey on power spectrum techniques for signal processing has proved to be useful. The paper that he and I wrote is my most quoted paper, I'm sure. It pops up quite a lot. I also did some work, motivated by Oskar Morgenstern indicating the lack of information in economic data. There were a number of things I did after that. The things I did fairly recently at Stanford were to develop much more precise ways of using semiconductor material to sense whatever it was you were trying to measure. The best result in that area was due to one of my students who devised an incredibly clever technique for sensing whatever it is, a voltage on a capacitor or something like that, in a much more precise way than had been realized was possible before. If you want to know the detail, it's pretty simple. Amazingly, nobody had thought of it. The analogy is, you're trying to pour ping-pong balls into a big bucket, and if you just pour them in until the bucket fills up, there's a lot of bouncing around. A better way to do that is fill the bucket nearly full, and then just trickle in the last few balls that will fit, and then you get a much more accurate result. That's easily done in semiconductor circuits. You could develop a circuit that does that very easily and quickly, and that produces a really dramatic improvement in the accuracy of whatever it is you're trying to sense. That's now widely used in all the sensors that measure things in your phones or elsewhere. So, my helping with that—the student got the patent because he thought of it. I encouraged him, but he figured it out in detail. That's the most prominent result from the work at Stanford.

We did quite a lot of work on using semiconductor material for sensing whatever needed to be sensed, and that's widely used in various—one group that used it a lot was the group that makes pick-and-place machines. They need to be able to sense where the part holder is, to poke the part into the right place. They used our sensors because they were more accurate and smaller and better. I think that's the most recent thing that I think of—the other thing that I think is interesting, but I've not had much success explaining it to my academic friends, but it is true that people for a long time have said that photons have two degrees of freedom. Dennis Gabor said this a long time ago, and everyone seems to believe it. But they failed to read an early paper by Poynting that points out that photons have orbital angular momentum, which can take a large number of states. I wrote that up, and I don't think I, in any serious way, contributed, because other people had noticed it at the same time. In any case, the current version of high-performance optical fiber communication systems are using the property that allows them to put, for example, 64 or even more bits per photon into the channel and therefore get a huge improvement in channel bandwidth. I think it was helpful that I wrote some of this up. Of course, the serious fiber communications people have done the real work.

ZIERLER: Michael, what aspects of your career have been more on the engineering side, and what aspects have been more on the basic science or fundamental research side?

GODFREY: That depends on how you measure it. If you measure it by a monetary measure, it's certainly true that work that I did at AT&T—I started this work at Bell Labs—I was at Princeton, and at Bell Labs. But then I spent quite a long time at AT&T headquarters in New York before the divestiture. Let me go back a step. The reason this came about was due to the 1952 consent decree. AT&T had to file every year with the FCC to show them that the company was behaving properly, not doing things that were unfairly disadvantageous to competitors or things like that, or subsidizing one service against another. For quite a few years, they tried to do this by sending lawyers down to Washington to tell the FCC that the company was behaving properly, but that didn't go very well. The main reason was that they filed their claimed costs and tried to show that that matched the objectives, except it never came out that way [laughs] in the end. This was due to problems of collecting the data, problems of getting organized. So, the revenues and costs came out differently from what was expected. The FCC, of course, suspected that AT&T actually knew what was going on but was trying to put on a better face or even take unfair advantage. The truth was, AT&T didn't know what was going on sufficiently accurately to produce the results they wanted. The real focus at AT\&T was to provide universal service'' in what they saw as a fair and efficient way. They were very good at that. But, they were not good at explaining the costs. The network was just too large and complex. And, specifically, highly non-linear. This made approximations quite difficult.

So, Bob Breedlove and I and a small group of very talented people took on the task of figuring out precisely how the AT&T message toll system of telephone service for the United States actually worked from the standpoint of its long-term incremental costs, and therefore could determine appropriate prices and best provide the communication services. That was fairly hard work, because there were only a few hundred million phones around, and a lot of switches, and a lot of complications. And, not just some non-linearity, but discontinuous switches. But we built a system that produced accurate long-term costs and benefits for the message toll telephone system for the United States. That did, in fact, convince the FCC. This was largely due to the fact that our claimed predicted results were very close to what actually happened. We had a computing system at 195 Broadway that did all the computing that worked this out. (I wrote a lot of the code.) But, this was also dependent on a very large amount of data from the operating companies. People with substantial knowledge of large data problems described what we did as the then largest data collection and analysis yet. Finally, one day, the FCC wrote a letter to Sperry UNIVAC, who provided the computers that we used, that said, "We want a computer exactly like the one at AT&T." [laughs] We agreed to send them all the code and the data and stuff, so they could run the model themselves and convince themselves that what we were saying was accurate and we could go ahead with the proposed pricing, and everything was happy and the lawyers went home. A lot of other work was saved. One of my staff said, "Do you realize if you made a 1% error, it would be $100 million?" I said, "Are you sure about that?" Well, it's

$100 million a day. [laughs] So, real numbers were involved in that work. If you measure by dollars and by value to anyone with a telephone, that was by far the most significant work that I did.

It wasn't all that simple to figure out, really, exactly how the telephone system worked and where all the costs were, because it is a quite complex and non-linear system and it included data traffic as well as alternate routing of the traffic if the load was heavy and. But, we got answers that were exactly right, and the FCC agreed that we weren't making up stories anymore.

Sadly, as you well know, that came to a crashing end, because IBM decided they didn't like AT&T doing business that might interfere with their business, so they convinced the court to prosecute AT&T, and that led to the divestiture and the breakup of the AT&T company into separate companies as exist today. Most of us think that was a very sad day, because telephone service in the United States is nowhere near as good as it was when AT&T ran it. We know that for sure. So, that's kind of the end of that story. It didn't end as happily as it could have.

ZIERLER: Between your individual affiliations and your academic affiliations, has all of your research been applied? Have you always had applications in mind? Or sometimes do you do things simply out of curiosity?

GODFREY: That's hard to say. I certainly wouldn't work on anything that I didn't think was really fun and interesting, and I've always, in any case, had a rule that if I thought of anything that might be useful, I first tried to find someone else to do it.

ZIERLER: [laughs]

GODFREY: That, I think, is just common sense. There are other people who think, if they have a good idea, they really ought to run off with it, exploit it, and patent it, or get some price for it. I do the reverse. If I can find anyone that shows any sign of being able to figure it out, I say, "Just go for it. I'll help you if I can." So, there were a number of things that—some of them didn't turn out—but some have turned out that other people did the work for, and therefore they got the credit. I do love the story of—I knew Oskar Morgenstern very well, and I would have known von Neumann if he had lived long enough—. But they were sitting at a math meeting, and a person named George Dantzig came bouncing up. You may know Dantzig as the mathematical programming guy. He explained he had just figured out linear programming. Von Neumann said, "That's really great. You ought to develop that result. That will come to something." Dantzig went skipping off, happy as a lark, and Morgenstern said, "But last week you explained that algorithm to me already." [laughs] And von Neumann said, "Of course. But he's a young guy, and he's enthusiastic and he figured it out himself. He ought to just go for it." That's the right attitude, I think. So, if I can find anyone else to do the work, I do encourage it.

The Applications of Signal Processing

ZIERLER: What have been some of the technological developments over the course of your career that have really been game-changers, in terms of applications, in terms of capabilities?

GODFREY: I think the signal processing techniques that John Tukey and I developed have had the widest application. They are used in medical signal processing, in geophysical exploration and a wide range of other applications. It's our most quoted paper. People quote it in a wide range of fields, because a lot of things turn up as time signals. Fluctuations of various kinds. The power spectrum methods and complex demodulation that we explained in the paper we wrote a long time ago tends to be a reference for all that stuff. So, there's no one thing. The techniques are widely used in imaging, in particular in medical imaging systems, but they're used in lots of other applications, too.

ZIERLER: A question that brings us together—President Thomas Rosenbaum wrote an article about Bell Labs, and you responded to that, and then Tom said it would be wonderful to talk with you. What resonated so much about Tom's article about Bell Labs? How did it speak to you?

GODFREY: I just recognized it as the labs I knew when I worked there, and it was an amazing place. I don't know what more to say. It was perfectly open. Anyone who felt like it could just walk in and say hi and chat with anyone there. There were no barriers or restrictions. All the inventions and stuff got patented and protected, but the patents were made available, so the lab really was a service to the community. That went on for a long time, until IBM put a stop to it, is the basic truth. It isn't that widely known that the breakup of AT&T and therefore the Labs breaking up and just dying away, was in fact instigated by IBM. The story that IBM tells, it was CBEMA, the computer equipment manufacturers' association, that was afraid that AT&T was going to take away their business. The truth is, it was IBM that was afraid that AT&T was going to do harm to their business, and IBM wanted to enter the communications and telephone business. They lied to AT&T about their behavior and motives, and AT&T very sadly thought that the IBM chief executives were in some sense telling the truth. It was true that Charlie Brown, when the truth came out, when the divestiture was forced on him, came to me and said he was really sorry he hadn't listened to me about who was playing what game. But, it was too late, then. So that, in my mind, is—the inverse of any success I've had. This was by a huge amount the biggest failure, that I failed to convince the people that needed to know what was going on. That's that sad history.

From Oregon to Switzerland

ZIERLER: Let's go back and establish some personal history. Let's start first with your parents. Tell me about them.

GODFREY: That's a fairly long story. My father and mother were at university together in Oregon. My father went on to work in journalism. He taught at the university for some time and ran a newspaper. My mother was seriously interested in Shakespeare and the Romantic poets and things like that. The thing that I've never fully understood was how different they were from each other. If you remember ancient history, Wayne Morse was the senator from Oregon for a long time, and my father was one of his key political advisors and go-getter to round up votes and do exciting things. This was nothing like my mother, who read poetry and went off to England to visit the appropriate Shakespearean and romantic places, but they somehow got along. When the War broke out, my father joined the military, and therefore he disappeared at the time when I was hardly recognizing him. I didn't see him again for a long period. All that period was under supervision of my mother, who had somewhat non-standard views about how to bring up children. [laughs] That included, "If you want to go out and play, just do it, and come back when you feel like it." She had other definite views that made sense, to me at least.

One thing I remember well—she liked to travel around, particularly when my father was away. Anyway, she felt that this was an opportunity to see other parts of the world. So we'd drive around somewhere and we'd come to some town, and if she thought she might stay for a few weeks, she'd drop me off at the nearest school, and just said, "Go in, and register, and you can go to that school until we do something else." So, I did that for a number of times. My list of former schools got longer and longer. [laughs] The best school by far was when we were on a large ranch outside Tucson, and it had a school which was a single room. There were 15 or 20 children, all of which except for myself and my sister were somehow related, and of various ages. When I first walked in, I sort of said, "What's going on here?" One of the children said, "Well, the teacher may or may not come out. She has to drive all the way from Phoenix, so sometimes she doesn't make it, but depending on your age, you ought to look for a book with the right number on it"—like a three or a four or something—"and just, read it." So, I started reading books with a three or four on them. We were there for less than six months, but noticed that I had gone through, I think, five numbers [laughs] of books by the time we left. I learned a lot in that school. I really did. That began to cause me to form a view of how education works, which is not well suited to regimentation by age and so on. Some people learn a lot quicker and a lot differently from others. So, I learned a lot in that school.

Other schools that I happened to be in for short periods were not as good an experience. The worst was, the teacher announced that today's lesson was to memorize the multiplication table. I said, "I can't do that." She said, "You're the dumbest student I've ever had. You'll never go anywhere." That didn't influence me one way or the other. Another fun educational experience came about due to the fact that when the War came to an end, my father stayed in Germany, in the military government, and then the High Commission for Germany. When things calmed down a bit, we moved to Switzerland, which was safer than Germany. You may recall that the Marshall Plan took five years before implementation started, so it was a five-year period between the end of the War and any kind of normal life in Germany. People were still starving, and horrible things. During this period lived in Switzerland, in Montreux, the French-speaking part of Switzerland. My mother did the standard thing; she dropped me off at the local Collège and said, "Try to remember how we got here, to get home again." [laughs] I walked in, and said, "Here I am."

ZIERLER: How old were you, then?

GODFREY: Nine, I think. Yeah, nine's right. I have a hint. I'm really bad with numbers, [laughs] being a statistician. Anyhow, the first person didn't speak English, so there was a slight communication problem. Finally, the English teacher showed up, and we could communicate.

They said, "Fine, you can enroll, but you're not allowed to take an English course, because that would be unfair." I discovered that I was taking German and Latin in French, without knowing any of those languages. [laughs] That was a little bit challenging. One thing that surprised me—I somehow thought that I could do the vocabulary, which was typically French-German or

French-Latin, and know what German word meant what French word without knowing what the word meant. But, at least I could not do that. I had to look up in my little dictionary what the English for those words were, and then I could remember those two words meant the same thing, and meant the same English word, which meant I had to work a little harder than the other students who all knew French already. But, it was fun, and I began to improve and be more accepted—I learned enough French that I could communicate with my fellow students.

Then, winter came, and it was discovered that I didn't know how to ski, so I immediately dropped to the very bottom of the hierarchy in the system. That was in part why I decided, "I'm really going to learn how to ski." I may have mentioned this to you—at the same time, an amazing book called La Méthode Française de Ski, by Émile Allais, had just been published.

Émile Allais was one of the real top French Alpine skiers. He won lots of races and did a lot for French skiing. He also, amazingly, developed a methodology which us slightly more techy folks know as the minimum work principle. He said, "The way to ski down a mountain is in harmony with the mountain." That was clearly meant to mean, "Figure out how to do it with the least expenditure of effort. It's not only more efficient, but it's quicker." There's physics that explains that, as well, not just for skiing. He really came to an amazing result, and it really works. So, I went from being the skiing dunce the first year to winning the school downhill the second year, and I was sure that Émile Allais had helped me out. It was really good.

ZIERLER: Michael, did your father ever talk about his work? Was he able to talk about his work?

GODFREY: Some. There were really substantial parts of it that he certainly could not talk about, because while he never quite admitted it, I noticed that he was also doing intelligence work. I shouldn't even say that. But it became evident. But, there were parts that he did. The most important part—Munich was 60% totally destroyed by aerial bombing during the War. When we first arrived there, basically the rubble in the roads had been plowed away in big piles on the side of the road, and that was all that was left of much of Munich. You could drive through, but there wasn't much to see except piles of rubble. One plan was to just rebuild a new city, but my father thought it would be better to rebuild Munich the way it was. And that's what came out. Now, Munich is one of the really nicest cities in Germany, and it's a real tribute to my father that he made that choice. As a consequence of that, for quite an extended period, people in Munich did really like him, and he liked being there. He helped organize and joined in the Fasching celebrations, one year he led the Oktoberfest band and he organized the establishment of the city and Bavarian political systems. He had a TV program for some time. So, he was well-known in Munich and well-respected, for more than because he rebuilt the city.

He did occasionally talk about other things about the War: one of the somewhat amusing ones, he arranged a press conference for Patton at the time that he had come across Northern Europe at an amazing rate and crossed the Rhine before the bridges were all destroyed. However, when my father organized a news conference for Patten people started asking him questions, and one of them was, "How do you move your troops so fast?" Patton answered: "We don't take prisoners." Pretty soon after that, with a few swear words mixed in, my father had to say, "The press conference is canceled. We're not going to go down that route." [laughs] That was a kind of thing my father helped with. Patton also said—I'm not sure that there were still press people around, but he said to Eisenhower, "Just give me the word and I'll take Berlin." Roosevelt had agreed with Stalin that the Russians could take Berlin. It would have been a much better choice if the Americans had taken Berlin, and Patton certainly technically could have done it. But he then added, "And of course, I'll keep going until Moscow." [laughs] But, as you well know, it didn't happen. The world would be quite different if it had.

ZIERLER: You would have been young at the time, of course, but did you notice deprivations in post-war Europe?

GODFREY: Oh, absolutely. We spent some time in Southern Germany, in Munich and in Tegernsee, and around there, before there were any non-military personnel. Because my father was the commanding officer for the whole area, he chose that he could make a rule that said we could—our family, my sister, my mother—could come for a few weeks in the summer. So we saw Germany when it was really devastated, and it was not a pleasant sight. People were just dying in the streets. It was really quite alarming. Cigarettes were the currency. You could get a full set of Meissen china for a few cartons of cigarettes, for instance. That five-year period before the Marshall Plan started was really a difficult period. And, the military did things that weren't all necessary. All the trains were segregated. There were train cars for the Americans, and a car at the back for everyone else. Germans who somehow had the resources to have a car were not allowed to drive their car on weekends so the roads would be free for Americans to take trips around. So, there were rules imposed that were, at least to my mind, not entirely necessary.

The way accommodation was found—when we moved to Munich, we moved into a very nice home in a suburb of Munich—the way that those homes came about when American families were moving in after the Marshall Plan started—military police just picked nice-looking homes and told the occupants, "You can pack one suitcase, and leave in eight hours." That was how the homes were acquired for American families to live in. That was not the most friendly approach to the problem. And, they picked really nice homes—our home had been owned by a prominent antique dealer who had really spectacular furnishings and things, and they got to take one suitcase and that's it.

ZIERLER: Was the Cold War palpable to you? Could you feel the Soviet threat, growing up in that environment?

GODFREY: Yes. Not constantly. I mean, I was really interested in other things, like skiing and stuff. But we went to Berlin, and the day we arrived, the Russians had machine-gunned a bus with American tourists in it. So, it was pretty evident that there were unfriendly people about. It was also true, of course, that the German population was not entirely friendly to the occupation, and there was a lot of war material lying around, so I had grenades thrown at me a few times, and things that were a little uncomfortable. So, it wasn't the safest environment in the world. But overall, the occupation, once the Marshall Plan started, really did a lot of good, and the bulk of the German population accepted that this was the best they could hope for, better than they thought they might have gotten, certainly different from the Russian zone, where basically nothing was done. Overall, it was, I thought, a friendly environment, except for [laughs] the odd grenade and things.

Also, as I think I mentioned, I got pretty seriously focused on skiing. That was easy, in a way, because the only educational institution in Munich was a school set up by the military, which I went to for a day or two but decided that was not suited to my interests. So, I spent all the winter skiing in Garmisch, which was not far away. I started skiing there quite a lot. A guy came up and said, "Do you really want to be a skier?" I said, "That's the only thing I want." He turned out to be one of the coaches of the German national team. So, I joined the German national team, and trained and raced in competitions, mainly in Germany and Austria. For a couple years, I had a career. I did kind of think that that would last, but a miraculous event occurred: a person came through to do some business with my father, and my father must have mentioned, "I have this son who seems to just go away and ski and do things… I don't know what to do about it." The person talked with me and told my father "Well, I think he ought to go to Deerfield Academy." He went back to wherever he came from and got the school to send all their materials. That caused me to decide that that was a better choice than a few more years of FIS skiing. So, I did that, and it was definitely the right choice. Deerfield was and is an extraordinarily good school. It was the first school, after the one room in Arizona, where I really learned anything.

Early Interests in Science and Engineering

ZIERLER: Were you particularly interested in science and engineering growing up?

GODFREY: Yes, definitely. As a really small child, I was curious about how things worked. I tried to figure out how the sewing machine really worked, and how record players could make music, and how radios could function—some mysterious wave arrived and caused sounds to come out. It was just those sort of obvious, practical things that caused me to try to start figuring out how it actually worked. Partly because I was so isolated, I had to figure out ways to get information to be sent to me, so I learned right away to get Popular Science magazines and get books from far away. I started learning things at a fairly early age and built model airplanes and trains and radios. I think I mentioned that I flew some of the planes and had trains that ran around and worked pretty well, but I sent away for a pulsejet engine, which was supposed to be capable of being attached to an airplane and it would fly. This wasn't entirely suited for an 11- or 12-year-old. The pulsejets were the things that Germans used to bomb England, extensively.

Mine was a tube about two feet long, a model version. But when I finally got it to work, it really blasted out, and flame poured out the back of it, for about three feet, and burned a wall, and caused [laughs]... So, that was kind of a high point of my experimenting with things. At this time my father chose to buy a BMW roadster. I kind of took it over, learned to drive it and do maintenance on it. When necessary BMW people would come out and take it away for servicing or repair. My father mentioned that he could never get them to send a bill. My father had arranged the Marshall Plan financing for BMW which seemed to explain this.

I was always interested in geometry because I wanted to figure out how things fit together. I sent away for a geometry book so I could do the math and figure out—make drawings. Particularly I wanted to make contour drawings so I could draw potential ski slopes in the summertime, and imagine setting the control gates for a slalom down the hill that I had drawn, using contours.

That was sort of a naïve, simple introduction to having an interest in the mathematics that lets you figure all these things out. I started doing that at a pretty early age, and that did help when I went to Deerfield. The rest of the students had gone to school for extensive periods, but I sort of caught up, because I had worked on my own, didn't have to do what the teacher said, and figured things out.

For whatever reasons, I did relatively well at Deerfield. While still in Germany people said, "You really ought to go to school," and I said, "Yeah, right." Occasionally, I'd go, and I noticed that exams were pretty easy. You could think about it, and not try to work out the problem in detail. If you thought a bit, you could see that—they were typically multiple choice—you could get rid of almost all the choices, because of simple things, like the answer had to be an integer, and most of the choices were 1.5 or .3, so you could just throw those away. Then you could see right away what the right answer was and go on to the next question. That was just fun. I don't think that was because I was in any real sense smarter. I was just bored, and wanted to think of something better to do than just answer the questions.

That caused me to begin to think about how to do these tests that Deerfield, of course, had to administer the SATs and all that. The SAT, the tests that are still used today, are of the same nature. You just check boxes. Among the things I did, I asked if I could put a carbon paper under the answer sheet, for the practice exams. Among the things I discovered—you might think that the mean value was sort of in the middle; the mean value was actually to the right. So, instead of looking left to right, if you look right to left, you could see what the right answer was a lot quicker. Other simple checks could produce getting the right answer without having to spend a lot of time and effort. People began to notice that for the exams, some of which were three hours, I would get up and walk out in less than two hours, sometimes quite a lot less. People at first thought I just got frustrated, but when it turned out I got all right answers [laughs] there was another reason.

Curiously enough, when I went to Princeton, where ETS had their offices, John Tukey, who was my mentor and was a consultant at ETS, said, "We have to go over and have a chat with them." So I went to the ETS people and said, "I've got to tell you a few things about your exam mechanisms." They were really shocked. They didn't realize that they had made it simple to just get the right answer without any due process of figuring it out. For example, you could eliminate all but one answer and then pick the last box that was left. It was also true that the answers were biased, in those days. The right answer was—there were five boxes to check—the right answer was predominantly to the right. Somebody at ETS sort of thought that that was OK, because it would mean that the person had to think about the different boxes until they got to the right answer. So I'd just start at the other end, get the right answer anyway, but quicker. The ETS people were a little upset. But that kind of testing I honestly think was and still is not a good way to choose who gets to go university. It's a different—checking the right box isn't all there is to learning. But it's an important factor to at least learning to play the game.

From Deerfield to Caltech

ZIERLER: How did Caltech get on your radar?

GODFREY: I had heard about it in Munich. I was trying to remember exactly how that happened. I read these technical periodicals and things, and something that was done at Caltech must have come up in one of them, and I thought it was really interesting. By the time I got to Deerfield, I had sort of thought that Caltech would be a good choice. I learned more about it. But that did cause trouble, because Deerfield, as you likely know, really likes its students to go to the Ivy League schools. Because I seemed to be kind of techy, they really thought I should go to MIT. I really thought I did not want to [laughs] go to MIT. MIT is a big school. It's regimented. I know people who went there. Feynman went there, incredibly enough. But a lot of other people struggled, and it's not a happy place, I think. It's better now than it was. Anyhow, I looked at Caltech and found that it was a small school, and there were nice people there. A Caltech Professor came to Deerfield to interview me. This confirmed my view and I accepted with, as needed, a full scholarship.

The practice at Caltech has been that real faculty go to interview the people who apply, or whoever comes is interviewed by real faculty. Most universities have admissions people who decide on admissions. Caltech, I'm not sure they do it as much now but they certainly have in the past, used real, serious faculty to go talk to whoever applies and decide if they're suitable. So, one of the faculty came all the way to Deerfield, said he thought I'd really like Caltech. I said, "That's true," so I went. Deerfield was surprised. I was the first student ever to go to Deerfield who went on to Caltech. In particular, the headmaster, who was a spectacular person, just amazing, he just said, "If that's what you want to do, you ought to do it." Instead of going to MIT or Dartmouth or something, or scoring points in an Ivy League school for Deerfield. He had each student's interest in mind no matter what. He was Frank Boyden, who is well-known in some circles. He was headmaster of the school for 60-some years and just an amazing character.

Deerfield was a defining experience for me. When I came, I said to the Headmaster, "I gave up a serious career as a skier. I was on the German national team. So, I kind of made a sacrifice." I thought he'd just say, "How could you do such a stupid thing?" He said, "Well, I'll let you take weekends off to ski here, and if there's a race in New England, you can go and compete if you want." I was the only Deerfield student who was allowed out on that basis. So, that was a great experience. The next thing, his wife—he was well educated, but his real skill was in managing people like me and more extreme cases. His wife had scientific training, so she taught the chemistry course that I took. And that was amazing. First of all, she was an extremely good teacher. Second, she had found the chemistry book that Linus Pauling had just published as a college chemistry text, so she used that as the text for the course.

I really thought I had learned a lot. I really thought I knew a lot of chemistry. So, I went trotting off to Caltech and discovered that Linus Pauling was teaching the chemistry course. I thought, "Ah, terrific. It'll be a breeze." At the first meeting, he even said the text for the course is the same book that I had already read. However, when he handed out the first homework assignment, I could not work a single problem. This is an event that's burned into my mind. Nothing like that ever happened to me then or since. I literally ran over to the Chemistry Department and said, "Pauling handed out these problems. I cannot work them. It's impossible. Can someone help me?" They said, "Oh yeah, I think he can. He is right in that office with the door open." And that was Linus Pauling. He spent probably close to an hour—he didn't just say, "The answer is 43" or something. He really explained the chemistry and what the right answer was. Then he said, " You shouldn't have been surprised: I referenced the book, but I'm not actually teaching anything that's in that book. It's just background. I'm teaching quantum chemistry." But that was a dramatic experience for me, because I had never seen a problem I couldn't figure out in my whole life, and I couldn't figure out a single one of them. That was a defining introduction to Caltech as well.

ZIERLER: Was your plan to pursue chemistry from the beginning?

GODFREY: No, no. The chemistry just came up because—

ZIERLER: Required course.

GODFREY: Yeah, you had to take—the freshman courses were almost all required. Physics, however, I handled differently, because a notice was stuck up on the Physics notice board that said, "For people who want to do it, Feynman is going to give the lectures in the Fall, for whoever wants to come. But, you have to understand what's being taught in the standard freshman physics and take the standard freshman physics exams. But you can come to these Feynman lectures as well, if you want." This was well before the Feynman lectures that became the Feynman Lectures as they are now. Maybe 15 or so students decided to go to the first one of the Feynman lectures, and we all trotted over to the designated lecture room. Feynman came in with no notes, no nothing. He started talking. One of the students got out his notebook, and Feynman said, "Put that away. You can't listen to what I'm saying and write it down at the same time." That caused some people to decide maybe [laughs] they weren't going to take that course. But, I thought, "I'm just going to sit here and listen to this guy and see what I can figure out." What perplexed me a bit—I mean, not only did you really have to concentrate, because he wasn't just saying something about a bouncing ball or something; he was saying pretty serious physics—but he came into the room with no notes of any kind, he talked for exactly the right length of time, and the concluding sentence was exactly a conclusion, and he walked out. Aside from trying to understand what he was saying and learn some physics, I was really perplexed: How can this guy apparently just come in and do this, without any preparation? When I got to know him more, I found that he had practiced the lectures over and over, at home. He made notes and he arranged things so that exactly the right things he explained in the right amount of time, and would come to an orderly conclusion. But when he was giving the lecture, it absolutely looked like he had just walked in off the street and decided to do it, without any preparation at all. But he really worked hard on it.

That, in the end, became the Feynman Lectures as we know them today. A lot different, because of the full 2-year sequence and lots of other work. But it also caused me to have infinite admiration for Feynman. I got to know him better, and he gave me good advice as well as told me things about physics. He was an amazing person. His life was amazing altogether. He was not uniformly popular even at Caltech, he wasn't very popular at Cornell, just because he did what he decided to do. He didn't take direction from anyone. [laughs] That tended to cause some people to think he was not as well-behaved as he should be. But his mathematical and physics skills were just amazing. So, we learned something from him, what he really knew. But also—you may know that when he was awarded the Nobel Prize, there's controversy about what happened. He did announce that he had been awarded and accepted the Nobel Prize, but what I understand was the reality was that by the time he was informed, the L.A. Times had somehow found out that he had gotten the Prize and published it. He said, "By the time I was told I was to be given the Prize, I would have had to then publicly refuse it. If the L.A. Times had not published the fact that the Prize had been awarded but I was told first, I would have refused the Prize without it becoming known that it had been offered." And I believe that is true. I believe that if he could have avoided the publicity of refusing the Prize, he would have told the Nobel Prize Committee, "Give it to someone else. I do not want it." That's a characteristic of Feynman that is remarkable. And through lots of other episodes of people wanting to award him in some way—really didn't like that.

One indication—before he got the Nobel Prize, local high school science or physics clubs would try to invite him to come give a talk to the club. He would accept if possible. However, after he was given the prize the first rule that he established was that they weren't allowed to tell faculty that he was coming because if they told the science advisor or something, then there would be a committee to meet him, and there would be a lot of jumping around, and what he really wanted to do was just talk to the students and tell them some fun stuff about physics. So he decided on a plan. He would tell the local science club at the local school that they could invite him but use a pseudonym. So he'd go along to the school, but he was "Joe Smith," [laughs] basically. He'd talk to the students and really have a good time and disappear. The administration and faculty, either never found out, or it was too late when they found out. So, that was Feynman's behavior. He was amazing.

ZIERLER: How much interface did you have with electrical engineering as an undergraduate?

GODFREY: Well, I was in Electrical Engineering; I wasn't in Physics. I don't quite—well, I sort of know why I chose that—because I got involved in a lot of engineering work. My first job at Caltech was measuring smog, which isn't exactly like high-tech —but it's experimental work. So—

ZIERLER: Was this with Arie Haagen-Smit?

GODFREY: Right, yeah. Of course my job as a lowly freshman, this equipment would only run for I think it was four hours before it had to be refreshed, and things done, and measurements taken. Of course I got the night shift [laughs], so I got to trudge over to the Chemistry Building and write down the right numbers, and write things, and reset the equipment. That went on for quite a while. Then, for some reason, I met Paul Duwez who was a metallurgist, and he was developing high-performance metals for things like aircraft and demanding things. His experiments required a high vacuum, because the materials would oxidize in air. Measurements of their molecular structure and bonding strengths and things had to be done in a high vacuum and during temperature transients. That was a bit of a challenge because you had to connect all the wires and stuff, and then really pull the vacuum down and cause a temperature transition. So you had to figure out how to prevent leaks in the vacuum chamber during all this. Luckily enough, I got to be pretty good at running the helium leak detector. Helium is the right material to use to find a leak, because it goes through the smallest holes and you can sense it with a helium detector. Because of this work measuring and resolving leaks in Paul Duwez's vacuum system and measuring behavior of metal alloys it became known that I knew how to work a leak detector, and therefore I became fairly popular [laughs]. That came in handy, because—I just liked to do the work anyhow.

But another activity that became related was that in Ricketts House, a standard practice was to build a big monster slingshot that was mounted in the oval in the courtyard, which could launch a water balloon that would land on California Street. With suitable people watching the traffic and signaling, it was possible to land water balloons on things going by. Unfortunately, there was a misfire, and one of the balloons went to the top window of the synchrotron. I volunteered to spend time finding the leaks in the synchrotron and to apologize for having broken their windows. That was fun, and it pacified them a bit—they were pretty mad that broken glass fell into the synchrotron, I can tell you, so I said I'd really work hard. "Not only will we clean up glass, but I'll find leaks in your synchrotron." So, we got over that problem. But I somehow decided, and Feynman substantially convinced me, that it was not a good idea to go into physics, particularly experimental physics, because it takes a very long time to finally complete a PhD. Getting a PhD in experimental physics is typically five, six years of slaving away. A lot of physicists are pretty aggressive folks. They don't think being friendly to poor, struggling graduate students is high on their priorities. So, I decided that was not a good choice.

Connections with Carver Mead

ZIERLER: Did you have any overlap with Carver Mead in Electrical Engineering?

GODFREY: Absolutely. In fact, he's a lifelong friend. In particular, I took his first course. He was only two years ahead of me. Two or three. He had just gotten his PhD when I had been there for a short time. He taught his first course. I don't quite remember why I decided that I'd take the course, but it sounded different, so I went along and took the course. We disagreed about something, I remember. Also I didn't come to lectures. That upset him. I never went to lectures at Caltech. I went to Pauling's lectures, I went to Feynman's lectures, and that was it. I didn't go to any other lectures. It's not that I don't like people talking to me; it's just I don't learn that way. I learn by sitting down and opening a book and thinking and solving problems, so lectures just weren't helping me learn. So, I went to Carver's first lecture, I went to pick up the midterm, and he was a little grumpy [laughs] at the fact that I hadn't shown up to listen to his lectures. But he got over it, and I got to know him better, and in the end, not only did I support his work, when quite a few years later, after I got a PhD and went to Princeton, I was Bell Labs and later at Univac—his work was really relevant to work that we were doing, so it made sense for me to go and say, "We'd like to help support your work and send some people to join in, help you out." This was particularly when he was developing the technology that everyone said was never going to work, the CMOS stuff. He wasn't getting the support he needed, so I provided quite a lot of support and spent time there, too, because it was fun to do.

So, I have known Carver well. Of course we both shared a view of Feynman. He worked with Feynman quite a lot. We still correspond. He's older now. We're all older now. I shouldn't go into a lot of detail; he always—it's not quite fair to say he intentionally annoyed people, but he really often upset the conventional wisdom, and he tended to do it in a kind of direct way. There is a lot of this in engineering—and he worked in engineering. If he had been in physics, he would have had less trouble. But for various reasons—in particular he really liked to do neat experiments and discover the uses of CMOS and all that kind of thing. A lot of the engineering community didn't get along with him. It got worse when it consistently happened that he was right [laughs] and they were wrong. [laughs]

ZIERLER: [laughs]

GODFREY: They had to somehow get around that. So, there was a period when, among other things, he had trouble getting funding. He had trouble getting the kind of support that was needed. But there were people who helped. Gordon Moore was one, of course, and there were others. I tried to help as best I could, too. That was good. But Carver—it wasn't exactly that he just liked to annoy people, but he did like to explain to them how they were mistaken [laughs] in kind of simple words. A lot of engineering, even today, does not have as sound a, both theoretical and formal, basis in what is really going on. Engineers tend to be kind of hobbyists, and they make up kind of approximate formulas that work some of the time, and things like that. Carver was particularly good at pointing out how it didn't always work [laughs], and his own work was closer to physics than most electrical engineers would think. That made him not so popular in some circles.

ZIERLER: Did you have a senior thesis or a capstone project at Caltech?

GODFREY: I don't think so. From working for Paul Duwez, some people approached me who had a company called Consolidated Electrodynamics. They built mass spectrometers and other measurement equipment for high-performance systems. They sold transducers and recording equipment to the aircraft industry for flight trials and stuff like that. The main people were former Caltech people, and their place was just down the road. Somehow, they said I could come by and help out, which I did. That led to developing new instruments. They had mass spectrometers that needed solutions to equations to optimize their spectrometer behavior. So I did that work for them, and did other work for them. The fun part of that was that fairly often, when they had some clever new instruments that could do interesting things, the company that bought some said, "Somebody has got to come along and explain to us how it works and fix it if it breaks," and things like that. Particularly, I got to go to Boeing fairly often and help out with their flight test instrumentation systems. That was not only fun but instructive. I learned how these things really worked. They paid me something, too. And, it was fun to do. That was the important point; I wouldn't have done it otherwise.

There was one incident. The first thing that I built was—a problem that all the aircraft people had was that for all their instruments that they could fly on the airplane, you couldn't read what the readings were. It's being recorded on something, and then it had to be processed after the plane landed, so they couldn't make any choices about how to carry out the flight test based on any new information while in flight. So, we built a system that produced on photographic paper—the big roll of paper went in the machine, and it came pouring out the back, with all the signals on it, so they could see what was happening while the plane was flying. If something was interesting, they could change the flight plan to acquire more data so they'd need fewer flights to complete the testing of the aircraft. That was fun, because I was the one who built the instrument that did that. So, I got to go up to Seattle, fairly often. They wouldn't let me fly in the airplane. I tried that, but it didn't work. I got paid, but it was really a learning experience.

The most fun part—I had just forgotten—after I did the signal recording stuff, I was asked to look at their magnetic tape systems, which were kind of the next generation. The recording on big rolls of paper was becoming old fashioned and couldn't record really high-frequency signals and so on. I went over to the group that was making the magnetic tape recorders. The big problem for them was maintaining constant velocity so that what was recorded was actually passing over the recording head at the right speed. With various wheels, cogs, and things, there was a tendency for the tape to flutter quite a lot, because of vibration and friction and stuff. So, a really fun experience was, I noticed that the magnetic tape was elastic, so you could make the capstan that was winding it up turn—a little bigger diameter , it would move the tape a little faster, so it would stretch the tape by a fixed amount that we could easily calculate. But, the engineers just always made the two the same. Because they really thought that if one of them had a bigger diameter and it was going the same velocity, eventually the tape would pull apart. [laughs] A few lines of math showed that wasn't true. But they still kind of said I'm a nutcase and this will never work. But we built it the way I said, and it did work. That was the dual capstan control that went into basically all the tape recorders for computers and other applications. So, that was fun.

ZIERLER: When it was time to graduate, what were your prospects? What did you want to do next?

GODFREY: First of all, do you know who Dean Strong was?

ZIERLER: Yes!

GODFREY: I have to say that he and I didn't always get along terribly well, and he was dean of students, so when there was some serious crisis with the students, and there were one or two while I was there—you may know about the FBI episode—and a few others—I was the one who had to tell Dean Strong it would all be all right and he should just go away and forget it. [laughs] He didn't always appreciate that. But, I discovered at the end of my third year I had taken all the required courses, so I went to him and said, "I'd rather graduate this year rather than next year." I didn't really mean it; I just wanted to pull his tail. But he got really upset. That's how I knew Dean Strong. He could have been a more understanding and tolerant dean. His physics problems work—he did the physics problems for the standard physics course, which was lucky, because the problems were almost all the same. You worked them all the same way for all the physics exams. I mean, I took Feynman's lectures. I didn't go to the physics lectures, but it was easy to pass the physics exams for freshman and sophomore physics, because they were all strong problems. You knew exactly what he had in mind, therefore you knew exactly what the answer was. [laughs] But he and I didn't get along very well. There was more than one suicide while I was there, and I had tried to tell him, "You've got to deal with this in a more humane way." The next day, another student committed suicide. I went to DuBridge and said, "You have got to step in. This guy is just not dealing with it." So, me and Dean Strong kind of parted ways. And, a lot was done to provide more support—you can't say that you could somehow magically reduce the stresses and tensions of being an undergraduate at Caltech. It's not just spending your time lying on the beach. There's a lot to do. But, it could be handled in a much more humane way than just yelling at the person. Strong was not good at addressing this issue.

ZIERLER: Did this affect your ambitions, what you wanted to do next?

GODFREY: No, I never felt under any pressure at Caltech. I shouldn't exactly say that, because I wasn't the top of the class, but I did pretty well. I did it with taking a lot of extra courses and doing the work in the labs and stuff. So I didn't feel the pressures that a lot of the students did. I really felt confident at the end of the quarter that I'd get good enough grades on the exams, and this proved to be. It is also true that I had completed all the required courses by the end of the third year, so I went to Dean Strong and said, "Just put me down for graduation this year. I've done all the courses for graduation." He became very hysterical. I just did it to pull his tail, and stayed for the four years. I got to know DuBridge really well, too, which was a real honor.

ZIERLER: What was DuBridge like as a person? Obviously you interfaced as an undergraduate to the president, but you said you got to know him. What was he like?

GODFREY: He was just an amazing person. The academics tend not to be easy to get along with, but DuBridge could handle any of the screwball physicists, any of the people, whatever it was, just amazingly well. He and one other (I am pretty sure it was Lauritsen) were the complete administration! He had connections. Caltech—this was the end of the big McCarthy communism scare, so there were a number of things that were remnants of that. One was that to work in any of the defense industries in the area, you had to get a security clearance. This was really harmful to the students, because it took months to get the clearance through, so if you wanted and said, "I'd like a summer job," [laughs] by the time the clearance came through, the summer was over. That kind of problem, DuBridge tried to help with. He tried to help with everything. I'm a little hesitant about this because people were kind of upset, but there was some trouble with an FBI agent, and the FBI got really mad. As soon as it happened, I went to DuBridge and said, "Here's what happened. I just think you may need to know." He just smiled and said, "I'll take care of it." In the end, he had to speak to Eisenhower, and have Eisenhower tell J. Edgar Hoover to just back down and go away.

That was DuBridge. He was just miraculous. There were only really two administrators. There was—what's his name? I'm trying to think of it. He and one other person worked with him. They basically ran the whole school, with just two people and a secretary. Every time that I had any interaction with DuBridge—I was the person that had to tell him things he really probably didn't really want [laughs] to hear, what students were up to and things. But he never said, "Oh, that's bad. We have to do something ." He said, "I'll take care of it," and he did it. That was miraculous. I've never—well, I've met another administrator who was on that scale, but he was by far the best. I kept in touch with him. I went off to London for my graduate work, but as soon as I came back and had a chance to go out to the West Coast, I went to see DuBridge and chat with him.

New Directions in London

ZIERLER: What program brought you to London? Where did you do your graduate work?

GODFREY: The London School of Economics. That was pretty simple. I talked with Feynman. Feynman always gave really good advice. That isn't sort of the natural part of his being one of the smartest people around and clearly good at complicated math and so on. But he agreed right away that I should not get involved in experimental physics for a PhD. I said, "Well, there's people in London that do economics and statistics and stuff, and that looks interesting." He said, "Yeah, great. Go for that." I'm reluctant even now to mention it, but DuBridge rang up and said, "Would you like a full fellowship for your graduate work?" I said, "Would I like it?" He said, "Fine, you've got it." The Edward John Noble Foundation, which still sort of exists, had decided to give fellowships to some number of people, so they called up DuBridge and said, "Do you have a candidate for a PhD that needs financial support?" He said, "Yes, I do. Here's the name." It's a small New York foundation. They still exist, and they still support various causes, but they somehow decided that the right way to provide graduate fellowships was to pay the full cost of a typical U.S. university for the three or more years of PhD study. They contacted me and said, "Would you like this fellowship?" I said, "Yeah, sure, terrific. Except I plan to go to the London School of Economics, which is on a different basis. Their tuition is really low. The costs in London are low." They said, "We don't want to know about that. This is the money that would support you handsomely at Harvard." So, they did that.

That was quite the experience. It turned out the next best person to DuBridge was the then director of the London School of Economics. Lionel Robbins was his name. He was a quite

well-known economist but an amazing administrator. He was almost on the DuBridge scale. That was a happy experience for me, because on the few occasions I needed to explain to him what had actually happened and why I shouldn't be punished for it [laughs], he was really skilled and tolerant. The other thing that had extremely low probability was one of the junior staff had been at Princeton, working with Morgenstern for—he was there for at least a year. I think he just went—he didn't go—planning to get a PhD at Princeton. I think he just planned to take a year off and get his PhD later. He worked with Morgenstern for a year, so he knew about game theory. He gave me good advice about the aspects of economics and statistics that would be things that I should feel were suitable. The LSE experience was really good, and I really learned things. That meant that when I completed my thesis, I sent a copy to Morgenstern and to John Tukey, because I used quite a lot of the techniques that they had developed. They both responded with encouraging remarks. Tukey wrote something on nearly every page of the thesis. I was just stunned. But, I learned a lot, and the end result of that was I was invited to Princeton. So, that was simple.

ZIERLER: Princeton was a postdoc?

GODFREY: No, after a short period as a research assistant I was made assistant professor.

ZIERLER: In what department?

GODFREY: Originally in Economics, and then in Statistics, when—John Tukey caused them to agree to have a Statistics Department. In any case I worked in Oskar Morgenstern's Econometric Research Program and participated there the whole time I was at Princeton, including when I transferred from being in Economics to being in Statistics. I was doing statistics, in any case, because I worked with John Tukey. But I learned game theory. I learned game theory because of Maurice Peston who helped me at LSE, then Morgenstern, and learned more game theory at Princeton. In any case, game theory is, I think, a really important subject. In fact, even today, economists kind of poo-poo game theory and say unfriendly things about it. I think that's really misguided, because it is an important part of scientific knowledge.

While at Princeton, by some event that was really unlikely, I ended up as a juror in a court case. What I fully expected was they just pick people out of the electoral roll box and say, "You're a candidate for sitting on a jury." I went to the court and expected them to say, "We don't want a Princeton professor on the jury." Any reasonable lawyer, in such cases, really doesn't want anybody that knows anything about anything to be on the jury, because they have a better chance of tricking them into whatever version of the story they want. So the lawyers involved weren't too smart because I ended up on the jury, which was initially—I mean, I thought I'd go for half an hour and get out of that, and be back at work. Instead, I had to come sit on the jury. Luckily I had learned game theory, so I decided, "I'm going to convince this jury of the right answer, and I'm going to get it done in under two hours." It was pretty easy. As I kind of thought about it more, there were two main coalitions. The case was really trivial. A child had run down a hall in school and fallen down and hurt himself, and the parents were suing the school for neglect.

Some of the jurors said, "Well, the child was really hurt and unhappy, and the parents were unhappy." But it seemed to be obvious that there was no fault. I mean, all the other children ran down the hall, too. The case had dragged on for a long time. The doctor who treated him because he fell down and broke a tooth couldn't even remember when he did it, or if it was really the right person. [laughs] In any case, there was no rational case to demand the school pay compensation for the poor child falling down in the hall. I looked at the other 11 jurors, and there were two main coalitions. There was the group that said, "The school has money; they ought to just give it to the parents and forget it." There was another one that had a different—I don't remember exactly what their view was, but it was obvious that if I could break those coalitions, that I'd be on my way. I talked with them and identified the weaker coalition. There was one member of the coalition that couldn't quite remember whether he was in the coalition or not. I broke that coalition in like less than an hour. Then, I can't remember exactly how I leaned on the other members, but we were out of there with a determination of not guilty in less than two hours, and I was on my way. So, game theory works! And, it has worked in other contexts with a lot more at stake! Game theory is handy.

The economists still don't basically believe in game theory. What they teach is two-person, zero-sum games, which are really trivial and don't occur in nature and don't solve real problems. But the n-person coalition games, which are structurally quite a lot more difficult—I don't think most economists knew how to read that chapter in von Neumann and Morgenstern—but that's what really works. It does really work. I've used it for lots of things. So that was a useful part of being in Princeton. The thing I intensely regret is von Neumann had already died when I arrived in Princeton. I would have added ten years to my life or done whatever it took to have the chance to at least have a conversation with von Neumann. He was spectacular. I learned a lot about him from Morgenstern, who was really his closest friend and collaborator.

Early Interface with Digital Computers

ZIERLER: Going to LSE, the Department of Economics at Princeton, did you feel like you were leaving science and engineering, or were you approaching it from a different perspective?

GODFREY: No, I think I was really directly applying what I learned at Caltech to other problems. Part of that was that computers were just getting going when I was at Caltech. They were building an earliest prototype, and I helped out a little bit with that, and came to understand what in principle digital computers were. There weren't any that were sufficiently developed and organized to solve a lot of problems, but it was clear that if all the issues of how to make it work right in some sense could be worked through, that it would really be a major event. Luckily, in those days, the UK was far ahead of the U.S. in computer development. I didn't realize that until I got there, but I found out right away, because the University had a Ferranti Mercury computer that people like me could use, particularly in the middle of the night. So, I learned about digital computing from the people who designed the first computers. The Manchester folks did all that work, and Alan Turing went to Manchester after Bletchley Park and a short stay at the National Physical laboratory (NPL). For some time I could not understand why NPL refused to let Turing build his computer there. But the answer was simple: Turing could not tell them that he had already built a computer at Bletchley Park!

So, I got to learn about digital computers at an early stage, and that was terrifically beneficial, because I needed computer power to do the numerical work that went into my thesis.

It was also true, of course, that since all of that was new, it made it easier to portray my thesis as new research, because it was new research. [laughs] But it was easy to do, because I knew how the computers worked. They weren't all that fast, so I consistently had to come in at like 11:00 or 12:00 at night and work until 4:00 or 5:00 in the morning and get the computing done. They weren't the world's most reliable computers, but they really worked. The English were much more advanced in computers and in computer programming languages than was true in the United States as well. The language that the computer that I used was relatively advanced—when I went to Princeton, I found it harder to get the work done there than at the University of London computer, because the language on the IBM machine at Princeton was inferior, but usable. So that was good fortune. It was a lot easier than it might have been.

That caused me to think I knew something about computers, from the standpoint of having some engineering background but also actually understanding language issues and organizational issues and all that stuff which was needed to get the work done. That caused me to end up being involved in computing quite a lot. Really quite a lot. That used up a significant part of my time. I ended up being head of Research at Sperry UNIVAC and developed their new computers, which despite the fact that they did the merger with Burroughs, which kind of ruined both Burroughs and Sperry UNIVAC, the resulting company still builds computers for relatively demanding applications. They're not PCs or anything like that, but serious machines that are needed for a lot more high-performance computing, a lot more exactly correct and reliable. The high point, if I can be yet again be immodest, I worked out what it takes for a computer to be provably correct in its operation. This became the defining document at UNIVAC. And, it was taken up by IBM.

ZIERLER: Which means what?

GODFREY: Which means that if anything goes wrong, it fails in a well-defined way so that recovery can be completed without any residual error. There's no condition that can happen that the computer—the standard computers, and particularly the IBM machines, fail in an unrecoverable or incorrect manner. IBM was reluctant to even put parity in as an error check.

The early IBM machines consistently failed for mysterious reasons. I finally got kind of fed up. People were saying, "Oh, you're just…—they don't really make errors. You're just making a big fuss about the fact that there was an error last month," or something. So I put together a series of tests to run on the IBM machines within the Bell system, and the mean time to undetected failure was like three or four hours. AT&T said, "This is just unacceptable." Getting the right answer is actually important. And, the errors, they weren't just the last bit or something; the errors could be really significant and cause significantly erroneous results. Sometimes the computers just crashed. They'd excuse that as just bad luck. So, we built provably correct computers. IBM, of course, collected all the technical data from all their competitors, including the work that we did, so IBM builds provably correct computers now too, because they took our design documents. I gave a talk at Yorktown Heights about that. It was really fun, because a few of the guys who actually knew what happened were there. That, I think, was helpful, because there are—the PC level of computers, and the laptops and stuff, don't use any of that technology, so mistakes happen. But computers in really demanding environments, both from IBM and other sources, use seriously correct computers and they get the right answers consistently.

Princeton and Bell Labs

ZIERLER: You mentioned that you started in the Economics Department. Where did you transfer and how long did that take?

GODFREY: It was about two years. John Tukey, who—you must know of Tukey?

ZIERLER: Of course.

GODFREY: Tukey was one of the world's amazing people. He was self-taught at home. He went to Brown University in chemistry and then came to Princeton for his PhD in mathematics. That all by itself is a singular event that had almost zero probability. Because, I mean, being a Chemistry undergraduate at Brown, you could get by with barely knowing how to add two and two. It didn't require anything like the mathematical skills that are expected at Princeton—the Princeton Math Department is the world's most esoteric and leading math department around.

So, he wrote a thesis on topology providing important new results, and was therefore promptly made a full professor in the Math Department. But he really was a statistician. I worked with him and Oskar, so I wasn't in any sense in the Math Department, but I was doing statistics with John, and Oskar's work and stuff. That was my Princeton experience.

Oskar was amazing as well. One reason I think he kind of liked me—I just had some common sense at least—but because I had lived in Munich and was a skier, he thought that I should do helpful things. He was one of the people who noticed something that should have been more noticed, which was in Germany during the Nazi period, all the universities were repopulated with Nazi officials. All of them. Real professors were sent away or got away before something bad happened to them. So, the universities in Germany and Austria really collapsed. The so-called professors were just Nazi flunkies. In Germany, they got promptly replaced. In Austria, that didn't happen. Oskar decided that he should try to help straighten that out, and he decided that I should help too. [laughs] So, I spent extended periods at the—he got the Ford Foundation to put up the money for a facility called the Institute for Advanced Studies in the Social Sciences, in Vienna, which was populated by people who were not former Nazis, and knew something, at least. I knew about computing. IBM provided a computer. The University refused a computer also offered to them by IBM. They said, "We'd have to hire a technician to work it, and it would be too much trouble, so we don't want it." So, the Institute had the only available computer, and attracted more students, and that in the end helped substantially to reestablish the Austrian universities on a more rational basis. Eventually the Nazis got old and retired anyway. The universities now are quite good, and we helped significantly to make that happen. That meant that I also spent time in Austria, which was handy for the skiing, so that was good!

ZIERLER: Being at Princeton, the proximity to Bell Labs, did that influence your interface there? Did that open up opportunities that might not have existed?

GODFREY: Definitely. John Tukey, ostensibly being in charge of me, he had in any case an arrangement—there was a standing arrangement that Princeton faculty could spend on the order of a day a week at Bell Labs, as a part of the University collaboration with the Labs. John had been doing that for ever since anyone could remember. Which was also amazing because he was in the Math Department. Nobody else in the Math Department could spell Bell Labs [laughs].

They didn't know anything about it. But John, the agreement was you could spend—John had a magic formula. He always had magic formulas. Princeton's week was actually five and a half days because they insisted on having classes on Saturday morning—so the male students wouldn't run off to New York and get into trouble, was the truth of the matter!—so they didn't have a five-day week. They had a longer week. John's magic formula produced the right number of hours you could spend at Bell Labs without infringing on your responsibilities at the university. I started doing that pretty much right away when I got to Princeton. And, because where John and I lived—we were on the same route to drive up to Bell Labs, so we would drive up to the labs together and drive back together. That had its strenuous features, because John just thought and worked continuously, and driving up with him, basically trying to solve really complicated problems while driving, was challenging, but fun, too. So, I did a couple days a week at the Labs starting fairly quickly after I got to Princeton. Then, the point came where people convinced me that I should move to the Labs altogether. That proved to be a good choice, partly also because in those days, there was an enforced retirement age, so Oskar was forced to retire. That was a step back from the standpoint of being involved in Princeton. Then I had that relationship with the Labs, and then I became a full-time member of the technical staff, and then went on to do other things at AT&T.

ZIERLER: What was the year you joined the Labs?

GODFREY: Hmm! Let me think about that. I started going probably within like maybe six months of coming to Princeton, so it must have been around 1967.

ZIERLER: It was only six months since you got to Princeton that you left?

GODFREY: No, no. No, I didn't leave altogether. The six months—I wasn't going to the labs a day a two or week in the first six months. There were gaps, too, because it was the second or third year that Oskar got the money for the institute in Vienna, and I started spending time there. But academics wander around, right? I think it's somewhat less true now than it was, but certainly in that period, before various other things went on, there was quite a lot more freedom for academics to just decide to take a month or two off and go to some other university, do something else. There was a lot more flexibility than I think there is now.

ZIERLER: When did you consider initially joining the Labs on a full-time basis?

GODFREY: The reason that came about—there was a statistician, Martin Wilk, who had been at Princeton and had gone to the Labs, and he wanted to have some really challenging work done. He somehow got into his head that I could help make that work achieved, so he convinced me to go to the Labs full time. I had been at Princeton at least three, maybe four years at that point. I'd have to look that up. Being a statistician, I'm really poor at dates! [laughs] If you really want the dates….

ZIERLER: No, that's okay. The more substantive question is, did your research at Bell Labs change from when you were there on a part-time basis to when you joined full time?

GODFREY: No, not—first of all, the Labs—nobody was told they ought to do anything. The Labs was a serious research place. Shannon and the guys who invented the transistor, and the guys who did cosmic radiation stuff, they were just—they just worked there. Nobody told them anything about what they should do, or when they should do it, or anything. So the issue of what you're doing didn't come up. You just did what you thought was best and would be most helpful. I was sort of hired because they thought—well, they certainly thought I might do something useful [laughs], but they thought I had some skills that might be helpful, so that's what happened.

ZIERLER: Tell me about some of the big research projects that you got involved with at Bell Labs.

GODFREY: I did other things. I had been doing work on data that had actually been acquired at Princeton on—there was a guy named Pittendrigh, who studied circadian rhythms in mammals, trying to figure out why there's a light/dark cycle, and what drives it and things. He instrumented mice [laughs] and measured their behavior on a 24-hour basis, and he had a lot of data about how mice wake up and run around, and what happens if you force a non-24-hour cycle. If you put them in a light/dark cycle that's 23 and a half hours instead of 24, what happens? Experiments like that, to try to figure out how the circadian clock really works in mammals. He had a lot of data for that. It also turned out that the Labs had better computing equipment for studying that data. So, part of what I did at the Labs initially was compute on that stuff, which ended up in part also being at Princeton. But as I said, the Labs was perfectly happy with that. They said, "We just think you're a nice guy, and you can do what you want." They said that to everyone. I wasn't privileged in any way.

There was also a group there that said, "It would really be helpful if AT&T could understand how its business works." There were people at Holmdel trying to work on that, and there were other people at Murray Hill, and they weren't getting very far. Exactly all the reasons for that is a longer story, but businesses, which are made up of people and rules and complicated things, are not like measuring the temperature in a room or something. It's definitely a more challenging problem to say, "This is how the Bell system works." How all the switching equipment works, how all the people who manage it work, and how the rules are set, and what difference they make. And, of course, how much everything costs which also is a function of time. Because, well, the engineers that work on it, they can only kind of solve one problem at a time. For instance, they had to figure out how to deal with the overloads that occur on holidays, and how to put in things like alternate routing. The switched network for the telephone system has a lot of tricky rules that are intended to achieve a goal of a low value for blocked calls. The goal is—whatever it was—97% or 98% of calls should complete, somehow. They want to do that in a way that actually works and doesn't cost a huge amount in extra equipment. Engineering people thought of a whole lot of tricky things, including alternate routing. If someone picks up the phone in California and dials some number and they dial some number in New York, deciding how to route it through the network is something that takes some thought and some engineering. The objective is to achieve the limit on blocked calls. Occasionally, when you dial—it still happens, occasionally, when you dial, and it says "beep beep beeps" because somewhere in the circuits it was blocked and it couldn't get through. The percent of failed completions is in the rules of how it's all supposed to work.

They somehow got into their heads that I could help figure that out. [laughs] It took some work, because the rules are not only—the rules are changed depending on other events, like holidays, or load, or various factors. Not only you need to really understand the rules and what implications it has for the probability of blockage, but you have to be able to figure out what will cause the rules to change, [laughs] and therefore—anyhow, that's the aspect of that problem, so that's what led to my being involved in not only the modeling of the switching behavior but the idea that it could be shown to be consistent with the FCC's regulatory process. Because, as I've said, consistently the FCC thought that the lawyers that AT&T sent to Washington were telling them a story that they weren't sure was the whole story. So, there was definitely conflict. The FCC was not confident that AT&T was willingly and correctly carrying out the rules about how they should not behave in an anti competitive way.

Divestiture and Move to Stanford

ZIERLER: When did you start to see the writing on the wall in terms of what IBM was trying to do to AT&T?

GODFREY: That was still at Princeton. I had not been comfortable about IBM for a long time. I know their history. I know Watson Sr. was building fake cash register machines with a competitor's (NCR) label on them that were unreliable and broke down and therefore IBM could convince the customer to buy an IBM one. So they had done basically illegal but also bad behavior. IBM has been like that their whole career. They did that in the computer business.

They had all of the competitors' records. They knew all of our design information at Sperry UNIVAC. They knew all the Burroughs information. They just had spies that came in and Xeroxed all this stuff. And the financial records. So I was never entirely friendly with IBM for those reasons. They really have never been a well-behaved company. And they technically were not very good. Their early computers were really pretty junky. That meant me and IBM didn't get along very well. I was surprised that not everyone had all the relevant facts. There were people who thought IBM was better behaved. I did go to Bob Santos, the AT&T controller, and point out that the AT&T data centers were almost entirely IBM equipment. All on rental. IBM never sold a computer. They were all on rental, which meant they could take them back anytime they wanted. So, it was true that if IBM got mad at AT&T, they could pull all the computers out of all the data centers, and AT&T would have serious trouble getting the work done. The AT&T controller noticed that—I was a little surprised he hadn't thought of that. [laughs] Therefore, he ordered all the data centers have at least one other vendor on the floor in all the data centers, so that if IBM got mad, they'd still have equipment to continue the business. That meant IBM didn't like me, because I lost them a huge amount of business and control.

ZIERLER: [laughs]

GODFREY: And they didn't have the power they thought they had. So, that was a part of the developments that led to all the stuff you know.

ZIERLER: When this started to happen, when did you begin to think about your next move?

GODFREY: Well, when the divestiture was forced, then I didn't have to think about a move; [laughs] there wasn't any choice.

ZIERLER: You were not proactive; you left when you had to?

GODFREY: Well, that's not quite true. The divestiture was, to my mind, a disaster. I don't know how they could have done it better, because they had to split up the company into regional operating companies. Therefore, there was no possibility they could convince the regional operating companies to pay the bill for Bell Labs at Murray Hill. That was a huge operation, hugely talented people, doing whatever they felt like. The operating companies that just operated in Alabama and a few places wouldn't in a million years put up the money to support Bell Labs. So, that was the end of the Labs, so I didn't have any choice about [laughs] being involved there, because it disappeared. That's when I went back to Imperial College, in the Math Department.

ZIERLER: What aspects of your research could you take with you, and what was so unique to Bell Labs that you had to start fresh?

GODFREY: Oh, I've never done anything that I couldn't take with me. [laughs] Scientific work is scientific work. If someone told me I could work someplace but it all would have to be secret, I'd say, "I'll do something else."

ZIERLER: I wasn't thinking so much about secrecy; just in terms of the unique research culture and all of the resources at Bell Labs.

GODFREY: Well, it disappeared anyhow, so there wasn't any question of trying to stay around. Murray Hill was sold to some company that didn't have a clue what to do with it. Almost all the people left. It is gone now.

ZIERLER: What did you take up or what did you continue when you got to Imperial?

GODFREY: Well, same old stuff, doing statistics. I had worked with a person who waa a serious authority on the behavior of the heart and heart attacks. He was in electrical engineering but he had a medical background as well, so he knew a lot of medicine. His objective was to try to figure out how to reduce both fatalities and heart attacks. That's the kind of data that I knew about anyhow, so I worked with him for a long time. He had a lot of data from various sources. He improved the defibrillators. He was one of the inventors of the defibrillators and improved them. So, if nothing else, he saved thousands of lives, because he helped develop the defibrillator and restart people's hearts. I think that was fairly useful work, and what I know about statistics and stuff helped to understand both the failure mechanism and how to get it to restart. So, I just walk along and do these things.

ZIERLER: When did your affiliation start with Stanford?

GODFREY: That was 1990. That came about because that was the Thatcher days in England, and I had been given an endowed professorship at Imperial by Schlumberger, because they knew my work in oil exploration and that kind of thing. But the climate under Thatcher was just unbelievably awful. I try not to say bad things about bad people, but she was just mean. One trivial example—the universities—there's a value added tax on all of buildings and maintenance costs and things, that the universities were exempted from, because they were not profit making. That sort of made sense; why should they pay extra taxes? As soon as she got the power, she cut that out, and said, "You've got to pay the tax." The naïve people tried to go to her and say, "But where do we get the money?" She just said, "That's your problem." So, she was just mean. She cut off all the research funds. She really hated the universities. That was partly because Oxford always gave the prime minister an honorary degree at Oxford, and Oxford turned her down. That helped her to be really mad [laughs] about the universities. So she just cut their funds every way she could. Even if people had tried hard to be tolerant, just being beaten over the head, the rug pulled out from under them day by day made it just—so, I decided, "Off to Stanford," which was a good choice. I had a good time at Stanford.

ZIERLER: What program did you join at Stanford?

GODFREY: There was a group called the Information Systems Lab, which contained some of what I thought of as the best engineering people there. One of them was one of the key people—he had been at RCA and helped develop television systems. He developed the imaging systems for medical diagnosis. He did a whole lot of neat work. So, really fun people doing the kind of work that I enjoy doing, too. That worked out great. That went on for quite a long time. The end of that story is sad. The lab has changed a lot by now.

ZIERLER: Just being there in terms of all the excitement in Silicon Valley, the dotcom boom and bust, what was that like for you?

GODFREY: Well, I wasn't involved in the boom and bust much. We did real useful technical work, including—the part that I worked most on was using semiconductor materials as imaging devices. All the imaging things in the phones and stuff like that in part derives to some extent from work we did to make silicon circuits that sense light in a low-noise, high-performance way and can be used for all kinds of image sensing. One fun application was the pick-and-place machines that assemble stuff. They're all in China now. They have machines that grab a part, stick it in the right place, grab another part, and stick it in the right place. They have to have an imaging device to figure out where to find the part and where to stick it in, and they have to be pretty precise, because they're tiny little parts, of little things. And they have to go really fast or they don't make a lot of money. They really liked our parts because they're very low noise and low power. So I did that kind of thing, which is fun. And I helped around the labs, because the statistical techniques that I learned from John Tukey and know something about are pretty widely applicable to a lot of engineering stuff, so I could help out.

ZIERLER: Were you witness to, or did you take part in, the way computation became relevant to so many disciplines around these times?

GODFREY: Yeah, I think so. Certainly I tried to do things that made computers easier to understand and use, and more reliable, in particular. In fact, as I will claim, that we (this was back in Univac) built the first fully reliable computer, that if anything went wrong, it would report, "Something has gone wrong" and stop using the failed part. No errors would propagate into the system. IBM learned about that and adopted that technology as well, so that's a hint that it was actually effective. It has kind of gotten lost now, due to the PC episode and due to you-know-who, Bill Gates, who did exactly the reverse. You may know he went to Intel, who made the memory chips for the PCs, and he noticed that they had a parity bit in the memory, to detect errors. Gates said, "That means it costs more and therefore I can sell fewer licenses and therefore it's a bad thing, and you have to stop doing it. Use cheaper memory that has no parity check. If you refuse to do that, I'll cancel all your DOS/Windows licenses, and you'll go out of business." Intel did what they were told. That means that PCs that everybody uses today have undetected errors, day in and day out. That's the way it is. Some people's bank accounts don't add up right. Worse, people die in hospitals since the monitoring PC goes wrong. I find that not a happy arrangement, and I know who to blame.

The Relevance of Computer History

ZIERLER: When did you start to get interested in computer history? Was that something always close to you, or is that more of a recent development?

GODFREY: I'm tempted to say I think I was part of it! [laughs]

ZIERLER: Of course. But I mean talking about Babbage, and von Neumann, really people who preceded you, the early computer history.

GODFREY: Well, yes. I was interested because von Neumann made major contributions, but Alan Turing and the people at Manchester were still alive, and also made major contributions, so I could actually talk with some of the people who made the really important contributions. I wrote some stuff about that. The history isn't very well understood. There aren't a lot of people who have some suitable skills and serious understanding of how computers really work. It's also true that a lot of the important work was done at Manchester. Incredibly enough, when we came back to England, I had worked with the Manchester people. I was actually on their faculty for a while and helped with their computer development. It didn't occur to me that would all just vanish for some reason. I knew a number of the people—Kilburn and Derrick Morris and other people—had retired, and some of them had died. It would have been natural for people who worked with them to take over. But when we got back, I tried to ring up Manchester. First of all, I looked in the University records: there was nothing that said Computing Department. I found someone who seemed to know something about computers, and I said, "What happened to people like Derrick Morris and the people who invented computers in the first place?" "Who are they?" They didn't even know their names! Knowledge of computing at the University of Manchester has just disappeared! Alan Turing was there.!

ZIERLER: To clarify, was Manchester a temporary appointment, or you left Stanford for Manchester?

GODFREY: No, that's the other way around. When I was at Imperial, there's a standard thing of university people being on various forms of membership at other universities.

ZIERLER: I see.

GODFREY: I was a consulting or something professor at Manchester while I was at Imperial. That's a standard thing. It's less used now than it was, but it was a good thing in the sense of having an outside person involved in the examination process and things like that, and reviews. That was just standard practice that pretty much everyone had some status at one or more other universities. The primary university was wherever it was. I forget my title—something consulting professor, janitor [?] or something, [laughs] at Manchester for quite a long time, and, I worked closely with the people who did all the work. They did spectacular work. There's a really sad story. The last computer that went into practice much was a thing called the Atlas, which was the first really high-performance paged machine, all kinds of good things. Ferranti built the computers. They were designed and the prototype was built at the university, but Ferranti was intended to be the commercial company to take it over. This was when computing in the United States was nowhere near this advanced. IBM machines were just dumb.

The Ferranti Atlas was a spectacular machine. A friend of mine, who was responsible for the engineering computation at Boeing, got friends together who said they would buy a total of, I think it was, 20 machines, to be delivered in the United States. Twenty Atlas machines. He went off to Manchester, and he went to the great Basil de Ferranti, and he said, "I'd like to buy 20 of your machines at list price, but they need to be delivered in the United States, and we need to have support there, because they may break down and need repair, and so on." Basil de Ferranti said, "You can always call us in Manchester," and he dropped the deal. These things happen.

That machine had all of the architectural structure and design structure of the machines that are being built today. It was really advanced. It had paging. It had well-designed buffer memory, structural organization ... If they were building them today, they'd be right up to date. [laughs]

ZIERLER: The DOE's embrace of supercomputing in the 1990s, were you involved in that at all?

GODFREY: Well, I was. I looked at it. It wasn't that interesting.

ZIERLER: Not interesting because it was not impressive to you?

GODFREY: I take a sort of human ground-up approach to useful things. Building a huge complicated expensive unreliable thing and claiming it will help pursue research is just not a part of my experience. The whole "build one monster supercomputer and it will satisfy the world" is just not consistent with what interests me.

ZIERLER: When did you start to think about leaving Stanford?

GODFREY: I gave a talk there, which was titled, "Has Anyone Heard of Manchester?"

ZIERLER: [laughs]

GODFREY: I described the founding of the Industrial Revolution and how it came to an end. I said, "It's happening here. And we're leaving." So, we left in—when was it? 2012, we left Stanford. Quite a few of the people I worked with were retiring anyhow. We all get older.

New York and Returning to London

ZIERLER: You saw this as a retirement, or was there a new opportunity for you to pursue?

GODFREY: I was at the point where it didn't matter a lot where I was; I could pursue the kind of work I liked to do. It was clear to us that we would be happier doing it in London.

In Silicon Valley—this doesn't seem quite rational, but—my wife just pointed out that sitting in California and finding a nice place to live in London is made difficult due to the distance and time difference.. We decided to move to the East Coast, which we did. We lived in the Hudson River Valley in a nice place. Good skiing, all that, and much shorter trip to London to look for a place there. That worked! How many years? Four years in Hudson River Valley, which was nice, and more good skiing and all that. Then we found a very nice place in London. The plan sounded a bit irrational, but it worked.

ZIERLER: During your time in New York, you were an independent scholar? Did you have any affiliations?

GODFREY: No, I just went on doing what I do. I'm not institutionally oriented. I just do what seems useful.

ZIERLER: The move to London, this is a much more recent development. That's what has delayed our talk these past few months.

GODFREY: Right. That's somewhat boring recent history. We found that where we had lived in London for quite a long time was not—it had all kinds of good features but it was not suitable for the longer run. So we decided we—we were on the Kensington Palace Gardens. There were some good features. The plumber who came to make a repair (free of charge, of course) said, "I've got to get off to Buckingham Palace now!" [laughs] But it had other features that made us think we needed to make a change. And, we did think that we should go back to the Bay Area for a period, because we had been there a long time, and we had a lot of friends there. So, we went back for over nine months altogether. But we didn't go thinking there was much probability that we would stay longer than just meet our old friends and enjoy things for a period and then come back here, which is what we've done. It is true, as I mentioned before, I did give the talk at Stanford called "Has Anyone Heard of Manchester?" I pointed out that Manchester was the source of the Industrial Revolution in the same sense that Silicon Valley was the source of the Information Revolution. It came to an end, and Silicon Valley is coming to an end, as we speak. It's not fun to be there anymore.

ZIERLER: How do you mean that it's coming to an end?

GODFREY: Nothing's happening, except people doing imitations or working tricks such as Elizabeth Holmes and Theranos.

ZIERLER: You mean from an innovation perspective?

GODFREY: Yes. I think of real innovation which usually comes about by serious people such as Gordon Moore and Andy Grove, with help from Carver Mead.. The Gordon Moore's are long gone. They are now replaced by people with no technical skill and only interested in doing an IPO and pocketing the money. Sadly, Google is an obvious example.

ZIERLER: Is it happening anywhere?

GODFREY: Yeah. China and India. China and India will likely be the colonial powers of the future. It happens. The English did it. [laughs] Someone's got to do it.

ZIERLER: Just to bring our conversation up to the present, what are you working on currently, or what is more generally interesting to you in the field?

GODFREY: This has been a long conversation [laughs]. But it is true that the ability to correctly transmit information using photons, it is light, is not currently fully exploited by orders of magnitude. There are properties of light that relatively few know and understand and see the prospects. This goes back to Pointing and people a long time ago recognized properties of coherent light that then got forgotten. Dennis Gabor said it's two degrees of freedom per photon. That happens to be physically incorrect. There are an indefinite number of degrees of freedom. There are advanced communication systems that already transmit in this manner, increasing the bandwidth of their transmission by orders of magnitude. That has been fun to explore, and I wrote a few notes about it. People who know about it are hard at work doing it anyhow, so they don't need me [laughs]. Because the math and the physics is fairly straightforward, you just have to do the engineering. Still, fun to think about.

ZIERLER: Now that we've worked up to the present, for the last part of our talk, I'd like to ask some broadly retrospective questions about your work, and then we can end looking to the future. To bring it all back to Caltech, what has stayed with you over the years—what you learned at Caltech—that really informed all the research you've done subsequently?

GODFREY: That's hard to characterize exactly, but let me try to phrase it. Just thinking carefully and making careful observations, and studying exactly how the physical world works, and in particular how light systems behave, is useful and can be made to do a lot of stuff. That's not exactly an answer to your question but it's the best I can do. If I knew more, I'd do it.

ZIERLER: When you look at all the advances in computation, where is your intellectual fingerprint strongest? Where do you see your contributions in what we have today?

GODFREY: I do think that at least some computational systems are both more reliable and more amenable to doing what people want them to do. They can be better understood. They can be made to achieve the objective with less trouble and without error. But that's a hard question. I'm not sure I can really answer it.

ZIERLER: What has been most satisfying to you, either because it was most interesting for you to work on, or because you saw the impact that it was making?

GODFREY: That's also hard. I do think that defining and designing and implementing a provably correct computer system was a really useful step. IBM now uses those results. They didn't ask me; they just took them.

Sadly, the Microsoft PCs don't make that choice because it costs money. As you know, Bill Gates only wants all the money. I think I said this already—he insisted on less reliable computers since they cost less and they'd buy more licenses from him. He insisted on reducing the cost regardless. That was not helpful. But it made a lot of money for him, so— However, this also has resulted in about 900 patients in UK hospitals dying each year due to Gates. Obviously, he could not care less. I also think that my work at AT&T was useful. The FCC liked it and it may have improved the ability to conduct business in a more effective manner.

ZIERLER: Last question, looking to the future. Either from you personally or on what you have done, where do you see things headed, and how might you have contributed to those future developments?

GODFREY: I think a more complete understanding of what amounts to the information content of physical systems is a good thing to spend time on, because it means more can be learned, and therefore maybe some good things will happen. That's the best I can do.

An Addendum of Caltech Observations

If I may, I would like to mention a few items about Caltech in my time.

One clear understanding was that all students were full members of the community and they had the same rights, privileges and responsibilities as all other members. This included access to Institute facilities for study, research, or assistance within and outside Caltech. An example was that a group of students set out for the beach and their car contained some suitable beverages for relaxation. Unfortunately they were stopped by a police officer who demanded to examine the car. Finding substantial quantities of alcoholic beverages in the trunk the students were promptly arrested. Word was gotten back to Caltech and Dubridge alerted the Institute lawyers. They quickly determined that the students came from several different states and as minors they would demand be prosecuted in their home locations. This at best meant that the local police would have to undertake extensive communication, transportation, and additional expense. The lawyers made clear that dropping the charges would be by far the best choice, and that settled the matter.

Another incident came about due to the McCarthy era continuing restrictions and nuisances. In particular there was an FBI agent who spent frequent visits to the campus questioning faculty and students about communist'' leanings or also asking about the loyalty of other members of Caltech. This was an annoyance to everyone. Unluckily the agent entered into the Rickets house lounge and started his questioning. He may not have known that the standard practice in Rickets for someone who caused significant nuisance was to take the offender and throw them into the nearest shower usually with just cold water on. This, therefore, happened to the FBI agent. The FBI was not amused and the matter reached J. Edgar Hoover. This was an event that I thought might test DuBridge's tolerance. But, he said immediately "I'll take care of it."

FBI agents did come to me demanding that I provide the names of the offending students. I said that there was no offense and, of course, there was no possibility of the names being revealed. This caused some fairly wild threats, but the agents got up to leave. I just said goodbye but added that it would be best if they avoided passing near Ricketts House.

I learned later that DuBridge had to speak with Eisenhower about the matter before the FBI backed down.

Another very important study was conducted while I was an undergraduate. A person (I think that his name was Wier) undertook to carry out a study of the performance of my undergraduate class. He started with each student's rank on admission and then computed the ranks at the end of each year. This continued until graduation. The key result was that there was no correlation of the ranks from year to year. This makes clear a fact that is not often taken into account: people develop in very different ways and at different rates. Thus, eliminating a student at an early age is likely harmful to the student and to society. About 15 years after graduation I met a classmate from Caltech. He had completed a PhD in physics at Princeton and appeared to have done very well. However, he came to me looking for a job. I suggested a few interesting opportunities. But quite soon he just said I could no longer do that kind of work.''

The reverse is also true. Some only appear to have high intellectual skills only later in life. There is hope for us all! It is a great loss to many individuals and to society that this evidence is not appropriately recognized.

ZIERLER: On that note, Michael, this has been a fantastic conversation. I want to thank you for spending this time with me.

GODFREY: So, to conclude, when the going gets tough, the tough go skiing! [END]