Oral-History:Clayton Lewis (Nov 2020)

About Clayton Lewis

Clayton Lewis is Coleman-Turner Professor of Computer Science and Fellow of the Institute of Cognitive Science at the University of Colorado Boulder. His research involves technology and disability, human-computer interaction, educational technology, new approaches to programming, and representation and computation. He holds a BA from Princeton, an MS from MIT, and a PhD from Michigan. He worked for IBM prior to joining the University of Colorado. He is a member of the ACM CHI Academy and University of Colorado President’s Teaching Scholar.

In this session, Lewis discusses his family, upbringing, and education at Princeton, MIT, and Michigan. He discusses working for IBM, first in Cambridge, MA and then in Yorktown Heights. He discusses some of the people who were important to his development: Paul Kolers at MIT and Marvin Minsky at MIT; Nat Rochester, Jean Sammet, Dick Goldberg, John Cocke, and Fran Allen at IBM; and John Anderson and Jim Greeno at Michigan. He also discusses his brief time at University of Texas at Austin.

In the second session, recorded in December 2020, Lewis discusses his joining the faculty in the computer science department and the Institute for Cognitive Science at the University of Colorado Boulder. He discusses the interaction of these two campus units, his extensive involvement and passion for teaching, and his various areas of research in human-computer interaction broadly conceived. He also gives brief remarks about other topics related to the University of Colorado Boulder: senior administration, ATLAS Institute, and Department of Information Science.

Copyright Statement

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE Computer Society. No part of the manuscript may be quoted for publication without the written permission of the IEEE Computer Society.

It is recommended that this oral history be cited as follows:

Clayton Lewis, an oral history conducted in 2020 by William Aspray, IEEE Computer Society

Interview

This transcript has been edited by both the interviewer and the interviewee.

Oral history Interview with Clayton Lewis – Part 1 (23 November 2020)

Interviewer: William Aspray

Aspray:

This meeting is being recorded by Zoom, and I also have a handheld recorder on as a backup. So, let's start by my asking you what year you were born and where you were born and what your parents did.

Lewis:

I was born in 1946 in Washington, DC. My parents lived in Alexandria, Virginia, across the Potomac, but I guess it was natural somehow for my mother to go to this hospital in DC to deliver me. My father was a farm boy from Maine, from quite poor circumstances, who earned a PhD in agricultural economics from Cornell and had a career in that field. The part I knew about was largely in international development. So, he was very interested in agriculture around the world, and had a lot of sympathy for farmers and their situation in society. When I was born, I think he might've been working for the fledgling United Nations food and agriculture organization, FAO, which at the time was based in Washington. It subsequently moved to Rome, where it still is, and my father chose not to follow it there. So, he moved into other work but was still involved in agriculture and development. So that's what he did. My mother was born and grew up and lived most of her life until marriage in China. Her father was a YMCA secretary, so a sort of a missionary. The YMCA was at least as much a kind of cultural as a religious organization. She came from that background and said later in life that she always assumed that her life would be in China. She viewed that as home, but with the war and whatnot, it didn't happen that way. My father happened to have his first job out of graduate school working in China, which is where they met and married. My mother never worked outside the home. The rest of her life was as a mother and housewife, with a good deal of volunteering with the League of Women Voters. I'm the fourth child of the family. I had two older sisters, one has since passed away much too young. And then a brother. So, there are three of us remaining.

Aspray:

Did any of the others become academics or do things like academics do?

Lewis:

Well, my brother earned his PhD in electrical engineering and had a career at Bell Labs. So, he was never an academic, but he did go into that field. I've reflected, even though my father and my brother both had PhDs, when I started my first round of doctoral study, right after undergraduate school, I had not clue one about what getting a PhD was about. So, something about the nature of our family, and the communication or lack thereof, meant that there was no transfer whatsoever of useful knowledge. In fact, my first go at doctoral study did not succeed. I ended with a master's degree and went back some years later when I had a little more idea about what was going on.

I would say that, apart from my not learning anything from my dad about being a PhD student, I grew up thinking that getting a PhD would be pretty much impossible. The reason I thought that was that one of the books in our house was my father's dissertation, which was a printed book, several hundred pages in length. Most of it was data tables, “Land Classification in Tompkins County, New York”. My father was one of the inventors of a technique called “land classification”, which looks at land from the point of view of its economic potential in agriculture, and considers all the myriad factors that influence that, so that rational decisions can be made about uses of land. I now know, (and there's some irony here; my dad was not bitterly, but substantially opposed to FDR and what FDR stood for), later in life, I've learned that FDR, when governor of New York, was, I'm sure, pretty directly responsible for supporting my father as a graduate student. Because I can see exactly where his work fit into stuff I found out Roosevelt was doing in the state at that time. But it wasn't just a dissertation. I mean, it was part of this big project. So, they paid to have it sat in type, you know, with all that that must have required in those days. So that's what I thought you had to do to get a PhD. And it was obvious I could never do anything like that. I suppose by the time my brother was in grad school and I was, of course, a lot older, maybe things shifted a little bit; but certainly growing up, I had no thought whatsoever that I could aspire to having a PhD.

Aspray:

As you were growing up, were you a good student?

Lewis:

I was a good student.

Aspray:

What subjects in school interested you most?

Lewis:

I think in fact that all of them did, but under the influence of Martin Gardner, I came to be pretty strongly focused on mathematics. But I also was quite interested, to the degree that one could be in my generation, I was quite interested in computing, and actually there was a little overlap there. Occasionally, Gardner's column in Scientific American, which I devoured religiously, would have some stuff. I remember, for example, there was actually an article-- it might've even been a cover article-- in Scientific American, on making a learning machine out of matchboxes and jelly beans. I did that, of course. And I had a, I don't know if you know about it, a GENIAC computer? You probably know about Edmund C Berkeley.

Aspray:

I certainly do. In fact, I was the person who went to his apartment and appraised his materials.

Lewis:

No kidding. Well, I didn't know that Berkeley was the person, or I didn't pay attention, but I now know that Berkeley was the person behind the GENIAC, and that was hugely influential on me. It was a Masonite logic machine. I could spend more time describing it, but basically, if you assembled the thing you had, I believe it was actually eight rotary switches that could make or break dozens of circuits, with each little, each little turn. So you could make an extremely elaborate logic machine. For example, you could make a tic-tac-toe player out of this thing. You would dial in your move, and a light bulb would light; and you’d dial in the computer’s move and another light bulb would light; and that sort of thing. In the GENIAC box there was an honest to goodness reprint, and the reprint was Claude Shannon's master's thesis on the logic of switching circuits. It included these ideas with just incomparable clarity. Like you have two switches in series, they've both got to be on for the circuit to be closed. So, that's it. Logical! And if you have two switches in parallel, then that's an OR, okay? In hindsight, I know that the inclusion of this reprint was a bridge from this thing sort of for kids into the world of grownups. But because the reprint was clear, it made me think, okay, I could kind of do this kind of thing. Whereas, you know, the Gilbert Chemcraft sets and all that sort of stuff, that was all for kids, and there was nothing to point beyond that. I mean, if you understood everything, you'd know there was a connection between that stuff and real chemistry, but, but nevertheless, there was nothing to say, this is not just for kids, because it was just for kids. So, I owe Berkeley a lot.

So, I was interested in computing. I went to Princeton as an undergrad, attracted by the reputation of its math department, although I took very little advantage of it, as it turned out. But there was no computer science department or computer science. So, math seemed like the obvious thing to study. And, I now know thinking back on what I actually did that, even then, I was interested in a lot of things. I used to spend a lot of time in the library reading about central Asia in particular. I took a wide range of courses, which you kind of had to, but I pretended that I was only interested in math. I know that was never true, but that's what I pretended. One of the decisions I made under the influence of that belief, about only being interested in math, was to choose to do only do three undergraduate years. I had gone to a good high school that had, even in those days, a substantial advanced placement program. So, I graduated with enough credits to start college as a sophomore. And I remember thinking very deliberately, well, you know, what you really want to do is to get on to graduate school. You know, that's what it's all about. Yeah. These were really stupid ideas, but that's what I pretended that I thought. Okay. I remember meeting with my bewildered freshmen advisor, who was trying to get me to be interested in things. He was a geologist and, you know, that's the way things worked at Princeton, your advisor would be a real scholar. For all I know he was a member of the National Academy. I mean, these were real people, but they were there advising freshmen. I could tell that he didn't really know what to make of me, but no one was going to push anybody around in those days. The distribution requirements did get me to take some things, which I remember quite a bit more about than the math courses that I thought I should be doing. But anyway, I took myself along to grad school.

Aspray:

You were also in college during the turbulent years of the student revolts. And my recollection is that the universities were particularly liberal about what they allowed students to do in terms of their courses of study,

Lewis:

As an undergrad, I don't think I was affected much by that. I'm embarrassed to say, when I got to my first round of grad school, I blush literally to think of some of the things I did then. Yes, it was that time. I entered college as someone who was probably sympathetic to the war in Vietnam, to some extent, to the degree that I thought anything about it. My father was really, because of his attachment to the world of farmers and peasants, he was bitterly anticommunist his whole life. We had many arguments about the war. He couldn't see the nationalist dimension or anything. He just felt, these are the communists, and when we see how things played out, it's not like he was wrong. Now, in the longer term it was wrong. I think that things have come out better. But it certainly went through as my dad would have would've predicted. Anyhow, by the time I finished college, the war was a factor actually. I think even I had some sense that, that maybe going right into grad school – what was I, 20 years old? – probably wouldn't have been the best thing, but there wasn't really much of an option then. You'll remember that, you maybe went through this, that, you know, this wasn’t maybe the exact timing, but there were certainly periods in which just by being in school, you could avoid the draft. I told myself that if I were drafted that I would refuse service, but I wasn't a conscientious objector. I wouldn't have done that. I would have gone to jail. That's what I tell myself. And, of course, I have no idea if I really would've done that. But at any rate I was willing to evade the draft by virtue of going to school. So, that sort of shut down various possibilities. I remember, I applied to one institution and for one fellowship, and the idea was that if I didn't get either of them, if I didn't get them, then I'd have to sort of see. But I did get them. So, off I went to MIT.

Aspray:

Okay. So I have several follow-up questions here before you moved to MIT. Many young people were fascinated by Martin Gardner, as I was. Were you also fascinated by E.T. Bell's book, Men of Mathematics? Was that something that you read, as I did?

Lewis:

I did. Yeah. I can't say I remember very much about it, other than it's kind of a reminder to me of the impoverished information environment that one inhabited in those days, even in a wealthy exurban community, which is where I was at this time. When I was 10, my dad got a job with an arm of the Rockefeller money, in New York, and became a classic New Yorker cartoon commuter. We became the classic cartoon commuter family. We were living in a place called Weston [Connecticut], which was hardly a town at all, with very strict residential zoning. The place you went for commerce was Westport. I don't think Weston even had a library, but Westport did. But anyhow, I remember I was very interested in game theory, and there was one book on game theory. Eric Temple Bell’s book was one of a tiny number of books about mathematics. So, what else were you going to read? But it didn't capture me; there wasn't anybody in there [Bell’s book] that I remember thinking of as any sort of aspirational model, but I did read it.

Aspray:

A number of mathematicians and a number of historians of mathematics pointed to that book as somehow seminal. Are there individuals either in your pre-college days or in your college days – both students and faculty members – who you thought of is having a profound influence on you?

Lewis:

One of my math teachers in high school, Walter B. Stevenson, was a World War One veteran, so, pretty old by the time of which we speak. He influenced me because of his emphasis on understanding things. You know, math teachers were, and I'm sure still are, sort of a mixed breed, with ones where it's all about particular stuff, and others that are about understanding what you're doing. I remember this was actually a matter of discussion among us students in the high school. Because there was a notorious teacher, who I never had directly -- and I've forgotten all the details -- but the assertion was that this person never gave partial credit for anything. Because in their view you either had it exactly right, or you didn't have it. Walter B. Stevenson was definitely on the side of understanding what you were doing. He formed a happy compound with the School Mathematics Study Group. I don't know if you went through that. Asperay: Oh, yes. I know about that.

Lewis:

That started in high school. Am I right about that? Well, I think it did, but I can't remember. Maybe we had an SMSG algebra one book, but I don't remember much about it. (I know our geometry book wasn’t SMSG. I really loved geometry. That was a wonderful course.) I think we must've had some of the SMSG material, because it was still in like, you know, reproduced typewriting Courier font stuff. This would have been, I guess, in 10th grade. I started high school sitting -- I can remember what chair it was in our living room –- reading, over and over again, the discussion of what if and only if means, saying it over and over to myself, and thinking about the if part and the only if part, and all of that. And then, eventually, coming away with the sense that I really knew what this was, what this was about. And so, I really loved the SMSG approach to stuff. I know there have been many critics of it, and I became one myself, in a way, much, much later, when on one occasion, I tried to teach some of the stuff. But back then it really connected for me. And again, it fit very well. Walter B Stevenson's outlook was that, by really developing the material in a more formal way, you could make sure that it all really fit together. You weren't just doing things because it was what you did. That's what occurs to me. I'm sure I was influenced by many other people, but I'm not very other directed. I'm a pretty selfish and self-centered person. So, I don't pay attention to who I should be crediting with things.

Aspray:

[Let me ask] the same question about your Princeton years, was there anyone [particularly influential] during that time?

Lewis:

I was a shockingly lazy student. In hindsight, I suffered from an illness that I only got the diagnosis of when reading Carol Dweck many decades later. I don't know if you know Dweck's work, on growth mindset and fixed mindset. It's actually becoming pretty much mainstream, but that's been a process. So, Dweck observes that, broadly speaking, people have one of two attitudes towards what one accomplishes. Some people feel that what you accomplish is a function of your gifts, and your gifts are pretty much fixed. So, the term fixed mindset sort of comes from that. You've got whatever abilities you've got, And some people have more gifts than others. Then a growth mindset. I don't think this was Dweck's original terminology, but it's what's kind of stuck. In a growth mindset. It's about the work you do. Whatever gifts you might have play a definitely secondary role. Well, I could have been the poster child for the fixed mindset. I believed that I was smart. It's only much later that I realized that, although I didn't think about it as work, I spent a great deal of time on intellectual work. And actually the time spent with Martin Gardner would be a fine example. So, I looked at myself and I said, “well, I don't spend very much time on school work, and I do really well in school. I must be smart.” What I didn't credit was that I was spending all my time on stuff like Martin Gardner and reading the Book of Knowledge and all this stuff, you know? So, I was working hard, but I didn't think of my accomplishments as owing anything to my work. So, when I got to college, I wasn't doing work. I still remember complex analysis, which I just totally frittered away. Now I realize that there were lots of really interesting issues there, but no, it didn't reach me. I was a huge underachiever and again, I just didn't have any clue. I now know. Actually, a lot of what I try to do as a professor is kind of a reaction to this. I know now that it's really good for students, if there's somebody who tells them things like, you go to people's office hours and you should look for an opportunity to get involved in research. I had some classmates who did that stuff, but because there was nobody saying you had to do it, and because I had this view, well, you don't do the things you don't have to do, I didn't do any of them. I didn't get connected with any of the faculty. Princeton had a senior thesis program. You had to do a senior thesis, and that should have been an opportunity for me, but my advisor, who was one of the few people on the faculty who actually did stuff with computing, was kind of what you would expect you'd get, if you have a leading mathematics department tasked with advising bachelor's theses for all of their majors. He made a few efforts to kind of push me to do some things, but never very serious ones. I should say here that this wasn't a case of not accomplishing anything. I had had an internship, which was very consequential, when I was a junior – an internship at IBM. I developed what I still think is really a pretty good idea about a way to solve differential equations, quite different from the usual way, by optimization methods, rather than by incremental integration. My advisor didn't make any real effort to help me connect with what real people would have been doing, which was a condition I didn't escape from until my second round of grad school… (well, it started happening slightly in the first round of grad school, but not really much.) I didn't see the need to connect one's work to a community of scholars, or how real stuff gets done. My advisor could have been an influence, but he wasn't really. I doubt very much that there's a single professor I had at Princeton who would've had any occasion to remember anything about me.

Aspray:

Did you have an interest in and take any courses in psychology during that period of time?

Lewis:

I did not. And I never would have. I would have laughed at the suggestion. I often reflect on that as someone who has a PhD in psychology. If anyone had said that that's something I would be doing, I just would have been totally, totally scornful of that idea. Nothing could have seemed to me to be more of a waste of time. Now, in my defense, what I thought of as psychology in those days was sort of clinical psychology, which I still have a pretty dim view of. Oh, actually there is one person that I remember who was an influence on me. This is a name you may well know: Julian Jaynes. I didn't know who Jaynes was. Okay. My classmates and I thought we knew him, and I'll explain the circumstances. We made a lot of wrong guesses. In particular, we were sure he was British, which he was not. Well, you may know the story. Having been a brilliant rising star, a graduate student in psychology at Yale, Jaynes dropped out of sight, perhaps as an alcoholic. I don't know the details, but anyway, he just dropped out; and then, years later, contacted one of his grad school classmates, who by then was a professor at Princeton, and asked them for any kind of job. I think it may be that he started out cleaning animal cages. Maybe that's an exaggeration, but I don't think so. Anyway, he did manage to get some sort of a position that allowed him to take his meals in the dining hall where my classmates and I took our meals. Princeton had these things called eating clubs, and they were awful, snobby institutions. If you didn't believe in that sort of thing, then you could join this alternative thing. Life is full of ironies: it was the Woodrow Wilson School (now renamed). At the time that stood for progressive thought and rejection of the old social framework. Anyway, Jaynes took his meals there, and acted as an interlocutor for us and taught us some version of charades, and other things like that. Anyway, he got me to read some Freud, and I agreed that Freud is a really good writer, which I still think, but otherwise I don't really have use for Freud. Jaynes did influence me in an important way. When he found out I was off to MIT, he said, “Well, there are all these interesting people at MIT. There's Noam Chomsky the linguist. (I knew about Chomsky because I had taken a linguistics course.) …and Marvin Minsky, and there's this thing called artificial intelligence.” This was at a time when my idea that the only stuff of any interest in life is mathematics was starting to kind of break up, and Jaynes helped that along. When I showed up at MIT, I took courses following his suggestions. Actually, you couldn't take a course with Chomsky right off the bat, so I had a course from Morris Halle, another prominent figure, and I did take courses from Chomsky later. I had at least one course with Minsky and Jerry Fodor and other people like that. But anyway, I'll draw back to the undergraduate period. So, Jaynes did influence me, not so much in what I was doing when I was at Princeton. He certainly influenced what I did next.

Aspray:

What was this program at MIT and why were you attracted to it?

Lewis:

I was a PhD student in math. Actually, I need to fill in one more anecdote on the way here. Okay. At a certain point in my senior year, the faculty in the math department announced that they would have a meeting to discuss graduate school for students who were interested. I was interested, and as I've already told you, I had this idea that serious work starts in grad school, and you want to get there as soon as possible. So, we all turn out. As I remember, there was a panel of three faculty up at the front, and some number of us students sitting there. The faculty call themselves to order and they say, “We've been sort of talking over, uh, what we want to say to all of you. And we've decided, we've all agreed on the message we want to convey. Our advice about graduate school for all of you is: don't go.” It was only many years later that I figured out, I'm sure, what their agenda was. I believe what their message was, if our telling you not to go to grad school keeps you from going, you shouldn't go. Which I think broadly is pretty good advice. I was, of course, quite angry; I'll show those so and so’s! But it comes to mind now that that's all the advice there was because I didn't have a connection with anybody in the faculty. I don't remember that I talked to anybody about where I might go to graduate school. I really can't remember why that's where I applied. I've also told you that, that I was playing that sort of Russian roulette with the whole thing, you know? Because I knew if I didn't go to grad school, something pretty awful might happen; but I didn't know what that really was. So, I didn't put thought that I remember into making a choice – just that it was a well-known university. Anyhow, I applied and got in. I'll also mention another odd fact. Okay. Given what I've said about myself and how little evidence of real promise I would have shown… I mean, I, I got honors, but I didn't get anything more than that. My grades were pretty good, but I hadn't shown any signs of exerting myself. We had orals as undergrads. Those were, at least my version, just a joke. I was examined on topics in algebraic topology that I had then only the faintest glimmering of. So, how could I have gotten into some place like MIT? Well, I got like a 900 on the math GRE. Okay. Since then, I think you can't get a 900, but anyway, it's all this scaling by standard error or whatever it is. I got this score. I'd always been great at standardized tests, and that was one of the things that that exurban school system did for me in spades. We were tested every week for something around there. So yeah, we had, we had bubble sheets for breakfast, you know, that's what we did – I was just a top standardized test taker. Right. So, I did really well on the GRE. The rest of it's a fog, I suppose I must've had letters. I don't know. Maybe I didn't have to, I don't know. Anyway, this same thing, that same portfolio of attributes, earned an NSF fellowship. It's another thing I learned the hard way: that turned out to be a huge problem. I would have been far better off had I not gotten that fellowship. If I'd been a teaching assistant, somebody would have cared about what I was doing. Certainly, if I had been a research assistant, somebody would have cared about what I was doing. But I had a fellowship, so nobody cared what I was doing. Yeah, absolutely. I had exactly one meeting with a potential research advisor, Manuel Blum, and actually I've met Manuel since and remembered this with him. I took one of his courses, recursive function theory or something like that. I found it pretty interesting. So, I thought, maybe he'd have something that I might work on. He suggested something, and for whatever reason I didn't actually find it very interesting. And that was the end of it. I didn't make any connections. That was the big thing I didn't realize about being a PhD student. You may run into this with people who have read the requirements. I was like, well, you take your courses and you take some exams and then you do a dissertation. The fact is that, from day one, you need to be thinking towards what you're going to do for your dissertation. That's how it works. You know, I had no idea of that. So, I didn't. I didn't make any progress at all. My nominal advisor (by which I mean my sort of administrative math advisor, not a real advisor) was George B. Thomas, the author of the famous calculus book. He called me in, and it was obvious to him that I wasn't making any real progress. He said, “You should think about doing a master's degree,” which I did. MIT had a form of master's degree, which last I checked they still have it. I think it's a very liberal arrangement. You can do a master's degree without course specification. You may know that “course” is MIT-speak for department. All you had to do was take a certain bulk of courses, that could be anything whatsoever, and write a thesis. So, that's what I did. And in hindsight, what I was doing was studying cognitive science. There was no such concept at the time, but if you look at the courses I was actually taking, that's what I was doing. I was taking the required math courses, but those were not actually what was taking up mind share. By this time… Should I continue with MIT or do we need to go back?

Aspray:

No, continue on MIT.

Lewis:

Okay. For the first time, at MIT, I got somebody who acted as a real advisor, and that was a guy named Paul Kolers. Kolers was quite a prominent psychologist, just beginning to make his mark when my path crossed with his. He went on to be pretty well known and died, sadly, quite young. Anyway, he had a career at mostly the University of Toronto. I don't think he was a postdoc, because I don't think he had any supervisor; I guess he would have been, in University of Colorado terms, a research faculty member. He was on soft money at MIT and doing his own research with grants. He taught a course in cognitive psychology. That was a real turning point for me, because this was psychology, but it wasn't the kind of psychology that up to now, I had thought psychology was. I still remember some of the papers that were on the syllabus for that course. So, historical note, this would have been ‘67, I guess, and the book, Cognitive Psychology (by Ulric Neisser) that launched that field was published in 1967. I can't actually remember if we used the Neisser book or not, maybe not actually. I think what Kolers did was to assemble foundational readings in what would today be called cognitive… well, cognitive psychology wasn’t moving quite as far as cognitive science. So there was “What the frog’s eye tells the frog's brain”, which was an early work in neuropsychology, and probably George Miller’s magical number seven paper, and other stuff. So, here was a perspective on psychology in which there was real conceptual structure, including quantitative structure and models of processes in the mind, and stuff like that, which I found really appealing. It was a great class, extremely stimulating, and Kolers needed a research assistant. I can't remember which of the summers it would have been, I guess it might've been the summer after my first year. I remember that he sent me around checking references on some of his papers. The MIT library system was highly distributed, so I had to go all over campus, tracking down these print volumes. Hard to imagine that anyone would have had to go around to libraries checking references, but that's what I did, anyway. Then, over the summer I worked for Kolers as a programmer. I'd mentioned that I'd had this internship, at IBM, and I’d learned Fortran programming then. I was too naive to have a sense that there was much of any difference among different forms of programming. What I needed to do for Kolers was assembly language, of which I probably had never heard, but not to worry. I became an assembly [language] programmer. It was a great project. Kolers posed the question, which actually I think is still worth studying, although we’d do different things from what we did. He was interested in the question, could you read more than one thing at a time? My job was to program an array of different ways of displaying text, looking for ways that might allow somebody to read two things at once. There were a lot of interesting things along the way. I learned that you can read well in a single character box, through which you just run the characters of a text in sequence. I mean, it sounds like the sort of thing that would be very hard, but when you try it, it's very easy, actually. One of the display setups was two such boxes arranged so that you could sort of have one in front of each eye, and then you could run, you know, texts through both of them. We did actually find one person who could “read” two three letter words presented that way, but it became clear talking with him that he could just remember the six letters and then sort out what the words were. So, nobody could read two things at once. I now think that probably if you practiced enough, you could. In the meantime, there's been work by somebody named Elizabeth Spelke, who did some phenomenal work, but the key to it, what you'd have to explore, would be truly heroic amounts of practice. I’d bet 10 cents that somebody could practice enough to read two things at once. This work led to my first publication, and it could have been more valuable than it was. I mean, it was valuable because publications are always valuable, but Kolers didn't involve me very much in the writing of it. So, I don't think I emerged from that experience with very much more of a sense of how things really work. Actually, one of the ways I know that, is that my master's thesis, which I was doing around that time, was something that could have been a real advance in the field, but was not – not because the ideas weren't sound, but because I didn't do anything with them. I remember that one of the really destructive ideas I had at that time was that there were some ideas that are so good that you shouldn't do too much to spread them around, which I now know is a completely cockamamie idea. Anyway, in the years since, I've acquired a copy of my master's thesis and it's really a joke. I mean the ideas are brilliant, if I say so myself, and I'll tell you what they were in a minute, but the execution, the writeup – and it's not just a matter of what things look like when they're typed by a bad typist on an ordinary typewriter. I think I had only one reference in the thing. It served as a thesis. I mean, it satisfied the requirement. I got my degree, but it had, well, I was going to say zero impact. That's not quite true. It did have impact later on, but quite indirectly. Anyway, what was the idea? Well, it was the idea that is now known as production systems as models of mental processes. What I conceived of in this thesis was that you can have a collection of rewrite rules that could play checkers and could learn things about playing checkers in a certain way. Not, not in any statistical sense, but by sort of reasoning. There were a lot of ideas in there that were really good ideas. There was another paper that I never got back from Minsky, which I turned it in to his class. I didn't keep a copy; that paper captured some ideas that have shown up later in other people's work. I think I could have had conversations with Kolers in which he could have said, look, these could be worthwhile ideas, but they're not going to go anywhere if you just have them in this form. But for whatever reason, we didn't have that conversation. It's possible that I turned him off, turned him away from that, on the grounds that somehow I needed to keep whatever fantastical notions I had about keeping ideas to myself. But the reason I say it didn't go wholly without any impact was that many years later, when I was a grad student at University of Michigan, and working with John Anderson, Anderson had his human associative memory work, which he had done for his thesis at Stanford. He was looking for a way to represent procedural knowledge. So, I was able to say, “Well, you could have these rewrite rules”. At about the same time, Alan Newell and others were also developing these ideas independently. So, it's possible that Anderson would've picked up the idea with or without any input from me, or that he already had. Anyhow, I was able to suggest to him that that would be a productive thing to do, which it proved to be. So, in Kolers, I found somebody who could turn my work to account and also became a sort of sponsor for me in later endeavors. So, um, just, just an outline to, to sort of jump ahead to that juncture. After working as a programmer at IBM for about five years, I decided to go back to school; and there's of course a whole story about that. At that time, I did have people that I had met at IBM. Notably, I got to know John Gould, who was a real mentor for me, and helped me understand, first of all, that studying psychology might advance some of the goals I then framed, which had to do with making a different kind of computer system; but also helped me understand, what are the good graduate programs to try to get into, and that sort of thing. He gave me proper advice, and then the fact that Kolers was willing to write a letter for me, and the fact that I had a paper published meant that I did get admitted to Michigan, which was probably the number two ranked program in psychology nationally. I didn't get into Stanford, where I wanted to go, not because I really knew much about what it would mean, but because Stanford had a course on how to become a psychologist. I recognized even then that that was really a good idea. Michigan didn't have a course exactly like that. It had something that was equally good in the end, but I didn’t realize that at the time. The Kolers connection was, I'm sure, important, because I didn't have a psychology degree or anything like that. Michigan was willing to take a chance on me.

Aspray:

Tell me the story about how you left MIT and ended up at IBM.

Lewis:

Okay. I mentioned that I'd been taking these linguistics courses and hanging out with the students. I went to some of the linguistics parties. I must've mentioned to somebody that I needed to get a job. I don't know who it would have been. The names of people that are coming to mind are people that I think I met after I went to IBM. There was a connection between this odd IBM lab across the street from the MIT campus and the linguistics department. Anyway, somehow or other, I found out that there was this IBM office there. I mentioned that I'd had an internship at IBM, which I had really liked. So, I walked into this IBM office and apparently even then they had their personnel records accessible. So, they could find me. I guess I looked okay, although I'll also say that at the time of which we speak IBM would hire pretty much anybody as a programmer trainee and pay you. So, you could be an English major and get a job. It wasn't like it was a great accomplishment, getting a job at IBM. But anyway, I guess I looked good to them and so they hired me, and I did it sort of in the same vein as what I didn't do about grad school. I didn't look at any other company. I didn't find out whether what they were paying me was reasonable or not, or anything. I just took their offer because it seemed good to me. I remember that I was making almost as much money as my dad had made a few years before. So, that seemed like it must be a good salary. I now think, probably the reason they hired me was that I knew something about programming and I knew something about linguistics. The first project they put me on was a natural language processing project, where I was working in support of a couple of MIT linguistics doctoral students, Peter Culicover, and, gosh, what was the other guy's name [John Kimball]? They were writing very early examples of what I now realize were utterly doomed natural language processing codes, and I was kind of their programmer. That was like the easiest possible transition: just across the street from where I had had been. That's how that happened.

Aspray:

Did you stay in that facility the whole time you were with IBM?

Lewis:

No. That facility had been a boutique set up, created by IBM to satisfy the preferences of two prominent IBMers, Nat Rochester and Jean Sammet, both of them people you could read about in Wikipedia. I come across Nat’s name in very unexpected places. I’m told some people credit Rochester with being the inventor of the assembler. I don't know that I can vouch for that, but he was a real computer pioneer and also an AI pioneer. It turns out he was behind some of the early IBM work, and may have been the backer of Arthur Samuel when he was doing his checker programming work. Anyway, that was Nat. Then there was Jean Sammet, who was a pioneer of the study of programming languages. Already by this time there were hundreds of programming language. Jean wrote this big book that you probably know, with the Tower of Babel on the cover. Apparently the two of them were working together somehow in IBM's Federal Systems Division down in Gaithersburg [Maryland]. And as it was represented to me, they were sort of musing that they both had some sort of connection in the Boston area. They were saying, “Wouldn't it be great if we could work in Boston?” I guess they went to management and said, “We like working for IBM and all, but we would like it even more if we could work in Boston.” So, IBM created a lab for them called the Boston Programming Center, which was in Cambridge. It was in the same building, I can’t remember whether upstairs or downstairs, as something called the Cambridge Scientific Center, which actually lasted longer. This lab ran as part of the Federal Systems Division for a period of a few years. I don't know how long it had been going before I got there, but after a couple of years, IBM's finances turned a little south and they closed it down. But these were the days of IBM's full employment policy when they would never lay anybody off. So, you could be fired for dishonesty, but otherwise the worst that could happen to you would be to be put on commission. Mostly, they would just find jobs for people. Nothing to do with me, but I benefited from the fact that, since this place was virtually on the MIT campus, there had been a kind of conveyor from MIT to IBM Watson through this place. There were alumni of the Boston Programming Center at Watson and they became sort of my patrons, when I made the move down there. So, the lab was closed. I was sent to Watson, which was an enormous break. I mean, the overwhelming majority of people there had PhDs. Not quite everybody, but most people did. Anyway, I was conveyed down to Watson after a couple of years at the Boston Programming Center.

Aspray:

What were your responsibilities at Watson?

Lewis:

The project I joined when I got there was called System A. I don't know if you've ever heard of it. One of my enormous regrets is that, many years ago, I had in my hand a book, or I was reading about a book, with a title, something like Great Software Failures. There was a chapter about System A, but I've never been able to find that book. System A was one of IBM's then-periodic attempts to have sort of a new generation of computing. This one was going to be built around the high-level machine concept. There were various reasons having to do with the demographics of the company, why IBM thought that it would be a good idea. They had an oversupply of hardware engineers and an undersupply of software engineers. They thought putting the hardware engineers to work, raising the level of the machine, so programming would be easier, would be a win. It was a very comprehensive program. It was going to remake everything about how computing was done. I got into the group that was working on programming language support. That was another huge break because I got to work with some wonderful people there, including two Turing Award winners. One of my managers was Fran Allen. You know, she was the first female Turing award winner and a lovely person. Then somebody who was never my manager, I don't think he was ever allowed to be anybody's manager, was John Cocke. For a while, I was a kind of informal sidekick assistant of John's, which was a blast and a half. Then it was suggested that we make it more formal. I asked John something like, I've got occasional things that I would like to work on of my own. Would that be okay? And he basically said no. I don't know what would have happened if I had gone the other way. I didn't cry salt tears about it at any rate. Another person who was my manager was Dick Goldberg. I don't know if you know that name. Dick was one of the great people, just a lovely guy. He was one of the original Fortran team. I also got to know Bob Nelson, who was never my manager, but I worked with… I had another publication with Bob, on something or other. These were palmy days at the Watson research center, in which the sort of freewheeling thinking that one thinks about and some idealized visions of what research might be like actually went on. We mostly spent our time talking about stuff, language design issues. I did have some coding duties. I remember writing a COBOL front end that was supposed to take in COBOL and spit out some sort of pseudo-code that was architected into this common, intermediate language, part of System A. It sort of foreshadowed something that Microsoft went on to do with their common runtime framework many, many years later. The notion was that you'd have many language front ends, and these would feed into a back end and then you could do optimization at the intermediate language level. There were a lot of sort of philosophical issues about language design, which is to this day a subject that I find really fascinating. And I worked with other people; Dave Lomet, who has been at Microsoft Research for many years, was one. Most of the people in the group were PhDs. Dick Goldberg was a prince of a fellow. Because of his role in creating Fortran, he was a person who could do no wrong in the world of IBM. In his group, you had to do what Dick wanted to do, which was to have a very comfortable group where we talked about a lot of stimulating things. There was also Les Belady, who went on to direct the MCC computing lab. I guess I might've had another publication with Les, but I was still not oriented really towards publication and what it meant. I was fortunate in working with people who had some idea about that, but I didn't buy very much into it. Another thing that that started happening around then was the dawn of the psychology of computing… I mentioned John Gould. Around the time that I was sent from the Boston Programming Center down to Watson, there was some sort of a corporate task force, on graphics or something like that. A guy named Bob Tabory who had been my manager at Boston Programming Center, was sent down to the Time-Life Building, where some of IBM had a serious operation. I guess I must've been put on this task force, likely by Bob. Through that, I met some of the people at Watson who were among the true pioneers of what would today be called usability or user experience - notably, John Gould and a guy named Lance Miller. Both of them were PhD psychologists. John had been probably an industrial psychologist. I don't know so much about Lance's background. Lance might've been more in sort of cognitive psychology, but anyway, they were among the first to realize that psychology actually had something to say about what computers systems should be like. John was a huge influence on me. I've already mentioned that he helped me figure out, when the time came, what would be a good program to get into, but more than that. I guess the work with Kolers had given me a leg up, but because of John and Lance, and there was later a guy named Steve Boies, another PhD psychologist, there was an environment there in which people were talking about these things. That was all happening in parallel with the work on System A and with John Cocke. With Fran, I was working on optimizing compilers, things like that. There was also a guy named Barry Rosen, who was a category theory and mathematics kind of guy. He and I wrote a paper, a couple of papers, on recursively defined data types. So, I got drawn into some things that ended up being published, but I didn't have that as sort of a focus. You'll see a theme running through here that, as a rule, I’m very oblivious to how the part of the world I'm in works. But I ended up just being seriously lucky to be working with people that make things work out.

Aspray:

Do you want to talk next about how you made the move from IBM to your doctoral program?

Lewis:

Sure. The time of which we speak was the dawn of the microprocessor. And as with so many other things, I didn't myself have very much insight into what that meant, but other people did. As you know, in the earlier years of the computer age, computers were very expensive relative to human time. The first programming I ever did was in high school, when somebody brought a Royal McBee LGP 30 to the school one summer, and there was a summer programming course that I was able to take part of. That was the first time I touched a computer. That was absolute machine language. I was already oriented, thanks to Martin Gardner and the GENIAC and so on. At any rate, that's the first computer I ever worked with. Many years ago, I gave a talk for which I went back and read up the specifications of that machine and did a little bit of economic analysis. I'm not going to get the details right, but that machine, which had much less computing power than it would be possible to purchase in any form today, cost several clerical office staff person’s years of salary. If you just looked at what you would pay for a clerk and what you'd pay for the computer, you could hire several people for a year for what it cost for that machine. So, under those circumstances, it made quite good sense to take the view that anybody who's going to be allowed to work with the computer has to be extensively trained, because the computer is a precious thing. I don't know if you're as big a fan of George Dyson as I am, but I really love Turing's Cathedral and the account of the Institute for Advanced Study machine. One of the things that comes from that is that if you view computation as a fluid, it's like incalculably precious. The efforts people were willing to put in to get a teaspoon full of computation! I mean, it's just staggering. So, now the microcomputer comes along and not me, but others could see that computers were about to cost 99 cents. And, you know, that's going to just turn the whole world upside down. In some organizational way, we knew that, we being the computer science department at Watson. I was on a project that was set up to try to conceive of a comprehensible computing system. I remember various things about that project. One was that we would get together in a conference room and we would dedicate ourselves to talking about comprehensible computers. In a little bit, we'd find ourselves talking about optimizing compilers, because we knew a lot about optimizing compilers and, and we knew not one thing about what a comprehensible computer would be like.

Aspray:

Can you explain the word comprehensible?

Lewis:

A computer system that an ordinary person could understand and get to do useful work. So, without knowing it, or marking it, anyway, I'm sure we were among the indefinitely many people who've independently invented the idea of the persona. Our persona, I remember it was a person who runs a dry-cleaning establishment. Our concept was that this person who runs the dry-cleaning established would be writing the programs that would be used to make the dry-cleaning establishment run. Okay. From a contemporary perspective, that's a totally cockamamie idea. But one has to realize that at the time of which we speak, there was no such thing or concept as a software market. Software was not something you went out and bought. Software was something that, in a very limited number of cases, was supplied to you by the hardware vendor, like the compiler. You didn't write that stuff, but other than that, you wrote it. You know, there were only negligible exceptions to that. So, we didn't have the idea that it's not the person running the dry cleaning, that person isn't going to be writing code. They're going to buy a dry-cleaning establishment program. We just didn't have that concept. We'd have these conversations about how could the ideas of computation be cast in an understandable form. Of course, this was before the spreadsheet concept, that came down later. There was very little to work on. I had some sense that this was kind of futile, the way we were going about it. Armed with the opportunity to talk with those pioneering psychologists, I had the thought, and they confirmed me in this, gee, if I went and studied psychology, maybe I'd learn something about how people understand things, which of course turned out to be right. That was most, but not all, of what led the decision to go and do that. That's what led to the decision to do the particular thing I did. But another thing was, as I've mentioned, basically the people at Watson had PhDs. There were exceptions of people I knew who were successful, who didn't have PhDs, but it was pretty, pretty limited. So, it just sort of seemed like if I intended to have a career there, that would make sense for me to do. The other thing is that I was starting to think that maybe an academic career would be something that I'd be interested in. One of the steps along that path was that I had an unusual opportunity to actually teach a sort of remedial math course at the college level. My wife was in the inaugural class at the State University of New York at Purchase. They were just getting it off the ground, meeting in trailers and stuff like that. Somehow or other, in putting together their entering class, they determined that there were some number of students who needed some sort of top-up in math. So, they decided they wanted to offer a course in that. I had been a math major and drunk at the spring of the School Mathematics Study Group. I thought, this will really be fun. And furthermore, I know exactly how to do it, because you start with the natural numbers, and then you develop the integers, and then you do the rationals and the reals. What could be more satisfying? So, I taught that course to these hapless souls, and it was, of course, a total catastrophe. I had no sense of the difference between conceptual learning and skill learning, or any of that stuff. They didn't learn how to do anything, and I'm flummoxed, that we've covered all the logic, you know, what's so hard?. But I came away from that catastrophe feeling that teaching was much more interesting than I thought. That's probably where the idea that I might want to actually teach had its origin. And then I think I was chafing at IBM's very stingy leave policy. At any rate I made a decision, which from some points of view was foolish, although I think I would do it again. IBM would have happily paid for my PhD study, but I opted to leave the company. The reason was that I didn't want to feel obligated to return to them, because I wanted to be able to consider these other options. As things turned out, I did return to them. So, I lost something like five years of seniority for my pains. Nevertheless, that was my thinking. So, I went off to Michigan and my advisor was a guy named Jim Greeno, who was quite a prominent psychologist. He just passed away within the last few months, at a ripe old age. He and his students, some of whom I'm still in touch with, were doing work, not on how people understood computer systems, but how they understood other things. I'm not sure anybody was working on the computing problem at the time, but shortly thereafter, Ruven Brooks at Carnegie Mellon started doing some work. There were a few others, but what Greeno was working on was how people understand mathematical concepts, including ideas in probability and statistics. Somehow, in connection with all of this, I got to look at some of those papers, and I could recognize that it seemed like work that was relevant to what I was interested in. In hindsight, it seems I might've had some sense that there'd be methodological carry-over, which I would say is mostly what there's been. Whereas, not that many of the concepts of cognitive psychology have transferred to the design of computer systems. The methods used by cognitive psychologists to study mental processes have proved extremely useful in the design of computer systems. I think I had some sense that that might be true. I'm sure it wasn't a very mature sense, but anyway that was the thinking that led to my first departure from Watson

Aspray:

As a psychology doctoral student, did you feel well prepared to be there? Did you feel like you had shared interests with your fellow students? It was an intellectual community, right?

Lewis:

Absolutely. The experience of the first time in grad school and the second time could not have been more different. We've already covered some of the dimensions. Having now worked with Kolers, and then later with John Gould, I had more of a sense of what was going on. I was helped by the way the Michigan program was structured. At that time, and I hope still – I don't really know – all students in what was called “experimental psychology” took an extensive proseminar course, which was many hours of credit. Essentially the whole faculty paraded through that class and taught us the fundamentals of whatever their piece of the discipline was, and all of us students were in there. So, all the things you just said, community. There was a sort of an odd twist. The program at the time was housed in a former elementary school, off the edge of the Michigan campus, which the university had bought, the Perry School. When I showed up there as a new student and met the graduate secretary, she showed me to my office, which had been the shower room in what had been a gymnasium in this school. The shower heads had been removed, but the pipes that the heads had been attached to and the faucets and all that were still in there, and it was tiled. But there was a desk in there, and it was really a weird situation. All the other students were in what was called the bullpen; all the other first-year students were in this other place. I didn't feel resentful exactly; I mean I wasn't inclined to question this, but I could tell that it would've been better to be in this bullpen because they're spending all their time talking about stuff. So, I would go down there and hang out there. Nevertheless, that wasn't where I was supposed to be. Sometime later, Greeno came down and he said, “what are you doing in the shower room?” And I said, “Well, that's where Laura put me.” He said, “No, we mean here.” In the gym area, adjoining, there was another desk. I was actually supposed to be at that desk, which I did move to, and it was better. I can't remember if I talked to Jim then or later, but it came out that, because I was somewhat older than the other students, they thought that I would want to have an office by myself, which wasn't true. Anyway, that's what came up, but it didn't matter very much. There was the nature of this course and how we studied for it. It really was a great way to build the community. There were other things in the program that really avoided the errors I'd made and the kind of vacuum that I had encountered at MIT. There was a first-year research project, and also there was culture that said, you can tell who is going to go on and be a success by the fact that they published their first-year research paper. So, there was a clear signal set there. I was very fortunate. This was another of many lucky breaks I've had. I mentioned Jim Greeno was my advisor. Jim actually was unfortunately away a lot of the time I was at Michigan. Somebody who just, by dumb luck, happened to get there probably about literally the same time I did was John Anderson, who, as you may know, is one of the most prominent cognitive psychologists ever. At the time, he was a fresh new PhD. What happened was that his wife, Lynne Reder, was a grad student, I guess a year ahead of me. And so it worked for John to be at Michigan; and Michigan had something modeled on the Harvard Society of Fellows or the Junior Fellows, I guess they are. Michigan had something like that. John got one of those appointments. So, during my student years, John was there. He was bringing in grants and all that kind of stuff. So, John was my advisor to probably about the same extent that Jim Greeno was. Both of them certainly influenced me, but in different ways. I had an idea about some of John's work that panned out. So, we were able to publish that first-year research project. I was off to a good start as a student. But again, because the Michigan program had those structures, that kind of meant if you were paying attention to what was going on, you would sort of get fed into the right sort of outlook and way of working.

Aspray:

Had your previous experience at IBM helped to acculturate you to this kind of community environment?

Lewis:

I suppose it had. I've mentioned that we spent a lot of time talking about stuff, and I think talking is really important. And, yeah, that was the atmosphere in cognitive psychology at the time. It was what, ‘67 to ‘73, just a half a dozen years old or so, so it was still sort of open ground. There was a lot of room for somewhat speculative thinking about how things might work, things like that. So yeah, it was a very congenial intellectual environment for me from that point of view. I've always liked, still like the feeling of ideas being exchanged. I especially like it when there's something that you thought had to be true and then you find out it's not, and that sort of feeling to it.

Aspray:

What did you do in the way of your research project, your dissertation at Michigan?

Lewis:

First thing I'll say is that the enterprise of developing a comprehensible computer system was essentially shelved in favor of the task of really pitching in to cognitive psychology as it was. I suppose if somebody had pressed me, I would have said, “Well, these processes we’re working on are all relevant to computer systems. It's just that that's not the setting in which we’re working on them.” This was the era of Newell and Simon's book, Human Problem Solving, which was very influential. I was very interested in problem solving, that was the focus of my work, apart from the work I did with Anderson. For my thesis, I did a project in computational modeling, accompanied by some experimental work, on something called the Einstellung phenomenon, which is one of the gestalt discoveries in the sphere of problem solving. The locus classicus of the phenomenon is water jug problems. You've not run across these? You're told you have, let's say, three jugs with certain specified capacities, but not graduated. You have an indefinite supply of water, and you have to measure out some stipulated quantity. This is a nice space of problems in which you can have very easy ones and quite tricky ones, and anywhere in between. One of the gestalt psychologists, Luchins, did an experiment in which he created a sequence of problems that could all be solved with a particular kind of routine. Then he also had a problem that could be solved using that routine, or it could be solved much more easily. “Einstellung” means “stamping in”, and what Luchins showed was that if you have somebody practice this routine a lot, and then you give them this problem that could be solved much more easily, they won't find the easier solution, they'll find the hard solution. So, I got interested in how one could model that phenomenon. I implemented a production system, and some learning mechanisms and actually learned quite a lot about that. Then there was some experimental work testing some predictions of the model, which actually weren't borne out in my model. Arguably, if you applied the model to the human case, you'd have a situation where even if you asked somebody to solve the problem the easy way, they wouldn't be able to, because they had reorganized their knowledge in such a way that the easily alternative wouldn't be available. Not surprisingly, that's not borne out in human experimental data. I'd say most of the value was in a bunch of theorems about how systems of this kind would work. Let's see, what would an example of one of these concepts be? Okay, so the question of safety, which comes from actually optimizing compiler theory. You have a program and you want to replace it by a more efficient program, but you don't want to change any of the effects of the original program. That puts a lot of constraints on you, and those ideas kind of transfer it to other kinds of learning mechanisms. So, if you require that all your learning, all transformations of your knowledge that occur in learning, are safe in the sense they don't change any outcomes, that turns out to be a very strict constraint. Anyway, the thesis was mostly about a learning mechanism called composition of productions, and in some form that survives in the work of Anderson, for example. That work did get published. It's been superseded by other stuff, but anyway, it did have its impact. I guess another thing, which is still kind of with me, is an appreciation of the role of modeling in trying to think about things like this. Actually, one of the projects I hope to get back and finish off in retirement is some work along this line. There are other things that I've been doing that draw upon the same kind of idea. In brief, I would argue that verbal argumentation, though it's so widely employed by people in trying to understand how things might happen in the world, is an extremely blunt instrument. It happens all the time that you can persuade yourself, by verbal argument, that things might work in a certain way. Then you try to create a computer model. At first you think it's going to be easy to do, because you think it's so clear what's going on. You try to build the computer model, and the computer model just doesn't do what you wanted. Actually, that happened to me when I was doing the dissertation. So, the insights I got in doing the work came from the fact that unexpectedly to me, the model didn't do what I thought it plainly should do. Since then, I've found many examples of this. I now find that I have very limited tolerance for just trying to have verbal arguments about stuff. It seems to me that it's so likely that you run into this problem that there are things that you think would work a certain way, but when you actually try to get them work that way, you find out there's something else going on that you just weren't able to appreciate. So, I'm a big believer in computational modeling as an adjunct to intellectual argumentation, and that got started with the dissertation.

Aspray:

How are you doing? Are you, are you wearing out a little bit?

Lewis:

I could talk forever about myself.

Aspray:

Okay, fine. Tell me about your thoughts about what you were going to do next, as you were finishing up your dissertation.

Lewis:

Okay. I thought that I was going to start an academic career. I went on the job market and I got some wonderful offers. There are probably people living who have never forgiven me for not taking those offers, but I didn't. Okay. What happened, there are a couple of things. One thing I should say was a difference between the first time in grad school and the second was that the second time I was married. Also, I was older. As I mentioned, I finished college extremely naive and immature, as the narrative has demonstrated. So, I was a lot older and my personal life was settled, happy. And my wife wanted to begin, did begin her doctoral studies. I guess I probably had a year to go roughly. She was admitted to the zoology doctoral program at the University of Texas. One of the reasons that we felt that could work out was that Jim Greeno had been away a lot, and maybe John Anderson had already moved on; I'm not sure. But at any rate, it just turned out that there wasn't really much to hold me at Michigan while I was writing out my dissertation. One of the many strokes of luck I’ve benefited from was that Jim Greeno had been, I think, a grad school classmate of a guy named Phil Gough, who was chair of psychology at UT Austin, which is where my wife was starting her studies. So Greeno contrived with Gough to get me a visiting instructor position. Now I know that that never happens. You never give somebody else's graduate student a visiting instructorship; you give it to your own students. So, I was in this really unusual situation, which in the long run panned out very well. In the short run, it was more mixed. There I am as a visiting instructor. I taught a grad course, even though I hadn't finished my own degree. The course was on memory, and fortunately there was a really good textbook that I could use. Then I also taught statistics to nursing students. I did a lot of that teaching, and I really loved it. But I also learned some things. I learned that, at a research university, you can't devote to your teaching the amount of time that you might if there were no constraints on you. I had some friends among the doctoral students, who are still friends to this day. I also was friendly with some of the assistant professors. I could see that the tenure process really made them miserable. I also saw that this had nothing to do with the objective likelihood of their getting tenure. There was something about the process. It doesn't matter how much everybody would say that you have nothing to worry about. You still worry. It was clearly an awful thing to go through. That was part of my thinking as I'm contemplating these job offers. Another thing was that I did apply for a position at Texas, but I didn't get it. So that meant that we would face some period, some considerable period in a commuting marriage. I knew that that would cost a lot of money. Another thing that happened again, no thanks to me, was that over the years that I'd been away from IBM, roughly five, four or five, of these ideas about the importance of usable or comprehensible computer systems had gained momentum within IBM. They had a company-wide task force to kind of decide what to do about it. They summoned people from far and wide to the Brown Palace Hotel in Denver. One of my sponsors, one of the people from the Boston Programming Center, who had preceded me down to Watson and had become somebody on the leadership staff down there, still kept track of me as a kind of patron and knew what I'd been doing. She said, “Clayton, maybe you'd like to come to this meeting, we're going to be talking about this stuff.” So, I came and…


Aspray:

Excuse me, who was that person?

Lewis:

Pat Goldberg, actually the wife of Dick Goldberg, who had been my manager. So Pat was, I don't think ever my manager, maybe she might've been my second line manager at some point. She knew how to get ahold of me and she knew what I'd been doing and studying, and thought correctly that I might have something to contribute to this conversation. They also had Alan Newell from Carnegie Mellon as an outside person. I can't remember who else might've been there from the outside. But anyway, as an outsider, I was included in some sort of closing discussions presided over by a guy named Lew Branscomb, who at the time was IBM's Chief Scientist. One of the things we talked about was how this technique of the experimental psychologist, thinking aloud, could be applied to computing systems. Methods for studying problem solving could be applied to understand what's happening when people are using computing systems. So, I said, this seems to me that that would be a good thing to try to do. I guess I must've made a convincing case. Maybe Newell endorsed it. I don't know. Anyway, Branscomb said something like, what would it take to get you to come back to IBM? He might even have said, so why haven't you come back to IBM? And I said something like, well, nobody's made me an offer. I don't remember exactly what was said, but what was done was that IBM made me an offer. Okay. So, all that came together in that I was offered roughly twice the salary the universities were offering. Again, this was facing a time when we knew we were going to have a lot of extra expenses, but more than that, it also was a way to avoid the tensions that I had felt in my visiting instructor role and observing the lives of my counterparts. That experience put me off my thought of academic life for a period of time. It did take time, but eventually I realized that a couple of my positions could be expressed this way: I care so deeply about teaching that I'm not even going to try to do it. I came to feel that that didn't on the whole make total sense. But it was a pretty accurate statement because I had felt, when I was teaching there, that I couldn't put the time into my teaching that I thought it really deserved. I knew that the system wouldn't permit that. So that's what I meant by “I'm not going to try to do it.” But then a sort of companion thing was that somehow it seemed to follow from my analysis that I thought there must be other people who, for whatever reason, would be better able to cope with these compromises than I would be. I came to feel in the end that I didn't think that was true. I didn't see that I would be any worse at dealing with these things than other people would be. These realizations, along with some other circumstances, led me to think that I wanted to go the academic route again. What were the other circumstances? Well, one was that my wife had finished her degree. We were living in Westchester County. That was not an environment that was congenial to us and particularly not to my wife, and also to having the sort of family life that we aspired to have. We had no children, but we thought we would like to have children, and upper Westchester, which is where the lab is, is sort of housewife-chauffeur country. There's no public transportation. Everything's very spread out. So, if you have kids, somebody – and it's usually the wife – is spending all their time, driving the kids around, while many of the men are working in the city. So, it's housewife territory. My wife is not a housewife, so it was not an environment that was congenial to us. And then another consideration was that somebody I'd met through my professional network was Professor Peter Polson, who has been retired in Psych here for some time. One of the things that happened was that I got connected with the environment of professional technical conferences and particularly the CHI conference. I was among the founders of that group. I mean, not one of the formal founders, but in at the beginning. Much to my advantage, I was kind of in on the ground floor of what's become a very powerful segment of the profession. I'd gotten to know a lot of people, including Peter Polson, and Peter let me know that the computer science department and the Institute of Cognitive Science here [at the University of Colorado Boulder] were recruiting, and that I might be a fit. Actually, they wanted to hire somebody in artificial intelligence, but that was such a hot field back then that they couldn't get anybody to accept their offer in that field. So, Peter was able to say, there's this other field we could move on. I did go back on the academic job market, and actually this time around, I didn't have multiple attractive offers. I had only the one. But that looked good. One of the things that influenced me was that one of my oldest friends, going back to sixth grade, had settled in Boulder. He’d spent time going all over the place with his wife, looking for a good place to live. I actually stayed with them when I came on the job talk and they said, we could imagine a better place to live than Boulder, but we've never found one. This wasn't based on nothing. They had really done a lot of looking. So that was a reassurance. So, I took that job in 1984 and, in the twinkling of an eye, it's 36 years later.

Aspray:

Let me stop you for just a minute. It's eight minutes to the hour of four, and that's when I promised you, we would stop for today. If you're willing, we have quite a bit more to do. We could do it in another [session], should we?

Lewis:

Yeah, I can't do it now. I do have to stop for today, but it's been really fun. So, yeah, I hope we can arrange another time.

Aspray:

So let's formally stop for right now, and I'll be in touch by email about setting up [the next se3ssion].

Lewis:

Perfect. I've really been enjoying it. Lots of things next time.