Adams: This is Judy Adams, with Henry Lowood, interviewing Douglas Engelbart for the Stanford Oral History Project, January 14, 1987. This is interview two.
Lowood: You still had your consulting company, and Digital Techniques. I think the place to start would be to have you just talk about how Digital Techniques came to an end. What happened to end the company?
ENGELBART: Okay. Well in the spring of '57, we were still building little prototype gas discHRGe devices in the laboratory, trying to collect data, and talking with our backers about applications. I started out assuming it could work; basically a shift register was the first thing I did. And if you think of how expensive the vacuum tube things were in those days, and what they were making shift registers out of, this would have been a lot simpler. We were looking at how you could make advertising billboard-like signs, because you could have these little moving lights come across with a message. Well, these little moving lights could make patterns. We were trying in some way to do that. But we had so little experience in anything industrial, putting out products and all of that, that we were just intuitively beginning to realize that we didn't know how to make the step from showing these pretty things in the laboratory to evaluating their potential--how much it would cost to produce them, how maintainable would they be, how subject would they be to all kinds of interference, and other problems. It's a very new phenomenon to apply out there.
Adams: . . . . let alone setting a production schedule.
ENGELBART: You can understand their interest in looking for a way to pay off their investment in it, but I was a little uneasy about being dislodged, or moved away from computer component design. The patent attorney said, "Well, you do the patents, and you work up some basic demonstrations. Then you approach some companies that could use it, and you sell the whole company to them, and make a profit that way." So it was all changing a lot, and becoming territorial. I was uneasy about the competency we had for doing it, though my partners were very enthusiastic, they were just that much younger than I am.
Lowood: Did you acquire competence in dealing with patents as this was going along; did you learn about the patent process?
ENGELBART: I learned a lot more, yes. We were concerned also about the emergence of the solid state devices, as far as computer components. What cost seventy-five dollars the year before as an experimental transistor might not only run for a few hours before something went wrong with it in the laboratory, now you could get it for thirty dollars, and the promise would be ten dollars, and on down the line. Also, we realized that the military and the commercial interests have so much interest in the solid state as a potential that there would be a great deal of money put behind it. Competing with that for the phenomena like ours, there were many advantages the solid state would have over the gaseous state. So the reverberation of these kinds of things caused our sponsors to seek someone who could help evaluate what we were doing. So I suggested Stanford Research Institute. They wrote a letter and made a little contract with SRI. They sent up a team of people who talked with us in the laboratory, and wrote a report. It essentially said that the solid state is going to doom all the work we were doing; it was interesting, but it can't compete as an investment.
Lowood: Do you have any idea who the author of the report was?
ENGELBART: Yeah, later, I was working right in the same building as all of those guys. One fellow was named Jack Bialik, and the other was Milton Adams. he's still there.
Lowood: We might be able to dig the report up.
Adams: Were you relieved about that?
ENGELBART: Well, it made sense, it spelled doom for our little venture, in a sense. And the sponsor said, "Well gee, fellows, that's bad news, but you realize it doesn't make sense for us to go on," and we said, "Yes, we realize it." We didn't know what to say after that. So they disengaged. They were very generous; they gave us a couple of extra weeks of pay, and they didn't have to. They said, "What do you figure you could get for that equipment?" And we said "Well, let's see, it cost about this much, etcetera, and they said, "You guys want it?" "Oh, yeah." "Okay, we'll just write it off." So they gave us the equipment. So we thought, "Oh, hey, that's interesting, we've got some lab equipment." There were four of us that were sort of turned on about all of this, to what we can do to carry on. I'd had a lot of interaction already with the Marchant calculator company, whose headquarters is in Emoryville. They had, inadvisedly, as it turned out, bought out some local inventor who'd conceived of a very ambitious data processing computer system. You know, it's just the kind that later everybody had, but this was before the technology could really get together to do it. They were in fact busy trying to get this ready as a product. All of this was about two miles from where we worked, in a special building. So I approached the guy who was running that, and asked him if he could use any of our work, so he hired me, and one of the other guys as a consultant to work pretty near full time.
Adams: This was the inventor?
ENGELBART: No. The inventor, as often happens, proved sort of hard to interact with the corporate people who bought it. They sort of disconnected him and sent him on his way, and brought in some younger guy from Los Angeles who had aerospace project managing, electronic experience. He was trying to whip it all together. When I look back on it, it's like saying, "My god, they don't have ships to cross the Atlantic, and let's put one together and start a freight business." There's so much technology, and so many details to work out that are very, very hard to perceive when you first launch something like that.
Lowood: The computer system itself was just a conceptual design, or was there. . . .
ENGELBART: Oh no, they had equipment. They had magnetic tape cartridges; you could put the cartridges in and out to hold your data. It had a lot of good concepts in it, it just was implementation-wise ahead of its time. Without a lot more money than the inventor and everyone else realized at the outset, you couldn't proceed. We had no idea of the failure rate. There wasn't even any established practice for determining that. The designer who built it was trying to design some test for it. So you go look at that test, and you realize that it's a little bit optimistic, and full of holes. How long can we test it before the testing equipment breaks down? (laughter) Things like that.
Lowood: Do you remember who the inventor was?
ENGELBART: George Greene, I think it was.
Adams: Did he go on to do other work?
ENGELBART: Well, he was the irrepressible kind. Later when we were at SRI, he came over to see me. He had gone off and innovated some other prototype. You know, they were using paper tape quite a bit for business records at that time, before magnetic tape came out. He made a paper tape typewriter that as you typed it would punch paper tape. You could read the paper tape and drive the typewriter; instead of running with electric power, it ran with compressed air, or a vacuum. I can't remember which, but it was quieter.
Adams: A player piano, sort of.
ENGELBART: It was the pneumatic equivalent of solenoids and actuators; simpler, more reliable, and less noisy. So anyway, he had that product and he thought SRI would evaluate it for somebody else who was interested in buying it from him. I don't remember ever hearing about him after that.
Lowood: So at Marchant, you were an employee of the company, working on this project?
ENGELBART: No, I was hired as a consultant. Three of us working could support all four of us, and then the fourth guy was supposed to be developing business plans and hunting up product ideas. Then we'd all get together and spend Saturday going over this. It was kind of exciting, we were gonna start our own company, and get into the production business.
Lowood: So the four of you stuck together as Digital Techniques, then, and were separately doing this consulting work? Eventually it broke down, of course, and Digital Techniques came to an end. How did that happen?
ENGELBART: I just started having more and more doubt surfacing within me. This internal struggle about the vector I'd internally committed to about augmenting the world was still on standby, and pushed farther down the wait list. You know, before, the idea was we could capitalize upon the inventions I'd made, and I'd have enough money to do sort of what I wanted. But now, that was set aside, and we were just going to start to run a company. It really became apparent to me during that year that if you start a company, and start to run it, you have to make a one hundred percent commitment and be dedicated, to make that thing profitable. You just don't have time and energy to do anything else. So if you say, the real thing I'm trying to do is I'll get this going, and it'll be a money cow for me to do something else," it just doesn't work; you just can't distract yourself. At what point would you disengage and then go after the other goal? I was also realizing that I'm not very effective as a manager. It was hard for me to turn my attention to the business issues. That internal struggle grew, and I was having less and less success in sleeping at night, waking up all sweaty. One Sunday I was up at five o'clock, tossing and turning, and our kids hadn't yet started demanding our attention, three of the little babies, and my wife finally said, "Let's talk this out." So about an hour or an hour and a half later, I just decided, "Boy, this just can't go on. I just can't do this. I can't give up my dream, it's locked in." As soon as I figured the other people were up and about, I called them and said, "I'm pulling out." They just sounded terribly crushed. "But you guys can go on." They all came dashing over, and sat around looking terribly dejected, and said that no, they didn't want to go on without me.
Adams: How did the dynamics work between all of you? Something must have clicked that they wanted to keep a hold of. What do you think you supplied in that formula?
ENGELBART: I'd say that I was six years older than they were. I guess I never realized what their ages were. So I was older, and I guess the big brother.
Adams: And you wanted another role than that?
ENGELBART: No, it's that I didn't want to go off on this path, whether I was the big brother, or what. The augmentation dream was just welded in, internally, and I would have nightmares the rest of my life if I didn't go after it. I've faced that same issue a number of times since.
Lowood: Was the break clean at that point? Did the company end fairly quickly? What did you do next, and how did you end up at SRI?
ENGELBART: Well, I tried to think of how I could go do what I wanted to do. The university didn't offer a very attractive place at all at the time.
Lowood: Any particular university?
ENGELBART: No, no particular university. I realized that if you came in as junior faculty, your job was to prove yourself by research and publishing, if you wanted to progress to get to be an independent full professor with tenure. So that what you had to do was do that which would be viewed as good research. Talking with different older people around the university, it became clear that, whatever the department was, it's suicide if you think you're going off in some independent direction that isn't popular or acceptable. You'll always have the low level jobs, and probably in a few years they'll ask you to move on.
Adams: It's like the pressure of the company.
ENGELBART: Sure. Then I added it all up, what would I have to do there. Well, they wanted me to really grab hold of the computer program. The guy who started it had been moved to chairman of the department, and was really distracted.
Lowood: This is Berkeley?
ENGELBART: Yeah, so you know, they wanted me to push that. I'd have to commit a lot of time to that. Whenever I find myself teaching, I found, internally, this strong commitment to doing a really good job for the kids I was teaching. I'd put a lot of extra energy into that. So I said, "Well, let's see, that's two jobs there.
Lowood: Your classes at Berkeley, were they entirely electrical engineering, or did you have a chance to teach anything that would today be called computer science?
ENGELBART: Oh, yeah. Let's see, I guess before that, I'd been teaching all electrical engineering. I would take over a class once in a while from another professor.
Lowood: There was no opportunity through your teaching, really, to somehow swing into your developing ideas about augmentation and so forth? You never had a chance, say, through your students, to bounce the ideas off them?
ENGELBART: Oh no, my gosh. This is a time in which they were trying to get over their fear of the mystique of computers. No one had ever seen a computer working. They were trying to build one, but it was still tied up in "completionitis." (laughter) You know, it was a mystical thing, and didn't have the impetus by any means. It's pretty hard to reconstruct a picture of what it was like culturally and technically, in those days, about computers. Never mind this huge conviction about what role they would play. It wasn't something you could communicate rationally to anybody else.
Adams: People were looking at the technology of it all, of the nuts and bolts?
ENGELBART: Yeah, and the programming was so primitive. The first symbolic assembler emerged someplace in there. Other than that you'd have to write your program in absolute numbers. People were talking about compilers, but FORTRAN didn't come until later. So it was very difficult to write any kind of program and get it debugged. If you made any changes or inserted a few new instructions, before the symbolic assembler came, you'd have to go down and look at every place that had access, and you had to have the absolute number. But the absolute numbers would have changed. There were no debugging aids. Anyway, so the kind of thing I wanted to do looked extremely outlandish. It would be like when people were first trying to get any fixed wing planes off the ground, and you were talking about helicopters.
Lowood: So what brought you in, then, to SRI?
ENGELBART: Well, I said to myself, "Universities are out, where else?" I realized it would be very hard. One of the difficulties was selling somebody on supporting my ideas. How would you know what company you could go to, and hope that that management would be able to sell it? I realized if you go to a place like SRI, you have a chance to approach almost anybody in the world to put up money, to do it there. So it seemed like a much better platform to try to do it.
Lowood: Was SRI at that time, 1957, mentioned in the same breath as Rand, as a think tank? Did it have that kind of mystique to it, or was it considered more like a university?
ENGELBART: It had established itself quite well. It had been operating for about ten years.
Adams: In association with the university as a context for research and innovation?
ENGELBART: No, it never really was connected in that sense to the university. The university played a fairly significant role in establishing SRI, along with other industrial leaders around, who wanted someplace on the West Coast that would be interested in things that would support the growth of industry in the West. This was the rationale. The university at the outset said, "Well, what happens if all this folds?" They set it up so that if SRI were to fold, whatever value was left over would go to Stanford. The Stanford Board of Trustees was always set up as a way to guide SRI. The university said, "The trustees are the ones to appoint the board of directors for SRI." But that was about the only connection. Some of the board members were from a university environment, but most of them weren't.
Lowood: So did you go to SRI yourself, or were you contacted by someone there?
ENGELBART: No, I had to knock on the door, and say, "Hey," and wait. They had had a secret computer project going for some years, funded by the Bank of America, to build a banking support system. The project was called ERMA.
Adams: There's an ERMA mainframe emulator system now. I wonder if that's a legacy from that.
ENGELBART: I don't think so.
Lowood: This was Hew Crane's.
ENGELBART: No, he had come relatively recently in that role. Guys like Bialik, and Milt Adams, and Bart Cox, and quite a few people who have been in there to design, the mechanical and electrical engineers. Actually, they had a prototype, and it was just phasing out when I came down there. They weren't doing so well financially, and weren't actually out looking for people. Then I came, and I didn't find any immediate enthusiastic reception. They said, "Well, we'll let you know." So I hung on for months. It was three months or more I guess, before they said, "Okay." A day after I got there, they were talking about having to lay people off.
Lowood: Is that the project you started on, then, ERMA?
ENGELBART: Oh no, they just hired me, and I was sitting in a lab, and they said, "Why don't you sit there and get acquainted with everybody?"
Lowood: Which lab were you in?
ENGELBART: I guess I was physically sitting in the areas where people had been doing the ERMA.
Adams: Were there "leftovers" from that project, personnel who remained on?
ENGELBART: Oh, sure, a lot of people remained on, and a lot of contracts would come in. But if they lost a project, they would seldom carve out the people that were involved in that and fire them like aerospace sometimes does. They would try very hard to carry people on overhead for a long time, and try very hard to get them projects.
Lowood: So, there you were in the corner, in the lab. This was around 1957. What was your strategy for getting your augmentation ideas off the ground at SRI?
ENGELBART: I realized that what I'd really have to do is find a way to earn my room and board there, from some kind of support for projects. This was the assumption. One of the guys I talked to said I should keep quiet. He didn't think they'd hire me at all if they heard me saying these kinds of crazy things. I'd learned enough by then, through all the years, that my augmentation ideas don't grab anybody else like they grab me. I can't remember how many weeks went by; I would put my nose into a lot of things going on, and get the reports and read them, and ask questions. One project that was going on had been started by Hugh Crane, was this multiple aperture device, magnetic. They called them mads, M-A-D, multi aperture devices. This was a way you could take up a hunk of ferrite that's had a core, and mould it with some other holes around the circumference, small holes where the stuff was shaped around them. By pulsing this in clever ways, you could actually set information in it. Then you could transfer that information off to the next one, just with the pulses doing the transfer, without having to have any kind of amplifiers between them. So we were making shift registers out of that. They thought that since I'd been in the shift register game with my gas discHRGe device, I might be interested. The manager gave me these reports, and said, "Read them before you go see the guys, and tell us what you think." I read all those things through, and said, "Oh, I've gotta figure out how they work a bit better," so I made an equivalent circuit sketch, and started to look at it. I started thinking, "Well, why did they need to have multi-aperture devices? Why couldn't they do it with two sizes of simple cores? So I sketched out things that would do that, and it looked like it had the same equivalent circuit, so to speak. I knew enough about patent things to realize that if Hugh Crane and these guys were getting sponsored by somebody, and if I got involved over there, that would rightfully belong to the sponsor of the project. So I showed the managers this thing, and said, "If this pans out, it would have a separate support and patent area. What would you do about that?" Oh, goodness, they got all excited. We talked to scientists and lawyers before I talked to the other guys about it. This produced a problem, because if a new guy comes in and before he even talks to the other guys, he goes off to the managers with something else, that's not a very good way to make friends. I always felt uneasy about that.
Lowood: It sounds like you had acquired some savvy about patents, and some of the ways to go about things.
ENGELBART: Yeah, I'd learned a lot.
Lowood: Were those what are called single aperture devices?
ENGELBART: Yeah, SAD.
Adams: Is there a BAD?
ENGELBART: I don't know. Anyway, I had been working fairly closely with those guys for quite a while, and I got a number of patents in that area. Then through them, through Hugh Crane mostly, I would get invited to some of the professional meetings, or special meetings of people interested in computer devices. Within a couple of years, I'd really gotten pretty well established as a contributor in that area. Then I got interested in that scaling effect. I got some projects on my own, and after a few years started really trying to find support for doing what I really wanted to do. There was a time in there, working with magnetic devices, and it was really a very nice sort of time, where I was contributing usefully, and there were very bright, good guys I was working with. Hew and Dave Bennion, and Charlie Rosen came to SRI part way through my first year there. It turns out I had bumped into Charlie when I'd been recruited by General Electric. They were trying to recruit me, and I'd met him the time I'd gone back to New York at the Labs. We all got to be good friends socially, too.
Lowood: SRI must have been a little bit smaller then than it is now. Was it small enough that you were all within the same division of the same lab, or were you scattered around?
ENGELBART: We were within fifty yards of each other. There were wings in the facility.
[END OF TAPE 1, SIDE A][BEGINNING OF TAPE 1, SIDE B]
ENGELBART: There was quite a large area, including what's now a big parking lot, and some other buildings that had been a hospital during World War Two. In getting ready for the invasion of Japan, they assumed they'd have a whole bunch of us eager military guys brought back on stretchers, and they'd need lots of hospitals fast, so that had been built. Then the university took it over after the war, and had lots of post-war student types living there in dormitories. Then SRI got a chance to take it over, as their low-rent, initial facility.
Adams: Was it one of these Quonset hut type buildings, or two storey?
ENGELBART: No, most of them were one storey, quite sleazily built. It had one central hallway going down, and wings coming off. I think we would at least be in adjacent wings, they'd put a hallway over and down.
Lowood: In the earliest report I think I have, it looks like there's a systems engineering department; Roy Amara is listed as the manager in 1962. Did that structure exist?
Lowood: Okay. Would it have been called just "electrical engineering department" in the late 50s? Was it that general? Or did they already have a focus on computers, in any sense?
Electrical Engineering Programs at SRI
ENGELBART: Oh, there were other labs in-house that were EE, so I doubt if that was the name. The name probably had something to do with information. Our division was run by a man named Jerry Noie, who subsequently resigned and got on the staff of the University of Washington. His chief administrative guy was named Bob Wing, Robert Wing, and his assistant was named Jim Norton. He was a young guy, and they were shuffling papers, trying to help make things work. I think he came just a little after I did, or just a little before or something, and so he worked in that administrative area for the next eleven or twelve years before he joined me.
Adams: What kind of budget did you have, or the division which you were in?
ENGELBART: I don't have any idea of the division budget. They have to create a budget every year for planning, but they're not given a budget from up above. They have to go out and sell research projects in order to get the revenue to meet their budget.
Adams: You said when a project closes down at SRI, they were attentive to keeping people on. Could you describe the atmosphere, intellectually? You talked a little bit about the sharing of papers and reports.
ENGELBART: It was a good, stimulating atmosphere. There weren't many places you could go, at least in the West, where people were there because they wanted to research. We had some fun gathering around the coffee machine at break time, and we'd drink coffee and have all kinds of great time talking and arguing. Usually, someone had some ideas that they were going to go out with and solicit a proposal, and that would get beaten around a lot by everybody. Or if there was a request for a proposal coming in. . . . we all sort of knew what people up and down the hallway were doing.
Lowood: You mentioned Charlie Rosen, and talking to him. He's a person, at least now, who has a vision of the way things are going in the future. I presume that was similar at an earlier part of his career. Was there a certain mass of people there who were in a similar situation to yours, who could provide support to you, to keep plugging away?
ENGELBART: Most of the people would love to work on whatever was interesting, and that's missing in most academic settings; unfortunately, that's what I hear. I rarely heard people talking about the strategic value of it, in the discipline and in society. I'd interview a fresh Ph.D. and start asking questions like, "What do you think the strategically most important research factor in your discipline is?" And their jaw would drop, as if they'd never even heard the words, or something. So you began to wonder, what kind of job are their professors doing? Then you realize their professors went through without anybody ever challenging them or getting them to think about it. So unfortunately, a great part of the research community just doesn't make a practice of thinking about the strategic investment in their career.
Lowood: Did anyone there in that time period, from the mid-sixties, strike you as being a little more resonant with the kinds of things you were trying to do?
ENGELBART: There were a couple of younger people, people that would come to meetings I started. I can't think of anyone who would talk with me about my ideas, what I wanted to do, in a strategic way. The artificial intelligence field emerged slowly; Charlie wasn't a computer-oriented guy when he came, you know; he was physics-oriented. He'd been working tangentially. I'm not sure what he was first working on there. Then there was the guy who published something about the Perceptron; I think he'd invented it. I should remember his name, because I've just had so much empathy for him over the years. It got very popular, and a lot of people were talking about it. That's what turned Charlie on. He wanted to build one, so he asked around, and got some money to build one. That was the very first activity in A.I. He conjectured that this was a model of how the brain works; each perceptor is something that can have a number of stimuli coming into it at different levels, and when the sum of those levels got to be high enough, it would trigger it. What would trigger it depended upon its experience. So if the net result of this whole network was good, it would come back and do something positive about upping the growing thresholds of those things, depending upon whether or not they contributed. An algorithm works for that. They made a very reasonable model of how the neurons in the head work. Was his name Rosenblatt or something like that?
Lowood: Rosenblatt. I always confuse them. I always thought it was Charlie Rosen...it was Rosenblatt.
ENGELBART: He was at Cornell, I think, at the Aeronautical Labs.
Lowood: You mention how he got funding from ONR and you've described yourself as being in isolation. Your first funding came from the Air Force Office of Scientific Research (AFOSR). How did you get it?
ENGELBART: No, not my first funding.
Lowood: How did you break that isolation, to approach AFOSR? Or did they approach you? How did you actually get started on the path?
ENGELBART: It was probably some kind of happenstance meeting. But you needed to write proposals to them.
Lowood: Did they have a standing representative at SRI?
ENGELBART: No, they were just a little tiny program. I think their total budget was just one or two million dollars a year. One man and his assistant and a secretary or two in his office back in Washington, that was the informational or mathematical part of OSR. The guy's name was Harold Wooster. Probably it took personal contact with Harold Wooster. I might have met him at some meeting. I had gotten interested in information retrieval, so I went to some conferences, and I may have met him there.
Adams: Would the journal you kept at that time elucidate that?
ENGELBART: It may.
Lowood: Maybe you can check that when we are going over the draft. I'm assuming this is about 1960?
ENGELBART: I think I started getting some money from them in 1959.
Lowood: Were they the only people to have information retrieval systems at that time? (The Advanced Research Projects Agency, Department of Defense )ARPA didn't have the information retrieval until the 60s.
ENGELBART: There was the Information Processing Techniques office.
Lowood: That was a little bit later than 1959.
ENGELBART: Right. In late 1962 they said they were going to do it, and it was January 1963 when they opened for business.
Lowood: You didn't get money from ARPA then, but maybe a year or two after that?
ENGELBART: Actually, I got money starting as early as 1963.
Lowood: I see, so you are a founding member?
ENGELBART: I was standing at the door with this 1962 report and a proposal. I had met Licklider before and heard about him setting up a program, and I thought, "Oh boy, with all the things he's saying he wants to do, how can he refuse me?"
Lowood: So there was a two or three year period when the OSR was keeping you afloat. Were you able to work on it full-time, with their support?
ENGELBART: Not with their support alone. What they gave me was about half of what it took. They didn't pay me any overhead.
Adams: What were you doing during that time, and what kind of progress were you making?
ENGELBART: It was one thing to do the thinking. They didn't have any kind of money to build anything.
Adams: It was all conceptual.
ENGELBART: Yes. They'd give out little bits of money to all these different wild-haired guys. Somebody was once commenting on it, that maybe I should be embarrassed to be in the company of all these wild-haired guys because, obviously they were backing all the kooks, hoping that something would come of it.
Adams: Was that part of the reputation of SRI at that time?
ENGELBART: No. It didn't have anything to do with SRI, it was OSR. One guy was studying the way a certain kind of gnat would cluster. This guy was saying he was sure there was something about self-adapting organization in the gnat. So he would write a program that would simulate the behavior of the gnat. They would say, "Well, that's certainly something." On the face of it, it was sort of interesting, in the biology or sociology of gnats, or whatever.
Adams: But it's lacking the larger framework; how your research fits into a strategy?
ENGELBART: There'd be a lot of guys who were saying, "Hey, what you ought to research is how you can make computer work at all, these days." Or something like that, and so they'd look at something like that and laugh at it. So here I was being sponsored by people who were sponsoring people who were kooky. "Gosh, Doug, don't you realize the company you are in." Or one of my friends was saying about then, "You know, if people really get to know you, it's one thing. But otherwise, you sound just like all the other charlatans." You've got to just live with that if what you're trying to propose sounds so far out.
Lowood: I want to ask you a little bit about the 1962 report, but first, I wanted to ask you about this article published in Vistas and Information Handling, which is, I guess, pretty much a summary of the report. In the end you list a lot of projects that you've heard about, a bibliography--I'm sure Licklider's article is on there, Bush, of course. If you look at the report, the way it is divided up is: there's a background discussion, there's the long example of an augmentation environment, and then there's the proposal. It looks like it's really a conceptual proposal at that point, the result of these two or three years of funding from OSR. Was that time spent absorbing other people's work, as is suggested a little bit by this, or was this more of an inward kind of creative process whereby you wrote and thought in isolation and eventually came up with a report?
ENGELBART: It was more inward. I started trying to reach out to make connections in domains of interest and concerns out there that fit along the vector I was interested in. I went to the information retrieval people. I remember one instance when I went to the Ford Foundation's Center for Advanced Study in Social Sciences to see somebody who was there for a year, who was into information retrieval. We sat around. In fact, at coffee break, there were about five people sitting there. I was trying to explain what I wanted to do and one guy just kept telling me," You are just giving fancy names to information retrieval. Why do that? Why don't you just admit that it's information retrieval and get on with the rest of it and make it all work?" He was getting kind of nasty. The other guy was trying to get him to back off.
Lowood: In the report it is clear, with the 20-30 page thing with "Joe", that you're already talking about hypertext structured data--I know people had different terms to talk about that, in pointing to information with some sort of layering of the information behind that. I'm guessing that in these conversations with people from information retrieval, where the concepts would have come up, they considered that part of standard information retrieval?
ENGELBART: I don't know.
Lowood: Was it?
ENGELBART: Pretty soon I gave up on going out and I just was thinking and working to try to develop a consistent and meaningful framework of my own. I already felt it just wasn't getting any place. I'd make contact with the AI group, Newell Shaw Simon group, RAND, and tried to talk with Newell, but to them, if they'd get smart enough machines, why would you want to worry about building something special if the machines are so smart?
Lowood: So you found that both talking to these other people didn't necessarily bring anything to you but also at the same time--and this is what I was getting at with the question--they were re-framing what you were saying, and putting it into another framework. So let's say in computer science, just to give it a name, as it was developing in the period, it seems like there were lots of frameworks floating around, and different agencies of different sorts, and there was really a problem of people communicating.
ENGELBART: Back when you asked, was there a singular influence on me.... There was the realization that if you're trying to do something that is different and doesn't fit into other frameworks, that you build a framework and work within that, or it just gets distorted, until your work gets recognized as a different kind of area.
Lowood: What interests me about this is that somehow, given all the problems in communication and different disciplinary frameworks, you were somewhat successful with OSR, then ARPA, then ROLM and other agencies. There was somebody who either on faith or some kind of similarity of interests, I suppose, such as Licklider, was willing to find support for the project. What was it about those funding agencies, do you think, that moved them to be more receptive to your ideas?
[END OF TAPE ONE, SIDE B][BEGINNING OF TAPE 2, SIDE A]
Lowood: I was just asking about ARPA because there's a big history of ARPA being written now. It is generally held up as the best example since Sputnik of American funding of really frontier kinds of research, especially of computer science and artificial intelligence and things like that. What do you think, on the whole, from your experience?
ENGELBART: I think that is right. Dramatic things, such as time sharing would have taken a much longer time to get out if ARPA hadn't just made it happen. I don't know when IBM would have developed it. They didn't believe in it, and started only after it had been proven. And computer networks, obviously it was ARPA that made that really work. AI. Yes. They carried it for many, many years; something that still has to prove itself in the world, and that's no comment on whether or not it's going to. They pushed a lot on super high-speed computers, the ILIAC 4. I'm not sure that everything they funded really panned out, but that's too much to ask, in any case. In a larger sense, like that, they've certainly made a huge difference. They were essential for some kinds of things, where there's a sort of threshold to go over in the resource and momentum, to provide something until the rest of the society's means for exploring and research can take hold. You have to have something that lifts it up like that. The way of working was highly dependent upon the particular kind of people who were there, their intuition and judgment in what they would support. You just couldn't administer the general plan for doing it uniformly. It's not something that you can give to the average civil service guy.
Lowood: Maybe we should double back now and talk about management at a lower level, which would be the management of the project itself. In the '62 report it's clearly a conceptual proposal for further research. In January '63 you got the ARPA support. How did you start from zero? How did you get a lab, language compilers, and user interface? There was no computer in a sense. How did you go about putting the pieces together, beginning in 1963?
ENGELBART: SRI had a good environment for that in a way, in that there were a lot of skilled people, engineers and computer-oriented people and machine shops and procurement people, editors to help you write reports. It was all geared for you getting something in that you had to get done and finding people around that you could employ. When you would get a project you would come in and get assigned, essentially an in-house accounting, control project number. You could walk down the hall for many kinds of people or specialty shops and say, "This is what I'd like to have done", and they would get somebody. There would be a person who would just start using your cHRGe number on his cHRGe card. It was very straightforward.
Lowood: I guess what I was wondering was, about 1964, with the Mouse, you already start having concrete results. The project, as it is outlined in the report, is a pretty big thing, and it has a lot of facets to it. It has software parts, hardware parts, user interface, there are low-level compilers to develop higher level languages--all of these sorts of things need to be arrived at in some order. How did you identify, "Here's where we start. Here's who I should hire to do this" and those kinds of questions, in terms of stepping the project?
ENGELBART: Well, the first thing that comes to my mind is the term "bootstrapping" that I wrote about. You build something that you can use. You start using it and then you improve that and build a different one. That was the projection. I would start with text-processing by the computer that could help me do the text work. It composes reports and memos along the way, and software.
Adams: Was that an innovative approach to the way that other projects worked?
ENGELBART: I don't know of another instance which followed that as cleanly and directly, but there are lots of cases where other people would say, "I will do something just to demonstrate it." I would say, "Well, what do you learn by that?" They'd say, "Oh boy, you'll turn people on and you'll get more money." Somehow, that didn't seem right.
Lowood: You're also talking about more than just an editor to write the programs; you were interested in the text processing system being the system.
ENGELBART: That's something I knew would happen. There are lots of things I wanted to do on down stream. But where do you start? There was no way I could talk somebody into giving me an army of people to build some final system, nor would it have worked at all. So we started slowly. One of the advantages of the environment I described is that if you said, "I think it will only take two months but maybe five months, and I need the money," they'd say, "OK." You'd find a guy for two months and if you got five months from him, good. When you're done, he goes back on overhead and someone else picks him up for another project. It means that you don't have to commit yourself to "This is exactly the staff I'm going to need and I have to keep them forever." That's a very difficult thing to do. So the first thing we needed was a computer, and we looked around and found out that we could get one.
Lowood: What did you start with?
ENGELBART: The very first money from Licklider had some conditions that proved to be unworkable. He had decided he was going to put a lot of money into Systems Development Corporation, which was a system implementation government "slave"--RAND and System Development Corporation couldn't do outside contracting work. They had to get their money from the government. So they were essentially government laboratories except they weren't civil service. The government gave them money to do things. So SDC had been set up down in Santa Monica, as an off-shoot of RAND. It got much bigger than RAND, and was doing actual system implementations for the Air Force. There was a command and control computer that had just been made by somebody, I'm not quite sure who made it--some big outfit. It was a big computer with vacuum tubes and it was just a real super machine. SDC had one there in house that they were trying to do certain kinds of things with, and Licklider came in and had enough clout to say, "If you want that machine, what you have to do is build a time sharing operating system in it so that it can be a time sharing machine." He believed in time sharing and wanted people to do it. He went to SDC. It was a huge machine, a many million dollar machine. It was just acres of bays of vacuum tubes and stuff; very fast and very complex and sophisticated instruction set for the time.
Adams: He was able to convince them to try the time sharing.
ENGELBART: He had enough clout to tell them they had to make a time sharing system out of it. That's a big project and it took them a long time. Then what he told me was, "Here's what I want you to do: you've got to program what you want to do in their machine." I said, "Yeah, but it's not time sharing yet." He said, "It will be." So, I said, "What does that mean?" He said, "Well, it means you actually have to get the programs running down there." So that meant I'd have to go down there to do it. We had one programmer, I think, working for us at the time, and he would write his code up and work on it and go down there and get on the machine and try to debug it and come back here. Sometimes I'd go down. What they had for us was a little tiny display and a keyboard. This was a long way from the computer that was in a secure area. It had cables that ran down to the computer. It was extremely awkward. They would get into timesharing mode for a couple of hours a day and they were so flaky at that time that it would just crash and crash and crash. It was an absolute mess.
Lowood: So you had no interactive system at all. If you wanted to test something, or if you wanted to do text editing, you had to go to Santa Monica to do it.
ENGELBART: Well, you could do research on it. But it just didn't fit the bootstrapping thing at all. After a while he began to realize that wasn't going to work. We got interested in remote telecommunicating stuff. We had good engineers around us, so why don't we fix it up so that we can get a phone linkup and modems connected back there? We bought this small mini computer that Control Data Corporation was just beginning to put out the CDC 160-A. They were machines whose purpose had been to be a peripheral for the big machines, for handling input and output and peripherals. I think the machine cost over $l00,000. It was much slower and had very much less memory than an Apple II.
Adams: But that was the best you could get.
Lowood: The acoustic coupler that you used, was that the one that was developed at SRI?
ENGELBART: No, this was many years before that.
Lowood: Reid Anderson's company, Anderson-Jacobson used a link that was invented at SRI. I was wondering if that was the same.
ENGELBART: I had been at his laboratory. He was hired to take over the laboratory that I had been in and after a while I moved out. Roy Amara by that time had set up a separate lab and I moved into that one. I had trouble communicating with Reid. I have a feeling that the Institute quietly pursued Reid. He wasn't cut out to be a lab manager there.
Lowood: But those projects had no connection at the time.
ENGELBART: No. I never did know who it was who devised the modem. But anyway, we had a terminal and we were using it remotely. The money was going to come in, and it was quite a bit of money for that day, a couple of hundred thousand dollars contract. Roy Amara and his boss, Jerry Noe were really fairly troubled about me and what I wanted to do. By then I had been passed up at least once for a review. I had gotten the idea that they were trying to figure out whether to fire me or not. They very, very seldom fired anybody. So this was probably a very difficult situation. They just thought I was way too far out. Later I met somebody who had left the Institute, who had found a memo with my name on it that was sent to him, and he said, "Doug, maybe you would want this." He didn't notice on the back of it, somebody had probably sent it to him and it said something like, "Fred, have you read this? My God, we shouldn't let this guy out of the Institute. Bad name." I think it was a proposal for which we got money from ARPA, to start a multi-plant [client?] industrial sponsorship for an augmentation program. Roy Amara called me in and said, "Well, this project is going to come in and we ought to talk to you about what you want to do after that." I said, "What do you mean?" He said, "What we're going to do is take this fellow (John) and he's going to be project leader and he's going to hire the people that work and we're changing your title to..."--some different title that was brand new and had no meaning to it, "and you'll be over here." I said, "Well, wait a minute. The reason I have been working all this time on this is to do this work. I don't understand." He said, "We think you'll be good at writing some proposals."
Lowood: They had you pegged as a "blue sky" kind of figure.
ENGELBART: Not only that, but dangerous. It just almost makes me ill to think about it. He said, "Furthermore, John has got to have a chance to do it and I don't want you in there interfering." I said, "What do you mean interfering? What's all the ideas, the framework and everything else for if I'm not supposed to talk to anybody?" He said, "You can talk to them, but only when I'm in the room." I remember him calling me a couple times later after meetings. He said, "You were in that meeting, and you were trying to push people around too much." I said, "I was just trying to describe the picture." He said, "You just can't do that." He was bawling me out for trying to speak up about how it ought to be done. He was a guy who had just moved up very rapidly to Vice President a little later. He was experimenting with modes of management style. He was being very hard-nosed about it. It was just unbearable. I tried to figure where I could go and what I could do. I thought of going to MIT or somewhere. It was miserable.
Lowood: This was the way 1963 went by?
ENGELBART: This guy, John, who came on to manage this was John Wensley. He later left and went up to Oregon and founded a company to make computers. He was a good enough guy for this position, because he had come from England and had quite a bit of experience with computers and programming and was a good, persuasive talker.
Adams: I'm assuming he never sought you out?
ENGELBART: Right. He didn't understand what was going on. For instance, here's an example: you get on the screen and the way he fixed it up, he would get on and do some editing. So you want to delete a word, so you type "DELETE WORD;" that's the way he set it up. He said, "Computer languages are like that." I said, "Hey, with an interactive computer all you have to do is put in a "D" and a "W." "No," he says, "They've tried abbreviations and it doesn't work." I said, "That was without feedback." I just couldn't get any place with him. This was the kind of thing; apparently it made him angry too. I felt so low that day. He called me up at home and said, "That's so bad that I'd be afraid if my boss saw that, that he should fire me for ever letting things like that get funded." I said, "Yes." I was just trying to be loyal. He said, " I went and talked to your boss . . . ."
Adams: Your boss was Licklider?
ENGELBART: I told him that and he ripped me up and down over it. He said, "The best I can do for next year would be to give you enough so that you and one or two guys could do something. It would be about this much money and I really hope you can do something. But I can't continue. You ought to have your own machine and just get a start on what you picture." They were very subdued about it and said, "OK you can do it, but you have to pay for the computer."
Adams: Out of what source?
ENGELBART: The money they were going to give me. There wasn't enough money. So it looked like everything was just going to fall apart. After a year of that kind of horrible thing I was actually thinking I was going to get money, but not in the end. So, right about then, I got a phone call from a guy in NASA headquarters, a psychologist. I had been talking to him the year before, knocking on doors, and he acted sort of interested, but so had a lot of people. It turned out that he had been trying to find money. He was in control of some money and was going down to NASA at Langley Research Center, which is sort of a sister to Ames. They were doing computer things there. So he said, "I fixed it so that there's a condition on it. I'd give them this kind of funding for what they want to do if this extra money I gave them could come back to you. It's only $80,000 but it's a start." Well, that together with Licklider's would do it. So, I thought, "Ah!" The psychologist was named Robert Taylor. It was this interaction that we had that introduced it to ARPA. Then a little later he walked over and got a job at ARPA and transferred to Civil Service. We go back a long way.
Lowood: That's the source of the confusion, probably, that he started the ARPA funding.
ENGELBART: I'm not at all clear, but he's the only one that's still active around there. I don't know where that thing came from. If it came through Rheingold, it's sort of natural.
Adams: Then it was perpetuated.
Lowood: I've seen it in other places too, before that. So, now you are on an even keel.
ENGELBART: Starting over again.
Lowood: Where does somebody like Bill English come in? Was he already aboard?
ENGELBART: That was a year or so later. We had to start from scratch. We had a stand-alone computer with a display wired to it. I guess we still had that other display, the l60 that had been used as a terminal computer for telecommunications. So we had to start with that. This is how crude it was: there wasn't anything like a disk. We didn't have magnetic tape until somewhat later. So the only input/output storage medium was paper tape. If you started editing something, when you were ready to quit, you would punch it out, and wind it up. When you were ready to go back to work, you would first load the parent program and take the tape and it would load your text to start working on it. If it crashed, we'd have to go back and start all over again. So you would back up and punch out the tapes again. It was very difficult to work. It had a very small memory. I was trying to do things like: with the framework of a word wrap, let's just fill a whole line and if it breaks and comes over here, we can read it well enough, so you can just break anyplace in a word when it comes to the end. The eye just scans and reads it. So this conserved screen space to do it like that. It took extra effort to keep track of where the words were, and things like that. Licklider came to see us another time and we showed him this and he would not believe it. He said, "Oh, it's not smart enough to do word wrap." He wouldn't believe that we had done it intentionally. He would just not believe it. I said, "Why do you think I'd lie to you?"
Adams: You had documentation of the decisions you made in your research?
ENGELBART: It wasn't written down.
Adams: It was in your head.
ENGELBART: In that kind of discussion of an issue, you might as well go to court.
Adams: Yes. Once he has questioned the integrity of your research.
Lowood: What you were doing in that period was bootstrapping and what you were bootstrapping was an environment. You initially started with an environment for programmers who would develop the system. The initial system was to develop the system further. So you create an environment for the program.
ENGELBART: Not just the program, but the documentation for the project.
Lowood: I meant program in that other sense, not a computer program.
ENGELBART: Yes, that's right. So anyway, when we were working, it was fun. Word processors. Then we were finding ways to transfer some of the files to the bigger computer. I was building an off-line system as well as an on-line system. The off-line one had a paper tape and you printed on it and here was a draft. Then you would want to go through and mark it up with changes, where you could sit and punch. You would make another paper tape to describe the changes you wanted in it, and change description language that would go and tell where to change it. I had it all worked out logically so you could simply point to what you wanted changed in here and it wouldn't lose track of the first changes.
[END OF TAPE TWO, SIDE A][BEGINNING OF TAPE TWO, SIDE B]
ENGELBART: How we named systems? We had an on-line and an off-line system, and they would both be OLTS. So I said, "Well, wait a minute, they have different second letters, so I will call one FLTS and the other NLTS." Later, we dropped the T out so we had FLS and NLS and that's how the name NLS came for on-line systems. And later, the off line stuff got merged into our one system, and we just called it NLS. We kept that off-line capability clear up until 1968. It was very powerful. I could go through and mark up something and my secretary could type on the paper tape the description of how to change it in much less time than the best word processor does today, because the screen will change. She'd just get it great. She'd run it through and it would come out. Everyone laughed at it. We've got a feature in Augment where you can say, "Hey, freeze the screen, lock it in, don't change it every time I make a change until I tell you to update it." It just freezes the picture. You can go back a paragraph and make changes and go forward. The things you point at are still there. You get all the changes and then say "Refresh." Otherwise, you make a change and you have to get re-oriented. It turns out that the human factors in that work a lot better if you just change your perspective and train yourself a little bit. Anyway, we had the start of both of those things on that old machine.
[END OF INTERVIEW]