Douglas Engelbart

Interview 4, April 1, 1987


Adams: It's April 1, 1987; interviewing Douglas Engelbart, tape one.

Lowood: Were the major pieces of the NLS system already in place by 1969, and if not, what were the major lines of development in the early and mid-1970s?

ENGELBART: From '69 into '70 we were busy transferring it from the SDS 940 to the TDB 1010X. During that process we cleaned things up and improved the programming language that we were using. That became L10. One of the ways we re-wrote it was an historic first: we used a network. We had our 940 tied to the ARC net and we still didn't yet have our own 10X machine. But there was one running on the ARPANET connected into the University of Utah. We were writing our source code and transporting it across and compiling it on the 10X. We fixed our compiler so that it would output, running on the 940. The guys could sit there at the same workstations and just go through working with NLS to write their source code, get it transported, compiled, coded and debugged at the other end, moving back and forth like that. So that was the first real use of the network; we used it to interlink facilities. It stayed there and you could debug it in the 10X, because that was our only machine. We could start putting our modules in and running them to get it to work. We weren't running the whole system. Module by module we could get the basic things working and debug them to piece together the whole system. This saved us a great deal of time. It worked out beautifully. Even though there was a 10X that the AI group had just gotten in and was only 30 feet away from our 940, it was much better for us because we had this network connection to use, the one in Utah. So Larry Roberts, who was the principal guru of ARPA networking, was very pleased with that as a testimony of how things worked.

Lowood: With ARPA net, what was the network information center? Did it fit in with your development at all?

ENGELBART: Oh absolutely. That got launched the spring of 1967. It's a very clear memory in my mind. Before that particular meeting, I had been interacting with my group some, telling them about how the picture of augmentation I could see ahead would move. As we got our own NLS tools working, how were we going to learn more about it and involve other people? I was trying to tell them that the only way I could picture it was by developing a community of users that was distributed around the world. I had started thinking about how that could be promoted and arranged. The people out there probably wouldn't be equipped with the kind of tools we had, but they could start getting value from their use somehow, or from our using them and handing them the products that we could gradually build up. More and more of them could potentially transport the actual tools into their environment and get it to evolve like that. I didn't think you could just merely say, "Here it is." It's too big a transitional step for someone just to adopt a really radical, whole different set of working tools. It would be a step-by-step way of doing it. It just interested me anyway, the collaboration among distributed people. As a separate train of events, the ARPA office would get together its principal investigators at least once a year and sometimes more often. They would all have a "show and tell" session that would be quite fun. There were eight of us, and then ten and then twelve. By '67 there were 13 or so. We met in the spring of '67 at the University of Michigan in Ann Arbor with Bob Taylor, who was the director of that IPT office, and Larry Roberts, who had come down some months before from Cambridge to be his deputy. Larry came with the idea of networks; he had been experimenting with them at Lincoln Labs. They said they were going to start a research project on networks. So everybody listened and said, "That's all very nice." But then they got to a point where they were saying, "We're going to connect all you people together with a network." Now that was something different. Everybody started sitting up in their seats saying, "Well, damn, I'm doing this very important research in artificial intelligence or in time sharing systems or something. I don't want to fool around and waste time getting all involved and getting my people involved with networks."

Adams: They weren't making the connection that if people got involved in networks it could facilitate their own work.

ENGELBART: Not at that point. They just got alarmed about it colliding with their interests. It was a very interesting dialogue that went on. They were being told, "Look, you can share resources." All these new kinds of concepts were coming in. It takes everybody a while to adjust. Bob Taylor happened to mention networks to me some months before, I guess the summer before. I was thinking about all that and said, "Why would anyone want to do that?" I remember saying that. (laughs) About an hour later I was thinking, "Gosh, what a funny reaction on my part." Because with a little reflection and a talk with him, I realized what it could do and how it would fit into the community goals I'd been thinking of. So anyway, I was much more prepared when it happened. I didn't realize they were actually going to start a program, so I was as surprised as anyone else when they announced it. I can paraphrase some of the reactions among the principal investigators. "You can share resources," says the ARPA office. So investigator A turns to B and says, "What have you got that we could ever use?" And this is very insulting to B, because of course his research is so important to him. Investigator B turns to A--they're all quick-witted guys--and he says, "Well, don't you read my reports?" This gets A because, of course he doesn't read the other guy's reports! And he comes back very quickly, "Do you send them to me?" And this gets B thinking, because he doesn't know where they go. Then they both realize they have a common problem. They both turn to the ARPA office and say, "You've got to set up a library of all our reports and all the resources, so we have some place to turn to know what's available." This stops Bob Taylor and Larry cold, because the two of them and the one secretary are already way overworked; how where they going to do it? So for a while that flounders around and they slough off that topic and talk about the different technical opportunities. I sit there and think, "Damn, that's a marvelous opportunity. If I volunteered to form the library, there's a community. But if I go back home and tell my people that I have committed us to that, it would be a problem. We worked things out by consensus." But it just got more and more intriguing. Finally I volunteered, "Well, I'm interested in it. How about if I form an on-line library (I don't know if they called it the information center at the time) and run it for this community?" It ended up that my research was not that big a distraction. It interested me anyway and everybody was relieved I was doing it. Then it slowly got the ARPA office interested in this as something that was relevant to their own pursuit. It was three years before things were really operating, so in the interim I did a lot of thinking and planning. By the time things started in 1970, we had our Journal up, and the mail system, which for me was a big part of how you support the people in the system.

Lowood: Can you talk about the NLS Journal system and how it was tied in?

ENGELBART: The Journal system is part of the whole Augment, NLS integrated environment. Along with that was the first really comprehensive mail system that I know about. The Journal feature is just one of the options. The concept is of making a permanent record, as though you'd published something, which then is always available. It's given a publication identifier that means you can always retrieve it. The operations of the system and some of the software that supports the operations and archiving and cataloging and all of that, are built to support that. This was essentially in place by 1970. That's when the very first Journal item was started, probably in August 1970. As far as I know, it had a lot of the features that are only now emerging in modern electronic mail systems.

Adams: And this was the first of its kind? Both the mail and the Journal?

ENGELBART: The Journal was just an added capability. As far as I know, it was the first general purpose mail system like we have now. We just gave an identifier, and we didn't have to worry about the process after that. We only had one machine at the time, but later it got extended with no hassles. You just gave the identifier and which machine the other person is on. If he moves to a different machine, he tells the service operators and they just change the catalog. I sent the identifier catalog to a distribution list. You could send it to groups: you just set up the group, named the group and it was sent to all of those people.

Lowood: Did this have to be developed in conjunction with the developing ARPA net, so that all the conventions were correct?

ENGELBART: We just built that, and you could use all of that in our own environment. People would have had to log in to us in order to use it, because all of that was built on the structured files we had. They could do mixed text and graphics. You could transmit a whole document as part of the mail; I could mail you a whole document as easily as I could mail you a one-line message. All that was handled effortlessly and graphics could be embedded in it. I remember writing, in those days, that a message should be considered a document; you want them to be part of the knowledge base that you are working with. You could pick up any part of what you are doing and wrap it up effortlessly and send it to somebody. So there were lots of features in that mail system that didn't show up for years in anyone else's system. It wasn't until almost '72 or '73 that the very first popular little message systems started. BBN wrote it. It was all based upon sequential files and telegram-like messages zooming around and getting in your mailbox. You would deal with them interactively. This was something that everybody could implement on their machine; it didn't use our complex files and all of that. It disappointed me immensely. Well, I'll tell you the sequence of things. Almost everything that we built and designed through the late '60s and early '70s for the network ended up not being able to be used because we were assuming we were going to have a community of 15 people at 15-17 sites all doing research. That was our community. We were committed to support the whole community. By '71, Bob Taylor had left and Larry Roberts was the director of the office. The network was his big thing. He was having the usual difficulty in keeping people interested in the funding with something new. Congress didn't know what all this was, and this had become a pretty big budget by the time we had that network going in all the sites. He needed to get support from DOD. His solution was to let more and more different DOD sites get on the network. We'd keep hearing about them even after it was done and we were supposed to support them with information. As the number started to explode there was no way with our budget and the kinds of the things we doing, to give the service we had been planning. It pulled way down to something much simpler. There were periods before all of that happened where ARPA would tell us, "Gee, you've got all of these neat things leading to document development production; you could do that for everybody." We had a lot of really neat plans that ARPA was getting excited about. But when it came to budget time, this was outside the technology they could justify easily. Not everybody was interested in what we were doing. Besides, the AI people started hacking away at doing their own documents. Their lobby was very strong: "Oh, we do it better anyway than the dumb stuff over at ARC." It was very disappointing.

Adams: Was there any schedule for nodes set up in your network? Would that have been a solution, somewhat, to the funding?

ENGELBART: It could have been, but everybody was subsidized by ARPA by that time, so that wouldn't have mattered. Things like how the marketplace will get established on the networks, still had to evolve. It's there now in the commercial systems; you pay so much to get your machine connected and so much per packet or byte for the communications through it, but that's as far as it goes. Then different people tie in and say, "I'll cHRGe you so much for storage and so much to use my mainframe." Then that vendor has to pay the network costs as part of his cost of operating, so the user just pays Compuserve and Compuserve has to pay its network prices. But there's a lot more potential in the future for the marketplace--how you can set up so you can do something for someone and know that he'll pay you for your services. There is a quick way to negotiate it, to cHRGe and bill him. I wrote a paper out of frustration in the early 70's, called "Intellectual Applications", something like that, where I was outlining some of that. But anyway, back to the Journal itself. It was designed as one of the options when you mail: you could mail something either unrecorded or recorded. As I began looking at that and we were about ready to implement it, I realized it is very hard to set up a policy about when you would record it. I wired it in that everything would be recorded; no one had the option.

Adams: Was that a consensus decision?

ENGELBART: No, it wasn't at that time. That caused a lot of trouble and dissension. People felt outraged that they were researchers and that they were going to have to do a service business: to run a network information center. I tried to tell them, "Look, that's an immensely important exploratory act. It's a tremendous opportunity." That just didn't go over with them. Finally, some of the really good, supporting people in my group were going around saying, "Good day, it's better to give than to receive. It's better to give service." They wanted to support me. Some of them just got grumpy. There were a lot of problems through those periods. It was like the people who didn't want to fall off the top of the world with Columbus, or who see a very interesting island with Captain Cook and would all want stop there instead of going on. The business of exploratory pursuits has a lot more of that kind of problem than is generally realized. That could be an important thing historically, to piece together. Remember, all the time, if you are venturing into new territory, there are more hazards. The conflicts among people were a constant drag on my energy, perceptions of some of my people and some of the administration in SRI who were obdurate and pig-headed and wouldn't listen to anybody. They wouldn't know how much I already had given them and how much I was sweating because I had only done about 20% of what could have been done with my funding in that year. I'd figure out that if I'd yield this much, then we would get to do what I wanted, and if I don't, then we'd have so much dissension that we wouldn't get much done.

Lowood: You mentioned this the other day too, that you had to compromise various aspects of the Framework or what you intended to build into NLS, because of this resistance that you are describing, and some of these things didn't get done. Can you recall some of the things that didn't get done that you would liked to have gotten done?

ENGELBART: Yes, sure. In the Journal, in '69 when I laid out what it was, I had made a deal with a young graduate student named Dave Evans, who was from Australia. He had been working with me. He would get a hold of something and he would be excited about it; pretty soon he would get excited about something else, and pretty soon he had too many things on his plate. So I told him, "If you can, settle down and just pick one thing. Let's pick something you can do a thesis on and get that off your back. I want to do this Journal, so why don't you do the detailed design for the Journal?" He kept going off on so many directions with that too, that pretty soon what he wrote was a 500 page paper that talked about all kinds of collections of information and what the world needed, and had done no design. But anyway, in the design, besides cataloging and making use of our links to point to something, I wanted what were called back-links. If in the catalog on an item you recorded all of the links in that item that pointed to prior Journal items, any time you wanted to look at any older Journal item you could ask, "What among a certain corpus of documents is pointing at this passage?" It would be very important to move forward like that, to see what's pointing at it. The backlinks never got implemented. The other thing called Set System--I'm not sure--that is something I'm still just eager as hell to do; it wouldn't be that expensive to add it.

Lowood: What you were describing is like a citation index, in that it refers in both directions.

ENGELBART: Right. This could be something dynamic. As soon as you put something in the Journal that was pointing it at, it would "go". With a little bit more work and conventions about irregular Augment files, you could include files people were working on, like graphs, that hadn't yet been put in the Journal, if those people made them accessible to you. It is also in the corpus that if they were pointing to you, you could know about it too.

Adams: Could this be retroactive, could it go back to earlier entries?

ENGELBART: Yes, you can go back through the Journal. This just meant that you might have to tighten up the conventions, so that the computer could find out everything that is in an older file, that represents a link. We had flexible conventions so sometimes something that is parenthesized could come out and look almost like a link, pointing to garbage places. There was another thing called sets, a way of analyzing a whole bunch of recorded dialogue. For instance, you could get a handle on the set of all the passages relevant to a given issue. Two people interested in a set of issues that were related could get information on where they intersect. The professional service people could go into the files and do a better job of maintaining things. You can imagine in the academic community the kind of payoff this would have. Both of those would be relatively easy to add. It's the kind of thing I'd want to add, involving higher performance teams of support people who know how to carry out that kind of stuff. Those are examples. There are lots more.

Lowood: There's been a lot of discussion about command syntax in some kinds of environments that you and the Macintosh used; the operation entity order. That has become such an often-discussed point now. When you brought it into the system in the 60s, was that a thing that brought up a lot of discussions and resistance, or were those kinds of decisions pretty easy to get through?

ENGELBART: I don't remember them as being major discussions. There would be disagreements, but people recognized you have to do it one way or the other. I remember it evolving in small ways, but the whole thing about Verb/Noun started out very early and was never seriously challenged.

Lowood: So that was part of the foundation of the whole system.

ENGELBART: Yes. The assumption was that people could experiment with wholly different command and syntax if you gave them the funding to do it. But until we learned and got a significantly valuable set of functions we could do (and the methods of working with them), there wouldn't be much that would teach anyone. If you are going off in your own direction to build a system that has a kind of functionality you think is neat and the kind of interface you think is great, how do you compare interfaces? After we had this independent front end so that you could make different sorts of interfaces, I put a clamp on that too, until we got more stability and experience in usage and some agreement on functions. I didn't want it to be some trivial thing. I wanted to seriously use and test it under meaningful usage. We just never had the money. In '74-'75, there was an editor called Teco, one of the earlier ones that had lots of control, like EMAX at MIT. Here we were saying that we'd like to have our front end so they could even give somebody an interface like that. Even now, I'd love it if we could fix it so people could actually build whatever interfaces they like. Then you'd really start learning something about where they wore it down. I have a good feeling that at some point in the future, it'll be some kind of mix of some of the conventions that we've found so valuable, with the pull-down menus, the icons, and all that.

Lowood: I'd like to hear about another change, again in the early '70s. You gave up the terminology of augmentation, and began to use terms like high-performance knowledge workers, knowledge workshops and all of that. I guess you probably borrowed the general phrase from Peter Drucker. From your point of view, what did that evidence do? Was there a change in your thinking behind that?

ENGELBART: Oh, no. It was more finding terms that seemed better suited to what I really meant. Augmenting the intellect, as somebody told me in the early '60s, sounds so precious. I try to talk about effectiveness. Even in that '62 paper, the first page of it was talking about being able to be a lot more effective.

[END OF TAPE ONE, SIDE B][BEGIN TAPE TWO, SIDE A]

Lowood: I'm more interested in ARPA's reaction to projects that they're funding moving into service situations.

ENGELBART: They didn't fund any of that.

Lowood: Had your funding stopped?

ENGELBART: Oh, no. But when we were negotiating that, Larry Roberts was still the director, and he thought that was a great idea. We actually could serve people out there and get moving. He juggled funding to get us started. It was like a cash advance that we could not otherwise have. The people that succeeded him accused us of not being interested in transferring our technology into the world.

Lowood: This was under Licklider?

ENGELBART: Yes. As a matter of fact, he got into this. He was just convinced that: a) we were ferociously overcHRGing everybody; b) he just didn't see that as a technology transfer; and c) we had too many people out there supporting and training. This was an admission of the weakness of our system, that we couldn't teach the people how to use it.

Lowood: So at that point then, you had the organizational problems you described; you had SRI with the accounting problems, and, after Licklider was back on board, you also had problems with your funders. How did you go about resolving that?

ENGELBART: We resolved that by dying. It collapsed.

Lowood: It collapsed before it was sold, or . . .?

ENGELBART: The actual sale took place about fourteen months after the day I was called into the division manager's office and told that I was being replaced. They were going to turn ARC into the kind of organization "it ought to be."

Lowood: The research direction or the service direction?

ENGELBART: Oh, research. I said, "How can you beat the system to get your funding, your goals?" My people just didn't even understand what he was talking about and started leaving in droves.

Adams: Who took on the direction of the project at that point?

ENGELBART: Bert Raphael.

Adams: Was he an insider?

ENGELBART: Yes, he'd been head of the AI Lab.

Adams: Did he have his own agenda or was he given directions: "This is what you do"?

ENGELBART: He just would not believe me for a minute about the whole system concept, why we were doing it. There was something he wrote once about NLS being glacial, non-changing, that we hadn't done anything new since the sixties. Here we had all these things we were talking about, plus the architecture which was so innovative and all of that. And he would just not listen to that. I'd say, "We've gotten these composite files," and he didn't believe in that. You could do the same thing with MSG and a line-oriented text editor, he felt. He had totally different concepts of the way things ought to run, what we were building. It was absolutely devastating.

Adams: How many people stayed on, and in what key positions?

ENGELBART: Most of the people in key positions left. There were quite a few people though, who felt really committed to it. As long as there was a chance of keeping all that alive, they really wanted to work on it. It was in November, this hit me on the head. That was two months after our house had burnt down, in 1976.

ENGELBART: In '77, I was hanging on to try to see if there was some way to save all of it. At home, we were trying to dig out from under that fire that was so totally devastating.

Lowood: What was your position at SRI?

ENGELBART: Well, that's what I asked them the day that they collared me. I said, "Well, what am I supposed to do?" They said, "You're innovative, start inventing some things." I said, "But this is the whole vector." One time I said during that terrible interview, "I'm really proud of what we built here and everything." The look both the division head and the vice-president he reported to, gave me told me that they were totally convinced that I was deluded about that, that I'd gone off on some really bad, barren approach. We stalled, we weren't doing anything creative anymore, that in fact I'd been cheating and doing all kinds of bad things. It was just astounding, the kind of feeling that suddenly you get when you realize that people aren't going to believe you and they stop listening to you. The next year was loaded with that kind of feeling. Finally, about March '77, within NLS, which had commercial rights in spite of it all being developed by the government, they realized that they probably had something of some value, that they could sell and let somebody else take it over. That really interested them, so they started doing that. They distrusted me so much that if somebody came who was interested in buying it, they wouldn't let me talk to that person alone.

Adams: Between the new director and the sale, what happened in the project? Did it produce anything?

ENGELBART: No, development dissipated very rapidly. About a year later there were fifteen of our people at Xerox PARC. There had been a few who had left before that.

Adams: But you stayed there to try to manage the resources?

ENGELBART: Well, yeah. If I'd thought I could go someplace like a university or Xerox or someplace and resume the kind of significant work that I felt I'd been doing, I would have jumped at it. But boy, there was no place. Everybody seemed to reflect the same kind of thing, "Oh no, that's way too complex. There were some interesting things that you did in the sixties, but . . . ." It was so negative. Nobody was interested in talking about the idea of Augmentation. At Xerox I didn't feel welcome. The personal feelings between most of the guys who had left to go over there were fine, but they were all so excited about the fact that they were getting each an Alto on their desk, and get powerful response. They thought that was more fun and faster than what we were doing. One guy said, "Well, we've got all this technology now. We don't see any need for an ARC anymore, because we can do it all so much better with the technology that we've got here."

Lowood: You were talking about the impact of the new processors. Was there any contact between the people around you and yourself with the microcomputer revolution? I'm thinking of the people like Homebrew Computer Group and Jim Warren, maybe even somebody like Ted Nelson. Did all of the ideas that were coming up in the mid-70s about people's computing have any impact on you?

ENGELBART: Well, some of it did, but it would have so far to go. If you look at what could it do for lots of individuals, you could realize that even if they're the small capacity machines that they had in those days, that you could employ parts of our system into them and it would have been useful. But the whole vector I was on was saying, "How are you going to learn about how much value you can get for serious knowledge work and for community support functions?" There's just no way those microcomputers in those days could do it. We could learn so much in the environment we were in. That's what I was trying to do, learn how you get value. I wasn't trying to get it out to everybody.

Lowood: Was there any personal interaction that was surfacing?

ENGELBART: Yeah. We got acquainted with Bob Albrecht, who had two or three little companies going all the time he was publishing books. There was a Portola Institute on Santa Cruz Avenue in Menlo Park. Albrecht was part of that before he split off to do his own thing. I'm trying to remember the name of the guy that knocked himself out for years keeping Portola Institute alive. But anyway, Stuart Brand was part of Portola. It was a genesis of very interesting things, like the Briar patch stores came out of that. My kids even went over and sat in on Albrecht's evening classes.

Lowood: At the People's Computing Company?

ENGELBART: Someplace like that. He was still doing that at Portola I guess. Then he was doing work with big desk top calculators and other things that HP got into. I remember having this class come over to my lab sometimes, and he got them playing on our system.

Lowood: The reason I asked that question was your comment about the Alto. I was wondering whether part of the reason for some of the skepticism about your project was rooted in the feeling that there were new technologies coming along that were going to be replacing the whole foundation of timesharing, networking and all of that, that everyone would just have their own microcomputer.

ENGELBART: Well, I think the reason we got cut was not that. But one of the reasons was that there was an element of contention in our lab. A lot of those guys kept wanting me to put the system into a mini. I kept saying, "I know you can, but that's a totally different trip from where I'm going." If we cram ourselves in those little spaces, we'd have to give up a whole bunch of what we're trying to learn. A lot of those guys never did see that, and don't to this day. It was like something in Howard Rheingold's book towards the end of the chapter about our lab: "It's too bad that Engelbart kept hanging on to that old technology." It totally missed the point of what I was trying to hang on to. Still, you just could not have an integrated environment like we have, in those days. Nowadays the PCs are big enough (in the last five years or so they've been big enough) so we could build all this in their environment too.

Adams: I don't want to get us out of the chronology, but is that now something that is interesting to you?

ENGELBART: Oh, sure. Absolutely. We started making the plans about '76 for transporting into a different environment. We wanted to keep it modular so we could, as time went on, transport more and more of it. Around '73, we actually took a microprocessor and built the box for it. The very first four-bit Intel microprocessor was built into what we called a line processor. That too was a big innovation in the world. You put that between the modem and a very dumb terminal and it would give you all its windowing and everything in the dumb terminal. The mouse and keyset support both plugged into this line processor. You could also plug your printer into it, and stuff could come down a line and it would separate what goes to the printer from your display.

Lowood: This was the 4004?

ENGELBART: Yes. This is the decision I made about how we were going to start supporting people out there. Well, one of those I wanted to support was that we could display NLS."We made these things and gave them to our customers, ARPA and that was really interesting. When some of our customers and our users' customers started using display NLS, the traffic back and forth became much different. It revealed a bug, a flaw, in the design of the protocols that they built into the earlier ARPANET. You send out something very long to the guy at the other end and this long packet, they wait, the little mode that you're talking to here waits until it hears back from that mode that it has delivered it before it lets another packet get in here. We were sending stuff like that and waiting like that, our stuff backed up like crazy! Our people looked at our timesharing systems and said, "There's nothing wrong with it. It's the network." This created a terrible crisis for weeks.

Adams: How did you resolve it?

ENGELBART: They had to change the protocol.

Adams: It accepted what was sent as it flowed through rather than wait for it to complete?

ENGELBART: I can't remember the details. I made the commitment of saying, "We're going to make ourselves live like the people have to live out there. So we're going to move to that same kind of display for our own work." My people said, "Give up our beautiful video with the graphics?" "Yup." That was a hard decision, but it was on principle. So we lost our graphics and a lot of the stuff, that we never quite could get back in again.

Lowood: Well, first just one clarification. ARC doesn't exist today except for the part that was sold to Tymshare. Did any continue at SRI?

ENGELBART: The only thing we left there was Network Information Center. That had grown pretty big and had specialized.

Lowood: The sale to Tymshare happened in late '77?

ENGELBART: I think January 1st, '78 was the official transfer of ownership.

Lowood: Could you describe what kind of company Tymshare was; what it was like to move there and what changes it meant for you?

ENGELBART: Another thing of historical note was that in July of '77, SRI terminated me, fired me, instead of waiting like they did with the rest of the crew. They singled out Jim Norton and me and fired us, saying that we were in conflict of interest because we were trying to see if we could find backers that would buy us or get venture capital to buy it ourselves. But that didn't get very far at all, and we couldn't. There were some terrible scenes with the remaining people deciding that the whole system wouldn't be of use to anybody without the whole crew. They got together and decided that if we sell it, here's how we would divide it all into equal shares. They were really dictating a whole bunch of terms that would change the nature of the group immensely. Anyway, the transition to Tymshare was a relief, because everything didn't died; we could continue. The interaction was mostly then with this guy Laszlo Racozzy who was one of the four group vice-presidents in the executive staff at Tymshare. He was in cHRGe of the technology side. He was pretty flamboyant, but later he created a hell of a lot more problems for us than he resolved. It was exciting to go on and change. They needed all the people to come to make it run like a business. Their idea was that this is a very cheap, low cost way for them to get into office automation. They formed a new thing they called the Office Automation division, and that was us. I wasn't specifically invited to come along at all, but I didn't have any place to go. I said I'd been too burned with trying to be boss and manager and I never wanted to be that anymore. It turned out no one asked me. What I began to realize later is that they wouldn't have considered me for a second, because I came over with all of this trouble with the SRI management. I didn't realize that for quite a while. All the rest of my stay in Tymshare was clouded and people who successfully managed a group like that would go to great lengths to keep me from talking to customers. This became clear.

Lowood: How big a company was Tymshare then?

ENGELBART: Oh, two hundred and fifty million a year or something like that.

Lowood: So the Augment group was certainly not half the company.

ENGELBART: Oh, it was just a tiny little thing over in the corner.

Adams: How many of your staff were in your old group? What was the total staff?

ENGELBART: Twenty came, out of the forty-six we had before.

Adams: Now, you were out of the project but you negotiated with Tymshare nevertheless?

ENGELBART: I thought I would be a fairly important figure in all of that negotiation, but as it turned out, no, in fact the Tymshare guy would always turn to the other people to ask questions. It was just weird. Most of the top level development guys had left. The guy who was current manager of development had been several levels down and had not been a terribly critical part in design decisions. But if there was anything about the technical nature of it, they would turn and ask this guy, not me.

Lowood: Were you brought in as a research scientist or something like that, or did you develop that role for yourself?

ENGELBART: I was given the name `Senior Scientist' then, which I still have.

Adams: Were you under contract that specified your duties?

ENGELBART: No, I was just brought in as an employee, and for a while I guess my salary was part of the salary billed in the office automation division. After a while I realized they were not giving me any role. I thought I could help a lot with the technical planning but the guys who were in cHRGe of the technical planning didn't want to talk to me. I did help do some things, like trying to get a productivity program going, almost all for development people. Because he wanted to do that, he just asked me to help him. I ended up organizing all of the development centers and getting some things moved in that could have been very productive, but pretty soon I began to realize the reasons why that was selective productivity. It was just to look good; the actual increase in productivity didn't matter a damn bit to the guy. It was just all what it looked like, that he had a program like that going. It turned out that form was very appealing to a lot of people and the president of the company thought that was a neat idea. So for a while that was all very good, but then we'd get into some things that required funds for the company to do something about it. Oh, he'd promise them, but that never materialized. It was a real lesson for me.

Lowood: You were continuing to get some company papers and so forth . . . ?

ENGELBART: Yeah, but that was hard too, because I just felt, "Who's listening?" Just so terrible. Friends of mine would encourage me to come out and say, "Look, I'm running this session here, you give a talk, huh?" Then I'd write a paper. But if it weren't for that, I probably would have just sat in a hole all the time, because you can hardly imagine what it's like, feeling that nobody wants to listen, that you've already been dealt within their minds as not being of consequence. It's really amazing.

Lowood: What happened to Augment during this time? Did it evolve in any way?

ENGELBART: Yes, it was totally evolved with some interesting things that were added. I kept digging away at it, bringing ideas in and trying. Somehow, I never died. At one time I made a really serious assessment of whether I should launch off into a new career path, because this one was so unproductive as far as income went. I've had about two raises in the last ten years, and still my salary is less than what some of the other guys have been earning. One guy was in cHRGe of four or five development programmers, and he was earning more than I was.

Adams: How long did this situation continue?

ENGELBART: It's still the situation. It's the strange situation McDonnell-Douglas is in spite of all the interest now that's gone clear up to corporate vice-presidents and the people that report to the chairman of the board. They were all talking about a whole bunch of these things, and that got excitement reverberating across them and into DEC, but I'm just hanging on by a thread. The last time different people got fired, there was no place for me to be tied to anymore in this information systems group, so some guy up in the marketing staff took me on as long as it doesn't cost him money - if I get my money from the aerospace side. He's not about to give me my review, and the aerospace side is falling all around us. They are crossing their fingers and saying, "Out of all this excitement in the world about what we did, there's still a chance our work can be reharnessed as a prototype."

Adams: If you can keep working on it . . . .

ENGELBART: You can save it. It still works; month after month that's been more clear to me. It still looks like there's a chance we'll make it.

Lowood: Augment is still marketed as a service, isn't it?

ENGELBART: That's totally dead-end if they can't get money to really transport it into other systems; it can only run on this machine that Digital Equipment doesn't make any longer. It would probably cost five or ten million dollars to do a good job of transporting it. It's a big system. There was a thrilling chance; a major customer for the last six years has been an Air Force communication command that was looking for some way to automate their organization. These are command headquarters people on an air base in Illinois about twenty miles from St. Louis. The idea has been around for a long time and they have a lot of organizational support. We had some very novel things, like a signature capability: actually using your very secret password to process the signing of a document. That process uses modern encryption techniques. It takes your secret password and combines it with the check-sum derived by scanning the passages to be signed. Then a public key can check it to see if that portion of the file is the same as what you signed. It approves if it is the signature. But if the file had been changed one bit, it would come back and say, "Doesn't check. It's either not your signature or the file has been changed. Then there is also the Journal function. Every government organization has to keep records in the archives. It's a huge job; tons of paper have to be moved and kept. They'd just gotten an okay formally that this mode, with the Journal and signature could be used as an exploratory system to send magnetic tapes instead of the paper.

Adams: Is that still being developed as a possibility?

ENGELBART: The Air Force said, "Alright. We like this design and with organizational support, we want to implement it widely." So they were going to get a request for a proposal out. Over a year ago we were still building it, and it looked like it was going to be wired just from the Augment base system. But the McDonnell-Douglas people, who were trying to assess whether or not to keep Augment, went out to check on that. They were so totally wishy-washy about whether McDonnell-Douglas was going to support Augment, that the General in the meeting just stomped out. He said, "I'll be damned if I'm going to risk putting out a RFP [Request for Proposal] based on something that doesn't have that much corporate support behind it." And no kidding, that RFP came out a couple of months ago, totally rewritten.

[END OF TAPE TWO, SIDE A][BEGIN TAPE TWO, SIDE B]

ENGELBART: It has been like that year after year after year.

Lowood: What is the system called in this period? I'm sure it's not being called Augment anymore. What was being tossed around?

ENGELBART: You mean within McDonnell-Douglas?

Lowood: I mean within these negotiations with the Air Force.

ENGELBART: Oh, the Air Force buys it as Augment, but in the Air Force, when they write releases and flyers, it's their Office Information System that they've developed. They don't even mention Augment.

Adams: Interesting.

ENGELBART: We've got two guys, Norton and Don Young. Don Young was a top master-sergeant or whatever it is Air Force that was involved in their first planning. Before he retired from the Air Force a while ago, he had been our chief guy.

Lowood: That was the path A. Another path I wonder about is if there's some capacity for an internal development of this system as a support for McDonnell-Douglas itself.

ENGELBART: Well, one of the problems now is the fact that it's an integrated system. If they're already using IBM hardware for their computer-aided design, or Hewlett-Packard, or DEC, and then you bring up something that's so totally different - even though it was built to go in there and integrate - it's very hard. I did sell a pilot program for use in the community sense. They've got all kinds of standing committees and important people involved, the corporate vice-president for engineering has engineering vice-presidents. They all have projects in a matrix he has to keep track of. It's still a very old-fashioned interaction. There'd be lots of places that are potentials within the corporation. But I kept saying to them that you'd have to invest in this at the corporate level as a strategic thing; you can't go and see whether those guys want to buy it now, because how do they know? So finally that got sold to the gigantic sum of two hundred thousand a year, which may seem a lot, but on the network it was just peanuts.

Adams: What kind of connections would the pilot have?

ENGELBART: Well, that was finally launched about a year and a week ago. This was to connect people who were interested in AI technology as applicable to internal things, like AI support for machine automation. From our experience, it seemed like a good idea. We trained a bunch of them, working like crazy. I would have been able to staff it if they hadn't made a final cut. The ASD thing cut our last, real experienced field people. I just got the funds in time to take them on, and now they're being threatened with being cut off.

Adams: Are you the only support person then, for that existing system?

ENGELBART: No. I think it was really a failure. We trained people and got all excited about it. They'd go back to do the work but it wasn't like around a university or ARPA, where almost everybody has access to the system. It's very hard for them to get on line and do much interacting. They get excited about it, and their manager keeps putting more pressure on them to get their projects done. He says, "Look, we're not in the business of collaborating. We've now got this project going." The biggest impact seemed to be at that level.

Adams: What seems to keep coming up is the difficulty in this environment of researchers to do collaborative work, to see that sharing resources doesn't take substantial time away from your own research task, but rather augments just a basic problem.

ENGELBART: Well, the only way it's going to start working is in some situation where you can sell the support for this as an exploratory thing and not try to make the people pay for it out of their own pockets. You've got to be able to get into their managers, their organization and make arrangements. There are situations where people are desperate to collaborate. There was a tremendously exciting conference two weeks ago in Dallas, called "Informatics International Access" but it's Third World countries who desperately are interested in the telecommunications potential for them to get access to a lot of information they need for all sorts of things, especially health information. They found out about what Augment can do; it's so much more than conferences and bulletin boards and those other sort of things. They had ideas of building a community of interest that's active in participating in building, developing and collaborating on a lot of these issues. That seemed to turn enough people on that they were hoping to write a proposal to do a pilot project. We have to get some philanthropic support back in our world in order to get the energies and capabilities for some exciting matching and adapting. Who can they approach that's going to understand their whole meaning? Yesterday I was in a really interesting meeting with people laying out the plans for a world center, in the city. You guys remember enough history to know that the United Nations was essentially born in San Francisco. I happened to be sitting on Treasure Island as a "countrified swabby" [sailor] that springtime. This world center is trying to design its different segments by enlisting committees from people around the community who have expertise. It just happened that I got approached about a communications facility they wanted, and they had the picture of dishes pointing to satellites and providing that kind of intercommunications. So I start telling them about this kind of networking. Yesterday I was in a meeting with the four of the volunteers in this committee. The potential there is really high. I went to Singapore two years ago, to give a week's lectures there. I demonstrated by connecting Dwayne Stone in Washington, D.C. It took me a couple of days to get the PC working for their network, but I could actually demonstrate the new projector. People can watch my screen, and when I called up Duane Stone, his voice came out loudly. Pretty soon the screen cleared and there was his screen, with me talking and showing people what was on it. "What time is it there, Duane?" I asked. There's twelve hours difference. The demonstration had a very high impact. They decided they want to be the software brain trust of the southeast Pacific region, and they have to import people - it's extremely expensive. So talking to them about that kind of collaborative work was very exciting. I was in Norway and Sweden, small countries that still have to communicate with the world and tie-in to other systems. But you see that huge potential. How does it coalesce? It's frustrating.

Lowood: There was a proposal that you did for Stanford . . . .

ENGELBART: Oh yes. It was probably in the late 70s.

Lowood: Could you talk about that a little bit? I think it would be interesting to know why that didn't get off the ground.

ENGELBART: It turned out that there was no money behind it at Stanford. They put out a request for information that was just a way to stimulate interest. It wasn't a request for a proposal. It was a request for information. But a lot of people got excited. Xerox made them a pitch, and they just didn't have any dollars behind it.

Lowood: What information were they seeking?

ENGELBART: Well, just information about what could your product or service do to support their needs, and they have nicely developed needs for how many people around there are preparing documents all the time, lecture notes, research papers; how much communications they need, how much it could save. It was very well thought out. It was just exactly the kind of community thing that we'd been designing for a very long time.

Lowood: But it was a false lead.

ENGELBART: Yeah. I'm trying to remember the name of the guy in the computer service area that put out the request. He's still interested in a lot of these things so we started getting together informally. I was thinking of what could have gotten started. One day I said, "Well, the thing I dream of is a center; I'd call it the Center for High Performing Scholarship." And he wrote some memos and "think pieces." But it really names it. He tried to raise interest among at least one, but hopefully more, really senior faculty so that it could gain momentum. We'd get different people in, and they'd listen for a while, and then they'd start saying, "Oh boy, then this is what we could do . . . .," and I realized that the perceptions of everybody about what I was talking about were so different, that if we did get together in some syndicated way and create it, it just wouldn't be what I'd been waiting to do.

Lowood: There's another related professional activity, which I guess fits in with this. There was a National Academy of Sciences committee that you were on that would look into research libraries? How was that tied in? What was going on in libraries at the time that you got involved?

ENGELBART: Well, it could have turned into something exciting. I'm always an optimist. The only way I could go on all of those years was by hoping. I wasn't going to forget that. It's like a quarterback getting sacked every play, play after play, after play. It's the way I feel sometimes. No offensive line to watch me, and you kept hoping. How I could keep getting up and doing it over again is just some crazy mania.

Adams: But you do.

ENGELBART: Yeah. The committee originated inside the Academy of Sciences. It was already started, so they went out to recruit people. Their objective was to try to make a study of the state of research libraries, and the potential for computer support to improve their effectiveness. They would write a report on what was needed or could be done. They also wanted to appeal to people interested in funding or promoting things from the library side. For the first year we visited exciting research libraries - the Library of Congress, Harvard, MIT, Bell Labs, Berkeley and Stanford. It was the predecessor of Socrates, Ballots, RLG. Another idea was the generic file management of data.

Lowood: SPIRES.

ENGELBART: SPIRES. Berkeley had some interesting guides. It was in the late 60s.

Lowood: Oh, that early - Ballots wasn't even up yet.

ENGELBART: That's right. None of those were up yet.

Lowood: Ballots was really one of the earliest, as far as I know, of the academic libraries that automated. So this committee could have actually played some sort of seminal role.

ENGELBART: Yeah, someplace I have the report that committee published, and that report did not have the sort of minority report attached to it that I wanted to produce. I just gave up because I'd begun to realize the kind of politics involved. From the several years I'd already spent working on the network information center concepts and ideas, I just knew that networks were going to evolve and make a huge impact upon the economics. You don't have to think about putting a time-sharing mainframe into every library. Some of them were the maintenance of data bases. It would change the nature of the service potentials and the economics immensely. The committee wouldn't put anything about networks in the report. "Networks are just a gimmick that ARPA is playing with on the side," they said. He just wouldn't have any discussion on that issue.

Lowood: Well, you were vindicated on that one, because it went very early into the network.

ENGELBART: Oh, sure.

Adams: Who was on the committee?

ENGELBART: I'll need to find the report.

Lowood: We can cite the report in the transcript.

ENGELBART: Anyway, I enjoyed the visits to those sites a great deal and the subsequent contact. The very first paper I ever gave, in 1959, was to the Record Documentation Institute. It seemed like that's where I'd find the most relevance. I tried looking at other communities.

Adams: Did any of the research libraries you visited set up, subsequently, a program that you would have approved of?

ENGELBART: I don't know.

Adams: Did you get involved in any actual arrangements?

ENGELBART: No. I was still too radical. When a university tries to get the money, the people who review it ask their friends' opinions. If it's a wild idea, they're not going to risk their own reputation investing money on something that's crazy.

Lowood: Did you have much contact with the computer science or electrical engineering people at Stanford at all?

ENGELBART: Off and on.

Lowood: Where there any relationships that you developed?

ENGELBART: Well, I got to be friends with George Forsythe, and he just loved one of my game simulations. I was invited to an early Applied Mathematics and Computers Society dinner meeting in 1958. It was a hilarious thing, because there were some people who had been drinking and especially one guy was boisterous. I hadn't realized that he was pretty loaded. He did one of my simulation things on debugging. I had him get inside the concept of the computer, and I said, "You look around. You have to figure out whether you can correct it or you have to replace the elements." George and his wife Sandra were both there. George, in one of his programming classes, would stop part way through and as an exercise have them write a computer program that would simulate the people that simulated the computer, to see if they would come up with the right sequence. Once he mailed me an actual printout.

Lowood: Yes, we have that in your papers.

ENGELBART: Then I remember going in one time to talk to a colloquia in computer science. I was talking about interactions, about Augmenting and what we were building. George said, "Well Doug, I'm really puzzled. I can't imagine making use of any faster turnaround than twenty minutes when I'm programming." He was just very friendly about it, but he just couldn't grasp it.

Lowood: But his area was numerical analysis wasn't it?

ENGELBART: Even after I told all about that - and by then I had slides and movies to show interactions - he just couldn't see the value of it, or the potential way in which computers were going to get cheaper and support what we were doing. But his reaction was typical. I exhausted myself, year after year, carrying movies and other demonstrations around the country.

Lowood: Was there ever any question of getting involved with Stanford anymore?

ENGELBART: One time I asked him about that, and he said, "Well, I'll tell you what, come over for a meeting. I'll get so-and-so and so-and-so and so-and-so together and we'll judge whether or not some program like this could get going." I didn't even follow up because I knew so-and-so and so-and-so and so-and-so and their interests. There'd be different kinds of special interest, informal seminars, and I'd go sit in on them from time to time. But there was no way I could get anybody's attention. The AI people, I'd try to talk to them about structured files and knowledge content, and they'd look at me stonily and defy me to prove that that was verifiable. There was no overlap. People would be friendly enough, but it was obvious. They were friendly, as though they were saying, "Well, you are a nice guy and we want to be polite, but there isn't any relevance between what you're doing and what we think is important."

Lowood: This may be completely irrelevant, but what about with someone like Donald Knuth and his text formatting, you know, TEX, and those capabilities?

ENGELBART: That emerged after we were down here. I remember thinking about that at the time. His perception just wasn't aligned with ours at all. One time there was a professional conference in Portland on text processing, the very first. I thought, "Oh boy, finally." I went to that and, I'll be damned, nobody even recognized anything we'd done. I later went up to the organizers of the conference, and congratulated them. They looked at me like, "Who is this old fart trying to talk to us?" It was really weird. This was in '79, maybe even '80. It was so depressing for me. They were excitedly talking about things we'd done fifteen years before or had improvements on it, and published. We all knew about it. Finally Andy van Dam, who happened to be there, stood up. He's head of computer science at Brown now. He published books on graphics, and he'd done a sort of a hypertext thing with Ted Nelson and I. He got up just at the end of one of those sessions, and said, "Listen, how can you guys talk about yourselves being professionals if you don't read our history? That guy over there [me], he's done all that stuff." That's all there was at that. I can get so morbid about all of that. I learned a long time ago that I shouldn't.

Lowood: The reason I'm asking these kinds of questions is that I'm trying to map out some of the context.

ENGELBART: I appreciate that. As far as anything productive, there have been a small number of people around the country who would talk to me. If it ever came to their putting out any of their own personal resources to do something about it, I don't know; it wasn't tested. There were very few people until the last three years who would even talk technically about the substantive parts of what we'd been doing. It's just amazing. The whole development of personal computers, which had a marvelous impact on giving people the idea that computers could help . . . . They were so stuck on the idea that you're going to be able to have everything on your desk in a few years that you're ever going to need, that they didn't want to fool around with any of this.

Lowood: That's the whole liberating part of it. I suppose that the lesson there is that when people are liberated in one dimension, they tend to close up the other.

ENGELBART: Well, it takes time to explore what that liberation provides you, and I understand that. What I have learned is that I don't need to take it personally. It takes quite a while for the evolution of all that. I've found lots of cases in my own past where I just was blind to something. Something about my experience got in the way until I got to the point of accepting it.

[END OF INTERVIEW]