corbato:430:written up, it is clear that John had a broader vision. He really corbato:439:CORBATO: Yes, now they were a vision. corbato:733:where we had a grand plan -- a grand vision, I guess, would be a corbato:734:better description, since it was not so much a plan as a vision. corbato:756:the grand vision. This was early 1960. There were several flaws corbato:758:One, he had the vision of trying to do all the software with his corbato:766:to work with him, because he tended to hold his plans and visions corbato:778:had done this partly out of frustration, because Herb's vision corbato:841:hard. He also had this vision of too vast and grand a system, corbato:867:were not aiming for the vision of a large timeshared machine. They corbato:875:trying to do it was that we wanted to set the vision of what people corbato:887:McCarthy's vision so I do not think that they borrowed from us corbato:892:independently from us. However, he had a different vision. He was corbato:893:trying to envision a machine where each user thought he had the raw corbato:895:operating system at all. So that was a different vision. Out at corbato:896:Rand they built the JOSS machine, which again was a vision of corbato:1049:NORBERG: Yes. Now, if there was this vision that had been corbato:1351:change your vision is rather difficult. corbato:1373:today but I did not. It is a very interesting vision of what a corbato:1386:enough of a visionary that he recognized thatwe were moving in the corbato:1451:think Fano, in his own vision of the project,saw it as getting off corbato:1457:beginning there was the vision that we would do it right. It would corbato:1468:model. That was the vision. It started out by our trying to corbato:1471:vendors to see if they had any enthusiasm for this vision. The corbato:1574:hardware people were curious blends of great expertise and vision corbato:1602:people to recognize the vision of what it could mean to have a corbato:1618:introducing new ideas, and they all had their own vision of what a corbato:1679:vision that we were not going to stop on that system, that we were corbato:1751:new markets.We were looking to create a new vision of the future. corbato:1894:system together as though one had a vision of how everything would corbato:1903:adventure. Now, IPTO started out with Lick as a visionary. He was corbato:2175:is another argument. The vision we had was that of the individual corbato:2213:have to recognize in the MULTICS system is that we had this vision corbato:2222:We had it nestable,so that was our vision of how to deal with it. corbato:2760:original vision, in the hopes that it would been... a lot of corbato:2927:that if you haven't got some sort of grand vision of what you are corbato:3127:communications,and much less the sort of let me call it "visionary fano:941:the vision of RLE as a model, besides other things. What was feigenbaum:379:vision and so on.But actually,the big win was having Dendral, to heart:713:weren't just arbitrating between competing visions of other people. heart:980:that ARPA envisioned it, in your opinion, didn't take place, and heart:1093:O'NEILL: Okay. So that was envisioned early on in the process. heart:1095:HEART: Yes. It was envisioned, but it was not envisioned how lick:171:on in the brain.It didn't really envision what goes on in the brain as lick:781:command and control research. It was envisioned that that would grow. lick:1216:philosophy,my vision of what I wanted to happen here,and these people licklider:1217:had every right to have their own vision. lick:1411:whether others didnot have a similar sort ofvision.That we would have lick:1417:suggest that suchavision mighthave beenin your mind,butit isn't clear licklider:1418:to me whether it was a clear vision at the time. licklider:1461:to your question, it was not a clear vision, as certainly,not mckenzie:364:vision of how this was going to change the world. I came back from newel:1319:the way things should be. Partly, because I believed his vision, newel:1416:little vision community, an ARPA vision community, not a bunch of newel:1417:ARPA vision investigators. This is now. This is now for .... newel:1490:Larry of all people, who was one of visionaries who really newel:4190:a clear vision, because I never have any view of what money goes to newel:4513:equivalents didn't stand up seeing the vision of this thing even newel:4514:though Larry Roberts was there preaching the vision at them and say newel:4818:there isn't anybody building a science or envisioning a science of newel:5499:there, mixed in with them, whose vision is really related to the newel:5500:original vision, although it shifts, because when Lick starts it ornstein:210:people who envisioned it in the first place. Of course it's easy ross:168:envision a generalization of the problem, where things are sort of simpson:743:work in vision and natural language processing. Those became my simpson:788:the same: vision in strategic computing, vision in the basic simpson:979:of the DARPA vision. They promote the DARPA vision. They also simpson:985:how to talk to them, how to understand the mission and the vision, simpson:990:community" pursuing the vision of AI in development and basic simpson:1213:into this office, something about the DARPA vision and people who simpson:1214:are supporters of the DARPA vision. Then a few sentences later you simpson:1215:talked about the AI vision. I am interested in learning whether or simpson:1221:DARPA vision, and you had mentioned the AI vision, separated by simpson:1225:difference in terms of who is promoting the images or the visions, simpson:1230:it. I do believe that part of the DARPA vision is the AI vision. simpson:1232:vision is much broader. AI is literally a part of the DARPA simpson:1233:vision. I think it has been DARPA's historically from let's say simpson:1234:the 1960s onward - part of DARPA's vision was AI. AI was a simpson:1258:DARPA's bag. Therefore AI was part of the DARPA vision. Because simpson:1261:and the AI vision that I'll come back to. Different office simpson:1263:whatever, AI as part of the DARPA vision was viewed as being big, simpson:1265:it was a very large part of his vision, of the DARPA vision. I simpson:1268:or others, I think AI as part of the DARPA vision sort of fell simpson:1272:preeminent part of the DARPA vision. Hopefully that helps you simpson:1275:Now, having said that AI is part of the DARPA vision, there is a simpson:1276:corresponding AI vision. At various times during DARPA's history, simpson:1277:there have been either distributed managers of the AI vision, or simpson:1278:there have been single people who have been keepers of the vision, simpson:1279:if you will - keepers, promoters, constructors of the vision. It simpson:1281:DARPA who were the keepers of the AI vision. It is part of my simpson:1290:heat and keep the vision. So there is very much the case that I simpson:1295:unwanted responsibility to keep the vision alive, as promoting, simpson:1315:shareholders and promoters of the vision. Amarel left. Alan Sears simpson:1319:the DARPA vision. The net effect was there were not many people in simpson:1338:keeper ofthe vision at the time was to articulate as best I could, simpson:1353:were just referring to -trying to keep the vision. I see two simpson:1392:the DARPA vision, such as it is as applied to IPTO, is something simpson:1407:directions that were not only achievable,practical, visionary, and simpson:1419:part of what I would define to be the AI vision is this simpson:1426:concern in "keeping the AI vision alive" was that, in fact, with simpson:1495:Simpson vision over the past two years, has been the articulation simpson:1517:list. Part of my articulation of the vision is, "Look, AI is a taylor:139:the Licklider vision of interactive computing. The 1960 man- taylor:315:anyone who's been in that DARPA position since has had the vision taylor:322:were simply extrapolations of Licklider's vision. They were not taylor:323:really new visions of their own. So he's really the father of it winston:686:admiration for him for doing it. He was clearly a visionary in the winston:704:been run by individual princes of science, people with great vision winston:1303:difficult period. In Marvin Denicoff we felt we had a visionary ============================== corbato:430:written up, it is clear that John had a broader vision. He really : progress, do some computations, and interact with the terminal and at the same time allow the person to run a program to check out debugging problems. John has told me that he went to a UNESCO conference and heard Christopher Strachey say something very similar; however when I looked at the documents that had gotten written up, it is clear that John had a broader vision. He really recognized that you could have lots of terminals. He had a whole framework, a whole computational environment. Strachey, I think, thought of it mostly as a debugging terminal. So again, those are written documents. ============================== corbato:439:CORBATO: Yes, now they were a vision. : written documents. NORBERG: Yes, I have read the Strachey piece. I have the McCarthy piece, for that matter. CORBATO: Yes, now they were a vision. CORBATO: In the early 1950s, there were many universities that had been involved in building computers, but by the mid-1950s, as the ============================== corbato:733:where we had a grand plan -- a grand vision, I guess, would be a : CORBATO: Well, John was always viewed as a provocative and interesting wild man. He would state things boldly, and of course when he said, "All you have to do is this," you knew that "All that you had to do," was a loaded statement. But he was right too, so it was a combination. So there we were. We were in that situation where we had a grand plan -- a grand vision, I guess, would be a better description, since it was not so much a plan as a vision. Part of the specs were a million words of memory. I forget how fast the computer was supposed to run, but the notion was that it would be a pretty powerful computer. It was effectively timeshared. ============================== corbato:734:better description, since it was not so much a plan as a vision. : interesting wild man. He would state things boldly, and of course when he said, "All you have to do is this," you knew that "All that you had to do," was a loaded statement. But he was right too, so it was a combination. So there we were. We were in that situation where we had a grand plan -- a grand vision, I guess, would be a better description, since it was not so much a plan as a vision. Part of the specs were a million words of memory. I forget how fast the computer was supposed to run, but the notion was that it would be a pretty powerful computer. It was effectively timeshared. ============================== corbato:756:the grand vision. This was early 1960. There were several flaws : what we could do. Phil Morse got a contract out of the NSF... I think it was NSF; possibly ONR; or possibly both... for the computation center to do timesharing research. Herb Teager was the main leader of that effort. Herb started off trying to build a timesharing system on the 704. This is in lieu of being able to do the grand vision. This was early 1960. There were several flaws in what Herb tried to do (this is all with hindsight, of course). One, he had the vision of trying to do all the software with his little group. He basically wanted to drop all existing software packages -- like no BASIC, no FORTRAN, no nothing. He was going to design his own languages. He was forced to design his own hardware ============================== corbato:758:One, he had the vision of trying to do all the software with his : computation center to do timesharing research. Herb Teager was the main leader of that effort. Herb started off trying to build a timesharing system on the 704. This is in lieu of being able to do the grand vision. This was early 1960. There were several flaws in what Herb tried to do (this is all with hindsight, of course). One, he had the vision of trying to do all the software with his little group. He basically wanted to drop all existing software packages -- like no BASIC, no FORTRAN, no nothing. He was going to design his own languages. He was forced to design his own hardware to get typewriters in and out. That was a set of aggravating problems in its own right. To some extent, he could not keep ============================== corbato:766:to work with him, because he tended to hold his plans and visions : design his own languages. He was forced to design his own hardware to get typewriters in and out. That was a set of aggravating problems in its own right. To some extent, he could not keep everything in perspective. He was also working on handwriting input. One of the problems was that Herb could not recruit people to work with him, because he tended to hold his plans and visions so close to his vest that he would be a little headstrong for what the resources were. Herb was one of the faculty in the Computation Center, and he was working on his contract with Morse, I think as the PI. I was also in the Computation Center and I began to put together a prototype of timesharing. ============================== corbato:778:had done this partly out of frustration, because Herb's vision : CORBATO: Well, what happened was that I had started up what was to be a demonstration project where I was going to write a crude version of timesharing, just to dramatize what could be done. I had done this partly out of frustration, because Herb's vision seemed too grandiose to ever come to fruition in a reasonable amount of time. In fact, that turned out to be true. We could not get anything working in a reasonable amount of time. Actually, Herb let me borrow his typewriter equipment, so that was a help, just to get in input and output to his computer. We also recruited ============================== corbato:841:hard. He also had this vision of too vast and grand a system, : NORBERG: Yes. How similar was it to Teager's design? CORBATO: To my knowledge, not at all, because he was marching to his own tune. Herb wanted other people to work with him but only on his terms, and it was so difficult that people did not try very hard. He also had this vision of too vast and grand a system, where everyone would start over programming from scratch. And I guess I was closer to the people that already were using the machine hard for FORTRAN problems and all that. NORBERG: I assume that you did not have to have any interaction ============================== corbato:867:were not aiming for the vision of a large timeshared machine. They : CORBATO: In that first paper describing CTSS -- it was presented in the spring joint conference of 1962 -- there were a lot of ongoing activities of people using interaction in modest ways. What distinguished us from them, I think, was that most of them were not aiming for the vision of a large timeshared machine. They had various levels of special purpose applications in mind. In fact, one of the arguments that we had with IBM, I recall, was that they tried to argue that they had solved the problem of interactive terminal systems by building the SABRE system, which is an airline reservation system for American Airlines. We tried to convince ============================== corbato:875:trying to do it was that we wanted to set the vision of what people : they tried to argue that they had solved the problem of interactive terminal systems by building the SABRE system, which is an airline reservation system for American Airlines. We tried to convince them that it was not a solution because it was not general purpose. So there was a lot of confusion on that point. One of our goals in trying to do it was that we wanted to set the vision of what people were after. NORBERG: Because as you know, a number of these appear very soon afterwards in 1962 and 1963. I am wondering if they picked up the idea from your group or whether or not there was some sort of ============================== corbato:887:McCarthy's vision so I do not think that they borrowed from us : tried. CORBATO: Well, I think people were tracking one another. There was a venture at BBN, which BBN arrogantly claimed was first, but it was not. It was in parallel. The commonality there was John McCarthy's vision so I do not think that they borrowed from us directly. We both borrowed from the same source. John was a consultant on that, but I think others did the programming. Jack Dennis built a machine, again inspired by the discussions that we had on the long-range planning committee. He, in turn, did that independently from us. However, he had a different vision. He was ============================== corbato:892:independently from us. However, he had a different vision. He was : McCarthy's vision so I do not think that they borrowed from us directly. We both borrowed from the same source. John was a consultant on that, but I think others did the programming. Jack Dennis built a machine, again inspired by the discussions that we had on the long-range planning committee. He, in turn, did that independently from us. However, he had a different vision. He was trying to envision a machine where each user thought he had the raw hardware at his disposal, and he did not appear to have an operating system at all. So that was a different vision. Out at Rand they built the JOSS machine, which again was a vision of trying to build a machine that was kind of a supercalculator, but ============================== corbato:893:trying to envision a machine where each user thought he had the raw : directly. We both borrowed from the same source. John was a consultant on that, but I think others did the programming. Jack Dennis built a machine, again inspired by the discussions that we had on the long-range planning committee. He, in turn, did that independently from us. However, he had a different vision. He was trying to envision a machine where each user thought he had the raw hardware at his disposal, and he did not appear to have an operating system at all. So that was a different vision. Out at Rand they built the JOSS machine, which again was a vision of trying to build a machine that was kind of a supercalculator, but it was not a general purpose programming machine, according to my ============================== corbato:895:operating system at all. So that was a different vision. Out at : Dennis built a machine, again inspired by the discussions that we had on the long-range planning committee. He, in turn, did that independently from us. However, he had a different vision. He was trying to envision a machine where each user thought he had the raw hardware at his disposal, and he did not appear to have an operating system at all. So that was a different vision. Out at Rand they built the JOSS machine, which again was a vision of trying to build a machine that was kind of a supercalculator, but it was not a general purpose programming machine, according to my recollection of it. Again, you were not able to inject any arbitrary computer language. You had to use the one they had. You ============================== corbato:896:Rand they built the JOSS machine, which again was a vision of : had on the long-range planning committee. He, in turn, did that independently from us. However, he had a different vision. He was trying to envision a machine where each user thought he had the raw hardware at his disposal, and he did not appear to have an operating system at all. So that was a different vision. Out at Rand they built the JOSS machine, which again was a vision of trying to build a machine that was kind of a supercalculator, but it was not a general purpose programming machine, according to my recollection of it. Again, you were not able to inject any arbitrary computer language. You had to use the one they had. You could write programs, but not in any arbitrary language. So, that ============================== corbato:1049:NORBERG: Yes. Now, if there was this vision that had been : meeting that Fano began to organize a group to nominally explore the feasibility of getting support from Licklider, and out of that gradually came a group. We developed the plan of what came to be called Project MAC. NORBERG: Yes. Now, if there was this vision that had been developed through the earlier interaction with the Teager committee, why not just trot that out and try to sell it at this point? CORBATO: Well, first of all, the groups had no organizational ============================== corbato:1351:change your vision is rather difficult. : 1963 memo on problem areas of timeshared systems, in which you listed 13 different problem areas, which in a sense are highly detailed, for example when you start talking about problems with scheduling algorithms. This is highly detailed, and for people to come in and be faced with this sort of a detailed list and want to change your vision is rather difficult. CORBATO: Sure. NORBERG: One last question. This can go on for a long time, and I want to be parsimonious with your time. One gets very tired ============================== corbato:1373:today but I did not. It is a very interesting vision of what a : which he outlined what he would like to see from Project MAC? CORBATO: No. I do not remember. NORBERG: It is one of the MAC memos. I should have copied it today but I did not. It is a very interesting vision of what a timeshared environment ought to be like. What I am going to try to do is compare that with documents from MIT in the same period with his to see whether or not he drew it from the MIT experience, or whether there was some sort of double track here to be aware of. But he was definitely guiding things in one sense, and I am hoping ============================== corbato:1386:enough of a visionary that he recognized that we were moving inthe : CORBATO: At the time I felt that Lick was totally supportive. I was amazed at how wonderful it was to have a sponsor who believed in what you were doing. Subsequently, I think I have discovered that Lick did not always understand what we were doing, and probably would not have agreed with every little detail, but he was enough of a visionary that he recognized that we were moving in the right direction. It was actually a very benevolent way of managing things. I also felt that Lick had large confidence in Fano and vice versa. I mean, the two had a relatively good relationship. That was rather critical to making it work too. They trusted one another. ============================== corbato:1451:think Fano,in his own vision of the project, saw it as getting off : NORBERG: [laugh] The statement, or my question? CORBATO: No, the statement; not your question. No, it's a good question. It really goes back to the genesis of Project MAC. I think Fano, in his own vision of the project, saw it as getting off the ground with some sort of a demonstration, but he saw it as a kind of a keystone of the evolution to the development of a commercial quality product with the obvious intent not of going into business, but of influencing the vendors to do this, to make that the norm rather than the exception. So right from the ============================== corbato:1457:beginning there was the vision that we would do it right. It would : the ground with some sort of a demonstration, but he saw it as a kind of a keystone of the evolution to the development of a commercial quality product with the obvious intent not of going into business, but of influencing the vendors to do this, to make that the norm rather than the exception. So right from the beginning there was the vision that we would do it right. It would become a major effort within the project. Fano's view, from a managerial point of view, was that this would serve as a unifying venture, too, because there were many roles where all the different players could contribute. Now, because I had been the leader of the development of CTSS, that was first a natural prototype, which ============================== corbato:1468:model. That was the vision. It started out by our trying to : was the demonstration. It obviously was not very extendable in its initial form. It had both strengths and weaknesses and it dramatized what ought to be done to make it more useful. So MULTICS started out as kind of a wish list of what we would like to see in a big computer system that might be made as a commercial model. That was the vision. It started out by our trying to encourage vendors to join with us. So we spent about three or four months - maybe it was six - touring around, talking to major vendors to see if they had any enthusiasm for this vision. The responses varied. Some companies viewed it as an opportunity to bid in a military defense contract - a way to make money on the ============================== corbato:1471:vendors to see if they had any enthusiasm for this vision. The : MULTICS started out as kind of a wish list of what we would like to see in a big computer system that might be made as a commercial model. That was the vision. It started out by our trying to encourage vendors to join with us. So we spent about three or four months - maybe it was six - touring around, talking to major vendors to see if they had any enthusiasm for this vision. The responses varied. Some companies viewed it as an opportunity to bid in a military defense contract - a way to make money on the margin rather than affecting any product plans. That was my view of Burroughs. CDC was disinterested. That is my recollection, although I would have to check notes to confirm that. Again, ============================== corbato:1574:hardware people were curious blends of great expertise and vision : CORBATO: Fledgling from being in the computer business, and the management was not that wise in the ways of computers. The upper management didn't understand software very well, and they knew how to get hardware built. But to some extent the software and hardware people were curious blends of great expertise and vision and naivete. So it was funny mix of people. One of the reasons they went with this was that the few key people were prepared to gamble with us. On the other hand, that was one of their very flaws; they didn't have the checks and balances inside to keep them from doing strange things. ============================== corbato:1602:people to recognize the vision of what it could mean to have a : talked to at CDC, who you talked to at Burroughs, and so on. I want to come back to Burroughs for a moment. CORBATO: At CDC we went out and gave a site visit at least on one occasion. McCarthy gave a pitch for time-sharing, trying to get people to recognize the vision of what it could mean to have a machine organized in this particular way. NORBERG: Is this after he went to Stanford or before? CORBATO: He was still with us, so it was before. Seymour dropped ============================== corbato:1618:introducing new ideas, and they all had their own vision of what a : had to assume that he had made up his mind, as I reconstructed after the fact rather than at the time. So he had... we had similar... not quite so extreme, but we had similar attitudes of most hardware designers, who were all interested in doing their own thing. They really didn't want to be bothered with somebody introducing new ideas, and they all had their own vision of what a computer ought to be. And it didn't include the social environment of people trying to use the computer all at once. The situation of Burroughs was that the people we talked to... we weren't talking to the commercial side of Burroughs; we were ============================== corbato:1679:vision that we were not going to stop on that system, that we were : CORBATO: Oh, we did both. Clearly, because CTSS was an immediate object of use, we spent a certain amount of effort for the first year getting it to be more effective and more useful to the participants in the project. But from the beginning we had the vision that we were not going to stop on that system, that we were going to go on to something else. So it was really both. And one of the management problems we had was forcing ourselves off of CTSS and kind of freezing its development, because there was a temptation to keep going, to keep fiddling with it, and changing and improving it. ============================== corbato:1751:new markets.We were looking to create a new vision of the future. : acquisition for their laboratories, and they had been scouting out GE. There was some synergism: because they knew we were interested they got interested. I think they first began to look independently of us. But they saw the possibility of our developing a new operating system together. So GE was looking for new markets. We were looking to create a new vision of the future. And Bell Labs had their own form of naivete; we had ours. Where they thought that you just got a bunch of good system programmers together, and they build an operating system on top of the hardware, and pretty soon it was useful for everybody - good things just kept happening. My own view of this was that this was based ============================== corbato:1894:system together as though one had a vision of how everything would : bricks in place. The obvious way to build it right is to build it brick by brick and see what you have done to date, and then decide how to go on from there. That takes a lot of time. It's a very serial process, and it also presupposes that you have a well- ordered view of what comes first. Here we were sort of rushing a system together as though one had a vision of how everything would fit together, and you would build all the pieces and then you would try to assemble them. And as you come together you find things don't always work together well. And it is that trial and error aspect of system synthesis, which I think is the major factor that people underestimate and a huge amount of... the difference between ============================== corbato:1903:adventure. Now, IPTO started out with Lick as a visionary. He was : aspect of system synthesis, which I think is the major factor that people underestimate and a huge amount of... the difference between engineering projects to me is the degree that the participants have done it before. If they have done it before it's a piece of cake sometimes; if they have never done it before it can be a hellish adventure. Now, IPTO started out with Lick as a visionary. He was not an implementor. Oh, he loved to hack a little bit but he was not a detail man himself. So he thought it was wonderful we were going off on new adventures. The plan sounded great to him. I think, in general, when IPTO put its money into different places around the country they hoped that good thing would come out of it, ============================== corbato:2175:is another argument. The vision we had was that of the individual : think they are different. There are some differences in degree, and there are certainly differences in the way they approach solving the problem, but we did not have in mind the military needs first. I honestly think that a lot of the military needs are met by systems that we did have in mind, but that remains for... that is another argument. The vision we had was that of the individual person in an organization, like a person in an office or a researcher in a university where there is a combination of a need to have the benefit of privacy, just like when I lock my desk at night. I can do a first approximation and assume that no one is looking at what is in the desk, including the cleaning ladies, ============================== corbato:2213:have to recognize in the MULTICS system is that we had this vision : in on cellular telephone conversations. There is no way the ordinary user can do anything about it except not use the phone. [laugh] He is forced to use it as it is. We were trying to give a system where there is a choice. Now, this reflects back on that earlier discussion of the idea of autonomy. The key thing that you have to recognize in the MULTICS system is that we had this vision that there was a central authority managing the system and then we wanted to distribute the authority. The questions of privacy we were going to distribute. And our model was, because we didn't have a better one, was to make it hierarchical. Now, in fact, not all activities in life are hierarchical. But nevertheless, that ============================== corbato:2222:We had it nestable,so that was our vision of how to deal with it. : have a better one, was to make it hierarchical. Now, in fact, not all activities in life are hierarchical. But nevertheless, that was the model we used, so that we have the administrator followed by project leaders, or group leaders, or department heads (whatever name you want to use) followed by individuals within the groups. We had it nestable, so that was our vision of how to deal with it. It did not (?) make a useful framework. NORBERG: Would you use the same approach now as you used then? [INTERRUPTION] ============================== corbato:2760:original vision, in the hopes that it would been... a lot of : NORBERG: So that sort of puts in the decision of all or nothing for a manufacturer who might want to incorporate this, doesn't it? CORBATO: Yes, it does. It makes it all or nothing. I guess our original vision, in the hopes that it would been... a lot of versions of MULTICS would have been replicated, and that by virtue of its presence it would influence people and possibly generate competitors. But the pivotal thing that... our introducing the ideas of segmentation and also introducing the ideas of hardware protection and security in the way we did, at least, guaranteed we ============================== corbato:2927:that if you haven't got some sort of grand vision of what you are : sophisticated level about systems design. And I think we probably had some influence on the need to articulate the design process. And I don't mean just blindly writing down a bunch of detailed specs. The design process could be perverted by people who fundamentally treat it as a bookkeeping exercise. What I mean is that if you haven't got some sort of grand vision of what you are trying to do and some sense of structure to it, then it gets to be a real mess when you through the system. So we probably were better than most projects in trying to spell things out. It turned out to be crucial for our own survival, because, as we had to keep reindoctrinating people as there was a lot of personnel turnover, ============================== corbato:3127:communications, and much less the sort of let me call it"visionary : NORBERG: Well, I recall reading the 1986 proposal for two year funding from LCS, to IPTO (ISTO at the time). And it seemed much more project-oriented. Things were confined to specific areas of programming language development, and system development of communications, and much less the sort of let me call it "visionary aspect" that we find in the proposals of the 1960s? CORBATO: Yes. NORBERG: So that confirms what you said. ============================== fano:941:the vision of RLE as a model, besides other things. What was : NORBERG: How about procedures, though, in getting more money from them? FANO: Okay, this is part of a bigger issue. I pointed out yesterday that when Project MAC got started, Licklider and I shared the vision of RLE as a model, besides other things. What was different was there was no laboratory; there was a project that was supposed to involve the whole MIT community as users in this. But we shared that model, which included a great deal of emphasis on education. Well, when Lick left... well, I do not know whether he had just left... the attitude in the Pentagon changed slowly, moved ============================== heart:713:weren't just arbitrating between competing visions of other people. : some of the other people. O'NEILL: Okay. HEART: I mean they were technical people in their own right. They weren't just arbitrating between competing visions of other people. In some cases they were leading. In the host protocol area, the DARPA people played a very central role. Crocker and Cerf and others were involved, you know, very directly in the Network Working Groups, as kind of group leaders at some points. Have you collected the written material that exists? ============================== heart:980:that ARPA envisioned it, in your opinion, didn't take place, and : the community were using it right from the beginning, a lot. But they were using it without really realizing that the whole world was about to want it. O'NEILL: You mentioned that the sharing of resources on the level that ARPA envisioned it, in your opinion, didn't take place, and then the surprise of the mail having more of an impact than anyone would have suspected. Are those pretty much the differences from expectations that you would list? Are there other things that you are surprised about how they turned out? ============================== heart:1093:O'NEILL: Okay. So that was envisioned early on in the process. : O'NEILL: Was the network control center considered part of your contract? HEART: It was part of the contract. O'NEILL: Okay. So that was envisioned early on in the process. HEART: Yes. It was envisioned, but it was not envisioned how tough, and expensive and difficult it was going to be. People knew they had to do it, but they didn't have as much of a feeling for the degree of problems. Tremendous efforts went into that. I ============================== heart:1095:HEART: Yes. It was envisioned, but it was not envisioned how : HEART: It was part of the contract. O'NEILL: Okay. So that was envisioned early on in the process. HEART: Yes. It was envisioned, but it was not envisioned how tough, and expensive and difficult it was going to be. People knew they had to do it, but they didn't have as much of a feeling for the degree of problems. Tremendous efforts went into that. I mean, the early IMPs could cross-patch their lines remotely. We could, from here, turn around the output line of an IMP in ============================== lick:781:command and control research.It was envisioned that that would grow. : LICKLIDER: Jack Ruina was director of ARPA, and was given responsibility for a command and control project that got set up from inside the office of the Secretary of Defense. The contract was let by them through ARPA to System Development Corporation to do some command and control research. It was envisioned that that would grow. There would be more to it than that one thing. So Jack wanted somebody to do that. Simultaneously, there was a recommendation from the Defense Advisory Board that the Defense Department look into supporting some behavioral research, particularly through some institutes that would be set up to bring behavioral researchers together with better facilities. Jack ============================== lick:1216:philosophy,my vision of what I wanted to happen here,and these people : one, but I certainly was associated with them, and I spotted Schwartz as a really positive character. We supported him to the hilt. We got him to turn a machine into a time-sharing system, which I think he enjoyed doing. So, it went well. I did not really have a bad battle with SDC, but I was aware that this was cheating a little bit. I would insist on my philosophy, my vision of what I wanted to happen here, and these people had every right to have their own vision. NORBERG: That's fine. I am just trying to set up an extreme contrast to get you to remember some of these things more clearly. ============================== licklider:1217:had every right to have their own vision. : a really positive character. We supported him to the hilt. We got him to turn a machine into a time-sharing system, which I think he enjoyed doing. So, it went well. I did not really have a bad battle with SDC, but I was aware that this was cheating a little bit. I would insist on my philosophy, my vision of what I wanted to happen here, and these people had every right to have their own vision. NORBERG: That's fine. I am just trying to set up an extreme contrast to get you to remember some of these things more clearly. ASPRAY: You were clearly committed to this time-sharing, or interactive ============================== lick:1411:whether others did not have a similar sort of vision.Thatwe wouldhave : thinking of computing in in a way essentially quite different, maybe even new, from people in other parts of the academic community -- certainly different from people in industry. It was trying to strive for some sort of discontinuous circumstance where a new world of computing would develop out of the research. What is not clear to me, first of all, is whether others did not have a similar sort of vision. That we would have to inquire about. But secondly, whether or not enough of the required pieces to make that work -- the displays, the interfaces, the expanded memories -- were seen as part of an overall plan at that time, each piece of which would need to be attacked in order to achieve interactive computing. The way you just described your visits to the various groups ============================== lick:1417:suggest that such a vision might have been in your mind, but it isn't clear : to inquire about. But secondly, whether or not enough of the required pieces to make that work -- the displays, the interfaces, the expanded memories -- were seen as part of an overall plan at that time, each piece of which would need to be attacked in order to achieve interactive computing. The way you just described your visits to the various groups suggest that such a vision might have been in your mind, but it isn't clear to me whether it was a clear vision at the time. LICKLIDER: Well, it surely was not clear. Take this area here. Here's a mass; here's a keyboard. A fellow named Herb Teager thought that you get rid of that, and you put a sheet of plastic down here, because it was ============================== licklider:1418:to me whether it was a clear vision at the time. : pieces to make that work -- the displays, the interfaces, the expanded memories -- were seen as part of an overall plan at that time, each piece of which would need to be attacked in order to achieve interactive computing. The way you just described your visits to the various groups suggest that such a vision might have been in your mind, but it isn't clear to me whether it was a clear vision at the time. LICKLIDER: Well, it surely was not clear. Take this area here. Here's a mass; here's a keyboard. A fellow named Herb Teager thought that you get rid of that, and you put a sheet of plastic down here, because it was an inductive coupling. Go out to Rand and it was the same thing: "It is ============================== licklider:1461:to your question,it was not a clear vision,as certainly, not tha : how much money went to each part, and what the interactions were. That blew the minds of the accounting people; they had never seen that. Still, they didn't heckle me much about being the manager there. It was a different dimension. "Oh, go talk to him and he will talk your arm off for two or three hours about that. Better avoid him." So, to come back to your question, it was not a clear vision, as certainly, not that: "We'll plug them all together, and that will be the system"; but, rather, it was a matter of getting a supply of parts and methods and techniques, and different people will put together different systems out of it. NORBERG: Okay. Were you using advisors at the time, either ============================== mckenzie:364:vision of how this was going to change the world. I came back from : things and change the way people worked. Other people have said it was just a fun, interesting project, or whatever, but didn't really think about any bigger implications. MCKENZIE: I think I was in the second category. I didn't have a vision of how this was going to change the world. I came back from a long vacation, I was asked to do this as my next job, and I did it. As I was doing it, it was fun and interesting, but I don't think that I really had an image. I've never been very good at looking into crystal balls. I'm not good at it now, and I don't think I was good at it then either. So I didn't really have any ============================== newel:1319:the way things should be. Partly, because I believed his vision, : know. Decisions did get made. He would not spend very much time on administration, so he would make the decisions rapidly. We never fought so in one sense we just understood that things were okay. I'm sure, in fact, we did talk a little bit. But fundamentally, we never had to sit down and have big sessions about the way things should be. Partly, because I believed his vision, and he believed mine in some respects. There are some interesting things in the literature where he comments about my being the only guy who he has ever known who he has always agreed with. Every time I would do something, I would find out that that was in fact okay. So there was essentially no friction in the organization. newel:1490:Larry of all people, who was one of visionaries who really : part of that story is, if you look at the history -- which you already know all about -- if you look at the history of the ARPANET, Larry Roberts was totally permissive with respect to the research community and whether they got on that and used it. This was a major frustration. He said, "Here's this great thing" -- and Larry of all people, who was one of visionaries who really understood the potential of the ARPANET and "It's all free and I'm hanging it out in front of you guys and you guys sort of say, 'Don't bother me; I've got my own things,'" and he never did anything to make that happen. He went off and did some other things. He went off and created the Center for... I can't recall ============================== newel:4514:though Larry Roberts was there preaching the vision at them and say : because it's not actually clear to me that we could measure it. It could be just kind of a perceptual thing. But certainly some of the people did. I suppose that happened with networking, but certainly there was a kind of a reluctance around in which McCarthy equivalents didn't stand up seeing the vision of this thing even though Larry Roberts was there preaching the vision at them and say wow, I'm going to go off and do this. It happened naturally enough, but it didn't turn immediately. So much of the issue of the community arrives with the ARPANET, so there is, as I have said in answering a number of your questions, 1970 is a kind of a key issue. It is one kind of community afterwards and another thing ============================== newel:4818:there isn't anybody building a science or envisioning a science of : something that mathematicians do, but they don't make any science out of the program. The only thing they do is numerical analysis, and then there is some programming. The mathematicians never get an image of computer science. The engineers never get an image of computer science. And it would never occur to them to. And so there isn't anybody building a science or envisioning a science of computer science. And that's what goes on in this community. NORBERG: What distinguishes this community, this IPTO community, in the 1980s from the rest of computer science? Because now there is a much larger context that call themselves computer scientists. ============================== newel:5499:there, mixed in with them, whose vision is really related to the : have spent $35 million getting this started. Now, what have you done for me lately? Now it's time to deliver." There is a piece of that that isn't just, "I want the products." There is also a thing that says, "You have had your vacation; now it's time. The palmy days are over." But there's another set of people down there, mixed in with them, whose vision is really related to the original vision, although it shifts, because when Lick starts it there is no such computer science thing, and then, in the 1980s, it exists already. The issue is fundamentally one of this same long- term growth of information technology, and the application stuff is all tactics. The environment forces you to do this. Like the ============================== newel:5500:original vision, although it shifts, because when Lick starts it : done for me lately? Now it's time to deliver." There is a piece of that that isn't just, "I want the products." There is also a thing that says, "You have had your vacation; now it's time. The palmy days are over." But there's another set of people down there, mixed in with them, whose vision is really related to the original vision, although it shifts, because when Lick starts it there is no such computer science thing, and then, in the 1980s, it exists already. The issue is fundamentally one of this same long- term growth of information technology, and the application stuff is all tactics. The environment forces you to do this. Like the justification for quoting that thing in the Congressional stuff, ============================== ornstein:210:people who envisioned it in the first place. Of course it's easy : looked it over and came back a day or two later and put it down on his desk and said "Well, sure, I suppose we could build that if you wanted to, but I can't see what one would want such a thing for." (Laugh) Prophetic words. In fact, I think that as the network has turned out to be used, it has had some surprises, even for the people who envisioned it in the first place. Of course it's easy to look back and say, "Oh I knew that all along." But I think there's a certain measure of surprise for practically everybody involved. Anyway, I certainly confess to being absolutely dead wrong about the thing at the time. (Laugh) But, nonetheless, we thought it over and decided it would be a fun thing to do - which ============================== ross:168:envision a generalization of the problem, where things are sort of : ROSS: At all stages, right from the very beginning, I tried to do things in a general way and then make it special for the use... That's just the way I've always gone about solving problems. I'm no good at puzzles. So instead of looking for the chink in the armor that will allow you to unlock something, I do best when I can envision a generalization of the problem, where things are sort of neat and abstract and idealized, and then get a solution there that I can then map down to a special case, you see? That's always been the way that I've done things. And also my training, which was still pretty raw, but it was all in pure mathematics, which is all about, you know, functions and systems and making structures of ============================== simpson:788:the same: vision in strategic computing, vision in the basic : the one hand, the kinds of basic research you were supporting, and then this 6.2 activity that was going on in these various other places. SIMPSON: That's easy. Conceptually, the areas of research were the same: vision in strategic computing, vision in the basic research program. The distinction is sort of an orthogonal one where you look at the distinction - it's a fuzzy boundary - but it's between science and technology, between a focused, targeted recipient of the research results versus a less focused, more diverse recipient of research results. Strategic computing was ============================== simpson:979:of the DARPA vision. They promote the DARPA vision. They also : managers, and all of the strange things - the incantations - that were sometimes required, as well as understanding what it means to be a member of the DARPA research community. Because DARPA contractors are in one sense contractors - they get paid by the government to do things, but at the same time they are supporters of the DARPA vision. They promote the DARPA vision. They also help program managers and DARPA articulate and defend the rationale and importance of what we do. So it is in that sense that it is a community. Because, again, there were people leaving Carnegie, Stanford, MIT and forming their own research groups outside, there was a transfer of an appreciation for what DARPA was interested in, ============================== simpson:985:how to talk to them, how to understand the mission and the vision, : help program managers and DARPA articulate and defend the rationale and importance of what we do. So it is in that sense that it is a community. Because, again, there were people leaving Carnegie, Stanford, MIT and forming their own research groups outside, there was a transfer of an appreciation for what DARPA was interested in, how to talk to them, how to understand the mission and the vision, and then how to participate. So that, combined with the competition in contracting, encouraged and provided a mechanism to expand the support. So I'm quite happy that during my five years at DARPA, the number of groups that are "part of the DARPA community" pursuing the vision of AI in development and basic ============================== simpson:990:community" pursuing the vision of AI in development and basic : how to talk to them, how to understand the mission and the vision, and then how to participate. So that, combined with the competition in contracting, encouraged and provided a mechanism to expand the support. So I'm quite happy that during my five years at DARPA, the number of groups that are "part of the DARPA community" pursuing the vision of AI in development and basic research and technology is quite large compared to five years ago. New groups, new sites. NORBERG: There are two issues in there, and I don't want to lose either one of them. I'm trying to decide which one to go to first. ============================== simpson:1213:into this office, something about the DARPA vision and people who : NORBERG: I want to go to the second issue that was on my mind a few minutes ago when I said there were two of them. It will help me not only to explore that issue, but get back to something else that I asked you before that I think is still not clear yet, at least in my mind. You said at a couple of points since we came into this office, something about the DARPA vision and people who are supporters of the DARPA vision. Then a few sentences later you talked about the AI vision. I am interested in learning whether or not ... ============================== simpson:1214:are supporters of the DARPA vision. Then a few sentences later you : few minutes ago when I said there were two of them. It will help me not only to explore that issue, but get back to something else that I asked you before that I think is still not clear yet, at least in my mind. You said at a couple of points since we came into this office, something about the DARPA vision and people who are supporters of the DARPA vision. Then a few sentences later you talked about the AI vision. I am interested in learning whether or not ... ============================== simpson:1215:talked about the AI vision. I am interested in learning whether or : me not only to explore that issue, but get back to something else that I asked you before that I think is still not clear yet, at least in my mind. You said at a couple of points since we came into this office, something about the DARPA vision and people who are supporters of the DARPA vision. Then a few sentences later you talked about the AI vision. I am interested in learning whether or not ... ============================== simpson:1221:DARPA vision, and you had mentioned the AI vision, separated by : not ... NORBERG: I think what I was saying is that you had mentioned the DARPA vision, and you had mentioned the AI vision, separated by several sentences. I am wondering if there is a similarity or an identity between those two, or if there is a difference. And if there is a difference, do you see any implications of that difference in terms of who is promoting the images or the visions, and who is the leader and who is follower here? And how has that ============================== simpson:1230:it. I do believe that part of the DARPA vision is the AI vision. : difference in terms of who is promoting the images or the visions, and who is the leader and who is follower here? And how has that changed over time, if you know that? So there are three parts to this question. SIMPSON: Let's see if I can keep them on my stack while I unwind it. I do believe that part of the DARPA vision is the AI vision. I do think they are different, because DARPA's mission and DARPA's vision is much broader. AI is literally a part of the DARPA vision. I think it has been DARPA's historically from let's say the 1960s onward - part of DARPA's vision was AI. AI was a technical ingredient that DARPA viewed as something that both ============================== simpson:1232:vision is much broader. AI is literally a part of the DARPA : changed over time, if you know that? So there are three parts to this question. SIMPSON: Let's see if I can keep them on my stack while I unwind it. I do believe that part of the DARPA vision is the AI vision. I do think they are different, because DARPA's mission and DARPA's vision is much broader. AI is literally a part of the DARPA vision. I think it has been DARPA's historically from let's say the 1960s onward - part of DARPA's vision was AI. AI was a technical ingredient that DARPA viewed as something that both defined DARPA in terms of impact - "Here is an area of technology that we can point to that we've had an impact in" - and at the same ============================== simpson:1233:vision. I think it has been DARPA's historically from let's say : this question. SIMPSON: Let's see if I can keep them on my stack while I unwind it. I do believe that part of the DARPA vision is the AI vision. I do think they are different, because DARPA's mission and DARPA's vision is much broader. AI is literally a part of the DARPA vision. I think it has been DARPA's historically from let's say the 1960s onward - part of DARPA's vision was AI. AI was a technical ingredient that DARPA viewed as something that both defined DARPA in terms of impact - "Here is an area of technology that we can point to that we've had an impact in" - and at the same time, it has been viewed as one of DARPA's ingredients that it ============================== simpson:1234:the 1960s onward - part of DARPA's vision was AI. AI was a : SIMPSON: Let's see if I can keep them on my stack while I unwind it. I do believe that part of the DARPA vision is the AI vision. I do think they are different, because DARPA's mission and DARPA's vision is much broader. AI is literally a part of the DARPA vision. I think it has been DARPA's historically from let's say the 1960s onward - part of DARPA's vision was AI. AI was a technical ingredient that DARPA viewed as something that both defined DARPA in terms of impact - "Here is an area of technology that we can point to that we've had an impact in" - and at the same time, it has been viewed as one of DARPA's ingredients that it brings to, if you want to think about the style of ingredients that ============================== simpson:1258:DARPA's bag. Therefore AI was part of the DARPA vision. Because : piece of DARPA. It is in that sense that I mean, very broadly speaking, directors of DARPA over longer periods of time, back before George Heilmeier, especially during George Heilmeier's period in the 1970s, and then on into Bob Cooper have all understood that AI was one of the things, one of the tricks in DARPA's bag. Therefore AI was part of the DARPA vision. Because of that, they were always comfortable with whatever the office director decided was the appropriate resource level to support AI and the AI vision that I'll come back to. Different office directors, because of their interests and personalities, or whatever, AI as part of the DARPA vision was viewed as being big, ============================== simpson:1261:and the AI vision that I'll come back to. Different office : period in the 1970s, and then on into Bob Cooper have all understood that AI was one of the things, one of the tricks in DARPA's bag. Therefore AI was part of the DARPA vision. Because of that, they were always comfortable with whatever the office director decided was the appropriate resource level to support AI and the AI vision that I'll come back to. Different office directors, because of their interests and personalities, or whatever, AI as part of the DARPA vision was viewed as being big, small, in relative importance. I think in the case of Bob Cooper it was a very large part of his vision, of the DARPA vision. I think that is what led to the strategic computing program. I think ============================== simpson:1263:whatever, AI as part of the DARPA vision was viewed as being big, : DARPA's bag. Therefore AI was part of the DARPA vision. Because of that, they were always comfortable with whatever the office director decided was the appropriate resource level to support AI and the AI vision that I'll come back to. Different office directors, because of their interests and personalities, or whatever, AI as part of the DARPA vision was viewed as being big, small, in relative importance. I think in the case of Bob Cooper it was a very large part of his vision, of the DARPA vision. I think that is what led to the strategic computing program. I think in my time as part of the team during Ray Holiday (?), for example, or others, I think AI as part of the DARPA vision sort of fell ============================== simpson:1265:it was a very large part of his vision, of the DARPA vision. I : director decided was the appropriate resource level to support AI and the AI vision that I'll come back to. Different office directors, because of their interests and personalities, or whatever, AI as part of the DARPA vision was viewed as being big, small, in relative importance. I think in the case of Bob Cooper it was a very large part of his vision, of the DARPA vision. I think that is what led to the strategic computing program. I think in my time as part of the team during Ray Holiday (?), for example, or others, I think AI as part of the DARPA vision sort of fell down. It wasn't as important. It was known; it was recognized, it wasn't a large portion. Because of Craig Fields' background I ============================== simpson:1268:or others, I think AI as part of the DARPA vision sort of fell : whatever, AI as part of the DARPA vision was viewed as being big, small, in relative importance. I think in the case of Bob Cooper it was a very large part of his vision, of the DARPA vision. I think that is what led to the strategic computing program. I think in my time as part of the team during Ray Holiday (?), for example, or others, I think AI as part of the DARPA vision sort of fell down. It wasn't as important. It was known; it was recognized, it wasn't a large portion. Because of Craig Fields' background I think that AI has once again shifted into being a large, if not preeminent part of the DARPA vision. Hopefully that helps you understand my perspective on what DARPA thinks about what I do. ============================== simpson:1272:preeminent part of the DARPA vision. Hopefully that helps you : in my time as part of the team during Ray Holiday (?), for example, or others, I think AI as part of the DARPA vision sort of fell down. It wasn't as important. It was known; it was recognized, it wasn't a large portion. Because of Craig Fields' background I think that AI has once again shifted into being a large, if not preeminent part of the DARPA vision. Hopefully that helps you understand my perspective on what DARPA thinks about what I do. Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or ============================== simpson:1275:Now, having said that AI is part of the DARPA vision, there is a : wasn't a large portion. Because of Craig Fields' background I think that AI has once again shifted into being a large, if not preeminent part of the DARPA vision. Hopefully that helps you understand my perspective on what DARPA thinks about what I do. Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in ============================== simpson:1276:corresponding AI vision. At various times during DARPA's history, : think that AI has once again shifted into being a large, if not preeminent part of the DARPA vision. Hopefully that helps you understand my perspective on what DARPA thinks about what I do. Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in DARPA who were the keepers of the AI vision. It is part of my ============================== simpson:1277:there have been either distributed managers of the AI vision, or : preeminent part of the DARPA vision. Hopefully that helps you understand my perspective on what DARPA thinks about what I do. Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in DARPA who were the keepers of the AI vision. It is part of my appreciation for what they did, when they did it, and their ============================== simpson:1278:there have been single people who have been keepers of the vision, : understand my perspective on what DARPA thinks about what I do. Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in DARPA who were the keepers of the AI vision. It is part of my appreciation for what they did, when they did it, and their sacrifices and contributions that make me sensitive, in some sense, ============================== simpson:1279:if you will - keepers, promoters, constructors of the vision. It : Now, having said that AI is part of the DARPA vision, there is a corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in DARPA who were the keepers of the AI vision. It is part of my appreciation for what they did, when they did it, and their sacrifices and contributions that make me sensitive, in some sense, to my responsibilities as the current keeper of the flame, if you ============================== simpson:1281:DARPA who were the keepers of the AI vision. It is part of my : corresponding AI vision. At various times during DARPA's history, there have been either distributed managers of the AI vision, or there have been single people who have been keepers of the vision, if you will - keepers, promoters, constructors of the vision. It is easy for me to link back historically to my predecessors in DARPA who were the keepers of the AI vision. It is part of my appreciation for what they did, when they did it, and their sacrifices and contributions that make me sensitive, in some sense, to my responsibilities as the current keeper of the flame, if you will. Because it is very much the case that I saw during my five years at DARPA that there were a lot of supporters when times were ============================== simpson:1290:heat and keep the vision. So there is very much the case that I : will. Because it is very much the case that I saw during my five years at DARPA that there were a lot of supporters when times were good, but there were a lot of people that abandoned the ship when the times got tough. I found there wasn't anybody during the toughest of the times that was willing to stand up and take the heat and keep the vision. So there is very much the case that I felt like the lone stranger for a while, the only person that was willing to stand up and argue against what seemed to be the conventional wisdom, which at times I felt like was attempting to blow out the flame. It is very much the case that I felt almost an unwanted responsibility to keep the vision alive, as promoting, ============================== simpson:1295:unwanted responsibility to keep the vision alive, as promoting, : heat and keep the vision. So there is very much the case that I felt like the lone stranger for a while, the only person that was willing to stand up and argue against what seemed to be the conventional wisdom, which at times I felt like was attempting to blow out the flame. It is very much the case that I felt almost an unwanted responsibility to keep the vision alive, as promoting, encouraging, articulating in some sense why AI is important, why it should be invested in seriously and promoted, and why it delivers new capabilities to defense that cannot be acquired otherwise. NORBERG: Now, was this in the last five years? ============================== simpson:1315:shareholders and promoters of the vision. Amarel left. Alan Sears : SIMPSON: This is, remember, over this five years. When I came to DARPA strategic computing was a new program; AI was one of, if not the preeminent, technical area of investment. So during Amarel's period, during Alan Sears' period, during this, they were all shareholders and promoters of the vision. Amarel left. Alan Sears left. Jack Schwartz came in. The DARPA directors, Ray Holiday and before him... why am I drawing a blank here? You know who I'm talking about. Again, they did not see AI as a major portion of the DARPA vision. The net effect was there were not many people in the agency that were promoters or who could articulate the ============================== simpson:1319:the DARPA vision. The net effect was there were not many people in : period, during Alan Sears' period, during this, they were all shareholders and promoters of the vision. Amarel left. Alan Sears left. Jack Schwartz came in. The DARPA directors, Ray Holiday and before him... why am I drawing a blank here? You know who I'm talking about. Again, they did not see AI as a major portion of the DARPA vision. The net effect was there were not many people in the agency that were promoters or who could articulate the distinctions, importance... At the same time one could argue that there was balance occurring. I can obviously see that as an issue of resources distributed among many possible areas, there is a legitimate view that there would need to be some rebalancing of ============================== simpson:1338:keeper of the vision at the time was to articulate as best I could : NORBERG: If you weren't getting the support within the agency, where did you go for such support, or did you not? In the outside community? SIMPSON: Basically what I viewed as my responsibility as the keeper of the vision at the time was to articulate as best I could, in as many forums as I could, the urgency for people to step forward and be counted on - in other agencies, among the research community and among my defense colleagues - as well as continuing to fight internally in any opportunity that presented itself on new ideas, new opportunities. As I think you understand, because there ============================== simpson:1353:were just referring to -trying to keep the vision. I see two : my view, if there was an AI winter, it was literally during that period. NORBERG: I'd like to make a suggestion to you and see how you respond to this in terms of this historical continuity that you were just referring to -trying to keep the vision. I see two things as characteristic of IPTO programs, I don't want to go broader than that at the moment. Two things occur very early in the establishment of the IPTO program, back in the days when AI was a portion of a lot of other activities... ============================== simpson:1392:the DARPA vision, such as it is as applied to IPTO, is something : different, but in their core they are exactly the same sort of problems. NORBERG: Therefore my conclusion, if those characteristics are a correct interpretation of the program over twenty years, is that the DARPA vision, such as it is as applied to IPTO, is something that was generated by a very tight coupling between the research community, on the one hand from whom most of the program managers come anyway as we know, and the program side, regardless of how many constraints there are on funding, evaluation, and so on. Now what I hear you telling me is that in the late 1980s that changed. ============================== simpson:1407:directions that were not only achievable,practical, visionary, and : tightly coordinated interacting collection of people who share the best interests of the country from a technical, scientific standpoint but also from a defense and national responsibility standpoint, these people because they share that same concern could collectively be wise in making research, choosing research directions that were not only achievable, practical, visionary, and could deliver, if you will - since DARPA is a mission agency - deliver to Defense things that would be notable, of consequence. So you are right; I'm absolutely convinced of that. It appeared that, I guess from the standpoint that there was obvious stress going on, there was in fact some risk or some damage to that sort ============================== simpson:1419:part of what I would define to be the AI vision is this : best interests of everyone was being served. So I think that is true, in a more general sense, and in a particular sense in the AI part of the world, which is basically where I confined most of my interests. It was the case that the best interest, if you will, of the AI research community was no longer of consideration. Since part of what I would define to be the AI vision is this relationship between DARPA in caring for, being concerned about the health and well being of the nation's technology base - people, institutions, and research programs - was being threatened because of resource limitations. I mean, there is a point at which, even if you believe that important redirections are necessary, the pace ============================== simpson:1426:concern in "keeping the AI vision alive" was that, in fact, with : health and well being of the nation's technology base - people, institutions, and research programs - was being threatened because of resource limitations. I mean, there is a point at which, even if you believe that important redirections are necessary, the pace at which those changes take place needs to be factored in. So my concern in "keeping the AI vision alive" was that, in fact, with some of the directions we were headed in, we would endanger the survival of long-term health of what I still believe is one of our national treasures. NORBERG: With the expanding activities in AI across so many ============================== simpson:1495:Simpson vision over the past two years, has been the articulation : NORBERG: Two things then. Is that list available, that is, can it be made available? SIMPSON: Sure. Part of what I've been briefing, sort of the Bob Simpson vision over the past two years, has been the articulation of that information. NORBERG: Can I get some of that information? SIMPSON: I was surprised that your literary search didn't turn it up. ============================== simpson:1517:list. Part of my articulation of the vision is, "Look, AI is a : NORBERG: They didn't come up with that one. The student did not come up with that one. SIMPSON: What I did there was for the first time I wrote a paper and also gave briefings for which I sort of continued to improve my list. Part of my articulation of the vision is, "Look, AI is a strategic technology for the United States, and let me tell you, it is also a critical technology for defense. That has been acknowledged as DOD's critical technologies plan. And let me explain to you why it is a critical technology. Let me show you what AI has done in just the past few years." This is where I go ============================== taylor:139:the Licklider vision of interactive computing. The 1960 man- : your reaction to them? Did you seem to think that they were doing things correctly? TAYLOR: First of all, the reason I moved from the NASA position is fundamentally because over time, I became heartily subscribed to the Licklider vision of interactive computing. The 1960 man- computer symbiosis paper [that Licklider wrote] had had a large impact on me. Have you read that? ASPRAY: Yes, I have seen that. Are you uncommon in having had that paper influence you? ============================== taylor:315:anyone who's been in that DARPA position since has had the vision : TAYLOR: Well, it's all relative. Licklider is modest to a fault. What he said is true, but he still was a knowledgeable person about scientific endeavor in general, philosophy of science, and he was a very good judge of people. I don't think that Ivan, nor I, nor anyone who's been in that DARPA position since has had the vision that Licklider had. His being at that place at that time is a testament to the tenuousness of it all. It was really a fortunate circumstance. I think most of the significant advances in computer technology, especially in the systems part of computer science over the years -- including the work that my group did at Xerox PARC ============================== taylor:322:were simply extrapolations of Licklider's vision. They were not : testament to the tenuousness of it all. It was really a fortunate circumstance. I think most of the significant advances in computer technology, especially in the systems part of computer science over the years -- including the work that my group did at Xerox PARC where we built the first distributed personal computer system -- were simply extrapolations of Licklider's vision. They were not really new visions of their own. So he's really the father of it all. And you'll never get him to admit that, because of his modesty. ASPRAY: Okay. Let's turn to your tenure as the IPTO director. ============================== taylor:323:really new visions of their own. So he's really the father of it : circumstance. I think most of the significant advances in computer technology, especially in the systems part of computer science over the years -- including the work that my group did at Xerox PARC where we built the first distributed personal computer system -- were simply extrapolations of Licklider's vision. They were not really new visions of their own. So he's really the father of it all. And you'll never get him to admit that, because of his modesty. ASPRAY: Okay. Let's turn to your tenure as the IPTO director. Can you tell me what you understood to be your mission, especially ============================== winston:123:problems of vision. I was interested in semantic nets and : might become involved. So I took Marvin Minsky's subject, wrote a term paper he liked. He invited me to work here at the laboratory during the summer, and the rest went on from there. I was, during the subsequent year, here at the laboratory as a research assistant. Tom Binford, who is now at Stanford, was working on problems of vision. I was interested in semantic nets and wondering how I might do a thesis in the general area of semantic nets, a thought that seems rather naive in retrospect. NORBERG: Why? ============================== winston:341:summer vision project of 1967, or so, as an undergraduate. I don't : and Papert, I think, encouraged that. They were interested in ideas, they didn't give a damn about degrees or anything else. So one reason it wasn't strange was because of this egalitarian outlook. The other reason it wasn't strange was because it had a history of doing that. Gerry Sussman was in charge of the famous summer vision project of 1967, or so, as an undergraduate. I don't think Marvin ever felt that he had any great interest in running projects, and he was always into the delegating the day-to-day responsibility for such projects to somebody else. And I think he just did it to people to whom he for some reason or other trusted most. Russell Moscar (?) had a great deal of... He was very much ============================== winston:686:admiration for him for doing it. He was clearly a visionary in the : WINSTON: Yes, I think that's right. NORBERG: Carrying the torch when no one else would. WINSTON: That's right. And we all had a great deal of respect and admiration for him for doing it. He was clearly a visionary in the sense that he realized there was something very good going on here. It was very clear that he was working really hard to deliver that message. NORBERG: Did you take any steps to try to change the funding of the ============================== winston:704:been run by individual princes of science, people with great vision : I always felt that DARPA could do things that the National Science Foundation couldn't. It was very clear that the differing missions of DARPA and the National Science Foundation led to different kinds of sciences. NSF after all is run by peer review, which means that it tends to favor established sciences, whereas DARPA has always been run by individual princes of science, people with great vision or a good intuition who could do things that couldn't be done if things were slowed down by an elaborate peer review process. They can start new fields if they recognize a smart, young guy like Minsky was back then and support him. ============================== winston:1172:that approach to vision did more harm than good. Marvin's : results, because he was only part way through that process when he was taken ill with leukemia. He was a big influence. He was a big influence on his own students and on the other faculty, like me for example, who admired what he was doing. That was not universal either, of course. I think Marvin Minsky, to this day, thinks that that approach to vision did more harm than good. Marvin's impatient and wants it done and felt that those guys were moving too slowly and were not thinking... Well, Marvin, would have to characterize this himself because I can't speak for him very well, because on this particular point I think I disagree. But to attempt it, I think Marvin feels that those guys were focusing too ============================== winston:1178:much on low-level early vision questions and not enough on higher- : impatient and wants it done and felt that those guys were moving too slowly and were not thinking... Well, Marvin, would have to characterize this himself because I can't speak for him very well, because on this particular point I think I disagree. But to attempt it, I think Marvin feels that those guys were focusing too much on low-level early vision questions and not enough on higher- level, more cognitive things that in the end I think Marvin feels are more important to visual understanding. NORBERG: But you say you think you disagree with that position? WINSTON: Yes, I think I disagree with that. I think it is very ============================== winston:1184:important to understand those details of early vision as a means : level, more cognitive things that in the end I think Marvin feels are more important to visual understanding. NORBERG: But you say you think you disagree with that position? WINSTON: Yes, I think I disagree with that. I think it is very important to understand those details of early vision as a means toward understanding the whole of the visual understanding. Let me elaborate on that. I was never very enthusiastic about the early work on speech at Carnegie, because it was always billed as successful because of the blackboard architecture. And my view was that if you didn't get the early phonemes processed well, it didn't ============================== winston:1195:you understand what you need to know about the early vision : matter what architecture you had on top, you'd never succeed. But on the other hand, if you got them processed well, and once again it didn't matter which architecture you had on top, because the final process can be handled by almost anything. By analogy I think you can't do much in terms of understanding an image unless you understand what you need to know about the early vision processes well enough to support the higher-level reasoning processes. So yes, I guess I believe more than Marvin does in the importance of understanding things down at that grungy level of detail where you are struggling to deal with edges and things. ============================== winston:1205:away on all sorts of vision processes with a view toward : NORBERG: Is it possible that these two positions will converge? WINSTON: Yes, absolutely, I think they're converging now. As we speak, there's a connection machine upstairs that might be humming away on all sorts of vision processes with a view toward understanding their integration. And that's possible now because of all that work that was done on edges and shading and stuff in earlier times. So integration is very much a theme of today, and Shimon Ullman, one of our faculty, takes it to an even higher level and tries to understand how early vision procedures might be driven ============================== winston:1210:andtries to understand how early vision procedures might be driven : away on all sorts of vision processes with a view toward understanding their integration. And that's possible now because of all that work that was done on edges and shading and stuff in earlier times. So integration is very much a theme of today, and Shimon Ullman, one of our faculty, takes it to an even higher level and tries to understand how early vision procedures might be driven by higher-level requirements. He calls this a theory of visual routines. He is partly motivated by psycho-physical observation, namely that some aspects of a scene seem to be computed almost instantaneously and others seem to require a serial scan across the image. He would argue the things that seemed to require a serial ============================== winston:1303:difficult period. In Marvin Denicoff we felt we had a visionary : I have another thing on my notes that I meant to be sure to bring up. That is the last time we talked about the tremendous missionary work done by David Carlstrom in ARPA in some of the more difficult times. I meant to mention, and I don't think I did, that Marvin Denicoff at ONR was a tremendous influence on us during that difficult period. In Marvin Denicoff we felt we had a visionary ally who would always be there in our defense. So we always felt that although he wasn't at ARPA, he was a person who would be telling our story at every opportunity. ==============================