MIT/Brown Vannevar Bush Symposium - Fifty Years After 'As We May Think' (Part 5/5)

Search transcript...


MODERATOR: I believe too few serious scholars are studying the social impact of computers, the way they affect us as individuals in groups and in society at large, as we learn, as we work, and even as we play. Now, that's strange to me. Because here we have the most powerful tool mankind has ever known. And we've hardly begun to analyze how it is affecting us, let alone how it is shaping us and how we can shape it so that we get what we want in the way of affects.

Lee Sproull is one of those few individuals who has devoted her entire professional life to this all-important area. After 15 years as a professor of social sciences at CMU, she's now a professor of Management at Boston University. And she's published seven books and many research papers in this area. And it's a special pleasure for me to have someone address this key topic, someone who has devoted her entire life to it. Thank you, Lee.

SPROULL: I feel so old.


I'm not a computer science pioneer. I hope that was at least implicit in what Andy said. Although some of my best friends certainly are.


For the past 25 years, I have been pleased and privileged to be an observer of the community making parts of Bush's vision real. When Doug was talking yesterday about how human systems and computer tools need to evolve together, I was actually reminded of one of my early experiences as a user of this technology in 1977.

I'm trained as a sociologist. My doctoral dissertation had absolutely nothing to do with computing. But I typed it on an Alto computer at Xerox PARC. I printed the final document on EARS. And I think there may be a few people in this room who know about EARS, one of the earliest laser printers. I collected the signatures on the signature page of my dissertation. I took my dissertation to the registrar's office at Stanford to be certified, bound, blessed, whatever it is that happens to these things.

Now, this was in the days when the production requirements for dissertations were extremely stringent. The requirements specified a certain kind of rag paper, black carbon ribbon, no erasures. So I handed in my thesis, my laser-printed thesis. The dragon lady in the registrar's office flipped through the pages, ran her fingers over the words, fixed me with a steely eye, and said, you can't turn this in. It isn't the original.


I was highly evolved. My computing tools were highly evolved for that time. But the social system in which I was doing my work had a ways to go. Still true today. I'm delighted to be here to celebrate the technology vision that we are all pleased to honor.

Actually though, mostly what I want to do is talk about the human vision or models of human beings that are associated with Bush's technology vision, and also other visions that we're hearing about today. Every technology vision that really captures people's imagination is accompanied by a view of human beings using the technology, who they are, what they do, how their lives are changed, improved, we hope, by the technology. The human vision that is associated with the memex with Bush's paper is a view of a person working alone.

Could I have the first slide, please? Working alone, not doing housework or manual labor, but doing intellectual work. The technology, of course, is envisioned to support that solo work, to make it more productive, to make it even more creative. Now, in Bush's paper, we have evoked the view of the scholar, seated at a wonderful desk, interacting with the world scientific literature, alone.

Licklider, in his 1960 paper, "Man, Machine, Symbiosis," evokes a view of the scientist or an engineer solving problems more productively with a machine doing the clerical and mechanical activities. The scientist or engineer is alone with the machine. Alan Kay evokes a view of a person of any age, especially a child, interacting with a dynamic medium for creative thought, a Dynabook, to work and play more productively and creatively, alone. Now, of course, we have to note that Doug Engelbart has always been the exception to this fairly common view of the human being associated with the personal computer. As early as 1968, of course, he was describing computer support for people working together, sharing files, working in real-time meetings.

This technology vision of personal computing, of course, is extraordinarily important, extraordinarily worthwhile. There's a whole lot more to be done, et cetera, et cetera, et cetera. But science, and Bush, after all, was speaking about the scientific community. Science and most other productive work in modern society is inherently a social enterprise.

Could I have the next slide, please? Lick was an empiricist, and therefore dear to my heart. He reported real data in his 1960 paper. He reported, in fact, that 85% of his thinking time was devoted to clerical or mechanical activities-- searching, calculating, plotting, transforming. He used his data on human behavior-- data on himself, but still, data on human behavior-- to justify his technology vision. What he didn't point out though, is that while 85% of his thinking time was devoted to clerical or mechanical activities, probably only 15% of his day was devoted to thinking. The rest was probably spent in meetings.


You've been there? Face-to-face interaction with other human beings-- I don't mean only formal meetings, conferences, symposia. Hallway conversations count, as well. There have been many, many time allocation studies of people working in a variety of fields, including scientists-- bench scientists, technical supervisors and managers, people across all of these studies, all of this kind of work-- spend between 50% to 85% of their day in meetings. And in addition to that, they spend some fraction of their alone time recovering from meetings. I mean during followup work, but I guess there's other kinds of recovery, too.

Let me give you just some more evidence and reminders of something that we all know very well, that science is a social enterprise, one indicator of growth in team size over time. It took two people to build the first airplane. It took 7,302 people to build the Boeing 777.

Author lists for papers and journals have certainly been growing over the years. In the past 40 years, the mean number of authors for papers published in scientific journals across all disciplines has doubled. The mean number of authors per paper in 1955 was 1.8, in 1994 was 3.5. Last year, in '94, there were more than 700 papers published in scientific journals that had more than 50, 5-0, authors listed on the title page. The, I think, record so far, last year, the two papers announcing the discovery of the top quark last year, each listed over 400 authors on the title page.

Now, even in academic computer science-- computer science is not known as a big science field-- the mean size of work teams is increasing. If you look at the proceedings of SIGGRAPH, the computer graphics conference, the mean number of authors per paper at SigGraph in 1975 was 1.7. In '94, it was 2.4.

If you look at the ACM Computer Chess Championship, the mean number of people associated with the winning program from '74 to '84 was two. The mean number associated with the winning program from '85 to '94 was 3.8. These are fairly formal indicators of the fact that science is not a solo enterprise.

Here's a livelier one. This is from Lewis Thomas's description of the marine biological laboratory at Woods Hole, a human institution that, as Lewis Thomas says, is possessed of a life of its own, with an extraordinary influence on the growth and development of biologic science. Lewis Thomas suggests that, to learn more about the marine biological laboratory, you might begin at Stony Beach.

He says, "It's the most minor of beaches, hardly big enough for a committee, but close enough to the laboratories so that the investigators can walk down for a sandwich lunch with their children on sunny weekdays. On weekends in hot midsummer, it is so crowded that one must pick one's way on tiptoe to find a hunching place. But there is always a lot of standing up anyway. Biologists seem to prefer standing on beaches talking at each other, gesturing to indicate the way things are assembled, bending down to draw diagrams in the sand.

By the end of the day, the sand is criss-crossed with a mesh of ordinance, abscesses, curves to account for everything in nature. You can hear the sound from the beach at a distance before you see the people. It is that most extraordinary noise, half shout, half song, made by confluent, simultaneously raised human voices explaining things to each other."

Now that's what science is about. Personal computing, fortunately, has evolved to allow interpersonal communication, to allow people to explain things to each other, to talk with each other using computers. Could I have the next slide, please.

You might just do a little thought experiment. What would the world be like today if there were 100 million personal computers and no interactive communication, no ability to contact people or create and sustain human relationships with other people through these machines? Actually, I'm not sure we would be having this meeting. And I don't mean just, we wouldn't have gotten the logistics of it organized. I'm not so sure we would be here to celebrate the way the vision has come to be played out.

Of course, we shouldn't feel sorry for Bush that he didn't predict the power or importance of interpersonal communication in the context of his technology vision. He's in good company. Not only the history of personal computing, but even the history of computer communications features technologists who underestimated the lure and importance of human communication using this technology.

Everyone in this room is familiar with the ARPANET and email story. In fact, some of the people in this room were the heroes or villains or something in that story. But the conventional wisdom, anyway, has it that the ARPANET was conceived to link computers. And of course, we all know it turned out to link people, too, initially through email, distribution lists, and bulletin boards populated by members of the same intellectual community, who were housed in a small number of universities, industrial labs, a fairly homogeneous bunch of folks, truth to tell. Although they might not be pleased to hear me say that.

More recently, as internet access has soared, the human links have been formed through mechanisms, like newsgroups or usenet groups, that are populated by people who may have absolutely nothing in common with one another except for internet access and a shared interest in the topic of the particular group. Commercial electronic services providers, like CompuServe and Prodigy, also didn't anticipate the importance of the human communication services, like email, chat rooms.

Now, scientists certainly don't just process information alone in their offices. Moreover, they don't just communicate either, generically. They do these things within social contexts. That is, through multiple ongoing relationships with other people and organizations.

I must say, I have been struck by the serious observation made by all the speakers thus far that people are important and social issues are important. And obviously, I'm glad to hear the speakers say that. But there's a sort of funny gap between how the previous speakers talk about people-- by which they mean individual human beings who speak into a microphone or type with fingers on keys or whatever-- and society, which is big forces that are good or evil or something, but they're huge.

There's a very large middle ground that's very important. And it's the middle ground where most of the productive work gets done in society, through multiple ongoing relationships with people in organizations. Could I have the next one?

Each relationship, every one of them, is structured by roles. Roles have formal responsibilities and obligations associated with them, norms, accepted rules for behavior. Each relationship is shaped through reciprocated action of people working together. Each relationship is sustained through trust-- trust, by the way, doesn't have to mean liking-- but sustained through trust, obligation, and affiliation.

Now, people may complain that they spend too much time in meetings, often because meetings aren't always an efficient information exchange mechanism. But information exchange isn't the only reason for going to a meeting or for participating in an ongoing social context. People like to talk with other people. People go to a workshop, even when the papers are online. Showing up signifies the importance of the occasion and the importance of the community.

Let me give you an example of a scientist doing productive work through multiple ongoing relationships. And by the way, I apologize for the sort of wishy-washy relationship word. I couldn't figure out a nice, crisp alternative, so we're going to have to go with it.

Think of a scientist at the weekly meeting of his research group who's critiquing the research presentation of one of his doctoral students. This scientist is simultaneously helping to create new scientific knowledge, perhaps describing additional analyses that the student could do in order to refute the claims made by a competing lab, in addition, reminding the student to verify the analytic techniques he's just been describing with the new postdoc, who has recently arrived from a colleague's lab where these techniques were first developed.

That's one thing that the scientist is doing, helping to create new scientific knowledge. At the same time, the scientist is training the next generation of scientists through suggesting the additional analyses, through, perhaps, critiquing the format of the student's slides, through asking the student to present his results in the professor's graduate survey course, helping to create new scientific knowledge, training the next generation of scientists, deciding which of his colleagues labs to recommend as a good postdoc placement for the student when he finishes his thesis, deciding which journal to submit the student's research to, and remembering that he still owes the editor of that journal reviews on three papers.

Thinking about how these results are going to affect the next round of studies, and deciding he's going to have to apply for new funds sooner than he had hoped, realizing that he's going to have to say yes to the provost's request that he chair the university task force on graduate education, if he's going to have any hope of getting the provost to approve his request for additional lab space.

Now, the manifest task, when we're thinking about this scientist, the scientist's manifest task is to critique a research project. But all of the other tasks are equally necessary to being a productive scientist. They all entail information processing, absolutely. But each of these information processing tasks is embedded in a particular social context, a particular social structure of reciprocal human interdependence, a structure that's sustained through trust, obligation, and affiliation. Furthermore, the contexts the scientist is operating in simultaneously operate at different levels of social aggregation. Let me see if I can explain a little bit what I mean by that. Could I have the next one, please.

The social enterprise of science operates at three levels of aggregation. At the institutional level, we have the human conventions and organizations that certify knowledge and allocate resources. We have the Nobel Prize committees. We have the scientific journals with their editors and editorial boards. We have the research funding agencies.

At the middle range, we have the distributed colleagues and competitors working in a single area. These play themselves out in conferences, summer studies, postdoc exchanges, pre-print mailing lists. The third level of aggregation is the close in level, the day-to-day interaction with colleagues and students. This is the place, this is the level of aggregation, where the apparatus is built, where the measurements are taken, where results are initially puzzled over and discussed, and where the next generation is socialized. This is also the place where, as some people would say, if you don't have to get on an airplane to go somewhere, this is where you go, the close in level.

The productive scientist has to work simultaneously at all three levels of aggregation and has to process information and manage relationships at all three levels. So what can we say about what computer support for interdependent actors looks like today? What are the systems and applications that acknowledge interdependence? Next slide.

I think the big success stories are electronic mail and file transfer. I don't know if there's anyone in this room who would think they're glamorous or represent research frontiers for today. But they certainly, in terms of numbers of people who use them to get productive work done, are the success stories, unstructured general tools. People can and do use them in any social context-- well, probably not yet to the Nobel committees, but in most other social contexts. They require very little social and cognitive overhead to use. But of course, they're just the beginning of what we need.

What are people working on today? I feel a bit audacious telling this audience about what computer scientists are doing today. But I think some of these kinds of work are not ones that would be particularly visible or perhaps even particularly interesting to you, so let me just say a very brief word.

At the other end of the spectrum from unstructured email and file transfer is workflow, something that, unfortunately, business organizations are quite enamored of today. I don't want to dwell on it, but it does acknowledge interdependence. It does things, like attempts to automate things, like administrative processes, like travel requests or payroll or some business processes, like approving loan applications.

It's sort of a transaction processing thing. The focus is on the flow of transaction tasks across processors. Some of the processors may be human beings. But the focus isn't at all on all of an individual human being's responsibilities, obligations, and tasks as they're structured in roles in an organization.

Another kind of work that a bit of which is going on today, there is something called CSCW, which stands for computer supported co-operative work. And there is groupware, allegedly support for people doing intellectual tasks. CSCW, the acronym, was created, I think, in 1986. So this is certainly not an old and venerable subfield. But at any event, some of the examples of things that people are working on include working with shared files, shared design spaces, shared document editors, group awareness tools, group meeting rooms, to support face-to-face meetings or geographically distributed meetings.

Now typically, in much of this work, the focus is on a small group. I suspect a part of the reason is it's sort of hard enough to get the technology to work for a small number of people. I'm not clear that people working in this area believe that the small group is the only useful or important level at which people do work. But at any event, much of the work going on now focuses on groups from size, say, two to 10.

Also, much of this work takes a single task focus. For instance, there's work on the task of collaborative design. There's work on the task of brainstorming in meetings. These applications are often highly structured. They may force people to state assumptions or to specify commitments and to state their binary choices. You're either in a meeting room or you're not. You're either present for the video conference or you're not.

You're either explicitly making a commitment or you're not. I think that computer support for the productive scientist, who doesn't just process information but who works through relationships with other people and organizations, needs to have a whole different flavor. Could I have the next slide, please?

It shouldn't just support a task, a manifest task. Because the scientist, of course, is doing multiple different tasks simultaneously, only one of which is the manifest one. It should not just support a group, because the scientist is simultaneously participating in multiple groups at different levels of social aggregation.

So what should this computer support do? It should support the scientist who works in multiple groups simultaneously, who works at different levels of social aggregation simultaneously. It should move with the scientist as he or she physically and cognitively moves from group to group throughout the day. And it should allow the scientist to map his information environment to his social contexts, not to arbitrary physical locations like a meeting room or temporal constraints.

Now, since I'm not a technologist, it would be extraordinarily silly for me to tell you what the technology should look like. Instead, what I'd like to do is describe some social facts that, if we had technology support for them, could lead to more productive behavior. Your assignment is going to be to invent the technology to support the social facts.

Let me offer you three examples. Could I have the next one? The first collection of social facts has to do with the absent meeting participant. People have multiple ongoing responsibilities. Their responsibilities obligate them to participate in particular real-time meetings. Conflicting obligations mean that not everyone who should attend a given meeting will be able to do so.

In fact, I'll bet that every one of you who is sitting in this room today is missing one or more meetings back in your home spot that you should be attending. I think we need technology support for the partially present meeting participant. Now, I have to tell you, I've been in a number of meetings where the people in the room were partially present, but that's not what I mean. What I mean is for the person who should be at the real-time meeting but can't be.

There should be easy ways to review the meeting after the fact. And I think, actually, in Raja's videotape of earlier today, I caught a glimpse of how that might be able to happen. But that's only half of the story. That's good for the person who can't be at the meeting. But for the people who are present at the real-time meeting, it should be just as easy for them to signal the attention of an absent member as a present one, to ask questions of the absent member at particular points, to request action. The only difference will be that the partially present participants will be delayed for a bit. So I hope you can invent technology support for the partially present meeting participant.

Second social fact-- the meeting isn't over when it's over. Most meetings are embedded in an ongoing social context. A meeting doesn't end when it's over. People figure out, after they walk out the door, what they wish they had said. They think of new ideas to contribute. They do what they said they would do during the meeting, or they don't and they have to be reminded.

And so I want you to invent technology support for the post-meeting meeting. After the face-to-face meeting is over, it should be easy for participants to slide new ideas, insights, clarifications, contributions, into the ambient information environment of the meeting. These could be from fully present participants or partially present participants.

Third social fact-- group membership is dynamic. Groups add new members. These members not only come with a different set of skills, which presumably is why you put groups together in the first place, but they also arrive with different obligations, different ways of doing business. Bringing a new member up to speed is necessary for the member to function productively in the group and for the group to function productively.

Creating a new group is simply a special case of adding members to an initially empty group. I know that Marjory Blumenthal had been invited to attend this meeting and was unable to be present. Marjorie runs the panels that are associated with one of the National Academy of Sciences committees. And I think Marjory will particularly appreciate your invention of technology to support the dynamic group member, technology to support adding members to an initially empty group.

The process of convening these advisory committees is one that she certainly handles with expertise. But it takes a lot of time and energy. The rest of us who have to convene groups of people from time to time I think could certainly appreciate that.

So what would technology support for the dynamic group member look like? Well, one of the things that we certainly would want to have is a "who are these people" button. When a new person enters a group, he or she can easily learn about the accomplishments, skills, social circles of other members, and vice versa. Of course, it will be easy to go deeper if you wish by reading papers or examining other work products that are referenced in the first level of the "who are these people" display. So that's one piece of it.

Second piece is automatic configuration of group communication protocols. Now, Nick was telling us this morning that, in some number of years, we'll be able to shake hands and all of that stuff will pass automatically. That's one way of doing it. I guess I don't know that I always want to shake hands with everybody.

But it would be nice to have other way-- well, it would be nice to have other ways, maybe that computers, without having to get your hands involved, could exchange information about email addresses, video conferencing connections, the conventions for producing work products that each individual person brings with him or her into the particular group. What versions of what software do people use? I don't know. Maybe in the future, there won't be versions. But I sort of think there will be.

And all of that information should get exchanged automatically. That's automatic configuration. Then there should be easy ways to configure at least some of the relationship protocols. What do I mean by that? How open or closed is the group? When and in what form can its work products be sent outside the group? Are drafts confidential? Can drafts be circulated to colleagues outside the membership or the particular group? Who must review work products before they can be shared more broadly? Who's in charge? What's the calendar?

Now, these are [? gizzies. ?] I think that's a technical term. And I hope it's clear that I'm not asking you just to build [? gizzies. ?] Can I have the next slide, please? These are just examples of what we could hope to find in a technology that supports a scientist who is not simply a solo information processor, not simply a generic communicator to the galaxy, but a person who works through simultaneous relationships, structured relationships, with other people and other organizations.

I also hope it's clear that my vision will require new social inventions, too. For example, the partially present meeting participant is a social role. We'll have to develop norms and beliefs about this role, ways of proceeding. The "who are these people" button requires-- I can imagine this being used with [? Avida ?] for academic people. But for non-academic people, people who do not work in an academic organization, we need useful ways of summarizing a person's accomplishments, skills, social circle, and so on.

Although I have explicitly tried to focus on science and the doing of productive scientific work, I'd like to look beyond that domain for a moment and suggest that technology that can support scientists, who already are working in multiple simultaneous relationships, may also support people who need but do not have a multiplicity of ongoing relationships. Many human beings are physically alone or feel socially isolated. They perhaps could benefit from having more people in their ambient information environment.

Some usenet groups, as I suspect all of you in this room know, already serve this function to some extent for some people today. There is a newsgroup or listserv for practically every malady or misfortune of the human condition. And some of these have been extraordinarily important in the lives of people who participate in them.

Now, technology to support the dynamic group member could be used to help manage electronic support groups. Support for the post-meeting meeting could be useful perhaps after a parent-teacher conference, when the parents think of additional questions they should have asked during their 15 minutes allocated with the teacher. Support for the partially present meeting participant could be useful for someone who is unable to attend a spouse's medical appointment, but who needs to be involved in the treatment plan.

I think that the vision of the solo individual made more powerful, more creative, more productive through technology will continue to be beguiling, seductive, important, fundamentally important. New inventions for personal computing based on that vision will and should continue. Could I have the next slide, please?

What I've been trying to do, though, is ask you to think about how the person embedded in multiple social contexts can be made more creative and more productive. Perhaps the real challenge for the web is to support social context through socially structured information. I think that the challenge for technologists, more generally, is to invent support for people who are not just solo information processors, not just generic communicators, but who are embedded in and accomplish work through interdependence with other people. Thank you.


That was sort of different, wasn't it?

MODERATOR: Thank you very much.

SPROULL: You're welcome.

ENGELBART: Folks, question time. Let's see, can we have the house lights a bit so that Lee can see you?

SPROULL: Hello, Michael. Yes, sir.

AUDIENCE: Michael Lesk. Garrison Keillor writes about the rise of the shy, this feeling that, as computer technology means that interaction at a distance replaces or supplements face-to-face interaction, the personality type that I guess we associate with automobile salesmen will become less important in society, and the type associated with the wimp sitting at a typewriter will become more important and dominant. Do you think this is accurate or that anything like it will happen?

SPROULL: Actually, I've done some research documenting the rise of the shy. And since I am one of those folks, I'm delighted to report that computer communication does, in fact, bring with it benefits for folks like me. I guess, though, I think we'll always still have automobile salesmen. So I don't see it as one becomes more influential or present and one less. I don't see it as a 0 sum game, necessarily. Nor do I think that real human beings are going to stop talking to real human beings face-to-face.


AUDIENCE: Stuart Card. As communication becomes easier with larger and larger numbers of people around the world, and, in fact, as we succeed in being able to collect more and more information from the net and so on, there is more stuff to do. But the amount of time somebody has remains constant. So one of the real issues seems to me is the management of attention among all of these things. So can you conceive of ways to support the scientists to get less communication or to have more time, to be more solo. [LAUGHS]



No, Stu, I didn't make my point clear. The scientist doesn't do science solo. The scientist does science in a socially structured group. And the group may be able to figure out ways to sustain its boundaries so that information, or stuff, data, isn't totally swamping them.

I think that graduate students are wonderful information filter retrieval devices for professors. And I don't want to propose that we get rid of them or replace their functions with computing technology. I really was trying to make the point that, if you try to invent technology for the solo information processor or the generic society, you miss out on a lot of the helpful stuff that social organizations are already doing for us, structuring people's attention already and providing mechanisms and roles for helping manage information flow.

AUDIENCE: Samuel Epstein with Sense Media. We operate something called the sprawl. Over a little a year and a half ago, we took some technology developed at Xerox PARC called LambdaMOO. And we modified it enhanced it-- that's what we call it, some people think we defiled it-- to support hypermedia, by adding support for hypertext and being able to inline graphics and things like that. And right now as a result over a year and a half, we now over 4,000 people on our server in Hawaii. And we call it a collaborative hyper media server. And it integrates together text and graphics and chat and mail and news posts.

And what I'm wondering is if you've done any research into some of the social interactions that are going on there. It's a very rich and diverse society with a bunch of people really working together all over the world to create hypermedia presentations. And this technology is now spreading out into lots of servers all over. There's some stuff being done at Brown University, as well. And I was wondering if you studied any of these new societies are starting up. They're mostly populated with people under the age of 20, but they're doing some incredible work.

SPROULL: Let's see. The short answer to your question is no. I haven't yet studied these communities. I'm certainly looking forward to doing so. I mentioned, when I was talking about the inventions that I hope you folks will do, the fact that there have to be new social inventions as well. And at the moment, I'm not sure that we have good social inventions for how to have 4,000 people work together productively on anything. People that you're describing are doing that invention now. At the same time, you're inventing or modifying or whatever the technology tools. So I look forward to having the chance to try to systematically understand some of these kinds of things that are going on.

However, let me return for a moment to productive science. Whereas there may have been 400 authors on the top quark papers, there certainly were 4,000 people involved in some way the production of the research that led to those papers. And those organizations are highly socially structured. And I think it would be interesting to learn from how-- to take the best of how those organizations work-- and lord knows they do not work perfectly-- but to take the best of that and to try to understand how that can also be useful in your kind of new world. Ron.

AUDIENCE: Ron Becker, University of Toronto. Lee, I'd like to thank you for a very interesting presentation. And it strikes me, though, that the scientific communities, in general, are resistant to understanding and appreciating and believing in the social context of science. And I go back to Bush. It strikes me that he probably should have seen some of that, in terms of the Manhattan Project, which was sort of one of the first examples of big science. And I wonder whether, first of all, you agree that communities are resistant to believing this and appreciating this and why this is. Is it something about the way we educate scientists or whatever?

SPROULL: Thank you for the thank you. I think it's partly a function of how we educate scientists. I think it's partly a function of how scientists are. It's important that there be a hubris, that there be a belief that the solo inventor, thinker can change the world. Usually, it doesn't work out, but you have to have that belief. It's fundamentally important to the survival of the scientific enterprise.

So I'm not discouraged or unhappy when I see that. It's natural. I think that often, as scientists filled with hubris, age, they come a bit more to appreciate the importance and necessity of the social structure and the social enterprise that supports their work. Unfortunately, that sometimes happens too late for them to then proselytize for the funding of social science research.



AUDIENCE: Peter Denning. There is a book that I always like to take inspiration from to combat the solo scientist. And that's the book by Bruno Latour, Science in Action. Many of the other scientists I've given that to to read loved the book. But when they get back to their own academic departments and start considering merit and promotion, they always look for the solo contribution.

And they have all sorts of ways to dig and see if they can find it. So we really do have kind of a contradiction between where your heart wants to go-- your heart wants to appreciate the social context. But when you put it in action, you want to find the solo inventor contribution. Do you have any suggestions on how to change that?

SPROULL: No. And I don't think, actually, that we need to worry about changing that. I think what you need to do is continue to push, to invent the technologies, to invent the technology environments that will make it obvious to one and all that scientific collective work is where the productive work of science occurs. But again, I don't want to stamp out hubris. I mean, it's kind of obnoxious in faculty meetings. [LAUGHTER] But it's very important for the scientific enterprise to continue.

AUDIENCE: Another question for you-- you named a few names that are more on the solo inventor side. What are some of the names that you would hold up as examples to look at for some of the other parts that you were talking about there-- for example, workflow or CSCW?

SPROULL: Well, Doug Engelbart, Doug Englebart, and Doug Engelbart, to begin.


And I hope you heard that I did acknowledge him initially. All the work that all of the people in computer communications and shared files are doing today-- I guess since I'm not a technologist, I never feel the need to put down anybody's stuff. I say, yeah, do more of that, but let's do some other things, too. I want to expand your world, not stamp out some things. But let's add more people and more ways of thinking about inventing technology support for interdependent people.

MODERATOR: Thank you. You getting hungry?


Thank you, again.

SPROULL: Thank you.

MODERATOR: Terrific.

What's been really fun about this symposium for me is that I've been able to introduce so many insanely great people, in the words of the late great Steve Jobs. And certainly one of the very topmost of those people is Alan Kay. Most of you identify him with personal computers and personal computing, and at a time when that idea was still embryonic or, as some of us thought, lunatic fringe.

He's made seminal contributions to object-oriented programming, graphical user interfaces, and window managers via the Smalltalk language and its programming environment. He's always been interested in the appropriate use of computer technology, and especially in the area that appeals most to me, education.

Alan is also, as most of you probably know, a very serious musician. He was a professional musician. And I try to characterize him-- I first thought of him as a polymath. But many people don't know what that is. So then I thought about interdisciplinarian. And Alan didn't like that. So we came up together with calling him-- Alan, what was the word? Trans--


Then we had to fish for that.

AUDIENCE: [? Transcendalist. ?]

MODERATOR: So trans everything that I know about, here is Alan to finish up for us.

KAY: Thanks. Thank you. We had to use the trans prefix somehow since Ted is here. Well, when I was thinking about what to do for this talk, when I kept on thinking back to the period when I think that Bush was most influential in the ARPA world, and I was thinking that it might make a nice talk to just think about the '60s, just to cut it off at the end of 1969. It just happened that 1970 is when Xerox PARC started up. So cut it off before Xerox PARC. And when I was thinking back, the more I thought back, the more I thought of some of the interesting people who influenced that very, very interesting and exciting decade.

So I thought to start off with a little trip down memory lane. And as a kind of a context for the talk, I thought I'd start off by showing you Arthur Koestler's-- Arthur Koestler was a great novelist who, in the last years of his life, became a psychologist doing research in creativity at Stanford. And one of his characterizations of creativity sort of goes as follows. He said that, our minds are kind of constrained to think within contexts or within systems of belief. Sometimes they're called paradigms.

But he was thinking of something perhaps larger, that we have a set of commonsensical things that we accept as reality. And our thought patterns sort of jog along, staying in this particular, here, a pink plane. Every once in a while, we have little outlaw thoughts that are regarded as lies by the pink plane. And they're usually just suppressed and pushed back. And every once in a while, when you're taking a shower or out jogging or something, you are relaxed enough to let one of these outlaw thoughts flourish a little bit.

And sometimes what you find out is that that little pink idea, when expanded into the blue context, is much more interesting than it was in the pink context. So the idea here is, terrific ideas often hide behind good ones. Good ones are the ones that are-- this is like peer review and NSF.


They always fund good research, but they almost never fund great research. Because six people are trying to decide whether what you're trying to do is reasonable ahead of time. And usually the way it works is, you decide whether it is reasonable by looking what happened. And then Koestler said-- and what happens when you get one of these big blue inspirations-- oh, sorry.

Well, I goofed. I left out a slide, so I'll just tell you here. He said, there are three reactions. If you are telling a joke, then the reaction is ha-ha. If you're doing science, then the reaction is a-ha. And if you're doing art, then the reaction is ah.


Because he points out that, in each of these cases-- for instance, a joke is something that takes you down the garden path and then suddenly reveals that the context you're in is not the one you thought you're in. Science has this aspect, as well. You suddenly see it was there all the time. And often, besides saying a-ha, you start laughing. And of course, in art, the whole idea in art is to try and take you out of the world that you're in, your commonplace world, and reveal that there's more to the world.

And one of the maxims here is that, knowing more than your own field is really helpful in doing this. And I've always thought that one of the reasons the '60s was so interesting is that nobody was a computer scientist back then. And everybody who came into it came into it with lots of other knowledge and interests. Then they tried to figure out what computers were. And the only place they had to use for analogies were other areas.

And so we got some extremely interesting ideas from that. And of course, the reason being educated is important is simply because you don't have any blue context if you don't have any other kinds of knowledge to think with. So engineering is one of the hardest fields to be creative in, just because it's all about optimizing. And you don't optimize without being very firmly anchored to the context that you're in. So what we're talking about here is something that is not about optimization, but actually in kind of rotating the point of view.

Now, I was five years old in 1945. And I must have seen the Life magazine article, because I used to peel through Life magazine. But at that age, I was much more interested in the manila envelopes that life came in because you could draw on them. I'm sure that memex made no impression on me back then. I ran across it when reading a science fiction story by Robert Heinlein in the '50s. And Heinlein had this nice thing of taking terms that he had found out about and making them generic. So he spoke rather casually in the science fiction story of a memex.

And whenever he did that, I had found out that that was a good reason to go to the library and trying to find out what basis memex had. That brought me to reading Bush's article in the '50s. And I'm ashamed to say that, aside from being interesting in the science fiction way, it made nowhere near the impression on me that it did on Doug Engelbart and some other people. It wasn't actually until after I had met Doug Engelbart and had heard him talk that I realized that the ideas of Bush had tremendous significance. But I got that through the eyes and the ideas of Doug Engelbart.

So for me, I start off in the very early '60s, 1961, doing data processing for the Air Force. And back then, they didn't really have operating systems. They were very ad hoc. And Air Training Command had many bases. And they had to move files around. And somebody, to this day unknown-- I once tried to find out who it was that thought up this idea-- had come up with a wonderful notion for allowing data to be used from base to base without having to have lots of standards and formats and operating systems.

And the way it worked was this-- is that each file on the Burroughs 220 tapes had three sections. The first section was a set of pointers into the second section, which was a set of code. And the code knew about the third section, which was the formats of the data records. So in modern terms, what you have here is a generic interface, a set of methods, and then the data parts of objects.

So that is the first example that I know of in which this object-oriented idea was actually used. Of course, it wasn't thought of in those terms back then. But it was something that was used quite a bit in the Air Force until they decided to standardize on COBOL, in which case all of this stuff got thrown out, and they went back to processing data.

Another thing I saw that didn't make an enormous impression on me-- in fact, I should say, the other thing that happened was is that the third machine I learned down here in the Air Force was the Burroughs B-5000, which I think has seven of the 10 best software ideas directly in the hardware. And at that time, I only appreciated a couple of them. The fact that it was very easy to compile too was pretty much all I understood about that machine.

But it also had something like this technique of being able to deal with [INAUDIBLE] indirectly. They could do all of the things that you're normally used to doing with bits but without having to know what the bits look like or where they were. I didn't appreciate that at the time. In '65, I was actually helping debug the Control Data 6600, which is my vote for the first RISC machine.

It was about 1965. It was a 10 MIP machine with the kind of instruction set and parallel operations that we associate with RISC machines today. And while I was working on that machine at Seymour Cray's lab, I read this first article by Gordon Moore showing what had been happening in integrated circuits up to 1965. Up to that point, it had been doubling every year.

And so his first cut at it, was what if it kept on doubling every year? So that was interesting to think about. But from my standpoint, I really couldn't make any extrapolation or jump from this freon-cooled 6600 that could detect how many people were actually in the room at a given time, because of the capacitance on its doors from human beings move moving around, to what this might actually mean.

Now, a couple of years later, Gordon Moore looked at it a little more closely, looked at the physics of the silicon, and said, well, it probably won't be worse than a factor of two every two years. So that's that lower, less sloped line. And just for interest, what actually happened from about 1970 on is that magenta line there. So an interesting thing about Moore's law was, it was actually a 30-year prediction.

And at least the one that he did in 1967 or '68 or so, he was able to bracket successfully what was going to happen exponentially in silicon for about 30 years. Quite a remarkable, almost unbelievable-- and I think that most people in the industry that were going to be affected incredibly by it couldn't believe it at all. At the most, they thought it might mean that they would have more margins on their mainframes. But the idea that this might mean millions and millions of computers, I don't think affected anybody at the time, including Gordon Moore.

But when I went to the University of Utah sort of by an accident, the first person I ran into was Dave Evans, the head of the department. And the "this" that's referred to there was Ivan Sutherland's thesis. He had a stack of these brown theses from Lincoln Labs on his desk. And he had to take one of these and go away and read it before you got a desk.

And I must say, I didn't really understand Sketchpad very well. But it seemed to be about a system that was not just about making drawings. And that was an idea I hadn't really thought about either back then. But it was also the idea that the drawings generally had meaning, especially in engineering. And what you would really like to do with these, most of the drawings you make in engineering is to see whether they actually work.

So the underlying notion of Sketchpad-- in fact, it was conditioned by the fact that the display on the TX-2 at Lincoln Lab-- which was a machine, by the way, about the size of this room, maybe a little larger. You're just seeing one small part of it here. The display was so bad that it had discouraged previous people who had been asked to do computer graphics on it. And the difference here was that Ivan looked at this terrible display and asked the question we should all ask more often, which is, what else can it do?

And when he thought about it, he realized that, although the pictures were going to be terrible and the drawings done by the plotter were not going to be up to engineering standards, that what else it could do was to simulate. And this is an idea that was not strongly in any of Bush's writings, in spite of the fact that he was one of the developers of early analog computers, which in fact did simulate. And of course, he had funded ENIAC, which simulated up to a certain point.

But if you look in his writings, you'll see that he thought that certain logical reasoning would eventually be able to be done by computing. And he had some real jumps, as far as transcribing speech. But the notion of being able to deal interactively with simulations was not something that was really in his writing. This was an idea that came about from other people. And I thought, just for the fun of it, to show you one of the movies made the late summer of '62, the first set of movies made of Sketchpad.

And in reviewing this, there's one place on the tape where there's some tape damage. So we'll just ignore that as we go along. Let's roll that first tape. See you dip the light pan into the ink. And this rubber band technique Ivan invented. And why they call it Sketchpad is that you could put in something very quickly and loosely, and then give it some rules. And here, the rules are to be mutually perpendicular. And what you saw there is Sketchpad figuring out how to do that.

Now, you'll notice this is the first system ever to have a window. And in fact, the simulated paper that he's drawing on is about 1/3 of a mile on a side, I believe. And again, a couple of lines drawn in quickly, and then giving the constraints to make them be perpendicular to the lines that they're terminated on.

And now the constraint is colinearity. So he's using these two lines he drew as guidelines for these dashes. And now he will tell the guidelines to be invisible. So now he's drawn a hole in this flange. I think this is where the damage shows up. So now he wants to do a rivet. It'll clear up in a second. Sorry. Or maybe it won't.

So he's using the center of the crossbars there as the center of the circle. And again, he's saying, let's make these guys mutually perpendicular. Well, that changes the crossbars and that changes the circle, so you get a nice, symmetric rivet. In fact, Sketchpad could solve some very interesting nonlinear problems. That bridge that you saw on the slide, you could draw bridges in, give the rules for stresses and strain of beams, and it would calculate how much stress and strain there would be for given weights hung from the bridge. You could draw in electric circuit elements, put in Ohm's law, and so forth.

So now what he's going to do is make some rivets for his drawing here. The controls here, he's using one hand with the light pen. And on the other hand, he has some switches for commands for different constraints. And he also has some knobs that he's turning. So you can see this system led to an interest in better displays. It's actually not even a line drawing display. It's putting up each dot individually.

And in fact, there's a kind of a twinkling pattern that's put up there. So he says, oh, I forgot there's these crossbars on the rivets. So I'll go to the master drawing and I'll get rid of those guys. And lo and behold, we see that these rivets here were actually instances. So Sketchpad was really the world's first object-oriented system. And he really had pretty much all of the paraphernalia of object-oriented programming.

And now he's made a flange with a rivet in it. And he can use anything that he makes as a master. So now he's making instances of the flange with the rivet, and positioning them and scaling them and so forth. So I once asked Ivan, how could you possibly-- okay, stop the tape. That's enough.

So I once asked Ivan, how could you possibly have invented computer graphics, the world's first object-oriented software system, and, as you can see, one of the very first dynamic problem solvers that could do both linear and nonlinear problems? And Ivan said, well, I didn't know it was hard.


And in fact, if you read his thesis-- Bob, do you know, did it ever get published out in the world? No. Besides still being great today-- I mean, it's really saying something to be able to read something that is interesting today. That is, you can actually learn something from it by reading it today and comparing what's going on. What is really interesting is the diffident and even apologetic tone in Sketchpad because it didn't do more.

And this I think is true of most of the people that I think of as the great people in the '60s, is that they really wanted it all. So the image of what Sketchpad should do went far beyond what it was able to do. Although, when we look at it today, we are just awestruck by what he was willing to take on just from scratch and what he was able to accomplish.

Well, right after that, I found my desk. And they had a tradition at Utah that the latest graduate student got the latest dirty task. And actually, I was graduate student number seven. So there were actually quite a few dirty tasks left for people to do. And there was a pile of listings and tapes on the desk that were supposed to be the ALGOL for the campus computer. It turned out to be the first SIMULA system. And the documentation was quite incomprehensible.

And finally, another graduate student and I unrolled the listings for it down the hall of Merrill engineering building, some 80 feet long or so. And he crawled on it on one part of the listing, and I crawled-- imagine doing this on a display today. It was very useful that he could crawl over it down there and I was crawling over it down here reading the machine code, try and figure out what the thing did. And we finally realized that it was allocating storage exactly the way Sketchpad did. In fact, it was kind of a programming language version of the same kind of internal model that Sketchpad had.

And the good news was, Sketchpad couldn't solve every problem you posed to it. So it had a limited range of things it could do. But it did those very elegantly. SIMULA was a generalized programming language. And so you could pretty much do anything that you could write programs for, but none of it was particularly elegant.

Now, in my checkered past was a degree in molecular biology. And this book had come out in 1965. This visit to Utah started in '66 or so. And I'd read this book by Watson. And one of the neat things he did in this book-- this book is still in print, but in a current edition. The first edition of this, I think, was the first book that ever tried to do an assay of living things. So the favorite animal that molecular biologists looked at then was E. coli, which is a bacterium we have in our gut by the millions. It helps us digest certain kinds of food.

And by that time, they were able to tell us that about 70% of E. coli were water and small charge molecules, which actually were, of course, important for the reactions that were going on there, but didn't interact informationally. But there were about 120 million organic components that interact informationally, mostly enzymes, proteins of various kinds, and DNA and so forth. And if you work out what these are in terms of representing them on a computer in a fairly compact way, it's about 100 gigabytes of stuff going on.

And so I wrote down there, it's about 50,000 2-megabyte desktops worth in an E. coli. And one of things to think about here is how fast the reactions are actually going on. At the temperature 37 degrees centigrade, which is pretty much what your body temperature is inside of you, these medium-sized protein molecules, those little white things that look like popcorn there, are able to move their own length in about two nanoseconds.

People, when they take chemistry in high school, most chemistry teachers actually don't understand why chemistry works, and they will shy away from any questions from students asking, just how can atoms and molecules find each other in order to have these reactions. And the answer is, is that the thermal agitation at that level of scale is unbelievably violent.

I'll give you an example. If a carbon atom is about the size of a tennis ball, then one of these protein molecules is about the size of a Volkswagen. So in our scale, if you can imagine a Volkswagen made out of tennis balls, moving from here, its own length in two nanoseconds, and if you multiply that out, you'll see that that is a scale speed that's faster than the speed of light. So if we were actually transported down in size to being able to look at what an E. coli is doing at normal body temperature, we would be stunned by the violence. In fact, we could not see anything that's going on. It's just unbelievably violent. And that is why biology and chemistry actually works, something that's very hard to get into people's intuition.

So that is a very complex and interesting structure. And it happens to be 1/500 the size of a typical cell in our body. And we happen to have between 10 trillion and 100 trillion of these cells in our body. And yet 50 cell divisions are all it takes to grow a baby.

Now, think of taking something much, much simpler, like a 747 that Lee Sproull was-- she was talking about a 777. And imagine what you have to do to make it six inches longer. The manuals for it are bigger than the plane, if they were listed out. I'm sure they're on CD-ROMs now. And yet a baby gets six inches longer about 10 times in its life, and you never have to take it down for maintenance.

So what we have here is something that is a kind of informational machinery that is just in a different league than computers are even today, and certainly, if you look at the number of components in animal cells and think of how many we have. The other thing about them is that they're dying all the time. And not only that, they're actually exchanging their structure.

So another interesting thing that I knew from molecular biology is that, no atom in our body stays in our body for more than seven years. And in fact, more than 90% of the atoms in our body are recycled every two weeks. So if you think about this, this is really quite startling, if you think of yourself as actually a pattern moving through time and space, sucking up energy and material and reformulating it to be yourself. And yet every seven years, you have not a single atom, not even in your bones, that you had seven years before.

So this is a completely different way of thinking about things. And things are going wrong all the time. But in structures like this, more things are going right. Well, something about SIMULA got me thinking about computing. I suddenly realized that computers were tiny, tiny, tiny, little simple things.

Physically, they were big. But compared to biology, they were nothing. Hardly anything was going on. And the only reason programming worked on them was because people were hardly doing anything. And what came to my mind, it was kind of like making a doghouse. Which, unless you make it out of toothpicks, you can make a doghouse out of almost anything, because the structure is so simple compared to the strength of the materials. It's only when you go to big things that you have to really introduce architectures to make them work.

And it suddenly leapt into my mind that all of the programming that I had known how to do in the '60s was nothing, even though it seemed hard to do. But compared to what the actual potential for building structures was, we were going to have to shift to a different way of doing it. So I had one of my Arthur Koestler bingos. And the bingo was that the computers are not like the clocks they seem to be, where the clock, if anything goes wrong, it stops.

The clock is something you fix, and a clock is something you build. But in a biological organism, you don't fix them. You negotiate with them. And you don't build them. You grow them. And so this notion of, why shouldn't we be building everything on the computer out of self-healing intercommunicating parts the way biology had? And so this was very compelling to me.

Because of course, one of the reasons why cells are a good idea is that nature is very, very messy about doing biochemistry and requires enormous numbers-- for anybody who's discovered it, you might know that there are hundreds of reactions just to move electrons around. And the reason is that the energy margins are so small, and that there was no grand designer figuring out how to move these electrons along. Nature found out how, gradually.

But biochemistry is very kludgy. And it requires stable conditions. So what the cell did was allow this kludgery to continue in a protected way. Because nature rarely undoes things. It basically adds on more kinds of things. And this was a way of taking messy things that sort of worked and guaranteeing that they would work better. And then eventually, of course, these cells found out how to work together.

And now, looking back again at SIMULA, what SIMULA was was almost one of these things. And they had been thinking about simulating oil refineries. And the thing that I got out of it is, gee, why not simulate programmatic things? Let's get rid of data structures entirely. We'll simulate those.

We can simulate servers on a network. In fact, these kinds of systems are kind of like servers on a network. We can simulate control structures. We can simulate physical objects. We can simulate biology. In fact, simulation, again, thinking about what Sketchpad was about, is one of the dominant metaphors that is in the destiny of the computer.

And there's one thing you can do with a computer that biology doesn't do well. And that is, you can share the DNA. Biology does it by copying the DNA. And so when you want to replace something, it is really a complex operation. People are discovering how to do that now for cystic fibrosis. In the computer, you can actually share the DNA by extracting it and having all the cells look to a master copy. So this is very powerful, because it gives you meta control over what the system actually is all about.

So that was my first revelation. It came in the end of '66. And a little bit later, Dave Evans got me a consulting job with a guy by the name of Ed Cheadle, who wanted to build a little computer for engineers. And eventually, we wound up building a version of it called the FLEX Machine. This is a picture of it on its own display.

And we knew that it wasn't the first personal computer, because Wes Clark and Charlie Molnar had already done one called the LINC at Lincoln Labs in the early '60s. And we hadn't thought about this machine until Ed Cheadle wanted to do one. And then we realized that somebody had already done it. And we were able to learn quite a bit of what you might try and do on such a small machine. This machine, by the way, was made in some quantity. And a lot of fundamental inventions were done on it, like the things that look like deck tapes there. But deck tapes were actually invented for this machine by Wes Clark.

And here's a shadowy figure. Now, this actually isn't a picture of Bob Barton. Because, as far as I know, his pictures have never shown up on photographs. He was called the Ghost Who Walks because he was seen outside of Merrill engineering building, and he was seen inside of Merrill engineering building, but he was never seen entering or leaving. Suddenly he would be there. And then just as suddenly, he would not be there.

He was the guy who designed the Burroughs B5000, and I think was the greatest computer designer who has ever lived. And his influence in his own inimitable way had a lot to do with how this little machine of Cheadle's and mine executed. We made it execute directly one of the earliest object-oriented languages.

And you see one of his famous sayings up there. Systems programmers are high priests of a low cult. So his idea was, if you have to have an operating system, then you might be designing wrong. And we realized that if you had an object-oriented system that you really didn't have to have an operating system, because it actually would supply all of the things you normally associated with an operating system, all by itself.

And so as I said, Sketchpad had the first windows. The first multiple windows system I know of was Sketchpad 3 that was done by Timothy Johnson. It showed you different views of a three-dimensional object. And so we needed to have a windowing system. This was done at the same time as Ivan Sutherland and Bob Sproull at Harvard. We came up with a clipping divider idea that was rather similar to, I think, Danny Cohen's algorithm, multiple views and so forth.

Now, right as we were working on this stuff, we were visited by Doug Engelbart. And Doug, I'm sure you remember, but when you traveled around back then, you used to travel with-- you didn't travel light.


This was before video and stuff. So he traveled with a 16-millimeter projector and a film. And the reason he traveled with his own projector, he'd had it modified so he could do stop frame. And the reason he did that is that people were not used to looking at cursors. A lot of what he wanted people to see was what people did with the cursors on the screen.

As I recall, the movie on that one was a green display. Because I think what you showed back then was the 3100, the control data single user version of NLS. My memory has it being in color, and it was green. So I think it was that system that was shown. So this was the visit that actually changed our lives. It certainly changed my life.

Because what Engelbart brought was not just a vision of interacting with the system, but he had a philosophical underpinning. And that philosophical underpinning is as important today, maybe more important today, even than it was then. One of the phrases that he used that I particularly liked-- and I'm not sure I still understand it. But on the other hand, when you're listening to this stuff, total understanding is not necessary.

But his phrase was, thought vectors and concept space. And what he meant, I think, was that what you're doing is creating an extension of the kinds of spaces that you think in terms of inside of your head. So it was actually, you were creating an augmentation of the ways of thinking and the ways of representing and the ways of associating that was now going to be extended, in a way, somewhat analogous to the way writing is extended, but something more like the way we actually think.

And that what we're really interested in is, in particular, the collisions of different kinds of ideas. So if you think of Koestler's diagram there, you can see that one of the things that you would like to do for maximum creativity and getting out of your rut is to constantly have the system help you see different sides to things, to remind you that you're in a context and that there are other contexts.

But anyway, his talk was so fantastic that, like all really good talks, I think everybody came away from that realizing that this was an actual turning point in the history of design for computer science. And I think all of us were very interested in trying to do something with that. I'm not going to show you the movie because you saw yesterday, of the 1968 demonstration in San Francisco. Those of us who were there-- I was sick as a dog, but I went anyway.

I had strep throat and maybe pneumonia, but I wasn't going to miss it for the world. And it still sticks in my mind. It was one of the greatest expositions of a new set of ideas that I've ever seen. I think everybody who was there remembers. And there were a lot of people there. There must have been somewhere between 2,000 and 3,000 people witnessing that.

Actually, I remember, that was where I met Andy van Dam. And for those of you who have only known Andy in the last 20 years or so, I want to reveal to you that, in 1968, Andy was a hippie.


He had long hair, and at the very least a goatee, if not more. And he was a wonderful, angry young man at that point. And I remember him being both tremendously impressed with the Engelbart system and amazed that it actually did what it did. That was my impression of Andy back then.

And right around this time, I visited Ivan Sutherland at Harvard and met Bob Sproull. Bob Sproull then was an undergraduate who was simultaneously graduating from Harvard, I believe, editor of the yearbook. But he spent almost 24 hours a day helping to wire up the clipping divider for this first head-mounted display machine. And I understand that he slowed down quite a bit. He only works 20 hours a day now instead of 23 and 1/2.

Now just one little plug here for why all this stuff worked. I think these are the most enlightened funders that we know of in this half century. And I haven't heard Licklider's name mentioned too much today, although Lee mentioned it. A lot of this was because of his understanding, personality, and vision.

And the succession of the ARPA funders during this golden age, before it became DARPA, were Licklider, Ivan Sutherland, when he was only 26, Taylor, who later went on to be the founder of the computer stuff at Xerox PARC, and Larry Roberts. And their basic idea is to fund people instead of projects, one of the great ideas of all time, and fund directions instead of goals. I don't think anybody else would have funded Engelbart in those days.

Because I believe he must have appeared, to most people-- in fact, I know he appeared to most people-- as somebody who was completely deranged and out of his mind. And it's a real pity that the ARPA funding from the '60s didn't carry over into the '70s, or I think history would have been quite different today. Because sometimes the people who can see the farthest and see the deepest are the ones that are most difficult for the bean counters to understand.

So I met Marvin back then. So a typical Marvin comment, I'd like to say you don't understand anything if you only understand it one way. And he gave a talk on education and mentioned some of the stuff he'd been doing with Seymour Papert. And the talk was so interesting, I read his book. My other major was in mathematics. So I found in this book, besides a lot of finite automata stuff, some really interesting ways of thinking about universal computing.

Not just Turing machines, but Marvin did a very funny thing of, when he was explaining Godel numbers and how Godel made a kind of a list structure out of exponents of factorizations of integers. Marvin built an entire LISP system in one of the chapters in this book using that as the representation. It was a really weird way of learning LISP, but it got me interested in this way of thinking about things. And let's show the next little film. Then I went to Rand Corporation. This is still in '68. And I saw this system. Yep, go ahead and run it. It's got sound.


- First, we erase a flow arrow, then move the connector out of the way so that we may draw a box in its place.

KAY: Sees he wants a box. It makes ones for him. Now it's recognizing his printing.

- The printing in the box is being used as commentary only in this case. The box is slightly too large so we may change its size.

KAY: That's where modern day window control came from, literally.

- Then draw a flow from the connector to the box. Attach a decision element to the box, and draw a flow from it to scan. We then erase the flow arrows attached to the process Post New Area, and move the box to a new position. This allows us to draw a new box.


KAY: Okay, now you may have seen character recognition done worse on some products today, partly because it's-- as Roger was saying, is that they weren't trying to do the-- this was an 80-20 thing, where it did what it did essentially perfectly, almost perfectly. You had to learn a little bit. It did what it did, and there were a bunch of things that it wouldn't do. But the result of it was that the character recognition part on this particular interface-- which the system was called the GRAALL system-- the character recognition was almost foolproof and very, very transparent in the user interface.

And there were some further advances done at Harvard and elsewhere on doing this. And this really struck me as a very, I thought of as a kind of an intimate way of interacting with a computer. And then just a little bit after that, a couple of weeks after that, I visited Seymour Papert for the first time out here in Lexington and saw his work with children. And that was another one-- see, for me, the '60s was pretty much a continual series of holy shits, just one after the other.

I think part of it is the golden nostalgia, I'm sure, from old farts. But also I think actually this stuff is still pretty neat, even today. I think the GRAALL system was neat. I think Sketchpad was neat. I think NLS was really neat. It was really a joy to learn NLS have a sense of what it could mean to be a virtuoso in dealing with your external representations.

So Papert had another profound idea, and that was that, whatever the computer was, it wasn't really a tool or even a vehicle, in the sense of being an automobile, as compared to IBM's railroads. But Papert thought of it as being a medium for powerful ideas. Now Nicholas was hedging on the definition of a medium. I like Neil Postman's definition. Neil Postman says technology is a machinery.

Like the printing press is a machine, the computer is a machine. And he says, media are the purposes to which we put that machinery. So for him the media on the printing press are things like Bibles, indulgences, money, comic books, romance novels, pornography, political essays, scientific essays. So it was a wide range of media that is all printed, all represented, all brought forth, by the same piece of technology.

So if you think of media as purposeful aims, you see the strongest things that Ivan Sutherland and Doug Engelbart and some of these other people did were actually creating media. I didn't realize that until I saw Seymour working with the kids. And he realized that there's only one way of thinking about this, to me. And that was, it was a kind of medium being brought forth from machinery, and it was going to be very much like the printing press.

So flying back to Utah, I sketched this thing, which I later made a cardboard model of, which later became called the Dynabook. And it was hollow. I filled it up with lead pellets to see how heavy you could make it before people didn't want to carry it anymore. And the result was about two pounds.

So we had this notion that, if you were going to have an intimate medium, then you'd have to do mundane things, casual things on it. So we asked the question, what would it be like to take a grocery list and actually put it on a computer and actually carry it into the grocery store and be willing to carry it out with two bags of groceries? What kind of computer would that have to be?

And one of things that was happening in the end of the '60s was that ARPA had been doing the ARPANET. But Larry Roberts and some of the other people who were-- Bob Kahn and other people-- who were driving the ARPANET were also thinking of packet radio back then. So in '69, I remember driving around, I think it was a Ford van with a model 33 teletype in the back and all kinds of radio equipment, communicating with-- I think it was actually Engelbart's computer at SRI by radio at 10 characters a second on this model 33 teletype.

So there's a whole sense that computing was going to get to be mobile. And I think many you may know that Larry Roberts last set of things that he was interested in doing were to have a handheld radio terminal that would be portable and would connect you through packet radio to the ARPANET.

And the other thing is-- I actually tried to get the picture here. But Lee, I'm sorry, we actually were thinking of this thing being networked. Because I think networking was part of the ARPA dream. And the first cartoon I ever did of this machine, I'll show you later. It shows two kids playing Spacewar! with each other, each on their own Dynabook, communicating by radio.

So just having these ideas here led to something kind of interesting. And that is, it struck me that having a time-sharing graphics terminal, this desktop computer that Cheadle and I had been working on and this cardboard model, reminded me of something that had already happened hundreds of years before. And that was the development of printing. So on the far left is a manuscript book chained to its little desk.

And if you've ever been in a medieval library-- they have one still in Florence-- they're like the size of this room or maybe twice the size of this room. But they only have about 70 or 80 books. One of the largest libraries in Europe in 1400 was the Vatican Library. It had 372 books. The richest man in France, who was the brother of the King, was a bibliophile and had books made for him. He had 154 when he died in 1435.

So these books cost the equivalent of millions of dollars. And they are so rare that only institutions could afford them. In 1454, at the Nuremberg Book Fair, Gutenberg showed 20 bibles. And people marveled at two things. One is how alike the bibles were to each other. That was a really new-- you could look at one page and it looked just like the same page on another Bible. They'd never seen that before.

And the other thing was that the costs had come way down. These Bibles only cost three years of a clerk's wages, maybe $60,000 today. And so you could think of those as being workstations, kind of things that smaller institutions and towns and some individuals could own. And you'll notice that the Gutenberg Bible was illuminated because they made every effort to make it look like the manuscript books on the left. And then 50 years later, to me, is when printing actually happened. And we have books that, instead of being this size, are now this size. These are printed by Aldus, whose last name was not PageMaker.


His name was Aldus Manutius. But he was the person that the Aldus Corporation took their name from. And he was a Venetian printer who had been a scholar. And he went out and decided books should be this size because he went out and measured saddlebags in Venice, because the first offering from his printing press in 1495 was called the Portable Library.

And instead of printing just one book, one kind of book, the Bible, Aldus, over the next 35 years, printed 40,000 different titles. He and his sons went around and gathered up all of the writing that the Greeks and the Romans had done that they could find. They printed each one of them. And many of the fonts that we have today were invented by Aldus in order to make this stuff more readable.

Aldus is also the inventor of something quite unusual, and that is page numbers. Page numbers went in to books the first time ever 65 years after the invention of the printing press. So they went in in 1516, and we know why. They did not go in there for sequencing. Because the way sequencing was done was to print the word on the next page down in the last line so you could see how the pages went together.

It did not go in there in order to refer to other books, because there were no really standard editions of books. Each press run was 2,000 to 5,000 copies. Page numbers went in there partly at the instigation of Erasmus. And they went in there because they were just starting to write down arguments that were so complicated that they wanted to refer people back to what they had said previously.

So the first page numbering systems were done in order to, if you will, hyperlink a piece of text within itself. So as you were following along an argument sequentially, the author could say, and as I said back on page 162, blah, blah. And you could go back there and check it out. So page numbers came in as people started realizing that the reproducibility of the books allowed for much more complicated arguments to put in there and not be garbled by somebody who was trying to copy the book by hand. And this led, in turn, to this idea of trying to link the arguments closer together. That was a very interesting sidelight.

And given that it took 65 years to happen from the invention of the printing press, I also use it as a kind of a metaphor for what we haven't thought of yet. Because clearly, we are somewhere between the desktop computer and this Dynabook kind of machine. And it's sobering to realize that nothing in the history of printing counted in those 50 years. There was a complete transitional stage where almost everybody was looking backwards.

And if we look at the way most people use computers today, it's really pathetic. Because most people do not do hyperlinking. Most people do not use computers the way Engelbart thought they should be used. And most people do not use computers the way Ivan Sutherland thought they should be used. What most people are doing is imitating paper, either in prose, or imitating paper in terms of accounting.

And that echoes all the way back to Sumerian times. Because as far as anybody can tell, the invention of writing itself came about first to account for the king's wine jars. And then a few hundred years after that, they realized they could also write down some of their poetry. So what's been going on in computing has been very, very-- as McLuhan once said, we're driving faster and faster into the future, but steering only by looking through the rear view mirror. And I think that's been going on.

We can think of a corner being turned into the actual modern era. I just gave a talk in Korea whose title was, "The Computer Revolution Hasn't Happened yet." And I would say that it hasn't happened yet until people start using the computer for dealing with thought vectors and concept space and dealing with the idea that you can have active dynamic models of the things you're trying to represent.

So that led to a kind of a sense that there were three different religions. And we didn't know very much about what the right-hand hand religion was going to be. But we could see that they were certainly at a mainframe religion, of which Tom Watson, Jr. was the Pope, that there was going to be a desktop revolution and religion, because Moore's law said that was going to happen. And then we had this more mythical notion of pervasively networked systems with computers that were kind of like clothing.

So one of the last characters I'm going to mention here is Butler Lampson, who talked to even faster back in the '60s than he does today. And he once came to, actually, a meeting on the ARPANET standards and gave a talk on capability operating systems, in which he had been working on one at Berkeley. And what had happened is, the meeting went later and later and later. And finally, Butler said, well, I have a plane I have to catch.

And they said, when do you have to leave? And he says, I have to leave in about 15 minutes. And they said, well, would you like to wait until next time to give your talk? And he says, oh, no, I'll give my talk right now. And he got up and gave an hour talk in 15 minutes on capability operating systems. And it was one of the clearest talks I have ever heard.

And part of the idea here is something that is part and parcel of objects, which is-- but in this kind of operating system, you think of clients and services. The services tend to be generic, like printing and searching and updating and viewing and delegating. And the clients are different kinds of objects. Objects in this sense are usually large things. They're the size of Unix processes. But they have certain things that they can do that are as generic as possible.

There are a variety of these kinds of things. And then each one of these clients of the services has a different set of privileges. And the way it's usually implemented is by adding a little bitmap to a protected pointer that says which of these privileges you have access to. So this is a very compelling way of thinking about-- in my mind, what I was thinking about was, gee, everything should be an object.

And so the question is, how could you implement this kind of protection, which seemed very, very useful and necessary, even for ordinary programming. How could you do it in a fine-grained way?

And that leads to this notion of designing in terms of concepts, that you think of behavioral concepts and transmit it into object-oriented terms, you think of having different kinds of things, different classes of things, different types of things. But you're trying to design everything in terms of these generic concepts. And what you wind up with is something that, if you can limit the number of concepts that you're dealing with, then your design tasks become much easier.

And finally, we had-- although, again, I was only thinking about the middle panel here as opposed to the one on the far left. We realized that the ARPANET would actually need objects, for sure, because networking, particularly pervasive networking, means heterogeneity. When you have heterogeneity, you better not require people to have to parse things in order to understand them. You really want to be able to execute them in order to understand them so that they have their own understanding with them.

And we felt that something like objects should actually be adapted for the ARPANET. That was a little bit too early. That was brought up in one of the meetings back there. It was felt to be a little bit too early. But I think we can see from the state of the internet today that this is absolutely critical to get out of a data structure-oriented way of doing things, which is kind of like going back to MS-DOS and trying to get to a situation where the objects that we interact with don't have to be from the same object-oriented language, don't have to have been created on the same kind of machine, but nonetheless, can interact and interoperate with each other.

And then finally, towards the end of the '60s, I finally went through this half page of code in the LISP 1.5 Programmer's Manual and realized that McCarthy had actually come up with an incredible way of thinking about what programming should be. Now, he was doing it to prove a different kind of point. And as he says himself, it wasn't his fault that Steve Russell sat down and made it work. But once it did get going, it provided a completely different way of thinking about what it means to implement programming languages.

And I started thinking very seriously about what an OOP language would be like if it were set up like LISP. I think of this as the Maxwell's equations for computers. Because everything you can do with a computer and in a high level form is there on that half page. So that's a very, very interesting thing to spend a Sunday afternoon going through. And I think everybody who has gone through this and understood it, spent those couple of hours to go through it, remembers that time when they suddenly said, oh, wait a minute.

Now, it wasn't helped by the fact that this particular rendition in this book has a bug in it. It has one bug. And that bug can cause us an extra couple of hours to be taken until you're quite sure that you understand all of it and that there really is a bug. It turns out there is. And what is the bug? Well, of course, they left out one set of parentheses.


Then the last part of the '60s was realizing-- and this again, I want to say something about Doug Engelbart and NLS. The other part of his project that was so impressive to me as a graduate student was, when you visited the Augmentation Research Center at SRI, what you found was kind of the epitome of what a research project should be like. They had lofty goals, lots of committed people.

They were willing to do any amount of building their own hardware and software, which they did. They built their own display system. And they built their own IO devices. And they built most of their software. And they helped modify the operating system in order to make all of this stuff work.

And actually, the other thing I should have mentioned, I think everybody noticed yesterday that NLS was kind of zippy as far as response. Wouldn't you like to have your desktop computer be as responsive as that system was, instead of lumbering around trying to bring stuff up? Well, I realized that, in order to do a computer for kids, that what we had to do was to build a bunch of them. And this is one of the entries in my notebook back then of what I wanted to build. And at that point, Xerox PARC came along, and we got a chance to try some of these ideas out.

Now, to me, the moral of some of this looking back in the past is that it is incredible how powerful a new context is. And it's also incredible how stultifying it is after a couple of years. I didn't realize until looking back on this, but I think all of us have had this feeling that the very things that made our ideas work when we're in a context were also the ones that held us up from thinking about other ideas.

So these contexts wear out. But we can only think about things in terms of context. When we go to one context to another, we get life for a few years. But then there's a syndrome, which I call the subgoals becoming the goals, that sets in. Usually the subgoals are hard enough to require a couple of years work. And once you've been working on them for a couple of years, they become the goals of the project. And all of a sudden, it's hard to remember why you started the project in the first place. So I think one of the hardest things to do on these large scale things that require so many different things to be done over so many years is to constantly be able to renew what the actual dream was.

And it takes so long to get people to understand what these ideas are. This is one of my favorite slides, because it shows hand axes from to prehistoric periods. The one on the left, you might notice that when you chop with it, it chops you. Because the place where you grab it on the bottom there is pretty sharp. It's almost as sharp as the top.

And then a little bit later, somebody came up with a hand ax that was blunt around the bottom so it wouldn't chop you as much. And according to archaeologists, the period of time between these two was about 200,000 years. So you can imagine, every generation, probably somebody sitting around a campfire chopping with one of these things on the left, said, jeez, this is hurting.


I'll just chip this away. And probably got burned at the stake for going against the traditions of the tribe, which are probably something puritanical like, work is good for you, sure it hurts, and so forth. And then finally, probably by an accident, some chieftain of some tribe came up with the blunter end one and made it a rule that everybody should have blunt end ones. And that's how this change happened.

So there's this low pass filter to reality. And the biggest example of it is what's happened to personal computing since 1980. The low pass filter has been so dominant that all we get is about a 60 cycle hum for most of these ideas from the '60s. If we look at what most people are doing with personal computers, they have unbelievable resources. Engelbart's SDS 940 had 192 K and it ran 20 users as a time-sharing system with 192 K.

We've got megabytes. The system is much less responsive than it is. The user interface is worse than it was. And humankind has sat down for one of these 20-year slumbers right now, bedazzled by stuff that isn't very good.

So the last slide is a metaphor for this whole talk. And that is something that was actually found out about frogs at MIT here. And that is that frogs don't know what their food looks like. So if you take their normal food, which are flies, paralyze them with a little chloroform and put the flies out in front of the frog, the frog will not eat them. The frog will just starve to death.

But if you flip a piece of cardboard at the frog, it will try and eat it every time. Because what it thinks food is an oblong thing that is moving. And I think this is a good metaphor for us scientists. Because the way I think of the flies is that they're ideas. The ideas are in front of us all of the time, and we just can't see them.

What we have to do is periodically find a way of realizing that you can't see unless you admit you're blind. And every time we admit we're blind, we can actually see a bit. And then we can push on to the next stage of this grand adventure. Thank you.


MODERATOR: Thanks a lot, Alan. We'll take a few questions directed to Alan specifically. And then we're going to arrange for the panel.

AUDIENCE: I just want to go back to one of your early slides of the dog house, that I think actually the dog house is the better model than the bridge. Because if we look at the scale, the bridge is about a middle-sized construction, rather tiny compared to cities and societies, which are totally unarchitectured, yet somehow work. The thing is, we don't quite understand how they work. I think that's really a challenge for intimate computing.

KAY: No, but you're making my point. The whole point about biology is that it works in the face of error and not having to be strictly architected.

AUDIENCE: Okay, then I misunderstood.

KAY: I said dog houses work simply because they're so simple compared to the strength of materials. So that's a different point than the-- but the very point I was trying to make about, when you go up into real complexity, you have to find a way of getting things to work together, even when they don't want to terribly. So the Constitution of the US is a pretty interesting meta program since--

AUDIENCE: And the nice thing is it's small.

KAY: It's very, very small. And it's not a set of laws. It's a set of principles for dealing with millions of only semi co-operative components, namely us. Other questions? Yes.

AUDIENCE: Not so much a question, but a comment. It seems to me that you focused on the '60s. And it's probably natural to think of the golden age in the area where you get your own awakening and you start to become productive in a creative part of society. But for everybody who's in the field, their golden age is probably going to be some other decade or whatever.

As I went back and just sort of picked another one, like the '70s, and just sort of made the list of things that came out there. This was the era when ARPA had a D in front of it. I was thinking of Minsky's adage about not being right about something for too long. I made the list of the things that came out of that decade. And it seems to me they were pretty impressive. The first work on speech understanding in a systematic way started in that decade, with Allen Newell's push.

The work on image understanding started in that decade. The internet came out of that decade. The microprocessor was in the decade of the '70s. The personal computer really came out of the decade of the '70s. The work on LANs, radio networks, and the DLSI design, the thing that empowered the universities, came out of the '70s. And probably an awful lot of the object-oriented programming work came out of the '70s, as well, although you would probably know better than I. So it seems to me it's a little unfair to cite the '60s as the golden age relative to all other ages, because I think it's relative to our own contexts.

KAY: Well, I hope I wasn't doing that. Because of course, the most productive period in my life was in the '70s, not in the '60s. The reason I picked the '60s was that was when I think a lot of these ideas manifested itself in such a way that we could see that they were possible. A lot of the shoes were dropped in the '70s.

But I certainly think of the '70s much more as engineering. In fact, most of the stuff we did at Xerox PARC was kind of a second engineering pass on things we'd failed at. Because the things were bubble-gummed together. We didn't completely understand them. So for instance, one of the dicta that Butler Lampson got us to agree to in the early days of Xerox PARC is, we should never do something that isn't engineered for 100 users. And that made us much more conservative than we had been in the '60s.

But it also added much more reality to the things that we did. And of course, a lot of the success that we had was because we could make 100 of something or make a machine that you could run 100 people on or make software that you could move around. But to me, that was very different than the-- I think of it as being less creative and more engineering than the '60s.

But as you say, other people might-- it just depends on when you start. From my standpoint, it's silly to try and compare one person's gods with another. But I think that, to me, Ivan Sutherland and Doug Engelbart did the most towering and important stuff in the last 35 years. And I think that stands up pretty well today. Yes, somebody else was-- okay, great.


MODERATOR: Before we go into the panel, I want to do a couple of things. First of all, Paul Penfield has to leave for a memorial service, and I'd like to thank him on our joint behalf for a job very well done. He and a lot of people at MIT, like Ed over there and Scott and lots and lots of other people, put in real quality time making this happen. So thank you very much, Paul, and the rest of you.


The second thing I have to tell you is, Paul has an announcement.


PENFIELD: I just want to thank everyone for coming to this. This has been a real wonderful opportunity for me to see some aspects of this community that I hadn't been as familiar with as I should. And I want to personally thank Andy for putting this together. Andy says that we did the [INAUDIBLE] job. But I can assure you that Andy, in setting this up over the past several months, has put it in far more effort than anyone at MIT has. And I think the whole group deserves to thank Andy for organizing this.


MODERATOR: Thank you. Thank you very much. As you know, it was the three guys from EBT who really thought of it, and Lisa who executed it. But I got to invite the speakers. That was the best part.

Speaking about speakers, you all know what day it is today?


MODERATOR: Friday the 13th. Now, I'm not normally a superstitious person. However, I have to find some explanation for what happened. Here's what happened. Inadvertently, Ted Nelson's tape got recorded over. So we do not have an official record of his great speech here. I just found out about this a little time ago. This is not a joke. I'm sorry. I wish it were.

And we are trying to find people who may have recorded it. There are a couple of possibilities. There may be some folks out over the MBone, who recorded it. It's a classic point of failure. And despite all of the duplication that we had in equipment and people, something went wrong. In any case, we asked Ted what he would like to do about it.

He was a gentleman to the core and took it exceedingly well. And I think I would have gone straight through the ceiling. Fortunately, he's jet-lagged and so he didn't react quite as strongly as he might have. As I said, he handled it beautifully. The folks here at MIT thought that what might be fun and interesting is to let him have the last word, as it were, a reprise performance. And any of you are welcome to stay while Ted tries to lay down another track for the historical record.

And we'll do that after the panel. Some of you have to go. Some of you have other things to do. Some of you, whatever. You're all more than welcome to sit in on this as Ted tries his best to recapture the moment. We apologize.

AUDIENCE: Forget that moment. We'll do another moment.

MODERATOR: We'll do another moment, of course. So that's where we are. We're going to take a couple of minutes to set up the chairs for the panel, and then we carry on. Well, why don't we just get started. And you can fire at will at the people who are here. Fire when ready.

AUDIENCE: All right, I'll go. We started this discussion by looking at Bush's vision of memex. And we just heard Alan talk about the golden age of the '60s. And here we are in 1994, and we've ended up with Windows, which is kind of a bleak thought. So I'd ask any of you who were interested to postulate where we would be if Windows never existed. Would we have had to invent it, or would we be someplace different, and where would we be?

PRESENTER 1: The question is Microsoft Windows?

AUDIENCE: Yeah, Microsoft.

MODERATOR: Wait a minute, wait a minute. Use the mic. I'm sorry. Clarification-- is the question Microsoft Windows or windows in general?

AUDIENCE: If Microsoft didn't exist as a market force, where would we be?

PRESENTER 1: We would be missing nothing. Microsoft doesn't even claim that they're a highly innovative company. They take programs that have already existed and sold well from other vendors, such as Lotus and WordPerfect, and they do a commercially better job rewriting these programs and selling them, as Excel and WordPerfect. So we wouldn't be missing any important technology.

AUDIENCE: So you think we'd be in exactly the same place? Personal computing would look the same.

PRESENTER 1: As far as I can see.

ALAN KAY: At PARC-- because I didn't talk about the PARC phase of the thing. But at PARC, one of the goals was to do NLS as a distributed system. And all of the Altos there had the five finger keyboards as well as the mouse on them. We basically loved NLS. And we'd done a few modifications which we thought even sped up-- NLS, part of the interaction scheme on it, I believe, was because the analog mouse, there was some drift in it.

So one of the things that they did was to say what kind of a thing you were pointing at. So you'd say, move character or move word or move paragraph and so forth. It was kind of a procedural, where you gave the command first, and then you'd go bug, bug, and then Command [? accept. ?] And we realized at Xerox PARC that you wanted to have a speedy scheme for interacting.

And we thought we could go even one better by selecting the object. So you'd select the thing that you were going to do something to. You would give the command. And then in the case of move character, you'd go select, move, select. And that would do it with fewer keystrokes. Now the abortion that happened after PARC was the misunderstanding of the user interface that we did for children, which was the overlapping window interface, which we made as naive as absolutely we possibly could, to the point of not having any workflow ideas in it. And that was taken over uncritically out into the outside world.

So we have many systems, like Lotus Notes and many mail systems, that, when you say reply, it comes up with a window over the very thing you were reading, as though there weren't any connection between these things. So this is an abortion to me. But it's basically part of the whole field. Whereas our notion was that, you start the kids off on this fairly simple, naive thing, and then there would be an actual progression where you would get up to this several commands a second kind of thing that you could do with NLS.

If you've ever seen anybody use NLS, it is really marvelous. Because you're kind of flying along through this stuff, several commands a second. And there's this complete different sense of what it means to interact than you have today. So I characterize what we have today is a wonderful bike with training wheels on. And nobody knows the training wheels are on, so nobody's trying to take them off. So maybe Doug has a comment on that. So I just feel like we're way, way behind where we could have been, if it weren't for the way commercialization turned out.

ENGELBART: Well, strangely enough, I feel the same.


Yeah, that's part of the easy to learn and natural to use thing that became sort of a god to follow, and that the marketplace is driving it. And it was successful and you could market on that basis. But some of the diagrams and pictures I didn't quite get to the other day was, how do you ever migrate from a tricycle to a bicycle? Because a bicycle is very unnatural and very hard to learn, compared to getting on a tricycle and riding.

And yet, in society, it has superseded all the tricycles, essentially, for anybody over five years old or something. So there's a lot. And the whole idea of high performance knowledge work is yet to kind of come up and be in the domain. You got to make it easy. It's still the orientation of automating what you used to do, instead of moving to a whole new domain, which obviously you're going to learn quite a few new skills. And so you make analogies of saying, suppose you wanted to move up on the ski slopes and be mobile on skis.

Well, just visiting them for an afternoon is not going to-- you're not going to find-- I'd love to have photographs of skateboards and skis and wind surfing and all of that to show you what people can really do if they have a new way supplied by technology to be mobile in a different environment. And none of that can be done if people insisted that it was an easy to learn thing. Anyway, so moving your way around those thought vectors and concept space-- I'd forgotten about that.

AUDIENCE: You said that, right?

ENGELBART: I must have. That's so good.



PRESENTER 2: You did. It seemed to come naturally at that point.

ENGELBART: Anyway, that's a nice formulation of what I've been saying. But it's to externalize your thoughts in the concept structures that are meaningful outside, and moving around flexibly and manipulating them and viewing them. It's a new way to operate on a new kind of externalized media, to explain it like that. So to keep doing it in the model of the old media is just a hangup that someplace we're going to break that perspective and shift.

And then the idea of high performance and the idea of high performance teams, who've learned to coordinate to get that ball down the field together in all kinds of operations, very different operations. So I feel like the real breakthrough for us getting someplace is going to be when we say, all right, let's put together high performance knowledge work teams. And let's pick the roles they're going to play within our organizations in some way and such that, even though they operate very differently from their peers out in the rest of the organization, they can interact with them and support them very effectively.

So there are roles like that that would be very effective. And everyone else can sort of see, because they're interacting, these guys, what they can do. And suppose it does take 200 hours of specialized training, et cetera, like that. That's less than boot camp, something like that. So anyway, to me, that's kind of the only way the evolution-- one of those boxes in that paradigm map about deployment was really coming down and showing you that special purpose teams is one kind of thing, the way that they can propagate. And very different from moving a group of people who have an existing set of staff and processes and methods and skills and equipment, and trying to move them all together.

It's a practically impossible task to do that in any significantly large step without having casualties in there that just aren't all equipped to be mobile in that space. Anyway, so there's a lot to go with that thing. And it all stems from looking at today and saying, why do we accept that? That's the modern thing. And it's almost like a religion. In any other company, I'm afraid to bring that out. But maybe I have to run from this one, too.

NELSON: I hadn't heard you say this before. So you're basically proposing information--

MODERATOR: Wait, wait, wait, use the mic.

NELSON: I hadn't heard you say this before. You're basically proposing posing some kind of information SWAT team that can move swiftly through an organization or it's going to be some elite [INAUDIBLE] in the files. This is a very exciting and interesting concept. But how would that function organizationally.

AUDIENCE: We did that to your tape.


NELSON: Obviously, they got in there.

ENGELBART: No. I mentioned the other day that, if you're going to have a strategy to go ahead with your organizations, you just can't lift them all at once. So the strategy I finally worked out is that, to that improvement infrastructure term I mentioned, I didn't show you much below that. But in that improvement infrastructure, there are roles in the improvement process for high performance teams that would really be helping the information and the improvement process go about.

And then they're roles for them to be plugged into an organization and to function that would be very supportive. And this is all assuming that those are the outposts that everybody is going to start moving there as well as you can. But you just can't lift the whole organization in some very radical way. And so that's a part of the process. So yes, they are an elite team. But you're assuming that, in some number of years, you'll learn how to get the whole organization there.

PRESENTER 2: I think that the gruesome practical point is that finding a path of [? thought ?] vector through concept space to where you are may be one thing. But the transition path to get the world there may be very, very different path. So that you have to find a path of whatever they are, of pushing the world vectors, where each step is downhill for the person who has to make that step. So now, your original question was, has Microsoft's existence and the dominance of DOS and later Windows helped?

Basically, if at some point, Microsoft has to make a change, obviously that's going to be one of the more difficult people to put them in a position so that the next step is downhill. And it can happen. And one of the things that can make that easier is if instead of having just one operating system, you have two. You might ask, for example, what would have happened if the Mac hadn't been there? Maybe we'd still be using DOS without Windows.

So there's an argument which says, this is easier. I wouldn't say it's easier in a situation where you don't have any dominance, you don't have any commonality. Maybe that's what held up Unix, the people who are all developing things on the Unix, in fact, because they didn't have enough commonality. It was difficult to make code that you could pass around. It was difficult to spread things around.

So the fact that you've got DOS there at least means that, if you're not trying to change the operating system, you can roll out all kinds of bits of software very quickly. So that's got a great advantage. The internet and DOS, if you can do what you want with the internet and DOS, then you can do it very quickly and roll it out. It's when you want to change the internet or you want to change DOS that you've got yourself a big problem.

The way I hope it will maybe work for the web protocols and all the these web standards, which we hope we'll be able to evolve from their kludgy current state into the beautiful golden dream, is that in any case where you lay down a standard, you also allow there to be a second. And you show a path whereby if a third one comes up which is better, you can move in that direction. So you always have one on the heels of the first, keeping it honest. And you have a transition, a hook in it, to be able to unhook it all together later on.

KAY: So American television is kind of a counter example of, in the effort to make it really easy for everybody to watch television. I'm not sure I see any path back upwards towards enlightenment, because the advertisers are quite happy. And so is Microsoft. It's quite happy with the stage of blindness that everybody is in right now. So it's a great idea.

There's an economic law called Gresham's law, which says, the bad that's easily produced drives out the good. Because it resets the whole notion of normality. Personally, I think if somebody invented a bicycle and it didn't exist today, they couldn't get anybody to buy it. Because it takes more than five minutes to learn. That is really pathetic when you think about it.

NELSON: As far as television is concerned, the operative phrase a few minutes ago was "a comfortable step down."

ENGELBART: I'd say you want to make your steps up comfortable. It's kind of take stress but you have to minimize it. And the older you get, the harder it is to change. And I keep getting that thrown back at me, and says, you're going to ask all the old guys that are ready to retire that they're supposed to shift and change?

And the only answer you can really get is, A, look, there's a lot of potential for organizations to get a lot more confident in their knowledge work, just a lot more. And it's going to take a lot of change. And the organizations that learn how to do it are going to be successful, maybe the only ones that will survive, including mankind. So look, we've got to get there. So what you need to do is say, look, we have to find a strategy that's as smart as possible for accommodating all of the human and social characteristics that change has to cope with, as well as the dynamics and the expense of changing, et cetera.

So we have to find a strategy. So that's the major push on our bootstrapping thing, to say, yes, it looks like a very smart thing to do is to build an appropriate improvement infrastructure. And then as rapidly as possible, get the improvements and these capabilities so they can plug into the improvement and structure, too. So that you're improving your improvement process, maybe even faster than the other. But that just looked like the only strategy I could think of. And there's a practical way to go after it. So strange as it sounds, that's what we're after.

MODERATOR: Next question.

AUDIENCE: Okay. I wanted to thank Alan for his retrospective. Oh, I'm Mike Zyda, Naval Postgraduate School. I want to thank Alan for his retrospective of the '60s. And there was an important slide in there that I think got kind of muddied, which talks about the golden age of ARPA. And I liked the print on the bottom, which is, fund people not projects.

Try and point out, actually, I receive a lot of ARPA money right now. And ARPA today is very different from 30 years ago. I've been on my own money for 22 years, so I know this. I had an interesting experience about in June, July of this year. I received a note from my ARPA program manager saying, we have a new director of ARPA.

You need to take your proposal that I've already sent you the money for and provide a three and four word title for each task, a task number for each task, a start date for each task, an end date for each task. Were you over budget on budget or what for each task? And the director of ARPA was going to look at the tasks and say which ones he was going to fund for money that was already there. And so when I saw your, fund people not projects, and then I'm saying, fund projects not tasks.


So I'm a little bit nervous about the way we're going right now. So this is not an attack on ARPA. ARPA's been extremely good to me over the years. But just the orientation of the bottom line is a little bit out of whack, I think. So I'd like to hear some comments.

ENGELBART: That really isn't just in ARPA, in our industry, too. It's like, if you can't explain it to me in five minutes, I don't think I should even consider it, somebody will say to you. And you say, well, you can't explain something very novel in five minutes. And they want the bottom line.

And you say, I'm convinced in the general thing. There couldn't be anything else. You have to get there. But that isn't the kind of thing that sells. And so it's sort of some kind of modern type way to manage. But I think it's going to get our country and our industry into real trouble.

AUDIENCE: Yeah, this presupposes I know what the system is going to look like precisely at the end when I'm handing in the proposal.

ENGELBART: Well, it's the kind of thing that can work when you're making incremental improvements. And so that the guy that's evaluating knows the system and knows that, yes, if you get delivery that much faster to the customer, and that's what your bottom line is, yeah, I can see it. Okay, good. But it really doesn't work if you say, I got a brand new material that'll make your products very different. And I can't tell you how much different, but you've got to explore it. You can't do that kind of sell.

PRESENTER 3: Yeah, if I could respond to that briefly. I guess I would start by saying that one person's bug is another person's feature, often. And that there are different models of how the system ought to work, just as there are different models of life, depending on what stage of evolution and involvement in it you are. ARPA [INAUDIBLE], I would say. Not to try and defend any specific action, because I haven't been involved with them for quite a while.

It's still an organization of people. And people do what people do. And if different people were involved, you'd probably have different actions taken. Now, a lot of that takes place in a context-- a political context, budgetary context, state of the world context, and so forth. But it seems to me that they have never been more open to really good people going down and helping out as I'm sure other parts of the government are.

And if you have the right kind of view and mentality and capability, then you have the ability to actually change things. And so rather than a comment on the way it is, I think the more helpful thing is to say, how can we become involved to change it to be more the way we think it ought to be.

AUDIENCE: Well, in fact that was kind of the comment, is to generate a little bit of discussion. And perhaps some of these people in this room filter that back that direction. I chat with Duane Adams all the time.


ENGELBART: I think that you have to realize, those guys, who they're between. They have to go to a new kind of Congress now that's very different and tough on this sort of stuff. So they're not

But the whole general process of the quick return and the way it is an industry. If your stock shifts a little bit in the next three weeks, that's bad. So you can't do any long-term changes that may say, I'll drop for a while and I'll make it up really later. It's just isn't allowable to the CEOs.

AUDIENCE: Okay, [INAUDIBLE] [? Frankston, ?] despite Microsoft. I'm not going to defend Windows. But what I want to try to understand is why Windows is such a problem, in that-- if you have to change the world all at once and you can't coexist with what exists, you've got a problem. The web is a great example of, independent of Windows, it adds great function and capability.

And there are lots of examples, like Andy's Rollerblades. People learning AutoCAD or graphics systems and interacting are great examples where people have decided it's worth investing major amounts of time to learn new paradigms and new systems. It doesn't mean they give up the rest. What are you real problems in getting-- do you really feel like to change all the world at once, or what are the problems to get people to try out pilots or new ideas in industry or the world at large?

ENGELBART: The only way I can see is, you have to get pilots operating. And there has to be some kind of conscious pursuit of that future that you can't really guarantee it as there. But to say, look at the potential.

AUDIENCE: I was going to comment-- Notes was mentioned, and trying to get people to adopt that had similar barriers. Again, far from perfect, but it had a certain utility which some people eventually found. But I was just interested in whether there are any examples of really sort of pilots in that kind of new systems.

KAY: I don't think it's worthwhile commenting at length on it. But I think my main point is that I think MS-DOS got millions of people used to, as a normal mode of interaction, to something that would have been barfed at in the mid-'60s. But they didn't know that they could interact better with a machine. And so it really set a low standard.

Just like television sets a low standard if you don't know what theater can do. You're never going to find out by watching television. That's the problem. And so it's not that you can't do other kinds of things. But for instance, if any of them-- let's put Apple in there, as well. Because from the standpoint that we're talking about, the Apple interface isn't any better than MS-DOS is and Windows.

They're basically part and parcel of monolithic-- I think, for instance, if you wanted to put an Engelbartian scheme in your application, in the Apple or Windows user interface, it's almost impossible to do. Because the theory of interaction is buried so deeply in there that you can't say, okay, I want to interact in a different way. And yet in any reasonable object-oriented implementation of these things, as in fact there was at Xerox PARC, you could sit down and design your own interactive. In fact, NLS, which Doug probably didn't mention, had this interesting thing where they actually-- you could drive the user-- in any part of the system, you could make up a new user interface if you wanted to experiment.

They had this grammar-driven-- so you could put in an interface grammar yourself and use it to drive commands if you were experimenting. So it was what I would think of as an open system. It didn't force the user to go just the way it did. It was actually extensible in some very interesting ways. I probably have never even shown that in a talk because it's more complex than the other stuff.

ENGELBART: Yeah, it's interesting, when he was telling us about the emergence of Smalltalk there. Well, what do you think we've encapsulated that all? And now that's going to [? augment. ?] And now we've got a Smalltalk encapsulation package that gives you a very flexible kind of interfacing that you can go conventional if you want to that you can shift. But anyway, I sort of take my hat off to Alan about providing the way the Smalltalk [INAUDIBLE]

MODERATOR: Microphone.

PRESENTER 1: I get to listen to a number of researchers--

MODERATOR: Turn the mic on.

PRESENTER 1: I get to listen to a number of researchers complain that the development groups or the users in general aren't willing to adopt their ideas. Sometimes I ask these people to look at their own desk and ask, how willing are they to say, I want to be the first user on this floor of bubble memory instead of magnetic disk? I've just abandoned the C compiler because I just heard about [? IFIL ?] and I think it's really wonderful and I want to try it. I don't hear that very often from people. And they ought to accept the idea that the rest of the world thinks that they're not different from the rest of the world. They're thinking the same way.

AUDIENCE: Samuel [? Ed ?] Epstein with Just kind of a setup here and then the question-- and I guess I'm young enough to not totally believe that there isn't any validity to the conspiracy theory that Ted was referring to the other day. And things are pretty screwed up, if you look at what was going on 20 years ago and even a little bit longer. And I have to say that I've had my own experience, like it was mentioned earlier today, about not being able to turn in a laser-printed report. The excuse that I was given was because, well, the computer wrote this. And for this English class, you have to write this.


And I have to say that, when we're looking at the big picture here, I find that, in my own personal case and I know there's others like me-- with very few exceptions of actively searching out the people that are up on this stage here and others like you, and finding this knowledge and trying to do something with it. The current educational system in this country seems to be doing a really good job of helping to create and accrete these established self interests, which seemed to prevent the dissemination of this information. And also seem to encourage this, well, if it takes more than five minutes don't waste your time to create this environment.

Do you have any recommendations or suggestions on how we can restructure, improve, modify, the established educational system so that access to the information that you're talking about is available to people that want to do things with it? One of the things that Alan mentioned was, he held up in a slide the book the Molecular Biology of the Gene. And I find this very funny.

If I had to name four books that did it for me, that basically set my career, the first one was Computer Lib/Dream Machines. The next one was the Molecular Biology of the Gene. The other two, which I kind of do together were Newman and Sproull and Foley and van Dam. The last one, which I just read a few years ago, which was a really good one, was The Decline and Fall of the American Programmer by Edward Yourdon. And this is also really good reading. But these are the kinds of things that have allowed me to build up a career on this, and other people, too, often at the expense, in spite or maybe despite of their academic career. How can we integrate these things together so that they don't seem to be so conflicting?

KAY: Just a quickie is that-- just looking back, to me, one of the paradoxes is that I think we made a complete mistake when we were doing the interface at PARC. Because we assumed that the kids would need an easy interface, because we're going to try and teach them to program and stuff like that. But in fact, the kids are the ones that are willing to put hours into getting really expert at things.

They'll spend hours shooting baskets. They'll spend hours learning to hit baseballs, learning to ride bikes. And now on video games, the amount-- I have a four-year-old nephew who is really incredible. And he could use NLS fantastically if it were available. He would be flying through that stuff.

Because his whole thing is to become part of the system he's interacting with. And so if I'd had that perspective, I would have designed a completely different interface for the kids, one in which how you became expert was much more apparent than what I did. So I'm sorry for what I did.

AUDIENCE: Let me rephrase my question a little bit.

SPROULL: But wait, wait, before you rephrase it, let me point out-- you're stuck with it-- that Alan and Papert actually did do a very wise thing, which was to focus on kids. Because not only do they have the time to learn weird and strange new things, they love doing it. And that counts for a lot.

NELSON: The design of media, as soon as you establish premises, you pretty much lock it on track. For example, the time slot determines the system we call television. Once you have the time slot and commercial television, the rest is given. Forget it. Similarly, once you have the curriculum plus the implicit necessity of keeping the dance cards of all those teachers filled-- in other words, it's a system for employment of teachers. And so each teacher has to have an excuse for what he's doing all the time. Then basically you have the educational system as we now know it.

As I heard your question, Sam-- and by the way, Sam and I go back to when he was 12. If I heard your question correctly, you said, how can we fix the educational system as we now know it? And of course, I think that's a contradiction in terms. The point is to get rid of what is in the way, and that is the curriculum. And that is the guaranteed employment for teachers. Because that essentially is what is in the way of the students.

If we could have a system, for example, where you simply say to a number of 14-year-olds-- right now, we say to them, you must sit here for four years enduring the duke of history, the mistress of mathematics, and whatever you dislike about them because they will fix your point of view. Education is a process of ruining subjects for you. And the last subject to be ruined determines your profession.


And each subject is personified with the face of this local ogre, or perhaps, in some cases, a wondrous person, who then represents that subject. And you never find out that every subject has something in it of interest to you. Every subject can be something that you take to your heart. I thought I hated history, I thought I hated mathematics, because of the people I encountered.

And so some other way around, I would like to see a school system where you simply say to the kids at 14, not, you have to sit here for four years enduring and enduring and enduring and being endangered and insulted. But rather, you can get out of here as soon as you present for examination any of these 80 mini courses, any 80 of these 1,000 mini courses on this sheet.

You can take more if you want and learn more before we send you out there. Have your choice. Then I think we would see real motivation. Because right now, there is no way students can exercise initiative, except A, if they're male disobeying, B, if they're female getting pregnant, and C, totally obeying and outdoing whatever the teacher wants, which is done by very few. Whereas, all systems are based on initiative. And if we give people a way to learn by initiative, then we'd fix the problem. But nobody wants to do that.

AUDIENCE: Yeah, that was what my question was more directed, not at the technology, but at the current system, and how can we make it more conducive to bringing people online and getting the kids involved.

KAY: It's like, cops need criminals. so. It's a very complex thing. Seymour and I were in Washington yesterday morning talking to two congressional committees about the justice problem. And of course, both of us were conscious of the irony that we were actually talking to the people who were in business to stay in business. And so the interest in changing things in any big way was something that's very, very far from what they're trying to do.

For instance, the average expenditure for children in the US is about $6,500 a year. If you multiply that out on a classroom, you get to around $185,000 to $200,000 a classroom. Well, those classrooms are not receiving anywhere near $185,000 or $200,000 a classroom. From 1/2 to 3/4 is soaked up by bureaucracy. It never gets down to the school or the kids or anything else. And I think that is an enormous problem.

And basically we said exactly what you are saying. The problem is that there is a curriculum. And we invoke the name of Montessori, whose idea was that school should always be an extended kindergarten. And it's the job of the people who designed the kindergarten to make what happens when you use it for your own reasons more interesting than the regular world is. And that's what Montessori did. And I think that's an excellent way of thinking of designing a learning system.

NELSON: Do you know what curriculum means in Latin?

KAY: Yes.

NELSON: Little racetrack.

PRESENTER 2: Ted, you suggest basically breaking the curriculum up into small pieces. And the Montessori system, it allows you to do whatever you like, so long as, whichever Montessori school, you use the same colors for the same length for the same blocks. So you're put within a very constrained world there. And I think one of the things I took out of your question, which was clearly a good question, is, there are four books which, for you, could have been your education. Given a computer, the network, and those four books, you would have been happy. And you would have achieved a lot of things and wasted a lot less time.

AUDIENCE: And I was.

PRESENTER 2: Right, you were. You managed to find them. Question is, how can other people find them? For you, well, one of the things you can say, that should be a curriculum. The reason that I feel that the curriculum won't go away-- there's a neat television program in which the BBC took four poor scientists who had been denied acceptance by the academic community. Their papers hadn't been published.

These beautiful ideas, wonderful ideas were just outside. They were just too different and interesting. And what they did, in fact, was to show that, by the end of the program, that they were also total junk. So the problem is that if you don't have the curriculum, then you leave somebody-- it's like throwing somebody out straight away. You say, right, guys, you're here to be free, to learn. And here's the net. And basically, they might as well do that at home.

So you're really asking is there a case from education in the sense of leading somebody. Do you want to lead somebody? Is there a sense for having something which is established as being reasonable for people to spend their formative years studying. And your problem is that it takes too long to get really good ideas into that. Now people are looking back at things that Ted said, and said, hey, that's cool. That should be taught really early on. That should happen sort of first grade.

But it's taken a long time for people to do that. And that's because it's a filtration process. That if you leave people to look at all the information, you also want to lead them and say, hey, this is really interesting. And sometimes it seems that the situations we've got, just that the mechanism we have for filtering this and reviewing it just takes generations. Now, my only suggestion if we're going to change that filtrating system, in some cases, for example with computing, we can move to a biological model.

So you can leave all your programming people around. And you let people pursue projects. And hey, you give them CPU power, and if they make something which demonstrably works and bubbles and enthusiast people. And now you've got the net you can actually use biological methods, in some cases.

You can't do it for learning good history. But in some cases, you can use biological methods and get a faster turnaround, because you can use computing and a whole lot of people out there who will participate during the hours of darkness to try these things out.

ENGELBART: It just occurred to me that giving its equal time is not fair. Because in the same amount of time, he can say so much more.


KAY: That was the problem with Butler.

PRESENTER 2: So many words, so little time.

ENGELBART: I just want to say that all these things are very relevant to look out about how you educate children and that they can learn more and differently. But the thing is, that the institution of education does not like to be changed by the children to be changed by the people that are there in the current state in order to change it. And we can't wait for the 20 years or so till those children are out integrated into society where they have the roles that they can start making changes that change driving things about both world events, et cetera, and the technology are moving too fast.

So we're just faced with the fact we have to learn a better way to shift organizations with the people that are in them now. So that's really why you need a strategy for it. And it's exciting to think of what the children can become and to flourish, et cetera. But the daunting problem that I think we really have to face is, how do you deal with the change of the adult world?

AUDIENCE: Thank you. And also, thanks for the last 20 years, guys.

AUDIENCE: My name is Ricky Goldman-Segall-- I've met some of you already before-- from the University of British Columbia, the faculty of Education. But I'm not going to pursue the issue of education, because I so vehemently disagree with some of the things that you just said, Ted. But you've heard my story before with that. Well, I remind you is that, A, how many of us-- I guess, raise your hands. How many of us would have been here if it weren't for one fabulous or several fabulous people in your life who inspired you, and were those people your teachers, if they were? I know they were for me.


AUDIENCE: Some of them were.

AUDIENCE: Some were.

AUDIENCE: Well, it's not just the teachers, but they had to perform within a system.

NELSON: You mean designated teachers-- a very important point.

AUDIENCE: Well, perhaps. I'm going to respond to Lee Sproulls notion that science or scientific investigation is this relational and interactive thing that actually could happen in the schools much more. And I'm going to do that in a way as a sort of summation, perhaps, as well, because we're at the end of the conference.

And what I'd like to know is what each of you learned from each other that you didn't know before you got here, and in some sense what you think your biggest similarities are. But also, what do you think your biggest differences are? Because I think within that, we sort of know where some aspects of the field are moving. So I guess that's my question.

MODERATOR: We'll start [? with you. ?]

KAY: Yes.


SPROULL: I learned that Ted Nelson and I are alike because we both have purple socks. I'm not wearing mine, but he's wearing his. And I learned that Alan Kay actually did believe that kids would be working together using their Dynabooks over the network. And I apologize, Alan. I should have known that you would have believed that.

I don't want to be only facetious, but that's a rather difficult question to deal with. So I think that most importantly, I wouldn't say I learned sort of bits of information. I experienced, I guess, a reaffirmation of the importance of the ideas and the importance of the scientists who have been pursuing them, sometimes with reward and honor and sometimes without. But that didn't matter because it's the ideas that were important and continue to be important. And so what was most important for me was the reaffirmation of that community.

NELSON: Well, I learned that everybody here is more idealistic than I thought. Well, no, some people couldn't be more idealistic.

KAY: I'm just glad to see all these old characters still hanging in there.


PRESENTER 2: I guess one thing, which actually Vint Cerf pointed out at a similar sort of mix of people, said that one thing that he pointed out everybody had in common was persistence, having ideas, and even though people telling you not to do it, hanging onto it and being really stubborn. And doing it even though you're told not to and even though nobody's giving you money to do it and what have you.

That seems to be something which everybody has had in common. I don't know whether it's something intrinsic to us. Or it's just that our ideas, which once they've got you, they don't let go. Apart from that, I guess it's pretty amazingly heterogeneous. And the differences are part of sort of-- and everything else. Which is good.

ENGELBART: What am I supposed to do? If it's to say goodbye, well, I enjoyed it thoroughly. And it opens my eyes to something that-- it was almost embarrassing. It's that I sort of lost track of a lot of things that are going on that are important and that I wish I could keep track of. So maybe I'll turn over a new leaf or something, or redivide my time differently. But anyway, I heard those remarks about idealistic old farts. And I thought, okay, that's why I have to say something.


Anyway, thank you for the meeting.

PRESENTER 1: There's nothing to say after that. I guess what I thought I learned is, once again, there's people who are laying out broad visions, and there are engineers who are building guts. And I'm led back to my preference for the engineers. And I'm glad to see that Bush was one. And I realize this is one of the great engineering institutions of the world, and I thank it for running the meeting.

MODERATOR: Sounds like a good note to finish up on. I'd like to thank our wonderful speakers and the audience for hanging in. Sorry about that. Sorry, Robert. I'll promise to do better. Oh, boy, don't like to cross that guy.


I want to thank our wonderful speakers and wonderful audience for a really amazing event in my life. And I'd like to invite any of you who would like to hear Ted do a reprise of his wonderful talk of yesterday. And for the rest of you, godspeed, and see you at the next one.