Part 3 of 3
Click here to read the second part of David Bennahum's interview with Doug Engelbart.
Doug Engelbart: Let's assume that the vendor world has to be driven from profits and market share: when you find the right hill you're going to climb, it can help you climb better and better towards cheaper and better product. But, hell, it's not going to be what helps you find the right hill. What you have to have is real experimental working groups. This is going to take a new kind of institution. So we ended up formulating -- for the last thirty years -- this bootstrapping consortium.
David Bennahum: The Bootstrap Institute.
DE: The Institute is there to say: "Hey, people, what you really need is a bootstrap alliance and the right kind of consortium." There's beginning to be movement about that now, which is very exciting. We're actually getting a not-for-profit corporate setup for a Bootstrap Alliance; a formulation for it and how it's going to do; and a conduit through which member organizations can participate.
DB: So Bootstrap is a kind of private initiative to recreate the kind of research and development that's no longer funded by the government?
DE: It's the kind of thing that's pretty hard to fund directly by the government. It's sort of like saying: if you want to find out how you can colonize the floor of the ocean, the government will subsidize quite a lot of the endeavor. Pretty soon, if you're serious, you've got to go into combination with the government -- because it gives the subsidies that support it. But someplace, you've got to find communities of intrepid people who'll say: "We're ready to go down there and start living now."
DB: Is that what happened in the late sixties at the Augmentation Research Center? -- did you guys totally live this system?
DE: That made a big difference. Then in the mid-seventies we started saying "Hey. Now we're ready to reach out to other organizations and start giving support to those who want to change" -- that's the way the Alliance is planned to be: it's there to find how you can give the best support to member organizations trying to improve their capability. Our web site has an increasing number of things that elucidate the stance and the goals, and some of the dialog that'll be on Electric Minds will be helping to develop those thoughts.
DB: What role do you see for everyday people in the attempt to design these tools for the future? Is that something we can all participate in? Or is it mostly up to the scientists?
DE: It depends on what kind of scientists you're talking about. But to my mind, it's a lot more pragmatic than that. If you really start looking at the way people can shift the way they think and work and collaborate, you might say: "If we had skills and new methodologies and new conventions between us like this and this and this, we could employ tools that would do this and that" -- for which there's no market yet today, because there's nobody doing this and that.
DB: What's one of the biggest things that's missing right now from the web and from our computers? If there was something you could fix tomorrow, what would you do?
DE: There are several categories here. One category are the things that we evolved in the Augment system. We found they were very, very valuable, and yet they apparently haven't shown up.
Look: it's stupid to have a separate editor and a separate browser. In the beginning, we envisioned reintegrated editing browsers. Then you say: the moment I create any fragment of a document, I want to be able to let people point to it. So I want every object in a document, at every stage of its evolution, to be addressable by a link-server, so you could cite it and talk about it from anyplace.
DB: So in a way it would be like having a domain name-server that would name every document separately.
DE: That's just a start. Every object in a document intrinsically has a name and an address. Several categories of addressing came about. One: it was given an independent identifier number as it was created; no matter where it was moved in the document, it still could be addressed by that number. Another was that we learned that the documents had a lot of payoff if you structured the document explicitly -- basic hierarchy was very valuable.
In any event, a document would also have a location indicator. So you could use either one. If the document was still evolving, the location was a little bit shaky. So then we said, there ought to be all sorts of optional ways in which to view a document. This is something where WYSIWYG was totally in the wrong direction. You don't want it to look like what it does on paper. Maybe that's one of the views you want, but you want optional views. The very simple ones, like folding up at the first line of every paragraph: boy, that was really neat! And you said: well, also I could control according to how many levels I want; then I can also filter by content. You could say these things have properties that I may want shown or not shown. These are other optional views.
The journal system that we built was very powerful. It circumvented the issue of somebody saying: "Yes, I can cite it, and yes, it would be very nice to cite each passage of something where I want to talk in detail about it and discuss that with you. But what if you modify that document or it goes away? Then my document that points to passages in yours will look stupid, because people will need the reference, and I didn't supply a great deal of context because you could go look." We figured that it was important to be able to consign documents to a frozen state (which was to be called published tape), and we had this publication environment called the Journal where you could set up any number of journals, each of which was like a library that guaranteed it would give it a unique identifier forevermore; and forevermore, a link to that library would get you that document as it was published, with the date, hour, minute of publication on it!
DB: That's crucially important now, because on the web stuff disappears and evolves. You can't really have an established research base or dialogue because the context keeps evaporating beneath your feet.
DE: We installed that system in 1970 -- the first year we had our email with hyperlinks in it. It took a year or so for people to feel comfortable using it, but boy, it just got so important.
So there are still things in our web site, from 1970 on, that have journal numbers we put there. When we put them out, every paragraph can automatically have its location number target tag on it. Those are just a few of the quite different things we implemented.
DB: What is your impression, right now, of where the web and the Internet might be heading in the next few years?
DE: I think it's headed for a spiral of increasing utility and utilization. It's going to be that new social nervous system, for sure.
DB: What would Licklider make of all this?
DE: Oh, he'd be delighted -- he used to talk about the Intergalactic Network.
DB: After he visited your lab, Licklider wrote this paper with Bob Taylor, "The Computer As A Communications Device." It predicts that by 2000, we'll have this online community -- this world of people who are online -- and the big issues will be privacy and security: who will and won't have access.
DE: We had one big difference in all of this that surprised me terribly, in about 1976. He had come back to ARPA, and reviewing what we were doing, I was telling him: we have this system now and we had it on a server that was commercially run for us, and it was supporting customers out there, and we had actually recruited and trained a set of young women who had liberal-arts educations --who would be the sort of facilitator-trainers out there in the field. That upset him badly.
He said, "You've just admitted that your system is no damn good." He believed that that if the system were designed appropriately, it would teach humans all they needed to know to harness it. And we couldn't get him to say, "Well, at what time in the future do you think that will be the case?" In every installation we'd put in so far, you had to adapt and learn and adjust both how the system worked and how the people worked in order to comfortably get them started. He was adamant. He felt we had failed.
DB: Because it required too much training to use the system?
DE: Because it required any humans out there to help train. The computer would be so smart. One of the paradigm things that delayed current usage a lot was a belief that started in the artificial-intelligence world that you'd be able to understand human speech and all kinds of stuff -- how humans are problem-solving -- and make a model. The computer would watch the human interact with it for a while, and make a model that could adapt to the human.
DB: People still cleave to the "intelligent agent" metaphor. It seems a similar idea: that somehow these agents sense what we want and the computer adapts to our needs.
DE: Smart agents are going to play a good part in the future. But it's like automatic pilots in airplanes. They play a role, but they don't yet take off and fly the plane.
DB: There's a core debate in computer science that's been around for a while. Some people (I guess Licklider was one of them) had this idea that eventually computers would be very intelligent beings, able to anticipate our needs. The other side thinks that will never happen. The issue is the interface between the person and the machine, more than the hope that the machine will ever become able to think for itself or anticipate your needs.
DE: Well, there is the sort of in-between place where I fit. I say: there's no way of saying that computers won't get as smart as we are, or smarter. But if we as humans want a life and a society and all of that of our own, we can't turn it all over to them. We have augment ourselves as much as we can so that we can shape our own destiny. That has been my goal all this time.
I tell people: look, you can spend all you want on building smart agents and smart tools, and I'd bet that if you then give those to twenty people with no special training, and if you let me take twenty people and really condition and train them especially to learn how to harness the tools, the people with the training will always outdo the people for whom the computers were supposed to do the work. To learn what high-performance human teams can do is, I feel, one of the really salient challenges to which we should give a lot of attention and focus.
So much has come about. A lot of it is residue from the artificial-intelligence image, but a lot of it is in the marketing world. The idea is that simple and easy-tolearn-and-use are important when you're selling to someone new. But I tell everybody, "Hey, look: if you really believe that, I'd like to see the tricycle that you ride around on. Because you'd never have learned to ride a bicycle." The value of learning to handle special skills in order to harness some artifact -- a bicycle, skis, a skateboard, a sailboard -- those are important examples of what you can do if you coevolve your skills with what the technology can provide.
DB: I guess the tradeoff is that, down the line, a tool may not be that useful when you've mastered it.
DE: Look, we can change the system so it requires more training. You don't want to make it harder to learn than is necessary, but you don't want to limit it. I use a little example: "That's like saying, 'I'm going to design my whole automotive transport system -- cars, the way they're controlled, the way our highways are, the rules of the road, everything else -- in accord with the views that people had about such a system in 1905.'" Think of merging onto the freeway; you're going 60 miles an hour, and you've got to check over your shoulder, and keep checking in the rearview mirror and side mirror as you're merging, keeping track of stuff. If somebody in 1905 had said, "Well, yeah, drivers will be doing that every day," they wouldn't have been believed. In the first place, no one would have believed that people could handle stuff more than 30 miles an hour, and nobody used mirrors for anything like that. And if you said, "Women will do it too," you really would have been laughed off. So I'm trying to get people to say: "Look, let's start putting some teams together and really compete to get high-performance teams together." Then we'll find out the kind of skills that people are willing to learn.
DB: I think our conversation will give people a much better sense of how the tools they use came into being, and also of what some of the challenges are facing us right now.
DE: I appreciate the opportunity.
Join the conversation about the Doug Engelbart interview.
It seems to me that we should address the needs of our educational system using the net as a highway to provide the needed classes. I envision a combination of television, computer and in-class lectures that will offer top of the line information to all, cutting away the need for holding some students back for lack of funds while giving the rich the better advantage of higher learning.
Most Active Topics:
Topic 7 Cybernetics and Psychology
Topic 17 FYI: Fallback Plans
Look: it's stupid to have a separate editor and a separate browser. In the beginning, we envisioned reintegrated editing browsers. The moment I create any fragment of a document, I want to be able to let people point to it. So I want every object in a document, at every stage of its evolution, to be addressable by a link-server, so you could cite it and talk about it from anyplace.
Smart agents are going to play a good part in the future. But it's like automatic pilots in airplanes. They play a role, but they don't yet take off and fly the plane.
Also in InterMinds: