WWW Journal, Issue 3

INTERVIEW

Tim Berners-Lee

On Simplicity, Standards,
and "Intercreativity"

There's not much white space left on the white board in Tim Berners-Lee's office. Berners-Lee, the creator of the World Wide Web and the director of W3C, has spent most of the previous afternoon fine-tuning Consortium priorities. Yet when Rohit Khare and D.C. Denison arrive for a morning interview, it's not entirely clear from the tangle of arrows and circles what areas have won out: HTML development or security; content labeling or demographics. So the conversation naturally begins with the topic on the wall: priorities.

Q. How do you go about setting priorities for the W3C?

A. With great difficulty. We do a certain amount of putting out fires and a certain amount of growing--nursing little trees. We've found that different technical areas, different political and social areas--each one has to be treated on its merits, because the timing constraints and existing situation tend to be different. In the final analysis, we are guided by our own perceptions of where we're going, and by feedback from the advisory committee. We put a lot of conflicts in front of the members. After all, our members do represent those people who are seriously interested and involved in the growth of the Web. They are the people best positioned to help us answer those questions.

Q. The Consortium's content-labeling project, PICS, has recently been at the top of W3C's agenda. Why?

A. The famous Time magazine "cyberporn" cover story (July 3, 1995). That topic was already on the back burner for us, because one of our members had flagged it as an area that we should start thinking about. So we had already started looking into it. Which was fortunate, because suddenly there was great public awareness of that issue.

Q. And even though the Time story was quickly discredited, W3C had to get involved?

A. Yes. Because when the article broke, not only did justice have to be done, it had to appear to be done. So we gave it an extremely high priority. Sometimes things get thrown into the limelight. There are always fancy new features going into HTML. But of course we also get involved with issues that we think are important, where there is no public pressure.

Q. What's an example of that kind of issue?

A. Internationalization is one. The pressure for that in the United States is not so strong, but we feel it's important for the world that internationalization should be solved, so we have to put some effort in ourselves. Also things like architectural changes which will relieve the pressure on the Internet. No one person seems to have an incentive to do it, so in that case we have to put in the push ourselves.

Q. It's interesting that during the last five years, some issues have gone up and down in priority. Security, for example.

A. Security has gone down a bit. That was a very big issue when the Consortium was first formed. That's what the press was talking about at the time. Now people have realized two things: 1) They've got some basic security for their credit cards, and 2) There are a lot of pieces to the security system, and there's going to be a lot of development over the next few years. The whole world will not change its way of working in the next two months. So the press fever has died down. That doesn't mean that the amount of work to be done has died down.

Q. Let's go back five years to the summer of 1991. You've just released the "WWW Program" on the Internet. Who are the first people to use it?

A. There were three groups. I tried to spread it through the high-energy physics community, because that's what I needed to justify my spending time on it, of course; it spread through the hypertext community, because I put it on the alt.hypertext newsgroup; and it spread through the NeXT community, because those were the people who could actually run the software.

Q. What did people use it for?

A. I've always known that the route to adoption of technology can take very strange directions--in order to get from A to B, you have to find a downhill path, where each step is downhill for whoever is going to take it. The path can wander around a lot, and it did wander around a lot. Of course there was the initial problem of bootstrapping the snowballing of clients (if you can bootstrap a snowball). We did that by making the first server a regular file server, but the second server was a server for the CERN phone book, which was a gateway into the relational database. That's probably interesting to the people who are saying that the new wave of servers will be gateways into relational databases. Anyway the phone book was a crucial problem that needed solving, so the Web justified its existence by solving that crucial problem. It allowed information which was only on the mainframe to be accessed from other platforms. But it left a lot of community people at CERN thinking that the World Wide Web was a rather strange phone book program.

Q. Any surprises at the way people started using the Web?

A. I was surprised that people were prepared to write HTML. In my initial requirements for this thing, I had assumed, as an absolute pre-condition, that nobody would have to do HTML or deal with URLs. If you use the original World Wide Web program, you never see a URL or have to deal with HTML. You're presented with the raw information. You then input more information. So you are linking information to information--like using a word processor. That was a surprise to me--that people were prepared to painstakingly write HTML.

Q. If people didn't have to write HTML, the Web would be different, wouldn't it?

A. Yes. There'd be more gray material, more material on the fringes of publicizable material. Whereas at the moment, it's still a lot of trouble to publish something. It's not just a question of hitting the save button. Because of that threshold, the only information that's published on the Web is information that's of sufficient value to a large number of people. So World Wide Web sites have tended to be corporate sites, corporations talking to consumers, rather than groups wondering what they're going to have for lunch.

Q. So, the bi-directionality is missing, because people are writing HTML.

A. Yes. Writing HTML is like a programming task; it is not a way of expressing your reaction to something you've just read. The result, from the process point of view, is that it's remarkably similar to the paper publishing process, with a great big sequence--from the idea, to the writer, to the code, to the publisher--going through the bottleneck of the person who runs the server.

The original idea, however, was that it should be totally bottleneck-free, something between people and information.

Q. Five years into the development of the Web, are you worried about the so-called second system effect?

A. The second system effect, if I remember it correctly, is that when you design the second system, you fix all the problems of the first and fall into all the problems that the first system avoided. So yes, we are in a position to have the second system effect; for example, we are in a position to make something that is inherently more complex, that has no clean architecture. That's a danger. If we produce something that's too complex, for which there are no simple, underlying rules, then we will be in a mess. I believe W3C has a mandate to keep an eye on architectural integrity and simplicity.

Q. Are you frustrated at the pace that the Web proceeds?

A. Not really. People are generally amazed by the way the Web has spread, not by how long it took to get here. Most people only got on board in 1992-93.

Q. 90 percent of the people who currently use the Web have only been using it a year or so.

A. Right, so to them it's a 12-month phenomenon. People are not worried that it's taking a long time. During the last five and half years there are a lot of things that have not changed, and that's okay. From the invention of the steam locomotive to the very latest automobiles, wheels are still around, there are still four in most cases. It is not that we expect that rockets will have won out, or that now we should be moving onto twenty wheels. So it's quite reasonable that there are databases, and hypertext use of databases, and maybe we shouldn't feel that we need to go beyond that as we spread into a larger community.

Q. From the start of the Web, you've been promoting interactivity as the ideal. Do you feel that we're any closer to that? If not, what's it going to take to make the Web more interactive, more collaborative?

A. The word "interactive" is terrible in a way, because people mean different things by it. To really be able to work at a distance, to use a knowledge space, you need all of it. So when some people say "interactive," they mean taking real-time audio and video and integrating it with the Web so you can create a meeting document and talk to people in it. Another form of interactivity is to be able to make a comment on somebody's paper, to put a yellow sticky on it and say, "This is really important," with a link to why it is. Yet the real-time video problems and the annotation problems are totally different, and they're both big problems.

Q. How do you sort them out?

A. Well, I've recently started using the term "intercreativity" instead of interactivity. By this I mean something like building things together, which is more than filling out a form and hitting "submit." Imagine, for example, a heap of objects--a compass, a magnet, and some iron filings. You come across them in a 3D virtual world, and you can use them to learn something about magnetic fields. Suppose you can take these magnets and their properties to another virtual world and with them create a little tower of magnets and discuss it with your friends. Suppose you can build with other people within the virtual space. That will be much more satisfying and more productive than any of the current forms of interactivity. Yet that sort of thing will also need a lot of engineering; you will have to roll in a lot of things.

Q. So how would you define "intercreativity" succinctly?

A. Building together, being creative together.

Q. Are we getting any closer to that ideal?

A. As usual, interface technologies are always further ahead on the viewing than on the creation. Intercreativity happens when you are able to build, make something, express yourself while you are in the same mode as when you are reading, absorbing, surfing. In other words there's no difference. When you have something that you need to express, the threshold is so low that you can move it out into the communal space.

Q. There's a democratic ideal in there, isn't there?

A. Yes. I believe in democracy, in people governing themselves. My belief is that for society to work, every individual has to be involved on a number of different scales. Each individual has to look after themselves, typically their family, their work group, their town, their county--there are a number of different groups. Sometimes these nest, which tends to make life easier, sometimes they're disparate, clashing. Also every individual has to have a lookout for the planet. People need a balance so society can work out. I believe this is programmed into people, that people feel fundamentally uncomfortable unless they have a balance. The Web has to be sufficiently flexible for that, so this variety of interconnectivity can happen. That's why a Web that's specifically aimed at the inter-corporate or the intra-corporate or family use is not going to work. You have to have a Web which allows the family photograph album to link up to the group picnic to link to the corporate home page if necessary. The hive is only one scale--a large one.

Q. How about the Web as a cultural phenomenon? How does that strike you, to drive down the highway and see URLs on billboards?

A. That's very difficult. For a lot of people the URLs appearing on buildings has suddenly happened. They've heard about the Web one moment on CNN, and then suddenly it seems to be everywhere. But the Web started off as a hard push for the first couple of years, and it slowly gathered momentum. For me there's been no one point where suddenly the Web seemed to explode, where I sat back and thought "Wow, how amazing!" Although every now and then something has struck me as being kind of strange.

Q. But none of them have been milestones?

A. None of them have been a milestone. You see your first URL on a building; your first URL on television; your first URL on a t-shirt, on a bar of soap. We had a pair of boxer shorts in the office with URLs. But I didn't regard them as a milestone. Actually, I'll tell you one milestone: when Frans van Hoesel put up the hypertext of the Vatican Renaissance exhibit that the Library of Congress was hosting. At that point he took some really good, high-quality material and used hypertext to produce a beautiful way of browsing around it. That showed a whole lot of people that you could put really gripping quality onto the Web. That was a milestone. The Xerox PARC map project was another early milestone. It was crude, but it demonstrated to people that URLs didn't have to be just a file name; hyperspace was infinite in as many dimensions as you choose to pick.

Q. How about the first baby pictures on the Web, or the lists of music CDs, or the confessional aspects of the Web. Surely you didn't see that coming?

A. No, but in all the initial talks we gave about the Web, we explained that you would have your own private Web, and that still hasn't come yet.

Q. The idea of the "home page" evolved in a different direction.

A. Yes. With all respect, the personal home page is not a private expression; it's a public billboard that people work on to say what they're interested in. That's not as interesting to me as people using it in their private lives. It's exhibitionism, if you like. Or self-expression. It's openness, and it's great in a way, it's people letting the community into their homes. But it's not really their home. They may call it a home page, but it's more like the gnome in somebody's front yard than the home itself. People don't have the tools for using the Web for their homes, or for organizing their private lives; they don't really put their scrapbooks on the Web. They don't have family Webs. There are many distributed families nowadays, especially in the high-tech fields, so it would be quite reasonable to do that, yet I don't know of any. One reason is that most people don't have the ability to publish with restricted access.

Q. You've said that the traditional browsers will eventually go away--

A. What I said was that this idea of a separate browser should go away, the entire user interface should be integrated. Clearly some of the browser companies worried that I was suggesting the browser companies would go away--that wasn't at all the suggestion. But the desktop metaphor and the browser metaphor have got to become one. Whether the browser software swallows the desktop software, or the desktop software swallows the browser software--that's up to the marketplace to decide. But there's no reason why both sides shouldn't have a shot at it. When you turn on your computer what you should see is information, what you should deal with is information. You should be able to create it, to absorb it; you should be able to exchange it freely in the informational space. The computer should just be your portal into the space, in my view.

Q. Do you see a difference between information and knowledge?

A. If we're looking at the engineering of an organization, we're talking about knowledge: the value of the organization, the knowledge of the people in it. You need to take that knowledge and share it, you need to make the organization be able to think. That means you have to make that knowledge exist, rather than having the individual knowledge of individual people. So, yes, I make the distinction between information and knowledge when I discuss this in regard to companies.

Q. So far technology talks about information.

A. Technology moves information around in order to share knowledge. In the abstract space, one sees information as a representation of the knowledge of the organization.

Q. What kind of developments do you see five years down the road? Do you think the Web will still be around in recognizable form?

A. If we stick to its definition as universal information, we know the Web will still be there, by definition. But the question of what sort of information will be out there, and how people will be using it--I think could change very dramatically.

Q. In what way?

A. When people are able to interact with the Web, you'll see it change the way very rapidly. For example, Java and downloadable code will make it possible to dramatically decrease the time costs for changing what software you have on your machine. One of the analogies I sometimes draw is that before the Internet it would take a few days to get software. You'd have to have it shipped, you'd probably get a tape and you'd ask the systems manager to put it on the minicomputer for you. With the Internet, you only had to spend the afternoon, the tape had been replaced with an FTP transfer, but you were still left with the installation. With the Web all you have to do is browse around and click. If you have a browser you can install another browser in about three clicks. That's a fairly short time cost for the market to be able to change. When you have agents on your machine, then the time cost will be even smaller because the agents will download the software onto your machine to prepare for your next day without you having to do anything. When that happens we're going to see technological revolutions happen extremely rapidly.

Q. That's a little scary. An Internet year is short enough.

A. What is a Web year now, about three months? And when people can browse around, discover new things, and download them fast, when we all have agents--then Web years could slip by before human beings can notice.

Q. These racing Web years could take a physical toll on those of us who work on the Web.

True. But the plus point is that we will be able to live for three or four hundred Web years, which will be very exciting.

Q. You often speak of the need for the Web to be open, free from exploitation by one dominant commercial player. How can we keep the Web interoperable with such rapid growth?

A. It's always going to be hard. As long as there's the Web and it's this exciting, there's going to be innovation. When there's innovation, there's competition; there will always be competition for markets. But so long as there is more that you can do building on top of the Web than you can by meddling with what's inside, it will be a win-win situation for companies getting together to agree that when they can both do something, they should do it in the same way. There is a danger of fragmentation, but there is also a tremendous incentive for companies to make sure that the Web does not fragment. There is always an incentive for one company to try to move standards, to change standards and leave other companies inoperable, but there's a tremendous incentive for the community as a whole to prevent that.

Q. Do you often feel as if W3C is behind the train laying down track?

A. In some cases, it looks like that, because necessarily with HTML if a company comes out with a way of making text diagonal, you can bet they'll generate a lot of press coverage out of it, in order to gain market share. Then within the next two weeks another company finds it's no big deal to make sloping text, and they come out with sloping text. For a while there will be sloping and diagonal text. That may last for a few weeks but then there is no interest from the marketing point of view in competing on whether sloping or diagonal is a more impressive way of describing text. There is a serious technical reason for having just one tag. There are serious motivations for sitting around a table to decide, "Okay, let's work on a single standard here. In other cases, such as PICS, we're way ahead of the train."

Q. So you're feeling more confident that W3C can counter this raging marketplace and keep the Web open?

A. I think a lot of the progress we've made has given us a lot of confidence in the model that we have. I think we will always be worried about the future. The day we aren't, we'll pack up shop. It will never be clear, never obvious; you will never take it for granted. Yet we've made a lot of progress. There have been programs like PICS which have gone from zero to basically global acceptance in a very short period of time. The fact that HTML converges at each stage even though there's a lot of froth on the top of the wave--that's been very satisfying for the community. The froth is divergent, but the body tends to be convergent. But nobody can ever take this process for granted. The future will always be difficult to predict. We will always have to be extremely flexible in how we deal with new events, and the ways in which we try to bring our own agenda--architectural cleanliness for the long term--into the marketplace.


WWW Journal, Issue 3