About Tim Berners-Lee
Tim Berners-Lee was born in London, England to parents who were both mathematicians. He was interested in electricity and electronics even in elementary school. He began experimenting with the use of transistors soon after its arrival on the consumer market and constructed two rudimentary computer in his home from spare parts and broken television sets in the 1970’s. He attended Oxford University from 1973 to 1976 before going to work for Plessy Electronics in Poole, England and later DG National where he laid their computer image systems. Much of the interview focuses on the beginning and growth of the internet and Berners-Lee’s vision of its future.
About the Interview
TIM BERNERS-LEE: An Interview Conducted by Robert Colburn, IEEE History Center, 9 April 2004
Interview # 439 for the IEEE History Center, The Institute of Electrical and Electronics Engineers, Inc., and Rutgers, The State University of New Jersey
This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.
Request for permission to quote for publication should be addressed to the IEEE History Center Oral History Program, Rutgers - The State University, 39 Union Street, New Brunswick, NJ 08901-8538 USA. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.
It is recommended that this oral history be cited as follows:
Timothy Berners-Lee, an oral history conducted in 2004 by Robert Colburn, IEEE History Center, Rutgers University, New Brunswick, NJ, USA.
Interview: Tim Berners-Lee
Interviewer: Robert Colburn
Date: 9 April 2004
Place: MIT – Cambridge, MA, USA
Education; intellectual and technological influences
So if I could ask some questions about your childhood and early schooling, and also any philosophical influences on you, either by works of literature or world events that perhaps might have had a hand in the democratic nature of the web.
Really that’s the sort of question which demands two weeks of thought and an essay! So historically, I was born in London, England. My parents were both mathematicians by training and by inclination. I was brought up as the eldest of four in a family where mathematics not only occurred everywhere, but was seen to be useful and fun wherever it occurred. So I went to elementary school at County Primary school in East Chine. My good friend all the way through that school was Nicholas Barton who read at an early age, and was also interested in science and read a lot more than I did and taught me a lot. He read about all kinds of things including dinosaurs and that sort of thing, physics and chemistry. So while everybody else was playing games on the hard top, we would on the playground walk around the edge just talking about things and occasionally winding electromagnets and things in the corners of the field.
I think one of the questions I had looking back is to whether my particular generation was particularly lucky in its timing when it came to the development of computers, and I feel as though we rode a wave. I feel that when we were in elementary school, and when technology such that you can wind a coil into a nail and make an electromagnet, but still you’re not really old enough to put the nail into the fire and let it die down so that the nail’s tempered so that the electromagnet won’t let it go then afterwards. That’s a problem I remember I had at one point. So the challenging thing is making things out of relays and burglar alarms and things out of coils, and that’s something which is available to you.
And it happens that as we made our way from elementary school into high school, at which we switched at age eleven, then Nick and I went to different schools, but fortunately we kept to our friendship and we started getting interested in electricity and electronics. At that point the transistor, which had been invented, became sufficiently ubiquitous that you could go to the stores down Tottenham Court Road and get packets of fifty of these things. But the point where we could make things like flip-flops for example, or relays, or we could start imagining quite complicated things made out of relays, it was going to get impractical to wind all of those coils, just at that point, the transistor arrived. But we got enough experience with the resistance and voltage and current to be able to understand the usefulness of the identifying function of the transistor. So, we could get a box of a hundred transistors, sort of empty them out on the table, and stick them to a little breadboard where we figured out which ones actually did amplify. These were reject transistors, but some of them the test would have failed, but they were in fact good. So out of that we’d get maybe a dozen decent transistors and we’d label them according to what gain they had, and use them for different projects.
Through the next few years we were making things out of transistors. I had a model train set. I went to a manual school, Wandsworth, which is actually stuck between two railway lines as the Brighton and the Bournemouth lines would leave Clapham Junction, which is a large junction in southern London. They fork out and it’s stuck in the space between them, very strategically placed for a prison; it’s a manual school. So all the kids obviously they were quite interested in trains because they were going past, disturbing the lessons all the time.
So I had model trains, and to control a model train it’s useful to make whistles, which we used to get Whitehall's [?] generators and we would make a ace [?] table multiplier for each of the back-to-back-to-back couple flip-flop, which made two transistors which will make a whistle -- a “wee” noise -- and you might want to have the whistle run for a certain period of time, which is another way of coupling two transistors back to back so that one is stable. And then there is a third way of coupling two transistors back to back so that they permanently stay in one position and the other is a flip-flop, which is the one with the memory cell. So there are a lot of things you could do with two transistors coupled back to back and sort of make some connections to effectively, and we made various sorts of intercoms to go between the top and bottom of the house.
So electronics was the fun thing, and mathematics was the easiest subject at school. Apart from that I was interested in outdoor things mainly. My parents had grown up in Birmingham and worked in Manchester and London. In all these places, their recreation was to get out into the hills. So we would take our vacations down to Cornwall and down to Port Isaac. So I suppose my relaxation was getting outside the town and into the sort of wilderness as it’s called in this country, and fun at home was electronics.
Now these flip-flops and things which you’d make when you worked with transistors, and in fact if you couple them up together, if you put two flip-flops together, you could make it so that every time one changes from one to zero, the other one changes state, whichever state it was in. So this in fact allows you to count in binary. If you have enough of these, you can count as much as you like. So that with counters, and if you can make counters you can make memory cells.
Also now my father was working in this area of computers, and I remember him trying to demonstrate to a general audience how a binary adder worked by building one out of water jets, such that if two water jets that go on at the same time, they collide and go into a funnel in the middle. If either of them is on by itself, it goes into a funnel at either side, which those two funnels are connected together and the funnel in the middle goes into the carry column. So if you work water jets for funnels you come to the effect. If you have enough of them, you can make an adder. Really an adder and memory and compiler is really all that the computer is made out of. So it was obvious computers were things which were more and more powerful.
As we went through high school, it became impractical to make what we could imagine out of transistors. However, ta-da, the little Texas ’74 series TTL logic packages came out just at the right time, just as we had figured out by ourselves with transistors, what nangates were, you could buy a 7400, which had put two nangates in one package, fine volts and ground, and that tie those together. And just when you got the point where you could make lots of things out of nangates, then subshifters and all kinds of things came out. So in fact, the end is that when I went up to Oxford, in my spare time there, making a computer from the human side in, I started off by getting an old television from AJM Electronics down the road for five pounds, which I went down there as a television repairman. I asked him if he had a television which had the radio part of it broken, but the monitor part working, and he sort of rolled his eyes and said, boy did he ever. He charged no more than five pounds. I got a couple of them and went home. In fact I put one up in my room in the attic and the other I put down on the hall table and I ran a video wire so the family could see my progress.
So I started up with a slow oscillator, which would give you horizontal bands, and then a faster oscillator will go up to vertical bands, and then some logic will give you a slot, and then the slot gets more. Then eventually I saved up for the memory to store the characters and the character generator. I think the character generator was a lot of money, thirteen pounds, twenty-five pence, if I remember. So I had to send away for that. It was just at the end of university that I had actually got the working visual display on that. It was 64 characters by 16, and after I demonstrated that I could optically isolate it—the computer was connected our window by a photoelectric isolator, demonstrated that to the lab technician at the Clarendon Labs. He let me bring it in and connect it up to BV8s so I could actually show it working with the real computer. Ta-da!
So now the next thing to do was to make the computer. I had all the TTL technology, but it would have been an awful lot of soldering of an awful lot of wires to have made the computer. It wouldn’t have been possible, it would have been worth it; however, at that point the Mac arrived, just as I left off for the port for that 6800. It’s a very nice architecture they did for the microprocessor, nice and clean simple architecture, so it was easy, so I made a computer and then programmed it to hexadecimal and then wrote a component for it and so on.
Employment at Plessy Electronics; building computers
At that point I could get a job. I didn’t really know a lot of people, have a lot of role models who’d gone on and done PhDs maybe at that point, but I also gone to Stanford and MIT done a Masters or done a PhD in computer science. I didn’t really know that route. I didn’t know anybody who had done that. At Oxford, computer science wasn’t a course. The computing was only there as a maintenance. But on the other hand, the electronics industry was very exciting because of the huge amounts of hardware which could now be replaced with the small board of microprocessors to do more thinking and just as much output. Now you have a designer, so I could go into any group which had a large rack of hardware and then design a piece of hardware to replace it, which then put in a microprocessor and program that microprocessor.
Now you’ve got Plessy Electronics in Poole. Plessy was one of the telecom companies in England. The others were ITT in Harlow, and GEC in Coventry, and the significant difference between them was that Harlow was a new town made of concrete. Coventry was bombed in the war and was largely replaced with concrete. The people there are very proud of how easy it was to get out of Coventry into other places, into the hills. Plessy was built during the war, well out of the way of the city, down south, a beautiful part of the south coast with seagulls wheeling around and daffodils and miles and miles of gorgeous coastline, so there was no contest there.
So I went to Plessy and had a job doing this fun stuff for money. Meanwhile, I had a computer at home which was the first of two home-built computers. Second one I built with a crazy math teacher, Mike Blanford, who built out of circuit boards. My first computer I did with a thing called variable, and you can probably still buy it, which comes with a lot of horizontal lines of copper which you can break and then you solder up the connections, so it’s half done for you. Mike had got into circuital technology here. One day a week he’d use his kitchen to cook a huge pie which he would then eat for the other six days of the week. During the other six days of the week, the stove was covered with a big bath of sulfate with which he’d be making the circuit boards. He had made himself a seven by seven board with a processor on it. Actually, after a couple of years at Plessy I went out to join DG National to lay their image computer systems. But while I’d been at Plessy I’d actually seen how printer circuits were laid out properly on a board, and I’d worked with some people in the printer circuit board shop who later spun off and had a little circuit board place in the industrial area. So I knew those people. So I professionally laid out on a drafting board a companion board which had a floppy disk interface and the other 32 K in memory, and so we had a fully made up system when you put these two boards together.
We actually put a small ad in a magazine. It was sort of at this point at which Wozniak and Jobs then set up a professional business. We never got the hang of really selling it, and we would have had to tap into more computer classes than we had around. There may have been something like, I don’t know, but we were part of the same excitement of that a computer could do anything you could program it to do.
What year was that, about?
I was at Oxford ’73 to ’76. I can’t remember the details of exactly when. Anyway, it’s complicated. I was at Plessy for two years, I think, so that would be ’78, and DG Nash, while I was there I actually took a break for six months to go to CERN as a consultant because the photo cyclotron division was stuck and in desperate need of programmers. That was in the second half, June to December of 1980. While I was there I heard that John and Dennis Nash had decided to split up and they were the partners of the company, and so this eventually split into DG Nash, which ended up Image Computer Systems. DG Nash was four people when I joined as the fourth. Supposedly small companies are a lot of fun, working in that environment with lots of small companies, largely seeded by Plessy Telecom.
Influences of technological progress on education and society
So that was really my education in computing. If a child now grows up and is put in front of a one gigabyte computer, which is to my mind, mind boggling, and so the question is: do they have time to go back and actually understand what’s really happening inside the computer? I asked some of the faculty here who said, “Well, actually it’s surprising. You teach just the same things, and then there are only a few sessions, of course, but we shall get people from where you were in ’76 to where computers are now. They are not that different, in fact, kids can follow.” So maybe they can understand. So I think the sort of cumulative value of the excitement of riding the wave of possibility, every time you imagine something more complex and more powerful being possible and figuring out how to do it, suddenly you are then able to actually build things with it and go onto the next stage. I have a feeling that I was just unbelievably lucky to be at that age.
There was certainly another wave, because about 1996, 1997, we were talking about the web and the possibilities, and one of the students in my class said, “This must have been what it was like living during the Renaissance.” And I thought, yes, it really probably was—that idea that suddenly so much is available and limitless possibilities and just the intriguing nature of what we were beginning to discover at that point. I thought that was a very apt phrase. So I think the wave is continuing, accelerating.
Certainly, yes, every bit as much…
Well let’s hope that we can catch the art and the humanities to the extent that we will. And it’s clear, of course, here at the Consortium where technology and society is in fact one of synergy, and that the work we do is very much connected with its impact on society, and we spend as much time thinking about social phenomenon as a technical phenomenon, and even the most technical things, such as the frenetic web which you could think of, which is really a bunch of applying logic, using the web as a web of logic as opposed to a web of human communication, which is now simply mathematical. In fact, when you look at it as a communications phenomenon it’s all about evolving culture, which is all about groups and the relative pain for a group of being isolated or being joined. So it’s about sort of harmony and conflict in ways that start out mathematical, but very soon when you try to link to bring into the enterprise become very real and social and very emotionally charged relating with the history of people. So that’s an interesting phenomenon.
Implications of privacy, commerce, and regulation for web design
Along those lines, getting into the places where the web has very much succeeded in what way and places where governments perhaps or cultures are trying to restrict it or feel it’s threatening, what sort of tensions has that caused and how is it making progress, perhaps, despite those restrictions?
Where to start? This is a huge topic.
Yes, I realize that.
It would be inappropriate to focus on just one area. I suppose the web design, the main feature of the web design is that it’s a generalization. The web is a generalization of all the different documentation control and access systems which had existed out there. So the driving point is that all the things like menus and selection bars and screens that you go through in talking to something like a library system could be represented as hypertext, as could all the documents. In fact, there’s no real difference between them. The hypertext was a fairly simple generalization—it was text with links. But boy, if you had text plus links as the humans interfaced, then you put the network there to make an abstract space out of the real space, and you could actually pretty much represent everything and the ability to represent anything is very important to it. So the technology has to be extremely unrestrained. It has to be like a white sheet of paper. It’s no good creating paper which can only be used for one sort or another, can only be written on in a particular language. So this has always then raised some questions when technology is abused, as that’s lots of pornographic sites, shouldn’t the power of the Consortium stop http being used for pornographic sites. So in fact, there are laws about porn and there are laws about fraud and there’s a lot of porn going on and there’s a lot of fraud going on and a lot of spam. We have a spam deluge. This may go down in history along with the Black Death and the Great Fire of London, we might look back at this and say this was the spam plague. For some reason, we haven’t managed to actually put the technology and the society pieces together to control the spam. We are losing huge amounts of time on it, and the spam brings viruses and viruses create spam and the spam now transmit viruses and they actually kill machines as well. So we’re in a bad way. In that way a lot of the spam is actually is fraudulent. The only reason it gets through the filters is that it’s actually lying about who sent it.
So we have rules about fraud. Part of it is a confusion about where the boundaries should be. Some people, because they feel that this plain sheet of paper access to the web is really important, but then they take it too far and say, “So on the web you should be allowed to do anything,” or, “You should be able to send email anonymously.” Why should you be able to send email anonymously? Maybe it’s an important part of society to be able to make the anonymous denouncement. Maybe that’s a core part of democracy. But on the other hand, when you denounce something, for example, you do it publicly; and if it’s anonymous, the public or other people can go off with it and talk about it and produce evidence to the fact that it’s actually wrong. Whereas the sneaky sort of spams which impact the scans sneak through there privately, they are anonymous.
I think the first thing that turned up was the porn, and the thought was that if nothing else happened, there would be pressure on the government and they would not put Internet into the school unless the government did something to prevent pornography being accessible. And so there was a pressure to make it illegal, for example, to have a wire which carried pornography, which was a misphrase. It was an attempt to put the onus on the service provider. The reason the Internet works and is so cheap is that the Internet service provider does nothing, they just sit there and watch the wire to make sure they don’t burn out, and every now and again reboot the routers when something goes wrong, slowly configure it. But basically they get out of the way, and that is not only why the Internet works so efficiently, it is also why new technologies can be developed—why I could develop a web browser and a web server, let communications across the net, only because the net itself knows nothing about what applications you run. So to actually start to put into the network some attempt to look at pictures to see whether they had too much pink on them would have been bad architecturally, bad socially, bad from the point of utilization from the web.
And also it would have been sensualized in another sense. We talk about desensualization in the sense of the technical. Technically, the web doesn’t have one single node that all the packets go through, the web doesn’t have one single server—even though some people would love for it to have one, they’d love to own one. To even attempt to try to take over net space, it doesn’t have any central bottlenecks. But the moment you say that Washington, DC, will determine the level of which something is deemed pornographic, you have got conceptual centralization. You’ve said that actually Washington is going to own the ethics of what is pornographic for the entire planet, and you’ve destroyed the diversity of human culture. So that, in fact, is the most serious threat.
What we did was to produce some standards to show that there is a technology which you could use, we use what’s called the PICS Standards, and they were in fact the forerunners and important by today’s standards. But the idea was to demonstrate that you - as a concerned group - such as a church, could write a list of all the bad websites you didn’t think were right, the ones which from your point of view were bad, and one could make a browser which would just check with the church website first every time it picks something up. There are various ways of making this happen more smoothly. So we demonstrated that you could work with a situation where the control of what the child reads is with the parent. Now that’s very traditional. It’s kind of built into the way we are—the parents look after the children and the children look to their parents for guidance and that’s much in families stronger then the power of the government. The governments look after the very seedy side of things, but in general for these questions of what children should read, that is up to the parents. We demonstrated that you could bring back the power to the parents if you have technology for having white lists and black lists. Of course there are a whole lot of products now which you can buy from different places which will use slightly different algorithms and some may be a little harder on sex and some on violence and so on. But the main point is we don’t have a centralized definition of what’s acceptable for a seven and a half year old.
So that’s an example of where we demonstrated that the net did not need the government to step in. The other is where actually there is technology but you still you need government stepping in. For example, fraud. Really, at the end of the day, if somebody lies and cheats over email and on a web page, you need a policeman to come and put them in jail, just as in the old days. What the Internet can do is introduce protocols so that you can check who people are. We could use digital signature on emails for example, or now there are sorts of features in the mail system which allow people to check whether the mail looks as though it came from the right place from a given email address. We’ve implemented these in the consortium already to try to lead the way. But all the same, if something is fraudulent, you have to have laws which are going to get back to the person to make it improbable for them to do it.
I think some people from the PICS case concluded that we can solve all this sorts of thing with technology, so government should just keep out of there. There was also a knowledge that for centralized government in a lot of cases wasn’t what Internet wanted, the Internet needs to be decentralized. But on the other hand, it was very clear that governments do have to be aware and we do need laws which work well on the Internet, and that’s not simple and it requires the law makers to have a good knowledge of the principals of the Internet, as well as the architectural principals of the Internet. That requires technology, so all the time to be thinking about how they are going to explain the social ramifications of the latest device to the lawmakers.
Do you feel that the technologists have been pretty good at being able to do that, or is the law of unintended consequences so huge that trying to predict that is almost impossible?
The way the web works is that there are some fundamental guiding rules about how two computers communicate, http to HTML and URLs. They are not just formats for data, they are rules about if I say this it means this, and the implication is that you’ll say this, and if I give you something which says that this page will expand in five days and you give it to somebody else in two days saying that it’s a valid copy of this, I can’t complain that they should have got a more recent one because the rule is that that’s what we call a valid representation of the page because it was guaranteed. So we have little protocols which allow all kinds of things like caches to work within the networks, and proxies to hold copies, and your browser to hold onto copies of things so that they work much more efficiently.
Things like network news is a great example of a peer-to-peer system which was put together by inventing just a protocol between two parties. So the art is to invent the protocol between two parties such that, if you do the mathematics, when you look up the results, it will be a harmonious and efficient and pro-human useful system. It’s a little bit like when a physicist looks at how billiard balls interact and says, “Okay, now let’s just imagine that we had a vacuum filled with billiard balls. What would its properties be? What would the average force on the walls be from the billiard balls bouncing off it? Oh, that’s interesting. When we do the system mechanics we find out that the effect would be that our huge space of billiard balls would behave very much as we see a gas behaving. All right, let’s say that we have a model that the air, obviously, is made of lots of small billiard balls. Now that gives us a way of thinking about what it that reminds me of it but it will justify calculations and things, and we work for a little while, imagine that is what air is made out of.” The interesting step is from the microscopic rule to the macroscopic rule. Well, guess what, laws are the same. Love thy neighbor as thyself. If your guide here is the Golden Rule, you walk around as an individual with these rules and then you act appropriately, and the result is that globally, everybody is harmonious.
When you make something like the web technology on network news technology or the Internet in general, linking something which is going to create a huge system, and connected to systems which have certain properties. The idea of the web, for example, that you should be able to read something, bookmark it, mail the bookmark to a friend, they should be able to click on the bookmark and see the same thing. That’s the rule of how it works. Sometimes people break that. So that’s the general property that it has from the users’ point of view and that which comes from various architectural rules about how machines talk to each other.
So what technologists guarantee is things like that. These will be the properties of the web, and I’ve spent a lot of time trying to explain to people which are the important properties and why it’s important that we keep them socially, why it’s important that you don’t go around changing all the URLs on your websites, because the fact that you can follow a link later is important.
What technologists don’t guarantee is that there will be no war in the Middle East. So there is another layer at which there are people working who look at this fact that you can publish stuff very cheaply, anybody can publish stuff. Let’s get 300 families in this country to publish websites about what daily life is like for them, and let’s get 300 families in this country to do the same, and let’s translate them all into the other people’s languages. So there are various layers at which you try to make things harmonious. There are people who are working to try to make sure that society evolves in a proper way. They are trying to use the web to make governments more accountable, to cut through the country boundaries. Patriotism is a great thing in America. Depending on which state you are in, you see an awful lot of stars and stripes. Everybody is proud to be an American, and at the moment we are seeing this is a problem in America is being very respectful of itself, but not sufficiently respectful of other countries. So why is this? It’s because there are a lot of very big communications within America. The federal system is put together to be very efficient. There’s a lot of infrastructure. Within contiguous states you can drive there, you can have strawberries shipped from ridiculous places at the wrong time of year. But outside that boundary there’s a language barrier, there’s an ocean barrier, there’s time shift barrier, all these things make setting up a relationship with somebody, understanding they may be in the same pain as you, much more difficult. So there is hope that the web just by its nature of breaking down geographical boundaries will be able to help us be more harmonious at the global level. So some people have just hope, just an opinion, some hold onto technology in itself inherently as making more productivity. Personally I think that’s a little simplistic, that the web won’t cure anything by itself; we have to cure things, and we have to do it by changing our attitudes, and as we do that, the web is one of the tools that we can use. So the law of unintended consequences means it’s going to run out of control.
I don’t think the web and information space itself is going to do that. But there are applications built on top of it and applications built on the Internet, for example, we are seeing intellectual property world being turned upside down, and the original system which was very much based upon a book being difficult to manufacture as a token of intellectual property has broken down. Even the one based on a vinyl record being difficult to press has broken down. This is really interesting. We are trying to create a completely new system of intellectual property which works with digital. The iTunes system has been very interesting. The Apple iTunes program and their website tackled it head on. They were not very weblike about it in that the software went only to one store, which is unfortunate, so it’s not setting up a very good market for music, it’s trying to monopolize and get vertical sites, which is always detrimental for the web when the software only works with one data source; that was preweb. In that sense it’s preweb. What’s interesting about it is it tackles the idea that it’s reasonable to pay a small amount of money, a manageable money, pocket money for music, and then it’s reasonable to share it to a certain extent. I turn on my laptop and I see, depending on who’s in the building, various sets of shared music which is shared over the subnet on the grounds that these are people I’d be swapping CDs with anyway.
So on the one side we've got record companies saying they want to absolutely own every computer—anything which has got a computer in it should be cryptographically protected from being a general purpose computer so that it can be controlled from Disney headquarters. Slight caricature of the situation, but only a little bit. Basically in this building, this is the Computer Science Foundation Intelligence Lab—computer science hadn’t even been born— and this original excitement that a computer is whatever you can program it to be is absolutely as ablaze as it ever was, and there are huge numbers of new things to be done. So the idea that somebody might take a personal computer and screw down the panels so that you don’t have access to it so that it, if it read an encrypted CD it would only give it to you and would not send it on a network and you wouldn’t have general access to the bits in your computer, is philosophically abhorrent to anybody who’s been involved in the development of computers. Clearly, it’s also, I think, people raise their eyebrows when they look at the history of the extensions of copyright laws for the very large interests, and so those lack trust of some of the large copyright holes there.
So we've got that at one extreme. At the other extreme there are people that believe that music needs to be free regardless of musicians’ need to be paid for their music as a livelihood. I think some people would like to see this being perhaps a step by which we could establish music and the artist being something where she could make a living for a change; whereas historically, it’s a really terrible way of making a living, at least until you die. So to be able to get money back to musicians and artists here in their lifetime for their works is something the Internet could do faster, using PayPal or pay with encryption.
So here’s a really interesting situation. I think it’s a challenge. Something we can design it from the ground up. Most of the copyright laws are very arbitrary, they are very much based in the wrong medium, they’ve got some good values between them. We’ve got huge problems with patents and the whole development of software is threatened by the fact that people put these ridiculous patents on lines of code, and then go rooting around, updating the program every year and that sort of combination of ideas in totally mischievous ways. So the patent system needs a total shakedown, and I think it’s starting to happen. You can see the ways that it’s falling apart at the seams. The whole idea of treating patents for software like patents for mechanical devices was a very crude first approximation, and we are seeing some interesting second approximations on how to treat patents and intellectual property and software and intellectual property and music. I hope we can get to a situation where the software makes it easy to do the right thing—it makes it easy to share music within the bounds, it helps you keep track of whether what you are doing is something you have license to do. But your computer is still your computer is still your computer so that you can program it to do anything. But if you go through all the trouble to program it to do something illegal, those people - the people that are in trouble - will be few and far between. They will only do it when they are going to do it on a large scale, and when people do bad things on a large scale, normally there’s ways of finding them. You find them when they put money in the bank. So the normal incentives of a large punishment and a reasonable probability of capture which we’ve used for centuries and centuries should work, and the people should have the freedom to write computer programs.
Absolutely. One of the things that you talked about in your book was the web of trust that makes the web work and also which seems to be developing. I was wondering about your feelings on that, asking as somebody who uses the web for a lot of commerce and is taking everything on trust, and so far I have not ever been disappointed. It seems amazing, I guess, but the human tendency is to trust.
Well, the human tendency, of course, is evolved to strike a happy medium where people who trust too much tend to end up sort of at the bottom of broken bungee ropes or in trouble, that in fact you feel that you trust. For example, if somebody walks up to you in the street, they are dressed in blue overalls with grease smears on them, and they ask you for your credit card or if they walked up to you in a restaurant and asked you for your credit card. You’d look at them and you’d say, “Excuse me? No way,” and you’d reach for your cell phone and push in 911 while looking for an escape route. Whereas if they walk up to you in a garage and ask for your credit card, then you assume that they’ve got the uniform, they look in place. If someone walks up to you wearing a smart black and white suit in a restaurant, you’d give them your credit card, but if they walked up to you in a gas station, you wouldn’t, you’d suspect it might be a scam. So you’ve got a lot of ways of figuring out who they are and you’ve got a lot of your brain that’s very good at looking at patterns and noticing things out of place, and then you’ve got the famous hairs on the back of your neck which prickle when something doesn’t work, even when your conscious mind doesn’t know what’s happening. The credit card companies have these programs which try to do this prickly thing. They look at patterns of people’s spending. Have you ever had one of those phone calls about your credit card?
Yes, I have. It involved some international travel.
Which you don’t normally do?
Which I do enough of, but I guess this was out of pattern enough.
So that it was odd. So they have computer systems which will look for changes in patterns. So there are some things which are really difficult to know, but in fact a lot of things when you do business on the Internet can be done with things like Robo Cops. For example, you can look, like when you go into a store you have groups of stores get together and check each other out, and they have their own entry criteria, and you just get to learn that the badge is good. There’s some of that that’s happened on the web. When a store uses a person, they have fairly simple rules, that if you are selling something in this category and it’s under this price and the computer is offline, just take the credit card. Some machines, credit card machines or ATM machines will function without connection to the big mother computer under certain circumstances with a certain amount of money to a certain number of people. Then at a certain point you have to go up a step, you have to make sure you’ve got authentication from the mother computer of the credit card company, and maybe they’ll ask you to check an ID as well.
So there are some basic rules. For example, a bit like the question of the parent and the child, the parent can set the child up with a set of rules for which they’re set to go through; similarly you could get from your local town or your friend something which gave you a set of rules of how to do business, or a set of rules for what you should open as an email attachment. When you are doing business, to say, “I’d like to check, have you been certified as an online site with wireless from these places?” And this I got from the particular list of one which I use, my brother. He told me that this is good list of things and this is the rules. So part of it actually I think will be when we've got the ability to move those rules around to be in one place and another. Part of the PICS effort that is causing a stir was the standard for being moved was to take the PICS rules, move from one place to another which some people worry that we had a standard for that, Congress would then write their rules that they wanted and say they must use their rules. I don’t think that Congress should do that.
So a good question for historians at this point is that we’ve had for 20 years now public cryptography, which is this amazing technology that allows you to do all sort of things. Why aren’t we sending our email, why aren’t we encrypting email? So I think the answer is that we haven’t got the infrastructure around it. We need to think of the concepts of trust. We need to create two systems. There’s the public key infrastructure which is a very centralized, preoriented hierarchy of trust, which I think a lot of people rebelled against. It was set up so that if you happened to be one of the companies in a chain dishing out certificates, you’ve got a good income stream. If we want to do an exchange of certificates, you give me a business card and I trust that it was who you were, and then I go to your affiliation, you go to my affiliation and we’d exchange these certificates, and you’d be trusting them because they’ve both been signed by a master signer which may be a large academic body in the United States of America or something that you trust, or a large company which has got a well known name and who’s certificate came with your computer when you bought it. This isn’t why we know each other.
On the other side there’s the pretty good privacy of the PGP, very much peer-to-peer thing, which wasn’t quite easy enough to use to get off the ground, I guess. Maybe it didn’t have enough software around for how to deal with people who you met. We should be able to exchange PGP keys very, very easily. Maybe because you can write a PGP key, a hash of it on your business card, but people don’t use business cards like that. So the whole business of introducing people to each other by PGP didn’t get fluid enough for it to actually pick up.
One of the things that I hope will come with this, and I think it will, is that we’ll have standards for our D [correct word?] rules, and those rules will allow us to write down who we really trust and why we trust them in every situation. We were talking about the IEEE earlier, I went to the IEEE website because I was asked what I thought about its organization and I went to look up, and I remember I was involved in the production of one of the IEEE standards, the fastbus standard and standard routines, so I thought I would look for that. I found it but I couldn’t access it because I’m not a paid up member of the IEEE I guess. I seem to get a lot of stuff for someone who’s not, but at least I didn’t have a password set up. But then I remembered that it would be very nice to have a rule on our website if you’d been to a meeting, whatever you were with, you should be able to read the minutes and check the minutes. So under most circumstances, you go to any secretary of any meeting and say, “Here I am, can I check the minutes please?” and of course you get access to the minutes; this is so obvious it doesn’t need stating. However, you can’t take it through a computer. I couldn’t say to the IEEE website, “Hello, duh, look inside ???, that’s my name, I wrote it. I’d like to, just for nostalgia, I’d like to read it again.” But if you are on the author list, then you get access to it. Because that’s a little trust thing that’s very reasonable and makes life work. And then obviously, other people may be on the editor list and they should be able to make a new version of it and so on.
So it would be nice to be able to take the rules which we’ve got which we’ve got written through our bureaucracy and give it to the programmers. Similarly, when you do consumer-to-business electronic commerce, you may want to have some rules running around to do simple things. Like if somebody asks you for your Social Security number, you don’t give it to them generally; but if you are doing some work for them, then you do give them your social security number because they are bound by law to write it down on the 1099 and they have to get it back to you. So there’s a protocol there, so they should be able to ask for it, your machine ought to be able to say, “Okay, but show me the order of work that you’ve done so I can guarantee that this is a reasonable request. So okay, we have a business relationship. In that case I can give you Social Security number.” As a protocol to ask, not to give it to anybody else. Also the protocol is that I’m expecting a 1099 from you at the end of the year. Is that okay?” Right. Now all these things fit together, so the fact that you’ve established a trust relationship. In fact, when it comes to go through your taxes, how do you find out where you expect 1099s from? What’s the list of people you’ve given your Social Security number to? It’s very nice that the computer will handle that for you and check.
So somebody was asking me recently, “With everything on computer, will we lose the paper trail?” And I think my conclusion is that actually the paper trail is going to get better, because your computer will keep more little trivia and it’ll be able to do more accountability. You’ll actually be able to sew things up much more completely. Certainly doing tax forms should be completely automatic. If it’s not completely automatic, your computer should ask you a few questions about your life, about what you use something for, but apart from that, tracking down all the various incomes and outgoings and doing tax rules…
One of the things that I’ve complained about for years is that most of the questions they ask you on the tax form, the government knows the answer to better than you do because it’s been reported to them first. About the only question they don’t know is, “Do you want three dollars to go to the election fund?” And your charitable donations, everything else they already have.
So a year ago I went through the US tax form instructions and wrote them in semantic web rule language. Then I wrote the rules which, from my accounts, I downloaded them, I tried to do as much as possible using my web languages, but put all my accounts from the bank in a standard which is called [???] and converted those into RDS data language and I wrote some rules which should explain, depending on who the payee was, what sort of category of spending, which related my category of spending to the tax form. I did this through the end of the year and on January 1st, pressed the button and produced the tax form, filled in. Of course, it didn’t have quite all the information from the 1099s; but again, the 1099s as well, it had the category of the income. So I had a pretty good guess at what they were going to be. But I didn’t have a machine copy of the W-2, which is what you get with a salary which you get in XML. That was one of the few things which I couldn’t generate already from slap adding machine little form. So in the various different files which represented different sorts of the social things, they were sort of my judgments as to what proportion of the telephone I actually spend for business related things, which is where I put in that judgment from me. And there were some other rules which were an encoding of the law as laid down by the Internal Revenue Service. Those really, I felt that was the last time I should write those rules out. What I should have done was sent them off to the Internal Revenue Service and said, “These are the 2002 rules. Can you send me the 2003 rules in the same language, please.” But hopefully in the future the rules will be published. At the moment there’s a certain number of sorts of software which you can buy which is certified and tested, but the test data sets aren’t public. I ought to be able to send them an encrypted, signed tax return by email without going through a special website or anything, or I should be able to prepare it myself using my own software based on information which I have. The interesting thing about doing this was you see where it was that you could trust-- I used some commercial type software another day and it made a mistake. It used the wrong date for a transaction, put it into the wrong category, miscalculated the tax bill which gave me a tax penalty when I didn’t need one. And I couldn’t ask it why. The “why” question has always been a very important question. Sometimes I describe it as the “Oh, yeah?” button. It should be the “Oh, yeah?” button. So when the machine has suddenly come up with some equation about your life, and the hairs on the back of your neck prickle, you press the “Oh, yeah?” Button.
And “Oh, yeah” means?
Take me through this. It means so why do you think that? Oh, so you take it from the tax rules. Why do you think this rule? Why do you think that was my income? Why did you think I got a check? So where did you put that information? Where did you get that? Track back, why did you put that there? It’s not got my name on it. So to be able to go back and see exactly where things have come from, who have you trusted for that? You know if it comes back and says, where it says because the Internal Revenue Service says that is what line 64 means, you kind of believe it. But where it says that because this door, I know it says that that one, it says that that’s what that costs when in fact it’s not worth that much, you go back to your bank account or something, you need to go to challenge it. So yeah, the word of trust is something we’re aiming for.
It’s going to take very socially aware technology. And there have been lots of times when we’ve tried to do that in the past, like with PGP and it’s not worked, but we just hope.
The State of New Jersey just introduced for 2003 a service you can use either your own software or go to their website and it would prompt you for certain questions, and then do your form for your calculation.
Both federal and state?
Just the state. And I used it and it was heaven, it was absolutely heaven. They actually caught the programming error that they had made on my return and called me back and said, “Actually it’s going to be more than you expected.” But that was a huge amount of progress.
The semantic we’re looking at all the languages from the basic philosophical and data and logical structure of this, and meanwhile people are trying to do cumulative interfaces for browsing through huge amounts of data. So when you have a question of, “What was that check for?” You can run it through and find all the photographs you took that day, for example, to figure out what you’re doing and that will tell you probably what you spent it on. So meanwhile, now when it’s a question of managing data and actually doing things with it, which affects you like for example in entering your taxes, we’re not there yet.
It was very nice to get that on the website, but the problem with the web form is that your browser, when you press submit, it’s gone. They may send you an email back, but the funny thing is, we’ve got an architectural way in getting to where the architectural problem is, but it’s an example of how technical architecture and social architecture connect very closely. When you send a message, it’s a social commitment. You say, “Okay, I’ll meet you at ten.” You send an email. If you’re not going to meet the guy at ten, you send another letter that says, “Oh, can we make it eleven?” You combine things by sending a message. As you edit a web page, when you hit save, if it’s going out to be published, then to a certain extent, there’s a commitment. But when you’re saving a file roughly that’s not really, it’s something that you can change later. When you’re browsing, when you’re following links, you’re not making a commitment at all, you’re just leafing through things. So the browser is in fact doing something completely different. It’s saying, “Get page. Get page. Get page,” to all these other web servers. And when you press submit, when you make it a buying transfer, you press this button which says, “I’m posting this form.”
Now a few sites get this wrong, so that you actually go through the website by pressing post buttons. So you’re browser doesn’t know you’re not making a commitment. But when you make a commitment, when you actually are submitting a form, that is just like sending an email. You should keep a record of everything you’ve submitted, which form it was that you've filled in, what it looked like at the time when you filled it in, the data you send in, and where it was sent to. Because it’s sent with a particular sort of parameter names and it’s taken as being representative of a certain commitment. You’d like to be able to demonstrate to yourself that you actually did pay your taxes, fill in your tax form for example. It would be nice if it was saved like an email message. You notice that when you save an email message, your email keeps the data; it does not allow you to edit the data anymore. It doesn’t allow you to edit an email message at all. You can take a copy of it, but the email message you sent, they are read only, buddy. That maintains the integrity of the system. It means that a message sent is a message sent, and the software means that because of the way the software works, nobody can edit them. So that this integrity is a fundamental sort of rock on which all email communication is based. Yes you could write an email path which allowed you to edit the sent messages, or even the received messages to make it look as though somebody sent something else. It would be technically possible. We trust people not to do it. It would be stupid, because anybody who does that loses all their credibility and the system falls apart. So that’s an example of how the technology allows you to do the right thing. This is normally what should happen with buying music over the Internet.
There’s a bug in it that when you press the web form, you didn’t have that information, so you fill in a huge amount of information about yourself which was actually totally reusable, in fact it had very well defined semantics. Most of the lines on the tax form correspond exactly to lines on the federal form which you have to fill in in a totally different way. So the semantic web is part of attaching semantics to that stuff, keeping awareness, and one of the interesting things is how to connect a loop to the person. So really, what would be interesting is to have a machine which said, “I’ve got to fill out this form. I need some information. Most of it I can handle, but I need your social security number and I need to know what on Earth you spent this on. All the rest is rules which have come from different places and data which has come from different places, the data which we used much more fully; and in return you as a person can find out what happened to the information. I mean you’re in charge. You’re in charge of where it is.
I could even start on the semantic web now, but we’re a little off track.
No, that’s all part of it, because yes, it has made it, although most of those sites - including this one - put it back up on my browser and said, “Print this for your records.”
Yes, responsible sites do. An awful lot of sites where you buy batteries for your camcorder don’t. Some of them do, some of them don’t. The good sites will give you something which is printable after you press submit, with the credit card number mostly inked out so that you actually feel good about printing it. Other sites, I find some that I have to use back up to get back to something which I’ve typed in and print that off quickly. Okay, I won’t be able to get it from the browser, but really the browser should keep a record of what you send—not because you want to repeat it, but just because it itself is can be a copy of a transaction.
So we’ve got a long way to go. It’s amazing what we’ve built socially at the moment, but when you look at some of these gaps, it’s not surprising that you wonder why we don’t have a way we trust. Well, just the fact that a browser doesn’t keep track of what who sent to what, especially on very simple steps which we haven’t got, and there’s a long way to go. I suppose a very important message is that we’ve got a lot more to go with the web. We’re doing our celebrating. We’re 10 years on the road, so 15 years of web and everybody feels like it was a great success. But in fact we really just started.
Well that’s the wonder of it, the hope of it, that as tremendous as it has been already, that the best I hope is yet to come.
It would be like celebrating electricity because of the light bulb alone.
Exactly. Hopefully we haven’t seen anything yet.
You haven’t seen the fridge, yes.
Internet and plagiarism
But I really do believe that student’s of mine observation about the Renaissance, I feel was absolutely true. I think I was lucky that I came along at a good stage also to appreciate it. Because I learned my computer programming in FORTRAN with cards and stacks and I never want to go back to that. But because of that, I was in a position when the Internet came and the web to say, “Wow, this really is going to be wonderful.”
Interestingly enough the headmaster of the school where my father teaches and where I help coach rowing, his main concern is not the students looking at pornography; it’s downloading other people’s papers. It’s the erosion of intellectual initiative that comes from being able to fake a term paper. So I found that was a very interesting view on that. He would rather they not download porn, but he doesn’t see that as anywhere as intellectually destructive.
That does seem to be a problem. That does seem to be a problem that the attitudes vary so much. There is this sort of attitude that the fastest way to the finish, if he cuts corners, is acceptable; whereas there are some problems in which that sort of attitude is wonderfully useful, but the way where it’s keeping, will students understand that this is what they’re here for actually is to come out with more in their heads than they went in, with as opposed to come out with having succeeded and having made the deadlines. I think yes. I suppose that is another area where the question is how much can you put in the integrity of the person and how much do you have to build into the technology? Just like with the forging received emails, you’ve got to lockup the cut-and-paste on the computer, and it’s going to get abused. I really hope that the students in general will get the idea that it is really dumb to cheat like that.
Some of them don’t even know that they’re cheating, though. Some of them think that is developed research. “I found this information. That’s what you asked me to do.”
Well, in some cases that is the goal. There are some assignments which are “go and find the answers to these questions on the web,” and that’s a very valid thing to do as well. Sometimes they are “go and get an original response from one famous inventor.” I’ve been given you and I must [unintelligible words] must be original, new.
And then you feel guilty, of course.
Yes, but I also feel like saying, “Hey, think about this from the other side.” So yes, you need to distinguish between draw this information and get to the point that you can discuss it. So presumably there will be tests where people are asked to discuss things or asked to sit down and write something, and there are tests like that, there will be tests like that. It’s like cutting corners on the athletic field, it doesn’t help you go faster.
Governmental and corporate control over the Internet
Exactly, but it is interesting what people may have initially worried about the most may not in the long run be the worries. A friend of mine who works for Amnesty International has as his beat Singapore’s attempts to control the Internet and the web.
Yes, it will be interesting to see how that pans out. It’s hard to imagine how something which has spread so far in fairly democratic areas will be able to do it. Technically, of course, you can do it. You can build filters which you can put on every point of entry into the net. But after awhile you imagine that your population will simply go abroad. People in Singapore are fairly well off—they travel, they know what happens elsewhere, they see reference to it. It’s impractical, just as it would have been impractical to draw a line on the web and say, “We’ll have an academic web and we’ll have a web for everybody else; or we’ll have a commercial web and a web for everybody else.” There’s a lot of questions actually I’ve had. Why didn’t you do that? Why didn’t you just have a high quality web where you can keep things clean, organize things according to Dewey Decimal System. And somewhere for everybody else. And of course, any time you draw a line, the line becomes a ridiculous anomaly and becomes under attack from everybody else. That’s why a lot of documentation systems weren’t successful because they were one person’s idea of how to organize information or how to cut what you have to publish and not another person’s. So when you try to draw that line around, there will be some information which is let through which makes a tantalizing reference to the information which is not allowed through, and it drives people crazy and there will be some friction and inability to put people to work economically, and also the effort of maintaining the political wall.
The political wall is a concern.
Actually, I’m more concerned at the moment, certainly in this country, about the large companies having a large amount of control over it. The Holy Grail for a big company is to give you the hardware for the software to be welded into it, and for that software to control which ISP you sign up with. Now for the ISP to then give you access to the Internet, but give you premium access to particular sites and run tons of load. You didn’t get access to your other sites, “Well sorry, they’re not our premium plan so pardon us,” and they also have political persuasions which are not like that. So already I’ve heard rumors of cable companies which will give you cable access Internet, but you’ll find magically you can’t pick up your email from a different email supplier. So there’s a pressure to, “Oh, all you have to do is change your email to email with us.” My cable company turned off Tracer Route, which is one of the Internet protocols which allows you to find out where the blockage is. Why did they turn off Tracer Route? Is it because there was a denial of service attack or a virus that used it, or was it that they are slowly starting to be selective about maybe they don’t want people to diagnose the network and find out where the problem is, find out who’s responsible for dropping packets. So we need some rules, general ethos that the Internet is the Internet is the Internet. An hour ago we were talking about how its fundamental property is that it just sits there and does nothing. You can make a lot of money if you break that rule, but you also bring the whole house of cards tumbling down.
And so far it really has stood up.
So far there’s been an outcry whenever somebody has tried. There have been lots of attempts to try and drop your competitor’s packets. There have been cases where you find that there is a proxy in the middle so that you are pretending to get-- When we went back and had a meeting back at CERN, CERN had put a proxy in and they said, “Yes, we’ll give you Internet.” Now this was an organization which was just trying to save money, but they had put in a proxy and their proxy wasn’t perfectly doing their protocols, so that when we connected through, we weren’t getting any of our authenticated pages through because it wasn’t getting authentication proper or something. We found some things were wrong and had the sinking feeling you get that this is a ten-day-old version of my site. I fixed it last night. “Hello, could I have the real one, please?” If somebody is saving time by giving you an old version without looking at it’s expiring date, or not recognizing headers which say, “Give me the latest one, really, really, really,” which are all implanted in protocol. So things like that, when the protocols start breaking down, it needs constant attention, constant out-flowing from people who understand how it works. Lawmakers need to understand that and be prepared to act on it, and the public also. There’s a loop with the public that the press writes out. “Such and such a cable supplier doesn’t give you proper Internet.” People have such a good choice in cable nowadays that that sort of thing can’t sway them.
Especially since as some of these technologies become more complicated and it’s even more necessary for the public to understand them, but it also may be hard if it’s only a number of specialists who are in the position to make that outcry.
Public education and publicity on new technologies
Yes, I suppose so. But the press typically does a good job. There are a bunch of people, some of them write blogs in semi-technical press, and they are in the chat groups of the real geeks, they get wind of what’s going on and they write it up. The Apple calendar product which uses standards which is wonderful, but instead of http column, mycalendar.ics, they use web calco [?]—it is really http but they use web calco which upsets everything else. You shouldn’t introduce new URI SPs, that’s a really major thing. That should happen every 50 years. So Tim Bray, a longtime XML guru, he’s been a consultant for many companies, and he’s got a blog which is called Ongoing, so he takes on these tasks. So hopefully that gets back to the management, “Oops, they are under fire.” It spends a long time explaining why that actually is really detrimental. There are all these links which will only work for one piece of software, and you are not getting the full benefit. If it’s got a web calco in front of it, people won’t know that they can use http caches on it, and so as a result these did low cache properly so they won’t be distributed properly. The web architecture is slowly crumbling. It’s okay, you can plaster it up for now, but what’s really needed is to fix the leak. That’s going to be a thing. We have technical architecture which is trying to customize these things, because sometimes they are a little bit of the sort that when you see it breaking you realize what’s happening, but it’s very difficult to write it down so that anybody else can understand it. We are trying to do that in the WPC technical architecture group, and we had Tim Bray on it until just now.
So in a way, the web has built in some of its own protection, publicity mechanisms, the very openness of it.
People who go to the web use the web. They also use email and Internet relay chat which is around another protocol which is what’s now called a chat room and it’s got all kinds of seedy connotations, but we have one for every meeting. So people, typically on a teleconference call, everybody is sitting on their laptops, the minutes just going past. The person taking the transcription minutes is typing real time so everybody can see what they are. It’s very quick. So we use the tools to build the web—to build the Internet tools, you use the Internet tools that you have, and that’s how you know how they can be better.
Any final questions which come to mind?
You’ve been very, very generous and you’ve given a lot of material.
A bit randomly.
But that’s the good thing about oral histories is that they do lead…
Some in the short term and some in the long term, but it will be interesting to see in 20 years to what extent the short term worries have now blown over and grown up.
But it is a wonderful, wonderful thing. Thank you for the interview.
- 1 About Tim Berners-Lee
- 2 About the Interview
- 3 Copyright Statement
- 4 Interview
- 4.1 Education; intellectual and technological influences
- 4.2 Employment at Plessy Electronics; building computers
- 4.3 Influences of technological progress on education and society
- 4.4 Implications of privacy, commerce, and regulation for web design
- 4.5 Internet and plagiarism
- 4.6 Governmental and corporate control over the Internet
- 4.7 Public education and publicity on new technologies