In 1981, he helped to found a company called Image Computer Systems, and ran its technical side for three years. Returning to Cern, he moved Enquire across to a Dec Vax computer. "I found it so useful that I put all the information on all the projects I was working on on it. It saved me having to tell people things, and allowed the people joining the project to get up to speed and do their own delving into it."
Berners-Lee speaks so fast that tape recorders have trouble keeping up with him. He is personable, but does not like talking about himself. He is, however, happy to discuss the Web at any length.
He came by his life-long interest in programming naturally: his mother was the first commercial computer programmer, working on what he says was the first commercially sold computer, the Ferranti Mark I, shortly before he was born. As a child, he built his first computer out of cardboard boxes and fed it with punched paper tape.
It was at Cern in 1989 that he proposed the Web in a memo he circulated. It took a year to get approval from his boss, Robert Caillou, to write the program. And once the nascent Web was set up, it grew slowly - at first, anyway.
Here is the second thing most people don't know about the Web: it was designed to be interactive. Not like it is now, where "interactive" means running the mouse equivalent of a TV remote control, but with an integral editor, so that you could pull whatever you were looking at into whatever document you were creating. If, for example, you were reading an article in one window, you could start writing your own in a second window, and automatically insert links across to the original. You still can't do this on today's Web. "It ran on NeXTStep," Berners-Lee offers as a partial explanation as to why not, "so not many people saw it."
Most people saw instead the line-mode browser he produced as a second effort, and were on the receiving end of his campaign to produce a graphical browser. This is what we now think of as the Web: at first Cello, Viola, Mosaic, and eventually Netscape. But all these programs were just browsers - not as sophisticated as his original idea.
Even Marc Andreesen, now starring on magazine covers for Netscape, said that building what Berners-Lee had in mind was just too hard. "He'd never seen the NeXT thing," says Berners-Lee, "and people just thought that was what the Web was, it was a browsing thing."
Roll on 1992, when the Web began to explode. On trips to the United States, Berners-Lee was talking to businesses pouring resources into the Web, and they all wanted to know one thing: did the Web depend on specifications held at Cern? If so, what was going to happen to them? How would the Web be maintained?
The result was the decision to set up the W3 Consortium, a neutral body to create and manage the Web and its standards. Berners-Lee is its director. It is based at the Massachusetts Institute of Technology, in the Laboratory for Computer Science. MIT had experience with the X Consortium, set up to do a similar job for Unix, it had the reputation of being trustworthy and acting in the public good when necessary, and it was willing.
So that's the third thing most people don't know about the Web: its development is not directed by Microsoft, Netscape, or Sun. Instead, the W3 Consortium works on new protocols in 15 areas to make the Web more efficient and more intelligent. For example, Berners-Lee logs on to the Web version of the Times before he goes to sleep every night: he likes to read British reports of American news. But what would happen if 5,000 other Bostonians decided to log on to the site at the same time? At the moment, 5,000 copies of the same articles have to travel across the Atlantic. With a little added intelligence, the network could send a copy over to a server somewhere closer - say, New York - for temporary storage and retrieval. Such a scheme would lessen the load on the network and significantly lower wait times for users.
Mirror sites do this now, but you have to choose to go to the site. "The user should never see anything," says Berners-Lee. "The user should just be click, click, click, and shouldn't know where the stuff's coming from." The solution the Consortium is working on is developing algorithms that manage the distribution of copies and how individual nodes talk to each other. That leads to the fourth thing people don't often realise: the Web is such a radical departure from the computer systems that had gone before because it is not centralised. Up until the Web, if you wanted to combine two databases, you bought or built a bigger computer and found a way to merge them into one. On the Web, you simply insert links.
"The Web is a massively parallel system," Berners-Lee explains, comparing it to the huge many-processor supercomputers being produced by companies such as IBM, Intel, and Cray Research. "And rather than a massively parallel computer, which is typically designed from the top down, the Web is defined by its protocols, so that anybody can add a new node on, and as long as it talks the Web protocols it becomes part of the Web. It's very similar to a kind of life. If you take a piece of mould or something, so long as everywhere there is water, and there is oxygen, and there is bread, then you spill one mould spore and it will spread across."