Cerf: It’s an interesting and somewhat remote connection. Tim started his work around 1989. By this time, of course, the Internet was not only in operation, but it was also beginning commercial operation. There’s a distance of some 20 years or so between the design of the Internet and the design of the World Wide Web.
Tim recognized that the Internet, which by that time was increasing in its scale and widely adopted in the research community and the academic community, was a perfect platform for his overlay, which he eventually called the World Wide Web. So, the hypertext transport protocols (http), which are the core of the World Wide Web, sit on top of TCP. It’s a layer above—it’s an application layer above the transport layer of TCP, which sits itself on top of the IP protocol.
This is a layered architecture, and Tim recognized that—he was trying to design something that would allow physicists to share their papers with each other. He was looking for a way of literally electronically or digitally conveying papers from one physicist to another.
You’ll notice that when you look with a browser, if you look at the display, it looks like a piece of paper. It’s formatted, you have images, formatted text. HTML, hypertext markup language, is a way of describing the format of the paper: which fonts are being used, where images go, and of course, eventually streaming audio and streaming video.
His intent was to make it easy—Tim recognized that this was a method for conveying visually the published work of his physicist colleagues. He built the first piece of software using Steve Jobs’s NeXT machine, a beautiful black titanium cube.
Cooper: I didn’t know that.
Cerf: It was irony in the way that the NeXT Computer company didn’t survive, but one of its machines was used to demonstrate the first node of the World Wide Web.
Cooper: Did they throw that away as well?
Cerf: (laughs) I don’t know the answer to that. If it’s anywhere, it would be at CERN. It’s very funny, in a way because Tim got no support from CERN to do this work. It was just sotto voce. This is not unusual. Skunkworks are often the way that dramatic new things happen because they are so weird or so outrageous that nobody wants to pay any attention. So, he did this work in spite of the fact that CERN gave him little or no support.
Of course, today CERN is so proud of the World Wide Web that when you drive up to headquarters, you see a big sign saying, “Home of the World Wide Web.” Tim did that work starting in ’89. In 1991 in December he essentially announced the World Wide Web, and nobody noticed, except for a couple of guys at the National Center for Supercomputer Applications at Champaign-Urbana, the University of Illinois, in the middle of a cornfield.
Two guys, Marc Andreesen and Eric Bina, saw the World Wide Web design and decided that they would design a graphical user interface browser because the original browser that Tim wrote was pure text. It wasn’t imagery. So they designed a graphical user interface to the World Wide Web, and they called it “Mosaic”.
Cooper: Oh, right.
Cerf: Mosaic, when they announced it, it was beautiful. Of course, it was downloaded a million times in I don’t know how many lots. Everybody recognized how powerful this visual experience was. Remember, at that point we did not have the kind of search engines that we have now. We still didn’t quite have the search capability that you would expect. This is 1993, right around there. Of course, I saw that Mosaic browser along with lots and lots of other people and was immediately taken by it. So did Jim Clark.
Jim had been the founder of Silicon Graphics. Here’s another place where ARPA has a role to play. Jim was at Stanford University and designed a chip called the Geometry Engine. It was designed to do very rapid rotations in three-space for presentation of three-dimensional objects. That chip design was done using the ARPA project called MOSIS, Metal On Silicon Implementation System.
He designed the chip, sent the design to the USC Information Sciences Institute, which packaged it up, and sent the design to a fabrication house that produced the Geometry Engine chip. Jim Clark—essentially, it worked—so he started a company called Silicon Graphics to build specialty machines for computer-generated imagery.
Silicon Graphics was the place to go for a high-resolution complex CGI for the entertainment industry. People who were doing animation and things like that. This was a very successful company. Jim saw the World Wide Web, saw the Mosaic browser and said, “There’s a huge business here. This is a giant opportunity.” And he went off and pulled Marc Andreesen and Eric Bina from the West Coast to start Netscape Communication in 1994.
By this time, I had come back to MCI and was building what we called the MCI Mall, which was intended to be an online shopping system. 1994, this is a little too early, but we did it anyhow. I went off and bought $7 million worth of licenses from Netscape Communications in order to build my shopping mall. And of course, we were their first big customer. As I look back on it, I wish we’d bought $7 million worth of stock—
Cerf: —because they went public in 1995 and the stock went through the roof. And it started the dot boom, which persisted until April of 2000, when it was a huge dot bust. And a lot of the companies that had anything to do with the Internet suddenly went out of business because they ran out of capital and had not ignited an engine to generate income.
Cooper: Revenue, yeah!
Cerf: So, the investors all got scared and dropped their—
Cooper: I remember.
Cerf: But a lot of the companies continued. And what’s interesting, of course, is that in 1998, while this dot boom was underway, Sergey Brin and Larry Page, the two founders of Google, were at Stanford as graduate students and decided one day that they were going to download the entire World Wide Web and index it. Sounds totally crazy, but they did it anyway.
And the search engine originally was called “BackRub”. It has a lot to do with the method by which they figured out to order the responses when you did a search. It’s one thing to discover every web page has a certain set of words in it. You have millions of responses. The question is, what’s the order in which you present them to the user in a list?
Their idea, which is very clever, is called page rank. Essentially, it said that if there is a page that lots of other pages point to, it must be important. So they rank-ordered them by the number of things that pointed to that page. Of course, that’s something you can game very easily, so there were all kinds of defenses that had to be built in, eventually. They started in 1998 officially. But they called their company Google.
Cooper: How do you spell that? (laughs)
Cerf: Right! In fact, they intended it to be Googol, which is a mathematical construct for 10 to the 100th. The attorneys didn’t know that, and they actually registered the company Google because that’s what they thought they heard them say. “Googol” is a mathematical term, but it’s “google” if anyone is not a mathematician. So they registered the company under the name Google, and of course by the time Larry and Sergey found out, it was too late.
Cooper: We created the first job board on the web for people with disabilities in 1995, and partnered with Career Mosaic, which was the first job board. I worked with them for a while until they were bought by Headhunter and then Career Builder.
Cerf: I remember that Chet.
Cooper: Where does Allan Weis fall into all of this?
Cerf: At the point where the National Science Foundation (NSF) starts to get interested in the Internet, he was interested in the ARPANET, which was turned on in 1969 and continued to operate until 1990. NSF seized the academic community that ARPA is funding. There are about a dozen universities doing research on artificial intelligence and computer science. They built the network so those universities could share their resources and their software. NSF originally was hoping that it could fund connection of other university to the ARPANET. But it got very complicated because the Defense Department and NSF are in two different branches of the executive branch of the government. So in the struggle to figure out how to pay for access to the ARPANET, they concluded that it would be faster if they just built their own network. On top of which, the ARPANET backbone was only running at 50 kilobits a second.
NSF offered a project for bid, did an RFP (request for proposal) for this NSFNET. IBM and MCI and Merit at Michigan won the contract. Allan Weis was the guy involved on the IBM side. Eventually that became Advanced Networking Services, ANS, which is a consortium that was formalized by MCI and Merit and IBM. So that’s where Al gets into this.
Cooper: Just a backstory: Allan created ThinkQuest.
Cerf: Yes, of course. After he sold ANS—
Cooper: —to AOL—
Cerf: Yeah, that’s right.
Cooper: So we are here at the National Press Club Washington DC and literally in this room is where abilityJOBS held it’s first press conference.
Cooper: And during the time set for my talk, Steve Case scheduled his talk about AOL the same day, same time, and we lost most of the press—
Cerf: Oh, my!
Cooper: (laughs) Timing is everything. And one other great memory. ABILITY Magazine partnered with ThinkQuest to include students with disabilities in their international student learning program giving more weight in their rubric for cash awards. Allan Weis invited me to speak at CERN during their event and that’s where I first met Tim Berners-Lee.
Cerf: Amazing! God, everything is connected, everything!