So that was, I think, a fundamentally different idea that wasn't... people had talked about building a semantic web, but it didn't have that idea. People had talked about building semantic networks, but it didn't have that idea. But that idea was, I think, the thing that made it work. So I started building it first. Actually, first for the intelligence agencies, as a way to get them to organise their information, but when I realised they really weren't going to get it together to do that, I just started building it myself for this education project, which didn't really have money for it. But it was an audacious enough idea to sort of capture all the knowledge of the world that we managed to get venture capital funding for it. And so I started a company, originally with people like Robert Cook and Kurt Bollacker. And we started building this, and we realised we needed better engineering and I didn't really want to explain to people what we were doing because it was so kind of in the stealth stage, but I wanted to find if there was other people that would be interested in this. So I made kind of an abstract drawing of, sort of conceptually, what would be a great goal. And I just showed it up in a whole bunch of slides. I was showing pictures of robots at some O'Reilly conference. And it was just kind of a troll, really, to see if anybody would pick up on it. And nobody at the conference picked up on it. So I thought, 'Man, that was a failure.' But then I got this call from this guy John Giannandrea, who was a fantastic person who had been one of the early people at Netscape, who said, 'Hey, I noticed you had this slide in your talk, I was wondering if you meant this by it', and he had, like, a whole concept and he saw what was important about it. And I said, 'Yes, why don't you come join us?'

So he turned out actually to be the key person that really put it together and made it work and so we built this thing and started putting knowledge into it. We made it all free, in fact we tried to come up with a name. We called it Metaweb at first, because it was sort of information that was about the stuff that was on the web initially. And that was the name of the company. But then we began to build this database, and we made the database free and open, so we joked around of what would you call it, and I suggested, initially just as a joke, 'Oh, we should call it Freebase.' Which at the time was the slang for crack cocaine. And everybody told us we couldn't do that. So we did. And I was like, 'Don't worry, someday, we're just going to change the meaning of the word and there'll be a generation that thinks Freebase means this.' And sure enough, today if you think for Freebase in Google, you get our Freebase comes up higher than crack cocaine.