And by then, computers were much faster. And I'd actually tried to do something like this at Thinking Machines. And I'd tried to do a project with Encyclopaedia Britannica, where I took all the information in Encyclopaedia Britannica and put it into a computer so that you could find if there were contradictions in Encyclopaedia Britannica. And that project got started and the guy that I had hired to work on it was named Doug Lenat. And who ended up having very strong ideas about his own ways of doing things which I disagreed with, and so he didn't work for Thinking Machines for that long. He went on and took that project with him and started something called Cyc, which really had started with that Encyclopaedia Britannica thing. But the wrong idea he had was that he wanted to make everything so that it was logically consistent, so that you could reason about things, you could make deductions and so on. And knowledge is too complicated for that. So I always knew that was the wrong idea, so when I went back to it, I did it completely differently. What I recorded in the relationship was that somebody asserted this relationship, so you didn't say that Taiwan was a province of China, to pick a controversial example. You said, 'This person said that Taiwan is a province of China.' But somebody else may have said it wasn't a province of China. So you said who said it and when they said it. And so you could have contradictions in it. And then you could... you didn't have to have consistent information, so it was very easy to add information. So unlike Cyc, which got more and more difficult to add information to without breaking it, this was totally scalable and many people could add information to it, and then you could decide afterwards how to deal with the contradictions and who to listen to and what.