Remember I said there were two ideas of artificial intelligence. One of them was kind of this neural network idea and the other was the symbolic idea. Well, I've always been interested in the symbolic idea, too, but the sort of big, emergent version of the symbolic idea, which is: if you actually had a semantic network that knew information, like it knew your name is George Dyson and it knew all of the connections to all the other events. You know, your connection to Bellingham, your connection to Baidarkas, your... that knowing all these entities and the connections between them is an important part of what intelligence is about, is knowledge. And so that was another thing I wanted to do with the Connection Machine, I was kind of inspired by this thing that Scott Fahlman had done, of a way of searching through networks like this, semantic networks, and reasoning from them. So I was very, very interested in representing knowledge this way in the Connection Machine.

But it never really played out, because the Connection Machine... there was no commercial market for it. But I always had that dream of building a big knowledge graph of all this information. And then, when Neal Stephenson wrote his book, Diamond Age, it had a story in it about the illustrated primer, which was a book that kind of knew everything and adapted to the student that read it. And for years, I had been interested in education, elementary education. And I just loved this idea. I was very inspired by it. And so I started thinking about how you would go about building it, and I realised well, one thing, it would need to know everything. It would need this giant network. And so it would need to know what there was to be known. And that connected with something else that I've always had as a dream, which is I've always imagined having a map of knowledge so that I could see a map of my ignorance. I mean, imagine... I remember looking at the globe and thinking, 'Well, I've been there, I know what that is, but I haven't been there, I don't know what that is.' And so in some sense the globe was a map of my geographical ignorance, but imagine you had a map of all the things that could be known and you got to see: oh, there's this whole region that I don't know about. You know, this is like the algebraic topology. You know, the mountains of algebraic topology are unexplored. And so that idea has always fascinated me, but then I was thinking about how to do a Diamond-Age-like book. I realised that if I made this network, I could finally print this map of my ignorance. And so I got very excited about that idea.