Topic: The promise of the semantic web.
Jimmy Wales: So for a long time I used to say about semantic web that it sounded neat, but I don’t understand it. And to go a little deeper than that, what I’ve found is every time I would start trying to learn something about the Symantec web, I would end up reading these sort of very long, dry, abstract, theoretical policy documents that I felt were unrealistic and didn’t tell me what I’m supposed to do. I always like to say, “I’m a carpenter; not an architect,” right? So sort of grand theory of architecture sounds great and all, but I’m a carpenter, I don’t know what to do with that. I actually need the architects to give me something, give me a plan so I can start hammering and nailing, and actually building the thing. And we’re actually getting closer to that now. And I think a lot of the ideas around the Symantec web are beginning to come to fruition. You know, we have people-- lots of people doing things like tagging photos at Flikr or adding articles to categories at Wikipedia. Those are the kinds of things that people are beginning to generate data, meta data, that’s machine readable, and useable by computers to help us find and navigate and do all that kind of cool stuff. So for me, I think that’s the first steps in that direction. And that’s hardly the dream of the Symantec web. But at the same time I’m a little skeptical of a lot of things in this area. I know there’re a lot of people who’re working on natural language processing, and Symantec search. And for right now, I haven’t seen anything particularly useful about that. And I’ve actually talked to people-- I’m speaking now just from the moment from the perspective of search engines. I’ve talked to some people at Google who share that sentiment. I don’t know if everybody at Google does. For all I know, they’re working on a top-secret natural language processing program. But to say, you know, keyword-based searching where the machine doesn’t even attempt to understand what the document is about, actually works pretty well. And it, you know, having the ability for a machine to read a document and understand it in some vague sense doesn’t seem to be particularly helpful for most of the real problems that we face in search. So at least for right now, given that the technology’s very limited in scope, I just don’t see it.
Recorded on: 04/30/2008