I'm going to go ahead and cross-post this from the other article about freebase.
Disclaimer: I admire Daniel Hillis greatly, perhaps irrationally.
That said, Metaweb sounds amazing. If it takes off, it will be not so much a useful product in itself, but an enabling technology for an entirely new class of products. Imagine the possibilities that arise from having a relatively complete, constantly evolving ontology of everything. All of those nifty tricks that computational linguists and proof theoreticians can do over finite domains, using very limited ontologies, become possible for all sorts of other applications. Armed with an ontology that closely matches the real world, it is finally possible to write code that can reason about the real world, rather than just throw Bayes law at it and hope something useful falls out.
While DH is a smart guy, one of our rules of thumb at YC is to be skeptical of any application containing the phrase "semantic web." It's a bad idea magnet the way AI was in the 1980s.
You're right, of course (and I need not point out the irony that DH was in deep with AI), but I'm holding out hope (and not just because I'm bitter that the Connection Machine failed!). The main reason (in my opinion) that the semantic web has failed so far is the reliance on 1) predefined, very limited ontologies, and 2) creator-generated rich metadata. Freebase seems to me like, if not the solution, at least a good step in the right direction in terms of addressing these. The real question, of course, will be whether a userbase springs up to keep it up to date, and whether the logic behind it can keep it reasonably self-consistent.
I think it sounds like a pretty interesting idea. I'm excited to get my hands on the API and see what kind of information you can really pull out of it.
Disclaimer: I admire Daniel Hillis greatly, perhaps irrationally.
That said, Metaweb sounds amazing. If it takes off, it will be not so much a useful product in itself, but an enabling technology for an entirely new class of products. Imagine the possibilities that arise from having a relatively complete, constantly evolving ontology of everything. All of those nifty tricks that computational linguists and proof theoreticians can do over finite domains, using very limited ontologies, become possible for all sorts of other applications. Armed with an ontology that closely matches the real world, it is finally possible to write code that can reason about the real world, rather than just throw Bayes law at it and hope something useful falls out.