Constructing a mnemonic circuit using stone knives and bearskins

Spock is the name of a new web2.0 company that uses technology to solve a problem that's only been partially solved--finding people. What they attempt to do is to disambiguate the multiple person-one name problem. They do this by using what I consider a pretty sound approach--use software to initially populate the database, then enlist the great unwashed, the webtwofer types, to refine the data.
So Google does do this, but it's oriented along the lines of pure content retrieval, it's not looking for people, but stuff. If the stuff is organized along people lines, great, otherwise you get non-standard results. What we want I think, is to search on a person's name and get the most famous or well-known person, not the first one on Google's search list. So Michael Jackson the musician, not Michael Jackson the beer expert.
(Technical note: We actually want neither of those approaches--we really want a context-driven search based on who we, the searcher are. If I'm a beergeek, then I do want the 2nd Michael Jackson. Neither Spock nor Google can do that.)
I haven't actually seen Spock work myself, but I would like to play with the beta. By the way, they're having a moneyed contest to help them with some technical stuff. More info here.
In theory, this should work, although I wonder about the practice. It's like the Wikipedia situation, do we really think that hordes of anonymous people can actually cook an omelet or just break a lot of eggs? I also wonder how difficult it would be for Google to add this capability to their engine, thus killing Spock yet again. Still It will be interesting to see how they do.
Posted on April 19, 2007





