Friday, December 8, 2006
Interesting post by my colleague Adam Kolber here. Reacting to both his post and some of the comments after his post, I want to make the following points:
Several commentors make the point that there should be something better than natural language search, maybe Google should enter this market, and so on. Well, there is something better than natural language search now for law. If you look at this you can see that the Precydent algorithm, which is network based, like Google PageRank, outperforms both Westlaw and Lexis natural language search by a lot. Could Google enter this market? Absolutely. But whether they will is another matter. They have hundreds of geniuses on staff, so they could figure it out if they wanted to, but it is not easy, very far from it. To do network searching on law, which you would have to do (at least until the great natural language search revelation comes, as people have been threatening but not delivering for years) you have to create a citation network, then create an algorithm that will function properly in the weird and wonderful world of the legal citation network, which is like, but not like, the web. PageRank won't do it. We tested PageRank on our USSupCt database, and the results were, to use the technical term, for shit. But we did it, so it can be done. We're patenting our approach, so Google (and Westlaw and Lexis) will have to figure out their own way. Unless of course they want to license ours (that's tacsmith at gmail dot com).
On the Wexis duopoly: all I can say is, it shakes my faith in the market, and that's pretty bad. One model Precydent is considering seriously is just to offer all state and federal cases, with state of the art search, for the very reasonable price of nothing. Just electricity and seeing some ads. It strikes me as quite doable. Perhaps not quite as profitable as other approaches, but doable. If anyone thinks that would be nice, please send me an email. I like to get email that isn't spam and love to be encouraged. A return to putting all this law where people can read it (and actually find it) which should not be underestimated, is that it would be fun. Not to mention educational, which, I should like to point out, is in fact related to what law professors are supposed to be doing. Why shouldn't any schmoe who can't believe what he just saw on Law & Order be able to type in "search warrant technicality" and get a bunch of highly relevant cases. All information may or may not want to be free, but when I commune with the Law each evening, it frequently mentions that It would like to be free, and so it should be.