On September 14th, Amazon.com finally went officially live with A9.com, its own search site.
Interestingly, A9.com has figured out a search verity (pardon the pun) that others haven't. Namely, that understanding/categorizing the user is just as important as categorizing the content. Relevant content is in the eye of the beholder -- therefore, if, based on the user's past search and browsing history, the system can figure out that "jaguar" is inclined to mean the animal rather than the car or sports team, this is a huge win for relevance. Typically, search technologies have concentrated on more brute force methods on the content indexing side, and have ignored understanding the user. This lopsidedness, I would contend, is forcing search companies to commit excessive resources on the search engine side, while just a little tweaking on the user side would offer huge gains.
Bill Gates gets it -- he made a comment several weeks ago that Microsoft was working to understand user behavior as a way to tune their forthcoming (who knows when, sometimes in the future) desktop search product.
Who knows, maybe there's something in the Seattle water. In any case, let's hope that if A9.com is successful that other search companies rethink their search architecture.