While reading what he wrote, and reviewing a paper he referenced, Improving Web Search Ranking by Incorporating User Behavior Information, a kind of scary possibility occurred to me. I just wonder if by focusing on trying to infer user preference we will be turning search into a billboard top 40 where queries yield results based on the vast majority of users rather than the precision of the query in relation to documents published on the web.
I should probably provide some background for my particular bias (as recommended by the Bloggers Guide to SEO). I think that the Billboard Top 40 is reprehensible. I find it amazing that people who profess to love music don’t know about anything that happened outside of this list. Moreover, they ignore the centuries of music that came before, across all continents, in favor of what is en vogue. (Try to google en vogue – screwed if you’re looking for anything not top 40 like definition, etymology, common usage, etc.)
When I was young I hated the Top 40, and it happened to be the counterculture thing to do. Decades later I am pretty far from trying to be cool and I still hate the Top 40 – perhaps with less vitriol, but I just think that it is a function of a production engine and I prefer less popular more creative music.
If you search in certain databases all you will find is what appeals to the lowest common denominator. Being a fan of things around the fringes, anything that moves the web in that direction scares me a little bit.