Why can't publishers predict bestsellers?

There's an article coming in the NYT's Sunday Business section on the vagaries of publishing, which reprises the usual "publishing as a crap shoot" theme. All the editors talk about how it's impossible to accurately predict bestsellers and the ones who got lucky pat themselves on the back for their canny intuition -- the editor who acquired the mega-million copy bestseller The Secret describes feeling a tingling in her spine. Yea.

Anyway, Shira Boss, the author of the article, rightly points out that unlike most other businesses, publishing houses have simply never bothered to try to figure out what their customers want. Which means that we'll all be reading essentially the same article next year and the year after and so on...until some smart publisher, who is sick of single digit returns on capital, steps up and says, "Wait a minute, we can map the human genome, why can't we figure out whether people will buy a book" and transforms the business of publishing with some market research data and analytics. Now, that'll be worth a story.