Netflix Ratings and the Anchoring Effect

There's an interesting article at Wired about about a competitor for the Netflix algorithm prize who is using behavioral economics techniques instead of mathematics-based techniques.  He doesn't reveal many specifics, except that he's using the concept of "anchoring," which is the idea that exposure to numbers will influence how a person makes choices (like whether to buy something, or, in this case, the how to rate a DVD).

The example given in the article is simple: if a person watches three movies in a row and gives them each 4 stars, then watches an even better movie, they would probably give that movie 5 stars, because it's one better, right?  But if they watched 1-star movies in a row, then followed them by the same good movie, they would probably give it a lower score, like a 3 or a 4.

(Also worth reading: a recent article at the New Yorker on the subject of anchoring and why we make systematic irrational decisions.)