Netflix: Movie recommendations now "more accurate"

The movies-by-mail giant says it's tweaked the formula it uses to guess which movies you'll like, with a little help from its "Netflix Prize" contestants (all of whom are still competing for the $1 million grand prize).

In an official blog post, Netflix execs wrote that they've "deployed a new algorithm" that will help the site "make more accurate guesses" about which movies and TV shows you'll love, hate, or shrug your shoulders at.

The changes "will be subtle," and indeed, I'm not seeing any violent mood swings over on my "Movies You'll Love" page on Netflix, which is giving me a thumbs-up for such movies as 1986's "The Fly" (one of my all-time favorites), "Taxi to the Dark Side," "Touching the Void" (which I saw a few weeks ago, and loved), "United 93," "When We Were Kings" (which has been on my recommended list for eons now), "Mad Men" (heartily agree), and "Miller's Crossing" (not my favorite, although my high ratings for "Fargo," "Barton Fink," and "Pulp Fiction" are tipping the scales in its favor).

The new algorithm is (we're told) based in part on the work of contestants for the Netflix Prize, the two-and-a-half year-old contest that challenges competing teams to improve the guesses of Netflix's recommendation engine by 10 percent. The grand prize? A million bucks.

So far, no one's collected the big prize yet, although at last count, five teams have come tantalizingly close, with the top team—"PragmaticTheory"—up to 9.65 percent.

Related:
More Accurate Star Predictions [Netflix blog]