YOU ARE HERE: LAT HomeCollections


A million thumbs-up

October 07, 2006

AT NETFLIX, EVERYONE'S A CRITIC. So who needs Ebert and Roeper? The online movie rental firm has turned over the process of critiquing movies to its subscribers, who rate them on a scale of one to five stars. So far, it has collected 1.5 billion ratings -- more than 20,000 ratings per title, on average -- into a huge

database of likes and dislikes. Netflix hopes to upgrade its recommendations with better software, and on Monday it offered a $1-million prize to the first person who can make them at least 10% more accurate.

Through some nifty software algorithms, the service uses its database to predict how much any given subscriber might like any given film. Unlike Roger Ebert's and Richard Roeper's thumb signals, these predictions are made on a one- to five-star scale, and they're not static. Instead, the number of stars that Netflix gives a movie will vary from subscriber to subscriber, based on how that person has rated other movies.

The company has left little room for film critics in its populist system. The underlying assumption of its system, and the key to winning its contest, is that people whose tastes are similar to yours will do a better job suggesting films for you to rent than someone who judges movies for a living.

Nor is the movie service, which had 5.2 million subscribers as of June 30, the only company fueling this trend. Many of the online outlets for entertainment use some variation of collaborative filtering -- a fancy term meaning software that compares individual users with others in the group to find common likes and dislikes -- as a way to help people pick songs, books, movies and games out of giant virtual catalogs.

The effect, ironically, is to use input from the masses to create experiences that are simultaneously more personal and more universal. The filters not only help weed out things you'd probably dislike but also expose you to a range of things you would enjoy that you weren't aware of. Such systems are much less expensive to launch and maintain than ones that rely on a staff of experts to analyze and classify every item in a music or movie catalog.

For some users, the results produced by systems such as the ones used by Netflix and just aren't as reliable as the recommendations from their friends or trusted experts. So Ebert and Roeper, who have the added advantage of seeing movies before the average person can, don't have to worry about being replaced by a computer anytime soon. Nevertheless, thanks to technology, their club isn't so exclusive anymore.

Los Angeles Times Articles