“Recommendation systems covering a wide variety of categories will play an increasingly significant commercial role in the future,” said Reed Hastings, Netflix’s Chairman and CEO. “Right now, we’re driving the Model T version of what is possible. We want to build a Ferrari and establishing the Netflix Prize is a first step.”
Netflix’s movie recommendation system helps subscribers choose rental selections my recommending titles based on the user’s ratings of movies they’re familiar with; it then compares those ratings with ratings entered by other Netflix users and suggests additional titles which might appeal to the user. Recommendation systems are generally useful in taste- and preference-based selections, like entertainment, music, movies, clothing, and accessories.
Netflix says it will publish a detailed description of what will constitute a winning entry for the Netflix Prize; briefly, the goal is to improve the accuracy of the recommendation system by at least 10 percent—although it’s not clear whether “accuracy” means that users agree with the recommendations, or actually convert those recommendations into successful rentals. Netflix plans to award a progress prize annually until someone wins the grand prize; to get things started, Netflix is publishing 100 million anonymized movie ratings ranging from one to five stars, which it claims is the largest data set of consumer film ratings ever released.
The idea for the Netflix Prize comes from the Longitude Prize, offered in 1714 by the British government to anyone who could come up with a practical way of computing a ship’s longitude while at sea. (The prize was won in 1761 by John Harrison.) The Longitude Prize also inspired the X Prize for the development of private space exploration.
(Note the lack of snarky comments about Netflix implicitly comparing the historical importance of movie suggestions to the practical computation of longitude. And people say journalists are incapable of impartiality!)