Sunday, June 28, 2009

Nerds Beat Netflix's Recommendation System, Qualify for $1 Million Prize [Contests]

Nerds Beat Netflix's Recommendation System, Qualify for $1 Million Prize [Contests]

In a very shrewd move, Netflix created a contest more than 3 years ago to award $1 million to anybody that could create an algorithm more than 10% more accurate than theirs. Now, a team has finally done it.

The two front-running teams, distressingly named Team Pragmatic Theory and Team Bellkor in Chaos, joined forces and have submitted an algorithm that is 10.05% more accurate than Netflix's. Now, the other contestants (it's a field of thousands) have 30 days to beat that score, and if nobody can (and no rules have been broken), Netflix will award the superteam a cool million.

This is a smart move for Netflix—improving their algorithm by that much would have been incredibly expensive anyway, and Netflix has said that the $1 million prize is actually a great bargain. Plus, it'll net them tons of press. Like the press I'm giving them, right now. Maybe in return, you guys can stop recommending 28 Days Later because I liked The Royal Tenenbaums. [Wired]




No comments: