Man versus Machine: Could a Video Store Clerk Win the $1m Netflix Prize?
Andrew Bergmann writes:
We'd like to let you know about our experiment in using a completely new strategy to tackle the NetflixPrize problem through the use of crowd wisdom.
Forty thousand people have been trying to solve the $1M puzzle without success for almost two years, using lifeless statistical algorithms which often end up suggesting movies like "Speed Racer" to someone who loved "Taxi Driver". We decided to try something different.
We've designed a game called Video Store Clerk (http://www.videostoreclerk.com) in which you are the clerk and must recommend movies to your customers based on their past rentals. The rating info is based on real customers, which makes the game a lot of fun to try and get into the heads of actual movie enthusiasts. You choose the categories that you know best and the data actually gets more accurate as you play.
We believe that that there are certain tasks that humans can still do much better than computers. So far our findings have proven that point, but to complete the experiment we need a few million people to try out the game.
Are you game to try?