Wednesday, January 31, 2007


NetFlix is offering one million dollars to the first person that can improve their system (which matches people to movies) by 10%. The company provides the contestants with a bunch of random data and a program that measures the attempt's accuracy. The contest is less than a year old but the leader is already at 6.75% improvement.

Here's the New York Times article which mentions two advantages to prizes: paying for results, not proposal (as grants do) and anyone--not just the usual experts--can try. I can think of two other advantages: allowing people to try whatever out-of-the-box idea they wish and creating a competitive atmosphere that demands constant progress.

Hat Tip to Donald Boudreaux.

No comments: