Netflix is taking collaborative software development to its logical extreme by offering a $1 million prize to the first person who can improve the accuracy of its movie recommendation engine by 10%.

Key to the effort is the release of  a dataset of 100 million ratings – 700MB worth of data. You can download the ratings and see full contest rules here.

It’s a great move for Netflix.  They win with the publicity boost for trying to improve the customer experience, and they win if some team of PhD statisticians achieve a breakthrough in collaborative filtering.  Read more from the New York Times.

Misha Cornes