How do researchers determine what ties folks to their community? Turns out it's sort of like how Netflix determines what draws you to different movies.
The answer is ratings. Netflix users have given the company hundreds of millions of ratings for lots and lots of movies. Netflix knows it's improved the recommendation engine when it can better predict how you'll rate a particular movie. So to test the teams who entered the contest, Netflix gave them access to a partial set of movie ratings by Netflix users. The company held onto another set of ratings by the same users, and the teams competed to see who could best predict what those other ratings would be. To figure out what rating a user might give to a certain movie, the teams had to look for obscure patterns in other ratings the user had given, and try to figure out which of those patterns were most important to the user's overall taste.