I caught a pretty interesting article in the New York Times (click on the picture above to go to it). I’ve been a satisfied customer of Netflix for a little over a year now and I have always wondered how they come up with the recommendations. For those of you not in the know, Netflix takes your ratings (1 to 5 stars) of movies you have already seen and tries to figure out which movies you will like. From a computer programming perspective this is an interesting problem. While it is easy to identify the categories of movies a person prefers (action vs. romance) there are some intangible aspects to many movies that are much harder to quantify. One movie mentioned in particular in the article is Napoleon Dynamite. Essentially it is a comedy, but it cannot be easily lumped in with other members of the genre because of the kind of comedy — hipster humor can be quite polarizing. As a result, Netflix’s algorithms cannot figure out which way a particular person will fall on this movie — even with a massive set of data to analyze. So Netflix has issued a challenge to all amateur computer scientists, statisticians, and brainiacs. If anyone can come up with an algorithm to improve on the predictions by at least 10% then they will be compensated to the tune of 1 million bones (that is six zeroes!). I have not read the entire article (it will come along in the NYT Magazine on Sunday) but it has been an intriguing read thus far.