Netflix Asks Who Wants To Be A Millionaire To Developer Community

October 2nd, 2006 Davis Posted in Disclosure - I own stock in co. mentioned, Netflix |

Last year Netflix partnered with ABC’s Who Wants To Be A Millionaire to give movie buffs an opportunity to win a million dollars. This morning though, the company announced a different version of the game except for the open source development community instead. The difference with their current contest though, is that instead of answering 15 trivia questions sucessfully, Netflix has instead turned to the open source community as a lifeline for improving their suggestion algorithms.

In a bold appeal for outside development support, Netflix is offering a $1 million bounty, if you can improve the efficiency of their recommendation algorithms by more then 10%. Netflix will be judging efficiency by comparing what their users actually end up rating a film compared to what your algorithm predicts that they will rate a film.

I really like this promotion for a few reasons. First and foremost, I am pleased that Netflix is working on improving their suggestion algorithms and isn’t afraid to ask for outside support with the project. Netflix’s suggestions have been a sore spot for me lately and as I’ve explored the service, I’ve actually been pretty disappointed by the many of the suggestions that I’ve received. By offering a million dollars to anyone who can improve the technology, Netflix is demonstrating a clear desire to not only offer the best filtering technology out there, but they aren’t afraid to risk losing money, if someone can figure out a better soluiton then what their internal developers have built. This creates a win win situation for Netflix because if they lose the suggestion challenge, then they’ll be out the million dollars, but theothetically they should have an improved suggestion algorithm as a result. If they win the development challenge, then it will offer proof that Netflix has the best filtering technology around.

Netflix’s suggestion algorithms are at the heart of Netflix’s ability to harness the power of the longtail and is an important piece of their technology. With such a large collection of movies available to consumers, Netflix’s ability to distinguish between what one person will like vs. another enables them to move beyond just the top hits and into archived content that is less expensive. In an interview with the Hollywood Reporter, Reed Hastings talks about how important the suggestions are to Netflix’s core business.

“THR: Has Netflix influenced the movie business in other ways?
Hastings: Our fundamental contribution is personalized movie merchandising, which creates demand for small movies that are otherwise hard to market. (It also) creates high consumer satisfaction because the movies are interesting to that particular consumer. For small movies, Netflix can often be the difference between creating a profit and not because we do generate significant demand.”

As part of the promotion, Netflix has agreed to release about 100 million anonymous ratings to the teams working on the contest, perhaps more importantly though is that rather then forcing competitors to keep their research a secret, Netflix is requiring that they “publish a detailed description of the winning approach for the benefit of companies, entreprenurs and academicians.” This collaboration with the open source community is a good step forward in improving the transparency and the accuracy of Netflix’s suggestion results. If this does in fact lead to better recommendation technology, then the $1 million that Netflix is putting on the line will be small change compared to the improvement that it will provide to Netflix customer satisfaction ratings. By allowing participants in the contest to talk openly about their research, it will help to benefit not just Netflix, but any company interested in recommendation technology.