Why I Like Netflix's $1 Million Challenge

I read with interest this article highlighting how Netflix (NFLX) awarded the $1 Million Netflix Prize and announced a second $1 Million Challenge. The challenge was to achieve a 10 percent improvement over the accuracy of the Netflix movie recommendation system, which internal engineers have been unable to do.

The prospect of $1Million to teams of engineers and statisticians was enough to prompt several teams of civilians to compete for the prize and ultimately, now 7 people from the winning team will split the winnings.

We've seen this tactic before, most notably, from the X-Prize.
What I find to be so innovative - and profitable, about this tactic of relying on the public for innovation is that it achieves incredible results for a mere fraction of the cost. Think about it.

When's the last time your company was able to do anything with a meaningful impact to the entire corporation and customers for under $1 Million?

If I'm thinking in terms of the Netflix prize, we'd probably be talking about an entire group of engineers, teams, infrastructure, support, etc. - for years - at a cost of several million dollars, not to mention the opportunity cost of the other projects they're not working on while pursuing the objective.

In the context of the X-prize, for space missions and novel technology, development, testing, etc., we're not even talking millions any more. These are multi-billion dollar projects. The X-prize was able to achieve some pretty incredible results (and additional leads/future ideas) for $10 million. Not only has there already been an initial winner, but new contests have since been announced. We may see the next mode of space transport borne from pure innovation, risk-taking and capitalism rather than a bloated government agency.

Wikipedia is another example of a "free" build if you will. Sure, contributors get some individual recognition and occasionally plug their own links elsewhere on wikipedia, but it's turned into a pretty trusted (some factual errors do end up on there occasionally) free source of virtually anything you can think of, the likes of which has never existed before.

If your company is facing a nagging complex problem from say, how to get the next optimal catalyst for your chemical process to how to improve the accuracy of a computation and you know it would cost your company millions to pull it off, do you think your corporation would ever consider the context/prize approach?

Comments

[Jeff B.:]

Although I admire Netflix's approach of having a contest to determine the best way to improve it's recommendations, I think they went about it totally wrong.
As a mostly happy Netflix subscriber for about 7 or 8 yrs now, I think I can say a bit about how accurate their rec's are. In short, as much as I like Netflix overall, their recommendations stink. I rarely agree with their picks for me.
They went about solving this problem as if it was a "programming" problem. That misses the point. This is first and foremost a business problem -- with programming as one part of the solution. I listened to the guys who won, talk about their approach. They have a few good ideas, but again, I had the feeling they asked, "Okay, how can we program this?" before they asked "What is the real problem here, and how do we solve that conceptually?"
For example, has anyone really thought about asking say, 1000 Netflix customers to complete a detailed survey (give us a free month or two & we'll do it) which seeks to understand why people like/don't like certain movies? Once Netflix understands that, then they can worry about how to program it. The way they approached it, I am not confident it will solve the problem. I guess I will find out.

Sep 23 10:41 AM