Coders Crack the Netflix Cinematch Algorithm
Filed under: Tech Stuff, Home Entertainment
Netflix began a contest in 2006 for coders to win $1,000,000 if they could improve Netflix's Cinematch algorithm by at least 10%, and because coders like a good challenge (and people like money), folks have been steadily tinkering with their 1's and 0's ever since. (You can read more about the contest and its rules here.) Three years later, two teams came together with the winning code, which was confirmed Friday. Netflix's VP of corporate communications Steve Swasey told Wired, "This has been terribly exciting. It's like watching the Belmont Stakes, the Preakness and the Indy 500 all at once -- for geeks."
There are still a few hurdles for the team to jump through; the Netflix contest site states, "your qualifying submissions must have the largest accuracy improvement verified by the Contest judges, you must share your method with (and non-exclusively license it to) Netflix, and you must describe to the world how you did it and why it works."
As Wired points out, Netflix isn't the only company to dip its toes into "so-called Prize economics," which is when these companies offer prize incentives to outsiders to basically do work for free with an outside chance of winning big bucks. But doesn't this say to companies that it's okay to try and get lots and lots of hours from (some) highly skilled professionals for free? It would have cost Netflix far more than $1M to get consultants and coders on the job, especially for full-time gigs with all the benefits and trimmings. I'm not a code geek, but if I were, I'd probably be interested in this sort of challenge as well. On the other hand, perhaps I'd rather have a full-time job at Netflix writing code all day along with, you know, insurance.
As a writer, it's hard not to get riled up by these kinds of deals. There are all sorts of reasons to work for free on occasion -- clips, resumes, bragging rights, free stuff, barter -- but I find it problematic when a large, successful company wants something for nothing, especially when that company started out as the little guy and is now the leader in its field.
Writers, film fans, techies, what do you think?
Reader Comments (Page 1 of 1)
6-29-2009 @ 6:15PM
Stefan said...
Even knowing they could probably get paid to do this elsewhere, these programmers still spent the time to try and do it. Whether it was for the money, the prestige or just the challenge, they had to know the ultimate outcome.
I think there are many people out there who still do stuff for the challenge of it.
Of course, $1 million is not too shabby of a pay day for completing the challenge. And wouldn't signing the code over to Netflix for money technically be a sale? Definitely a chunk on the resume.
Reply
6-29-2009 @ 6:35PM
Billy Soistmann said...
If the deal is a non-exclusive deal license with Netflix, I wonder if they're willing to work with any other movie rating sites.
Reply
6-29-2009 @ 7:09PM
Kurt Munro said...
There was $1m "Make Something Unreal" game contest but the money's split between the game engine etc etc...
Reply
6-29-2009 @ 7:27PM
Jon said...
I don't care how good it gets ... i will always pick my own movies
Reply
6-29-2009 @ 10:39PM
Chet said...
Was there anything stopping a "team" from being a company, hiring and fully benefitting a team of employee programmers?
Is the winner beholden to Netflix before they begin, or can they decline the prize and launch their own startup (or sell out to Blockbuster)?
Didn't Netflix risk creating its own competition by putting up this prize? Perhaps second place needs only some refinement to leap ahead.
Isn't there some value to society in fostering such widespread competing development with potential applications beyond Netflix's immediate corporate needs?
And doesn't this shorten the odds substantially for all those who may have sparks of brilliance but not a corporate-friendly resume?
Reply
6-30-2009 @ 1:20PM
Gordon said...
"Didn't Netflix risk creating its own competition by putting up this prize? Perhaps second place needs only some refinement to leap ahead."
I don't think they care. Second place would need millions of dollars worth of infrastructure to leap ahead of Netflix in the DVD rentals department. I was with GreenCine for a little while, but the fact that they took a week to get a disc to me as opposed to two days made Netflix far, far more valuable.
Setting up a new online streaming site would be extraordinarily expensive, too. An equal or better "Cinematch algorithm" doesn't mean a thing if you don't have movies people want to watch.
6-30-2009 @ 1:35PM
Jenni Miller said...
I had a very cool convo with my friend Brian Grosz, a coder and voiceover artist here in NYC via Facebook on this post. I'm reprinting his comments (without mine, since they basically reiterate my take on it as posted above) with his permission. (His website is here: http://www.briangrosz.com/).
"Crowd-sourcing for open-sourced development apps? I'm a fan, without question. It sidesteps groupthink, scope-creep and all sorts of other corporate nightmares."
"It was once said of jazz musicians that they were the only people who would play a paying gig and then go play for free at an after-hours jam session.
Honestly, I think the same applies to coders. I have yet to meet one who didn't honestly enjoy the tinkering, the problem-solving and, hopefully, the successful compiling of a script."
"Even DARPA does it (http://en.wikipedia.org/wiki/DARPA_Grand_Challenge).
By opening up your source-pool to a competitive environment of independent, unpaid contractors, you (ideally) wind up with a product that does the job elegantly... rather than what could be expected from a consultant team of paid contractors: a product that does the job 'good enough' for the asking price."
Thoughts?
Reply