Back in 2006, Netflix offered a $1 million bounty for any team that could improve their Cinematch algorithm by 10%. If I remember correctly, a Netflix honcho called that 10% the “NAPOLEON DYNAMITE anomaly” because, well, here’s what I remember him saying.
“If someone rents and loves NIGHTMARE ON ELM STREET and SAW, we can successfully suggest WOLF CREEK to them. If someone rents and loves PRIDE & PREJUDICE and REMAINS OF THE DAY then we can successfully suggest SHADOWLANDS to them. If someone rents and loves THE SIMPSONS and FAMILY GUY, we can successfully suggest THE OBLONGS to them. The anomy is that each of the above renters also check out NAPOLEON DYNAMITE and give it 5 stars. We want to predict that.”
It looks like someone finally did it. It took two teams combining their coding abilities, and the results still need to be verified, but I think we have a winner.
Or do we? The most interesting part of that Cinematical article is the question of Prize economics. See, by Netflix offering a prize to the winner, they get a lot of people working on their very specific problem, one that, if solved, would hypothetically generate more revenue for Netflix, for free. While this is a boon to the winners, couldn’t Netflix get the same results by hiring some coders, paying them a salary and offering them health benefits? Need another example of this? How about the Amazon contest for a user generated 30-second commercial that they plan to air on the major networks. What does the winner receive? One, possibly two, $10k gift certificates. Basically, you trade your winnings back to Amazon. Um, hey, look, if I had more time I’d enter but isn’t that just a bit fishy?
Or maybe I’m just getting all worked up since I have so little work at the moment.
Loading ...