A while back, Netflix announced a $1 million award for improving the DVD-by-mail service’s recommendation service, which suggests films to its users based on their ratings of films they’ve seen. The service provides enough of a challenge to Netflix users, who have to make hard choices about their ratings. Do I give The Curious Case of Benjamin Button one star? What about Duplicity? Both terrible movies, but generally the kinds of movies I like created by filmmakers and actors I like. I wouldn’t want to miss out on Children of Men because Netflix thinks I don’t like Clive Owen, or The Game, because it thinks I don’t like David Fincher.
I likely just revealed how little I know about Netflix’s rating system, but may also have illustrated the kind of foolishness lots of smart people are dealing with as they seek to make the recommendation system work. And if these smart people could improve the system by 10%, a $1 million prize would be theirs. It seems they have.
What’s the point to the LCRM community? The successful efforts of an international team speak powerfully to the possibilities of crowdsourcing, sharing expertise and data to answer formerly unanswerable questions. The Times article linked above suggests applications in the sciences, but crowdsourcing has a role in the humanities, too. We do it all the time, such as when we send questions to a listserv. One result and possibility for the future is the Espy File, the massive collection of data on the history of the death penalty that was built mainly by one man, M. Watt Espy (who recently passed away), but has since been taken on by other historians. The Espy File reveals both the power and the potential for crowdsourcing history data–what if everyone working in the area contributed what they learned about, say, the races of victims in these crimes. A remarkable history, one with real relevance to today’s civil rights-inflected discussion of the death penalty, could emerge that would go much deeper than names and dates.
But, as the article suggests, crowdsourcing has its pitfalls, too. After all, most of the teams competing for the prize did not win.
0 Response to “The Netflix Prize and Collaboration”