Netflix Cancels Plans For New Netflix Prize As Part Of Privacy Lawsuit Settlement
from the careful-what-you-crowdsource dept
Netflix, of course, received tons of attention and (apparently) a lot of valuable research, with its Netflix prize competition, that allowed anyone to take a bunch of Netflix data and try to improve on Netflix’s ranking algorithm. Of course, whenever you’re dealing with “anonymized data” there are questions about whether or not it can really be anonymous. In nearly every case, someone figures out how to “re-nonymize” at least some of the data. And, of course, that also happened with the original Netflix Prize data. This was especially troubling for Netflix because of the Video Privacy Protection Act (VPPA), 18 USC 2710, a special law that was passed after Supreme Court nominee Robert Bork ran into some trouble when his movie rental lists were made public, which made it specifically illegal to reveal movie rental data. And thus… a lawsuit was born, late last year.
Netflix has announced that it has now settled the lawsuit, but as a part of that settlement it is canceling the plans it had announced for additional Netflix prizes. While the company can still do contests in the future, it will need to make sure that the data cannot be reconnected to an actual person, which may be quite difficult in practice. This does raise some interesting questions for other attempts to crowdsource research. There are certainly benefits to opening up data to a community of smart people — but companies are going to need to be extra careful in those settings in dealing with privacy issues.
Filed Under: anonymize, crowdsourcing, data, movies, privacy
Companies: netflix
Comments on “Netflix Cancels Plans For New Netflix Prize As Part Of Privacy Lawsuit Settlement”
welp
people should have known netflix was going downhill the second they forcefully moved people to silverlight.
now they’re just circling the drain, really. I give it 2 years.
Oh well, mission creep affects every success that isn’t ready for it, and is what is happening to netflix right now.
Re: welp
Yes, clearly the company is in deep trouble because they’re cancelling a research competition[!] Oh and they use Silverlight? Well that’s another nail in the coffin[!]
But seriously, I think the investments they have made in getting their players on tons of set top devices and expanding their digital library are the smartest moves they’ve made for surviving the ongoing transition to digital distribution. They’ve even got a revenue model around it that it seems more than a few people are willing to stomach for the convenience granted.
Re: welp
I don’t understand why people don’t like the Silverlight player; it works fine on my laptop, my media center PC and my desktop..
Solution?
Why can’t they offer a free extra movie rental or some such for those who opt in to being included in such plans? That way, they gain the permission of whoever provides the data and they can still make in anonymous (or at least, as well as they can).
The only part of this story that I find surprising is that this idea was ever green-lighted in the first place, it being quite well established that there’s no such thing as totally anonymised data. Compliant with the letter of the VPPA or not, (and how did your video rental history end up needing its own special act anyway?) I can’t believe it never occurred to Netflix that people might prefer not to have their personal information turned over to someone not affiliated with the company withour their prior, informed consent.
I thought the contest was a great way to foster research and produce a better product. I can understand the privacy issue though. Wish there was a better way. I think the whole contest makes good sense for everyone involved.
What if Netflix randomized the data such as randomly change the zip codes and other data it gives to the public? Would that help to protect the data from being “re-nonymize”? I know this would not give that true real world sample but it should give the public the ability to test their own code against some sudo real data. Netflix can then take the submitted code and run it themselves against the “real” data to give the true real world score. This way the public never gets the real data but has enough to test their code and still see how their code did against the real data.
Re: Re:
If the public never gets to see the real data and run it themselves, how can they be sure the contest isn’t rigged? There’s no third party verification.
It’s not on in anyones book I’d have thought.