||[Nov. 12th, 2006|09:24 pm]
As I posted
a few weeks ago, Netflix released anonymous movie rating data which turned out to be not-so-anonymous.
What's new is we can now identify users based on their IMDb comments/ratings in the netflix database! We haven't done it on a large scale yet, so I don't have any stats like what percentage of users can be deanonymized or anything.
We've faced the criticism that the privacy of movie ratings doesn't matter, or that people who cared about their privacy wouldn't post on IMDb anyway. So we looked at one of the users we identified, and tried to see what we could deduce about them based on their Netflix ratings (on movies they didn't
rate on IMDb). It was more than we expected. Actually, halfway through this process, I felt very uncomfortable prying into this person's life even though I didn't know them from Adam. It's in the paper; I'm not going to post it here.
Up until now, I wasn't 100% sure that a large number of Netflix customers would be embarassed by having their movie viewing/rating record linked to them. Now I'm sure.Updated paper
(section 4 is new.)