The www.netflixprize.com data from 2009 is quantitative evidence about whether the recommender works (it is much better than random). I'm open to the idea that

The www.netflixprize.com data from 2009 is quantitative evidence about whether the recommender works (it is much better than random). I'm open to the idea that a curated collection might delight more people, but in my experience and based on this data, I can't agree that Netflix's algorithm is useless. Personally I find it's the unavailability of streaming content that makes Netflix frustrating.