Deciphering Recommendation Engines by Cathy O'Neil

Data scientist Cathy O'Neil provides a glimpse of the methods that Netflix, Google, and others apply to recommend or offer to users selections based on their apparent interests.
Read about O'Neil's MAA Distinguished Lecture "Start Your Own Netflix" bit.ly/1bqRs7U

[@MAAvideo]

Deciphering Recommendation Engines by Cathy O'Neil
Data scientist Cathy O'Neil provides a glimpse of the methods that Netflix, Google, and others apply to recommend or offer to users selections based on their apparent interests.
Read about O'Neil's MAA Distinguished Lecture "Start Your Own Netflix" bit.ly/1bqRs7U

[@stevepalmateer]

Ms. +Cathy O'Niel, I've heard you speak elsewhere about "authority of the inscrutable". In this video you describe algorithms that are computationally difficult. Then you mention certain assumptions ("you give something up, and you get something") that we can make about the data that allow us to be "much smarter, and much faster". These are precisely the kinds of remarks that sound intimidating to those who are not experts.
Why is "parallelizable" sufficient? And why is it so easy to see that it's sufficient?