This is a great video. What I find very confusing is the fact that Facebook has no spam detection to eliminate fake clicks/likes ? Given the obvious nature of fake likes, some of the heuristics seem very straight forward i.e.

- just filter profiles with an insane number of likes

- have very few friends but relatively a huge number of likes,

- certain regions in the world where the brands that the users have liked have no market. It looks like a straightforward matching problem.

- Suspicious Like activity and patterns from an account (like/click arrival rates, diversity in liked campaigns etc).

- De-duplication of profiles that may have liked a certain entity by clustering on some obvious signals e.g. if too many disproportionate likes all fall in egypt relative to other primary markets, that would be an obvious red flag.

All of the above is not difficult to figure out for the quality of talent Facebook has. Here is my real confusion: Is it a lack of inclination for improving their ad network quality or a fear of exposing the low ROI of ad spend that's stopping Facebook from implementing quality spam/fraud detection ?