Earlier quoted context omitted.
I see very little evidence that Netflix' data science has been a unique differentiator. For what they've developed, they have some hits and quite a few duds. Their licensed catalog is not exceptional--and even less exceptional when it comes to film. They'll likely mostly turn into another studio with a streaming platform.
To add to this, given the resources spent on it their UX is pretty bad and I don't think their data scientists are giving them much if any edge over the competition at this point. Features such as only using company generated ratings, aggressively cancelling shows, and the lack of ability to browse/sort/filter the catalog, in favor of Netflix pushing the recommendations they choose on users, is probably great for qua…
I think all of those companies fall into common problem of falling deep into A/B test local maximum. It's impossible to rework entire site at once, especially for an A/B test.