3 ms·

'(recommendations are useless)' You've watched 12 horror movies so, guess what, we're recommending this crock of ___ because it is also a horror movie... Seem

• magicalhippo 8y ago
  So first off, thumbs up and thumbs down is dumb. There are movies which are really great, there are movies that aren't but are still worth watching, there are movies that aren't great and aren't really worth watching, and then there's the movies with Adam Sandler or Amy Schumer in them... And reasons for liking a film could be that I liked the cinematography, or the script, or maybe just one actor, and not just that it was a horror movie. Take for example Beyond the Black Rainbow (2010). It's weird, the story wasn't all that for me, the ending is weak. But I loved the way the director shoot it as an 80's sci-fi film, really capturing the look and feel of those films. And well, I love slightly weird movies. So overall it's a movie I really enjoyed watching, and will likely watch again a few times. So just how would Netflix interpret my thumbs up for this movie? I've rated literally hundreds of movies in Netflix. I still get 90+% match recommendations that I can tell just from the poster and blurb that I'll hate (and my control for this are the movies I've watched elsewhere and hated). I've found some of the sub-50% match movies to be great, and really enjoyed watching them. I know Netflix had this machine learning contest a few years back, looking for winners who made better recommendations than their own algorithm. But IMO it's just gotten worse and worse.