• oyo@lemm.ee
    link
    fedilink
    arrow-up
    158
    ·
    7 months ago

    Remember when Netflix had a million dollar competition to improve their recommendation system by something like 10%? Remember when they had user ratings and reviews? Remember when they threw it all in the fucking trash?

    • Semi-Hemi-Lemmygod@lemmy.world
      link
      fedilink
      English
      arrow-up
      82
      ·
      7 months ago

      The original recommendation engine was amazing. Showed me stuff I loved that I would have never found myself. My favorite example is the French claymation acid trip called “A Town Named Panic”

      Now all I get recommended are shows that are cancelled and that I’ve already watched.

      • BarbecueCowboy@lemmy.world
        link
        fedilink
        English
        arrow-up
        35
        ·
        7 months ago

        The goal isn’t to give you stuff you want to watch, the goal is to do just enough that is just good enough to keep the highest amount of users subscribed. We like to think those are the same thing and they are usually aligned, but don’t have to be. It’s better for Netflix if you find those slower and just keep coming back to look around, kind of like some frustrating streaming gatcha game.

        • Semi-Hemi-Lemmygod@lemmy.world
          link
          fedilink
          English
          arrow-up
          9
          ·
          7 months ago

          I watched it because it was the top pick in children’s movies and ended up enjoying it more than my kids. I rewatch it every few years because it’s just so weird and good.

          Another one is the TV show Oscar’s Oasis. It has no dialog, but a very Coyote-and-Roadrunner feel. My kids loved it and it was actually pretty funny.

          But when we were at Disney World we ended up sitting next to a family from Brazil who knew very little English and their kids were watching Oscar’s Oasis and my kids immediately joined them.

    • bionicjoey@lemmy.ca
      link
      fedilink
      arrow-up
      4
      ·
      7 months ago

      That was such a glorious time. I recall they even held a competition at one point to develop a recommendation algorithm that could predict whether or not someone would enjoy Napoleon Dynamite