Netflix used to be famously good at suggesting films. Articles were written about it, and there was even a cash reward for anyone who could contribute to its performance. Then it just turned to shit.
And the funny thing is that it would have helped counteract the shrinking library. Sure, there would be fewer films on the platform, so you’d be less likely to find a specific title, but at least you could select a film Netflix recommended based on your past ratings and be fairly confident you’d enjoy it. Now? Absolutely not.
It’s not widely available and its only in Norwegian, sadly.
However, I will second @mkengine proposal for Letterboxd, I think it is the superior site to nerd out on. Discovery can be a challenge, depending on your own level of investment into the medium.
I’m a big ol movie-nerd, and I’m currently grateful to have access to most streaming services through friends/family/partner so I get to browse them if desired.
Apart from that my twitter algorithm is quite skewed towards movies, and I have a “list” on there (curated users you can browse, kind of like a community on here. That’s been great.
Other than that, I listed to podcast, sometimes check out our national newspapers reviews (but most of those reviewers are already in the aforementioned twitter-list) etc.
As for reading on recommender systems and the algorithm for netflix. My work was based around bias and “trust” when it comes to the recommender systems and how much it recommended/pushed “its own agenda” to users despite having differential tastes.
Good keywords I enjoyed was: recommender system bias
I also read some good articles on the spotify recommender systems. But those mostly centered around people growing attached to their algorhitms. It was a fun read.
Well this can get quite complicated to implement I suppose. I heard letterboxd works nice for discovery if you are lazy, but I don’t know if they have a jellyfin plugin.
Hell, I remember when they had ‘Max’ back when I watched on my PS3. I absolutely loved that it would ask me a bunch of questions and then give me a movie to watch. I’m surprised I haven’t seen that more around, that was an amazing feature.
I find the same thing with music streaming on Spotify. I used to discover lots of new music I liked on it but these days I can’t get it to generate an interesting playlist. It’s songs I already know interspersed with things that are boring. Seems like the recommendations got worse.
I quit Spotify when the “New Library Experience” completely fucked the music library side of the app. If you mostly use playlists, it was a lateral change. If you used it to collect some songs here, and album there, and keep them all sorted, it’s like it dumped your entire collection on the floor and expected a thank-you for the new organization system.
My guess, as others have mentioned, is that Spotify tries to squeeze more profits by pushing certain songs, whether because they get paid to promote them, or the royalties are lower. That’s easier to do with their playlists and recommendations, so they pushed people to that side of the app by making everything else dogshit. And now, apparently, the curated side took it too far and is awful, too.
I still use Apple Music, which is one of like two services that actually let you organize your music in a sensible way outside of playlists. That said, after I cut cords with video streaming services and set up my own library, I think I might do the same with music.
Discovery was always the thing that made streaming services better than buying recordings individually. If these services stop being good for finding new music, then there’s not much reason to keep using them.
And their recommendation engine sucks.
Netflix used to be famously good at suggesting films. Articles were written about it, and there was even a cash reward for anyone who could contribute to its performance. Then it just turned to shit.
And the funny thing is that it would have helped counteract the shrinking library. Sure, there would be fewer films on the platform, so you’d be less likely to find a specific title, but at least you could select a film Netflix recommended based on your past ratings and be fairly confident you’d enjoy it. Now? Absolutely not.
My bachelors thesis was basically about recommender systems like this. Netflix truly is a sunken ship.
How do we read it?
If you are just interested in Netflix recommendation algorithms, you could start here
Thanks.
I am in the process of setting up a jellyfin server and was wondering how I would deal with discovery.
It’s not widely available and its only in Norwegian, sadly.
However, I will second @mkengine proposal for Letterboxd, I think it is the superior site to nerd out on. Discovery can be a challenge, depending on your own level of investment into the medium. I’m a big ol movie-nerd, and I’m currently grateful to have access to most streaming services through friends/family/partner so I get to browse them if desired.
Apart from that my twitter algorithm is quite skewed towards movies, and I have a “list” on there (curated users you can browse, kind of like a community on here. That’s been great.
Other than that, I listed to podcast, sometimes check out our national newspapers reviews (but most of those reviewers are already in the aforementioned twitter-list) etc.
As for reading on recommender systems and the algorithm for netflix. My work was based around bias and “trust” when it comes to the recommender systems and how much it recommended/pushed “its own agenda” to users despite having differential tastes.
Good keywords I enjoyed was: recommender system bias I also read some good articles on the spotify recommender systems. But those mostly centered around people growing attached to their algorhitms. It was a fun read.
Well this can get quite complicated to implement I suppose. I heard letterboxd works nice for discovery if you are lazy, but I don’t know if they have a jellyfin plugin.
I will look into them thanks.
Hell, I remember when they had ‘Max’ back when I watched on my PS3. I absolutely loved that it would ask me a bunch of questions and then give me a movie to watch. I’m surprised I haven’t seen that more around, that was an amazing feature.
I find the same thing with music streaming on Spotify. I used to discover lots of new music I liked on it but these days I can’t get it to generate an interesting playlist. It’s songs I already know interspersed with things that are boring. Seems like the recommendations got worse.
I quit Spotify when the “New Library Experience” completely fucked the music library side of the app. If you mostly use playlists, it was a lateral change. If you used it to collect some songs here, and album there, and keep them all sorted, it’s like it dumped your entire collection on the floor and expected a thank-you for the new organization system.
My guess, as others have mentioned, is that Spotify tries to squeeze more profits by pushing certain songs, whether because they get paid to promote them, or the royalties are lower. That’s easier to do with their playlists and recommendations, so they pushed people to that side of the app by making everything else dogshit. And now, apparently, the curated side took it too far and is awful, too.
I still use Apple Music, which is one of like two services that actually let you organize your music in a sensible way outside of playlists. That said, after I cut cords with video streaming services and set up my own library, I think I might do the same with music.
Discovery was always the thing that made streaming services better than buying recordings individually. If these services stop being good for finding new music, then there’s not much reason to keep using them.
They probably discovered that they can get paid to promote certain songs.
So now what you have is the same as FM radio, except you pay for it now.
I think they have always done that, but maybe they’re just doing more of that now. It seems harder now to find interesting artists I don’t know.