The discovery algorithm in general has never been as good as say Pandora, but I guess I agree that it's felt worse recently.
Most annoying to me recently is how there seem to be three different ways to indicate I like a song, depending on which device I'm listening on (desktop app on my laptop or desktop, or mobile app on my Android phone) - I swear I've seen hearts, plus signs and thumbs-up symbols, without any clear indication of any differences. I know a sibling comment mentioned this change, but they made it sound recent - I _feel_ like I've had that point of confusion for over a year...
In part out of frustration with Spotify, and in part to hedge against streaming disappearing one day, I'm in the process of ripping my exiting CD collection to FLAC, and buying either digital or physical copies of my favourite albums.
I've found it pretty refreshing to open my music app (Finamp) and only see exactly what I want to.
This is bizarre to me. So to punish the record industry for ruining Spotify you're... giving them even more money?
I make it a point to buy a couple of albums a month from Bandcamp. The artists get a much larger cut than whatever I have been streaming of theirs, and in exchange I can listen as much as I want, completely offline, without any surveillance of my listening habits.
https://pitchfork.com/thepitch/how-much-more-money-artists-e...
I'm not sure any of my issues with Spotify (pushing podcasts and audiobooks, nonsense UI changes etc) have anything to do with the record industry.
I've been generally very happy with the Spotify app for years and it's disappointing to see the quality slipping.
To add, I think I’m seeing less discovery and more popularity bias. I used to discover a lot more.
There is a limit to "delightful serendipity". At some point, you will have heard everything you could potentially be interested in but not normally listen to. After that, it will be either not your thing or more of the same.
Also, algorithms don't read your mind, they just match your usage patterns (mostly what you are already listening to) to other people usage patterns and give you what they have and you don't. For example you may listen to psy trance, and the algorithm notices that many people who listen to psy trance also listen to folk metal, so it will give you some of that. And there is a good chance for you to like that even though it is a completely different genre you may not be aware of.
But once all the matches are done, that's the end. There may be a few paths left for you to explore, because you are an individual and not an average, and tastes change, but there is no easy way for any algorithm to find that. The only solution is to go with stuff you probably won't like, so, not so delightful serendipity. Or just give you more of the same.
I never like spotify's "for you" playlists.
Yeah, and she ain't givin' it back. Sigh ... guitar solo ♪♫♬
Everything is owned by the same tiny group of people and they need the number to always go up.
More and more frequently, as the blood is squeezed out of the stone, that means laying everyone off and dumping the work on fewer and fewer people.
I can choose pretty much any "song radio" and the same set of artists shows up. I mean, I like Alice in Chains, but not in every play list.
I ended up with Bandcamp+Spotify. I find new stuff on streaming and then buy it on bandcamp. Store all the FLAC files in a plex server and stream as much as I want from my library.
Plexamp even has some new ML features where they can generate playlists based entirely from your own library.
Need to show growth > Hire more and more people > A lot of them have nothing to do because companies always over hire > Change parts that work in the name of improvements because they have to show they are working > enshitification