Is this intentional on Spotify’s part? Do these covers have cheaper royalties than the originals? Have I somehow trained their model to think this is what I love? Does not skipping songs tagged as covers quickly enough train the model to think I like them? Maybe because I listen to lots of rock, jazz and classical the algorithm thinks a jazz interpretation of “Iron Man” is what I love? It feels like if you “over trained” and ruined a good Pandora station way back when.
Spotify used to be an excellent source for music discovery but these covers are straight up aggravating. Between this and their hostile UI changes pushing podcasts (I will never use Spotify for podcasts) I’m becoming motivated to jump ship.
I remember finding out that Freebase knew about 200 tracks titled ‘Sweet Home Alabama’ and these were copies from greatest hits collections, live versions, remasters, covers, etc. Finding the ‘right’ track in this case is challenging if only because of the large number of candidate tracks.
But people still like popular songs so we eventually ended up with all these covers being listened and recommended again and again.
I only remember it because I had the same reaction you did!
The short version is - different royalty rates + algorithm preferences.
While it may seem immaterial to each user, the ability to lower the royalty cost on billions of plays is likely material to Spotify's margins, which are already caught between the cost of the platform and the strongly defended royalty structure on most music.
[1] https://www.theverge.com/2015/9/8/9260675/spotify-cover-song...
What other streaming service is even comparable?