There's a lot to think about in this article... it's stuck with me and been present in my mind since yesterday.
One could argue that the algorithm in Spotify is doing the exact opposite of the radio, where you can only hear 1-2 songs from any band, and they will always be the same songs. Spotify is crate-digging for you... but also, at the same time, kinda misrepresenting that artist in a way...
I've made records where we just threw a track on there to fill it up... a simple little thing that we had no attachment to, but needed a space-filler. And what would I think if THAT was the most played track by my band on Spotify?
My metal band in high school got quite a bit of popularity, in our area. There was a dumbass ditty we made up at rehearsal one day to make fun of the type of "technical metal" (Iron Maiden, Judas Priest, etc.) that we were playing, and required a lot of skill to play.
PEOPLE STARTED REQUESTING THAT COMEDY SONG.
So... I like the algorithm isn't skimming artists, but doing some actually deep and interesting algorithmic analysis of songs and their relation to each other... but at the same time, doesn't the popularity of a song OUTSIDE the algorithm need to be considered as well...?
Or is radio just so corrupt and non-representative of what people want to listen to any more that completely ignoring plays outside of Spotify is the right thing to do? Or does your algorithm become to incestuous?
No right or wrong answers there... but definitely things that need to be considered by people who ARE NOT writing the algorithm...