Before conducting this experiment, I believed that there were essentially two broad types of music: the music you listen to, and the music you tell people you listen to. The second category comprises the songs shared on Facebook or Twitter, the shout-out to Neutral Milk Hotel in your OKCupid profile, the stuff you send your friends. As with most things, the image you try to present to the world is substantially more cultured and interesting than the mundane reality.
This article fascinated me, mostly because I feel like a) the music I tell people I listen to is actually the music I listen to and b) data analysis of my music listening habits is not new, and not especially limited to streaming services.
Yes, I’m weird and I look far more closely at this stuff than your average music fan, but why are play counts that new? iTunes has had them for more than 10 years now. We all know with one or two clicks what our most-played songs are, and it doesn’t take much to extrapolate our most-played albums and artists. Hell, Last.fm has been doing this for even longer.
The only revolutionary part about all this data is how it coincides with recommendation engines:
Normally when you create a radio station in Spotify, you feed in one song or one artist. Nestify feeds in an entire weighted cluster to generate three playlists: a “My Music” playlist, which is tightly focused on songs I’ve already played a lot; a “Discovery” playlist, made up of artists and songs that are outside the cluster but similar to ones inside the cluster; and a “Default” playlist, which comes somewhere in between the two extremes.
This is where streaming services will finally make their presence worthwhile. For now, the glorified “jukebox in the sky” remains nothing more than a be-careful-what-you-wished-for utopian vision.