Too much about music stats

There’s a web page called How Bad Is Your Spotify? that looks at your Spotify playlists and listening habits and judges you like the record store guys in High Fidelity. You cannot win. If your tastes are too mainstream, you’re mocked for being a mindless drone. Too eclectic/obscure and you’re mocked for trying too hard. It’s amusing. My report started with:

Your spotify was heavy-eyeliner-post-punk-suburban-80s-nobody-puts-baby-in-a-corner bad.

Thank your obsessions with Talking Heads, new wave, and (I’ve Had) The Time of My Life – From “Dirty Dancing” Soundtrack for that.

Mostly reasonable. I’m a 54 year old guy who grew up with KROQ as my background music, so new wave and Talking Heads fit pretty well. But I don’t remember the last time I listened to anything from Dirty Dancing.

Also, this comes up later:

You listen to these tracks too much:

• Dress Sexy At My Funeral by Smog
• Chelsea Hotel #2 by Leonard Cohen
• Stella By Starlight by Miles Davis
• Different Drum by Stone Poneys
• These Things by Looper

And I do listen to most of those, but Chelsea Hotel #2 only comes up because Spotify has decided to add it to my Daily Drive mix, and apparently counts playing any part of it (even after I skip it) as a play.

This isn’t a complaint about the “How Bad” page; it’s using the only data available to it. But it highlights how inaccurate Spotify is at knowing what I want to hear, or even what I usually play. Most of my music is stuff I play from iTunes (or whatever Apple wants to call iTunes these days). I have tons of music that doesn’t exist on Spotify, and I usually let it play randomly. The problem is that iTunes/Apple Music only keeps two data points about a song: when you last played it, and how many times you’ve ever played it. I wanted to have a list of my most played music of 2020, but it can’t be done without an outside assist. I used last.fm and got this list of songs:

That’s better, but it makes it look like I’m a huge fan of White Hassle when it’s really the iTunes shuffle algorithm failing to effectively randomize stuff mixed with my letting music play in the background without paying attention to it.

The top artists list is slightly more reflective of what I listened to in 2020:

But that one has some issues as well. For one thing, slight name changes count as different artists; if you add Elvis Costello to Elvis Costello & The Attractions, they move to second place.

…and this still doesn’t really reflect my “shuffle ’em all and let the algorithm sort ’em out” listening. I mean look at this list of over 4500(!) songs I listened to exactly once in the last year:

So, what have we learned? My main takeaways:

1. Music stats are difficult.
2. I can write long and boring posts.