Every couple of years, I change how I get music. It used to be CDs—actual honest-to-God discs full of bits. Then I switched to buying MP3s from Amazon, then the iTunes music store. About three years after the rest of the world, I caught up to the swiftly-departing Spotify bandwagon and dragged myself on. Then I hopped over to Google Play Music All Access, mainly because it gives me typing practice to write out its name.
Every time I switch, I end up forgetting about half of my music. It turns out my shelves of CDs serve mainly as a giant, exhorbitantly expensive, list of bookmarks. Once I ditched the CDs, I ditched those bookmarks too. I started rebuilding playlists in iTunes, then Spotify, then Google Play, but I kept losing stuff.
There’s a thing in the field of artificial intelligence—oh wait, we call it “machine learning” now—called “ratholing”. This is a real thing. I’m not making this up. It works like this. Modern recommendation systems, the piles of code and data that back seemingly simple features like Amazon’s “You might also like”, and Spotify’s automatic playlists, and Google’s, well, everything, are based on software that adapts and learns automatically. You probably didn’t know it, but you have a little robotic butler.
You tell it a few things you like, and it guesses at a couple of similar things you might also like. You tell it which of those you like, and it suggests more. This back and forth goes on and it builds up a picture of your interests. But this back and forth can also be a feedback loop.
Sure, it’s learning your interests, but it only gets positive feedback from things it already suggested to you, which in turn were based on things you already told it you like. So, instead of building an expanding picture of what you like, it spirals down into the same narrow interests you got started. It gets stuck in a rathole.
When I switched to Google Music, I told it a few things I was into right now. And now, after a few months of giving thumbs up to stuff, it has possibly the world’s most comprehensive list of glitchy house, nu-disco, and hipsters-making-80s-roller-rink-throwback-songs on Earth. Which is great, but it gets a bit tiresome even for me.
The other day, I randomly stumbled upon a video of Snow Patrol playing “Chasing Cars” to an adoring sing-along crowd, and it was like I’d clicked “Play” on a high definition clip of my own youth. There was a two year period in my twenties where Final Straw never left my CD player. I was in a local rock band in Orlando at the time, and that album formed part of the soundtrack to one of the happiest periods of my life.
(There aren’t many recordings of us around, but here’s one. I’m playing bass on the left.)
In a fit of nostalgia, I made the mistake of looking to see when Final Straw came out. 2003. How can eleven years ago still feel like yesterday, when I usually can’t remember what I did last week? Getting older isn’t the sensation of the past receding. It’s the opposite, an increasingly long stretch of time that still feels like “now”.
Whenever I discover a new band, I go through this process where I track down all of their footprints on the web. I want to see what they’re doing, what other releases they have, where they’re playing shows and when. (More than once I’ve found that the answer to the last two is a maddening “in my town” and “last week”.)
Rediscovering Snow Patrol reminded me of my simultaneous infatuation with Longwave. What I learned today, why I’m writing this down, is that going through that “where are they online?” process with a band whose peak was a few years back is a weird mixture of ephemera and permanence, present tense and past.
Longwave does still have an active Facebook page, but it’s mostly posts about what other projects the various band members are doing. A simultaneous reminder that they are all still here for some definitions of “here” but that “Longwave” is now mostly just a label I use on the filing cabinent of memories from my twenties.