It was less than a year ago that a lifetime of music collection went up in smoke. Actually, eyewitness reports suggest that the vinyl did not really go up in smoke so much as it dissolved into a molten bubbly mass for about twenty-four hours, eventually hardening into a shiny black orb.
Long-time readers of this blog know I had made being a “vinyl lover” something of a fetish, part of my identity. Having that stripped away was a little disorienting, though it’s hard to untangle my feelings about music from everything else that was lost in the fire. I haven’t even begun to build a new record collection. Part of that is practical — right now I live in a furnished place, and I relish the idea of my next move being something I can do in a single carload (including the drumkit), so I have been re-acquiring possessions sparingly. But more than that, starting a new record collection from zero just feels too overwhelming to even contemplate.
All this a roundabout way of saying that my music consumption is now almost exclusively streaming digital. WFMU remains my cultural touchstone, of course. I listen to it more than ever. What’s changed is my “on demand” listening. I am a Spotify Premium subscriber, which offers decent 320kbs sound quality (I’ve also been dabbling with higher quality Roon/Tidal options), offline access to favourite songs, and playlists. It’s not ideal, and I’m not thrilled with how streaming services treat artists, but right now it’s an affordable option that works conveniently for me.
Spotify has also been fairly aggressive in trying to exploit the data of its users for myriad effects and studies. Some of the implications are fascinating, some creepy. I usually tend to live on the info-paranoiac end of the spectrum, but in this instance I have made a conscious decision to let those concerns slide. Mostly, I don’t want to live without music. Also, music seems like an interesting place to observe how the algorithm-driven life might play out in reality.
A few weeks ago I read about Spotify’s new Discover Weekly Playlist, which is billed as “a completely personalized playlist”, constructed out of the data it has gleaned from your listening patterns, yet oriented to artists you aren’t actively listening to. What’s interesting is how the algorithms are constructed, and how much of what can only be described as editorial work goes into the playlists.
Ajay Kalia, who oversees the project, tells us they realized early on that there’s an important distinction between the music you listen to and music you actually like.
For example, just because I play a lot of instrumental, ambient music while I’m at work doesn’t mean that I have a particular affinity for those kinds of artists. And just because your significant other plays a lot of country music while you’re both in the car doesn’t mean you want a bunch of country playlists shoved at you.
Although Spotify has had humans making playlists for years, its efforts got a major boost last year with the introduction of Truffle Pig, an internal tool from The Echo Nest that breaks music down into thousands of categories like “wonky,” “chillwave,” “stomp and holler,” or “downtempo.”
Doug Ford, Spotify’s director of music programming, walked us through how a Spotify playlist comes together.
“Every curator has to come up with a hypothesis,” Ford says. “You brand it with an image and you give it a cool description that gives you an indication of what you’re diving into if you’re about to listen.”
I have listened to every song presented to me on my Discover Playlists at least once. What does big data think of my musical tastes? You can get a sense from this screenshot:
That’s only a sampling, but it’s fairly representative. Post-punk. Indie. Slightly off-mainstream classic rock. In short, dad rock.
In some senses, I am impressed by the playlists. I rarely hate a song they recommend. I even realized I liked Paul McCartney’s “Monkberry Moon Delight”, so in some ways Spotify knows me better than I know myself. And I suppose I am impressed that Spotify knows that if I am playing hip-hop, it’s probably because I am listening with my son.
But that also points to a defect in this thinking. Just because I mostly listen to hip-hop with my son, that doesn’t mean I don’t want to listen to it any other time. In fact, I would love to get turned on to some fresh and inventive artists so I am not always at my son’s mercy when we listen.
But worse than that, I feel like Spotify is really not playing close attention to me. When I look at my tagged favourite songs, and my play histories, I really don’t listen to all that much dad rock these days. (Honest!) Lately, it’s been a lot of 1970’s funk, freaker country (natch), old time country and folk (lots of Jimmie Rodgers), Latino music (traditional, electronic, and some prog that Pedro turned me onto). And the raw power of really smooth music, of course. None of this is reflected in any of the playlists so far. The closest its come is one track by Os Mutantes (which may have been the first Latin psychedelic group that I listened to, more than twenty years ago), and one track by Gram Parsons and the Flying Burrito Brothers (which again, was music that changed how I thought about country music, but that was long, long ago). More than turning me onto music that I didn’t know existed, the Discover Playlists seems very good at reminding me of the music I listened to when I was a student. Which makes me wonder if they are trying to shoehorn me into their finding of how our musical tastes “flatten” at a certain age.
Big data seems to be saying, “you may think you have adventurous tastes, but the algorithm knows better. Now get back in the minivan, shut up, and enjoy the Galaxie 500 we have selected for you. Remember them? Isn’t that nice?”
It feels like a glimpse of how more experiences will be in the near future. It’s easy to make parallels with the current trajectory of ed tech. Let’s just hope the human sensemakers manage to keep on keeping on.