Illustration: The Globe and Mail. Source photo of Elliott Smith: AP PHOTO
Steve Kupferman is an editor for The Globe and Mail.
For a long time, whenever anyone asked me what kind of music I liked, I would tell them that I wasn’t a “music guy,” which probably sounded ridiculous. Every guy is a music guy. This has been true since the first guy crawled out of the primordial sea.
The truth is that I was listening to music, but could not bring myself to reveal the details because they were – at least to me – mortifying.
I had moved to Toronto in my early 20s without any real idea what I was going to do here. During this period, when I was spending a lot of time alone in a basement apartment convincing myself that the scratching sounds I was hearing weren’t rats in the walls (they were), the one artist I could not stop listening to was Elliott Smith, a singer-songwriter known for his limber melodies and morose lyrics, and for dying at the age of 34 of stab wounds to the chest that may have been self-inflicted. (Fans have spent the past two decades poring over an inconclusive coroner’s report.)
A university roommate had introduced me to Mr. Smith’s music, a catalogue of bright and meticulously composed indie-pop songs that often ring with sadness and anger. (His imagery isn’t always relentlessly melancholy, but he does sing about cheap alcohol and amphetamines, rage against absent friends or ex-lovers, and occasionally wallow in outright nihilism.) It was almost instantly captivating to me, the sonic manifestation of life (or, at least, my life) as a young city dweller at the start of what was beginning to seem like a bad century: so much possibility, all of it clouded by a sense of doom.
But music is an easy proxy for identity, and I didn’t want to be known as a devotee of the sad dead guy, even as I listened to him so much that I got tired of the studio albums and started hoarding posthumous releases, live shows and outtakes. And so, when it came to musical taste, I chose not to be known at all.
Eventually, I moved out of the basement, built a career, got married and rose above the metaphorical doom cloud, which in the end turned out to be more of a ground-hugging doom fog. But I found that I couldn’t move beyond Mr. Smith’s orbit. It had been so long since I had sought out new music that I no longer knew how. I began to suspect, to paraphrase Mr. Smith, that I had become a symphony man with just one note.
And then came the pandemic, which condemned me and everyone else in the world to their own private vermin-infested basement apartments of the soul.
It was around this time that I discovered a music-streaming app. I won’t name it, because it doesn’t need free advertising, but it’s definitely the one you’re thinking of. I had been aware of the existence of apps like it, but had never used one myself.
I was spending a lot of time with my electronic devices. Should I go ahead and imbue them with the astounding and near-magical ability to sing me any song I can think of, whenever I want, for a low monthly price, I wondered? Okay, sure.
What I didn’t realize at first, but probably should have known, is that the app was built around a machine-learning-driven recommendation algorithm.
There has been a lot written, over the past decade or so, about the malign influence of the algorithms that decide what we encounter on streaming platforms and social media. These recommendation engines have invaded virtually every corner of media, and they have been blamed for radicalizing suburban moms into QAnon jihadis, impinging on Canada’s cultural sovereignty by unfairly boosting American artists, and even imperilling democracy by influencing the outcomes of elections.
And I had just delivered myself into the greedy data vacuum of another one of these things.
Once the app had learned all it needed to know about me, it went to work.
In the beginning, I listened to the same things I had always liked. The app was the first outside entity with which I had ever fully shared my musical taste. Even my wife, though she knew I liked Mr. Smith’s music and similar stuff, didn’t always know what was going on in my headphones. But the app knew.
Once the app has a sense of a listener’s likes and dislikes, it begins building custom playlists that include some familiar songs, but also some algorithmically selected tracks.
With no strong musical preferences of my own, I let the app direct me through moods and genres I had no names for. Is there a word for the kind of music where the melody is hypnotic and the singer sounds kind of bored? Who is “Stephen Malkmus”? Was Sonic Youth always this good? Wait, do I love psychedelia-influenced country music from the 1970s now?
I found that I was still developing fixations on particular artists, but now they were brief. At one point I spent a few weeks listening to a Chicago-based group called Finom, whose most recent work I can only describe as sounding as though it is written and performed by sexy androids with music-theory PhDs. The app guided me to an artist called Jim Sullivan, a spaced-out wannabe cowboy singer who cut two very good albums before literally disappearing off the face of the Earth in 1975. (He was last seen in a remote part of New Mexico.)
How we – by which I mean the app and I – got here from Elliott Smith is a mystery to me. The app’s makers say its algorithm weighs a number of different factors, including how frequently users group particular songs together on playlists they create. The app also automatically analyzes songs for qualities such as “danceability,” “energy” and “instrumentalness.”
But to me the process did not feel as though it was being directed by software. My mind felt like it was spreading feelers at random and sending up shoots.
And this is precisely what is so pernicious, and so wonderful, about the algorithm. I have a whole new musical sensibility now that feels as though it came from within, but that actually was imposed, at least partly, from outside. The precise ratio of algorithmic conditioning versus personal free will at play here is at best a trade secret, and at worst completely unknowable – a matter of philosophical debate.
Rewiring a person’s musical preferences was once seen as a social act, or even an act of love. It was something that used to be done by radio DJs, or cool older siblings, or mixtapes compiled by dorky guys trying to express mind-enveloping romantic obsession to their crushes without freaking them out.
Nobody gave any thought to what it might mean for us, as a society, to automate this process. For all we can tell at this point it could be the end of music as we know it. We could be entering a world where music is no longer a marker of identity, but rather a product of it – a world where songs are no longer recommended algorithmically, but are actually written algorithmically, to tickle the pleasure centres of each individual listener.
But that’s not where we are today. For now, even though I feel as though my mind may have been colonized by Big Tech, I also feel … great? Discovering new music after my long period of incuriosity had effects I couldn’t anticipate and can’t quantify.
I think my emotional aperture has expanded, ever so slightly. Listening to music on public transit or while walking through the city, which I had not done for years, is a cheap and effortless source of joy. For the first time in my marriage I can play music for my wife that isn’t “too depressing.” I went to a concert, after a decade of mostly avoiding them, and saw a crowd of a few hundred people who all have at least one thing in common with me.
And my relationship with music is no longer a source of weird, neurotic shame. In a small but important way, I feel like I’ve been transformed for the better. Everything else had changed; this was the last thing that hadn’t.
There are so many ways big online platforms have damaged the world. Even the streaming app is notorious for leveraging its market dominance to underpay musicians.
But as bleak as the future of automation sometimes looks, dealing with the music app has made me wonder if there’s still hope that these new systems will find ways to integrate with human minds that aren’t exploitative – that promote grace and humanity, rather than the opposite. I now think it’s possible, though I wouldn’t say it’s likely.
I still love Elliott Smith’s music. Even from the not-so-lofty heights of a semi-successful early-middle age, I can see the doom fog on the ground. It has a certain beauty when viewed from above, and it’s a part of me.
But I’m also part algorithm now, I guess. And I’m okay with that.