You Are Having Fun

Will Gottsegen considers what is lost in Spotify’s era of pandering recommendation.

By Will GottsegenFebruary 9, 2025

• Music

I CAN’T GET RID of my CD player. It’s a relic from 2003, a Sony CFD-E75—glossy, bulbous, and Ferrari-red, barely functional after two decades of overuse. Still, I can’t get rid of my CD player. What are my options? I’ve spent the past decade purchasing fewer and fewer physical discs, devoting more of my time to online streaming services that represent easier and cheaper alternatives to outright ownership. And yet somehow I’ve replicated the worst listening habits of the CD era, listening and relistening to the same music without ever feeling moved to branch out. Maybe there’s a reason I haven’t quite moved on from those dusty jewel cases.

One of the things that tends to get lost in the conversation about streaming platforms—Spotify, Apple Music, QQ Music, and the hundreds of other flavor-of-the-month competitors scrabbling for that last 10 percent of the global market share—is that most of them are in fact very good consumer products. That’s in part because the basic promise of streaming hasn’t changed all that much since its introduction in the early 2010s, at least from the perspective of the listener. For a lowish monthly fee, streamers offer access to a near-infinite library of recorded music. Where customers once paid a set amount for new albums they hadn’t even heard yet, all of a sudden they were paying that same amount for what felt like every song ever put to tape by any artist. There’s a reason streaming has accounted for the vast majority of industry revenue for the past decade, and it’s not because Spotify has any particular gift for product design.

It’s because streaming is still, unquestionably, a great deal. Actually, it’s too great a deal: as the journalist Liz Pelly explains in her definitive new book Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist, many of the existential problems with streaming as a business model, and the inequities created by its aftereffects (including, most notably, these platforms’ meager payouts to artists) are downstream of the bad deals struck at the beginning of the streaming era between digital service providers and the legacy music business. In their bid to get listeners to try a new way of paying for music, streamers wound up overpaying to get the hottest, most viable artists on their platforms. Pelly writes that Spotify ended up ceding “enormous advances, guaranteed minimum payments per stream, and shares in the company,” among other potential undisclosed perks. By the mid-2010s, streaming was huge, and first mover Spotify was best positioned to come out on top—but thanks to those initial contracts, it was still funneling 70 percent of its revenue to major labels. If Spotify was ever going to achieve profitability, it would need to figure out some other way to make money off its user base.

This has been Spotify’s fundamental challenge over the past decade, and you can feel its consequences in nearly every aspect of the product, from Discover Weekly—an algorithmically generated playlist that regularly repopulates with new music according to your personal tastes—to the so-called “Perfect Fit Content” program, a company-wide mandate to flood the platform with music that’s cheaper to stream than the stuff on the major labels. As it has sought to understand the way people listen to music, Spotify has been getting better and better at predicting what its customers want to hear. Armed with that knowledge, the company’s algorithms are able to steer us toward tracks that we’ll probably like just as much as anything on the majors, but which are much more lucrative on Spotify’s end. Often, this is just instrumental music commissioned for and owned by subcontractors, tracks expressly created to fill this niche, as opposed to music on independent labels.

Ultimately, the goal is just to keep us listening, locked into a constant flow of background jams from morning until bedtime. And the best way to do that is to feed us music we already know we like, or that we at least don’t actively object to. Today, my personalized “daylist” is called “tough stank face friday afternoon,” reflecting some of the buzzwords it associates with my typical Friday afternoon listening habits (“tough,” “baller,” and “808” among them). It’ll refresh in a few hours with something new for the evening. In the meantime, there are also at least 20 different made-for-me playlists tagged for “afternoon” listening, like “Cozy Afternoon Jazz Mix” with Chet Baker and Pharoah Sanders, or the similarly named “Cozy Afternoon Mix,” with cuts from Nick Drake and Bon Iver. The vast majority of the tracks on these playlists are culled from my own library, songs I’ve already proved that I enjoy.

Nowhere is this bid for total ubiquity more clear than in the company’s AI offerings. Take the interactive DJ “X,” so named for the real-life employee from whom it’s been cloned, Xavier Jernigan. Theoretically, it’s the most personal and engaging radio station on the planet, an infinite bounty of songs punctuated only by X’s introductions and changeups.

As a DJ, X is somehow both naturally convivial and eerily robotic, like an HR rep going out of his way to be your friend. During your first encounter, he makes you a promise: “I’m gonna be here every day, playing those artists you got in rotation, going back into your history for songs you used to love, and I’m always on the lookout for new stuff too. Just to push your boundaries a little bit.” Therein lies the Spotify credo, that push to stay within the bounds of what you already know you like. And yet, X never really seems to push my boundaries at all.

Here’s a sense of the kind of playlist he likes to feed me: “Drive” by SZA into “Screamland” by Father John Misty into “Rich Baby Daddy” by Drake into “CRYING IN THE KITCHEN” by Dave Blunts. “Respect to Dave Blunts on that one,” he said, flatly, after that little set. Then he played Gucci Mane’s “First Day Out” into Carly Rae Jepsen’s “All That,” before transitioning into “New York Subway” by Lil B. This is, frankly, an insane grouping of tracks—ultramodern R & B meets retro balladry meets meme rap—unified only under the nebulous heading of “songs I’ve listened to over the past few months.”

Which, of course, is why I love it. How could I not? Where real-life radio DJs, beholden as they are to the charts, are ultimately curating their own sequence of songs, X is just feeding me my sequences, unruly and ridiculous as those groupings often are. X’s implicit goal, maximizing the time we all spend listening to music, is a noble one, and yet actually engaging with X is the surest way to stay trapped in the laziest kind of feedback loop, listening and relistening to the same few tracks. The casual deception of this slick AI packaging lies in convincing the listener—emotionally, if not intellectually—that what you’ve already been listening to is in fact cool and fun and worth playing back. DJ X is an indefatigable yes-man, an imaginary arbiter of taste ready to validate your listening habits. And it’s no coincidence that this “pretend cool-guy” is discernibly Black (unlike, say, the less racially distinct voices of Amazon’s Alexa or Apple’s Siri): X is the Swedish product manager’s idea of a tasteful American.

It’s a similar story with “AI Playlists,” computer-generated mixtapes based around listeners’ own prompts. The amorphous prompt “give me something to listen to on my long flight tomorrow” yielded a playlist called “Sky High Chill,” populated mostly with songs I’ve already saved to my library. Maybe 80 percent are songs I’ve already saved, and the remainder are songs by artists I already know. Even when I asked it to “give me music that’s radically different from what I listen to now, something to really expand my horizons,” I ended up with a playlist called “Musical Horizons” consisting almost entirely of K-pop, which—even if I’m no expert on the genre—comprises some of the most popular and commercial music on the planet. Plus, songs I’d already saved, by BTS and NewJeans, were still slotted in alongside the unfamiliar tracks. Not necessarily horizon-expanding in the way I’d imagined.

There are also “AI podcasts,” which, as part of Spotify’s year-end Wrapped offerings, promise to translate an entire year of listening into a digestible four-and-a-half minute “show.” Anchored by two AI hosts, both of whom sound infinitely more natural and fluid than DJ X, the podcast is a way of keeping users engaged with their own habits. While the more feminine voice regales us with the data—things like the number of minutes we spent listening, or the genres we most enjoy—peppering its delivery with uhs and ums and even a little cross talk, the masculine counterpart is mostly just there to affirm and compliment. “You should be proud,” he told me during my personal podcast, right after his co-host explained that I was in the top 21 percent of Spotify listeners worldwide.

Part of the cumulative effect of these AI trifles is to dazzle us, to hint at even greater implementations of the tech down the road. AI is, obviously, kind of cool. But in their insistence on sameness and comfort, these tools remind us of the value of good old-fashioned music discovery. The process of finding new music is work, and it should be. It’s far easier to stay in your lane, listening to the same tracks, rather than sitting down and exploring a new album, a new artist, a new genre, which is why Spotify keeps giving us comfort food in the guise of something truly new. The irony of having an entire world of music at our fingertips is that we’re only ever encouraged to listen to the same few songs.

This isn’t some “gotcha”; it’s basic business sense, and it’s part of how Spotify has managed to post its first-ever annual profit. No one would listen to DJ X if he exclusively pushed outré experimental music to listeners who mostly want to hear Taylor Swift, or screamo to fans of what Spotify might call “chilled-out folk.” But listeners also contain contradictions. There’s always the chance that a real-life DJ might play something out of his typical wheelhouse, introducing listeners to something genuinely, radically new. Maybe the DJ even prefaces it with a personal story, a reason for listeners to care. In its effort to minimize friction, Spotify has inadvertently foreclosed on this kind of experience, the spark of genuine discovery that’s only ever ignited by an outside perspective: a stack of old records in a specialty store, a freshly burned mixtape, a stray link from a friend who won’t stop bugging you until you actually listen to some new song they found. Even the humble CD can be lent with intention.

DJ X isn’t another person. He is more like a homunculus, a shadow puppet locking us into a musical rut of our own perverse creation. He knows us because he is us, and this makes him irresistible. Listening is easy; true discovery is nonnegotiably hard.