Update, 2/3/22: I have grown increasingly uneasy about the economics and ethics of streaming services since writing this. I’m going to leave it up anyway; I think it’s a good piece and I stand by the methods I described. It’s possible to use them without relying on streaming services; you’ll just have to download songs (legally or less-legally) to put on playlists.
I cannot tell you if you should or should not use streaming services. You need to weigh the arguments and make that decision for yourself. When I wrote this piece, I had made a decision. I have recently started to reevaluate it. In any case, algorithms are still dumb and you should not rely on them.
Disclaimer No. 1: This article won’t rely as heavily on traditional research as my last few have. If you see an assertion without a citation, understand that I made it based on personal experience or experiences that other people related to me, or a combination of the two.
Disclaimer No. 2: I only recently learned that David Dayen, Ryan Cooper, and Matt Bruenig had all written about algorithms and playlists.1 Of the three, Cooper’s concerns intersect the most with the ones I discuss here. I swear I didn’t read any of these articles until I was finished with mine.
Anyway, I want to talk about playlists. I’ve been making them since I was eight years old. I’m not joking. I got an iPod Shuffle when I was eight, which would have been around the same time that I started making mix CDs to play in the car. I can’t remember which came first, the iPod or the mix CD, but either way, I started making playlists when I was eight, and I never stopped. I even made a couple of tapes before our cassette player broke. I have also seen High Fidelity a bunch of times. I’m not by any means a playlist prodigy—if you asked me to DJ at a frat party, I’d be completely out of my depth—but I know enough about playlists to write coherently about them.
My main intention is to counterpose the playlist to the algorithm, and my basic thesis is that you should stop relying on algorithms and start making your own playlists.2 By “algorithm,” I mean music-discovery algorithms like Pandora and Spotify Radio as well as algorithmically-generated playlists like Spotify’s Discover Weekly. Similarly, by “playlist,” I mean custom playlists created by individual people, and not the “editorial” playlists that are technically curated by human beings but whose creators rely heavily on the same kinds of listener data that algorithms also use. You should not rely on those, either; they lead only to sameness and staleness.3
I’m not totally opposed to the use of algorithms to find new music. (Or, rather “new” music—the scare quotes will make sense in a moment.) On the other hand, if you’re going to use them, understand that they can only do one of three things:
Show you more of what you already like. If I click on a Nineties Alternative playlist, for example, I will probably hear The Bends-era Radiohead, Dinosaur Jr., Yo La Tengo, and the Pixies, all of whom I already like.
Show you a progressively more cartoonish version of what you already like.4 If you like Lorde’s first album, you’ll hear a succession of young female singers who have ersatz New Zealand accents despite being from, like, Simi Valley. If you like Seventies rock, you’ll hear various contemporary groups with just as much bombast and coked-out pseudolyricism as the groups they’re imitating but absolutely none of the charm or charisma. Et cetera.
Attempt to throw a few curveballs at you. As an experiment, I created two Spotify radio stations based off of Big Thief’s “Shark Smile” and Nick Lowe’s “I Love the Sound of Breaking Glass.” I noticed that the former included the Replacements’ “Swingin Party” and the latter included Big Star’s “September Gurls.” Ha! the algorithm seemed to say. You thought you knew me, but I know you too well, you pathetic, predictable little man. But I suspected that something else was going on. I did, after all, have the Big Thief and Replacements songs on a playlist together, as well as the Nick Lowe and Big Star songs. So I created a burner Spotify account and started identical radio stations. As expected, Big Star was nowhere to be found on the Nick Lowe station, and the Replacements were nowhere to be found on the Big Thief station.
In short, the adage that algorithms only know how to show us what we already like turns out to be correct. The algorithm might do a better job of this when it has more data on what we like, but it’s still just showing us what we already like. Even when it outright invents new subgenres, it hasn’t truly invented anything.5
We can see, then, that the problems with media algorithms are twofold. For one, as far as the algorithm is concerned, you are a fixed, static being. When it shows you new music, it assumes the boundaries of your taste are permanently frozen. This is a bad mindset for anyone to have, but it’s particularly damaging for younger people. The algorithm makes no distinction between a sixty-five-year-old who’s been listening to the same ten albums since college and a thirteen-year-old wrong-generation type who could realize the errors of his ways with a little gentle prodding. (I used masculine pronouns just now because that thirteen-year-old was me.)
Similarly, in the eyes of the algorithm, you are only one thing at once. You are a fixed, static being. You can be a Belle and Sebastian fan or a Gang of Four fan, but never both at once. The algorithm would never think to put “Damaged Goods” and “Get Me Away From Here, I’m Dying” on the same playlist.6 My father and I used to act like we had reached some advanced stage of humanity for liking many different styles of music, but I’ve since concluded that this sort of hybridity is the default state of humanity. There is no law of nature that prohibits me from putting Van Morrison and 100 gecs, Stereolab and Ski Mask the Slump God, Steely Dan and Sweat on the same playlist. If this seems like an anti-climactic statement, good. It should be. I’m sure plenty of people like all six of these artists. But as far as the algorithm is concerned, they don’t exist. As we saw earlier, it has to learn that such persons exist.
This is supposed to be the part where I tell you how to make playlists. To tell the truth, I don’t have any magic formula, and I don’t think one exists. I just throw together a bunch of stuff that I like. I have one main playlist called “Current” that I periodically add songs to and take songs away from, and then various others based around specific themes. I usually make them for an audience of myself, though I’ve made a few with specific audiences in mind. I don’t spend too much time worrying about order unless I know for a fact that the listener will be hearing it from start to finish.
But before you can make a playlist, you need to find stuff to put on it. You can rely on algorithms, but frankly, once the algorithms get to know you better, there’s no point in making your own playlists. They’ll be indistinguishable from whatever the algorithm spits out. No: you need to make a real effort to find new music.
One way to do this—I don’t know if it’s the best, but it’s what I did—is to regularly read music reviews. Start with a music news source or forum that disproportionately covers the stuff you already like but is not limited to one specific genre. If you already like, say, Russian hardbass, don’t limit yourself to LegitHardbassUpdates.ru7—maybe branch out to other Slavic dance music. See what the scene in Slovenia is like.8
Let me clarify what I mean by “disproportionately.” As of this writing, Olivia Rodrigo has 44,949,598 monthly Spotify listeners; the U.K. post-punk group Dry Cleaning has 284,720. If we pretend for a moment that Spotify listeners are a suitable proxy for the world’s music listeners, this means that from a perfectly neutral music-news outlet, Olivia Rodrigo would receive about 158 times as much coverage as Dry Cleaning. In other words, its Olivia Rodrigo to Dry Cleaning ratio (hereafter OR:DC ratio) would be 158:1. The idea, then, is that if you’re a fan of Olivia Rodrigo, you’ll want to find outlets where the OR:DC ratio is even higher than 158:1; if you’re a fan of Dry Cleaning, you’ll want a lower OR:DC ratio. You don’t need to do any real math, but you get the point.
Once you’ve found an outlet or two, read it regularly. Listen to anything that seems even faintly interesting. If you see any references to artists or albums you haven’t heard of, look them up and listen to them. “No, but I’ve heard of them” does not count. Let yourself fall down Wikipedia rabbit holes. I don’t even want to tell you how many bands I found out about just from trying to understand the references that other people made.9 Yes, other people, not just music critics; I first listened to 100 gecs because I wanted to understand all the Twitter memes I was seeing.
Again, that’s just what I did. The old way to do it, by the way, was to follow your local scene if you lived in a major city and/or to make friends with cool people who knew more about music you did. Both of these options are still productive. Less productive, however, is radio. Any retail or service worker forced to listen to FM radio all day can tell you that playlists and formats are pretty strict and programmatic, but it wasn’t uncommon for some rock stations in the 1970s and 1980s to play groups like the Clash alongside arena rock. A few stations have at least tried to be eclectic, but as a general rule, if the station is corporate-owned, it won’t have much variety.10
No matter what method you use, the key is to avoid limiting yourself to one genre or style. Once you’ve expanded your horizons, the next steps are easy:
Make a new playlist.
Put songs you like on it.
Don’t put songs you dislike on it.
Any questions? I hope not.
Legal disclaimer
The views and opinions expressed in this article are those of the bylined author. They do not represent the views or opinions of people, institutions, or organizations with which the author may or may not be associated in a personal, professional, or educational capacity.
See:
Matt Bruenig, “What Is Lost in Post-Scarcity?” Matt Bruenig Dot Com, May 12, 2021, https://mattbruenig.com/2021/05/12/what-is-lost-in-post-scarcity/.
Ryan Cooper, “The case for music.gov,” The Week, March 26, 2021, https://theweek.com/articles/973964/case-musicgov.
David Dayen, “Islands in the Stream,” The American Prospect, March 22, 2021, https://prospect.org/power/islands-in-the-stream-spotify-youtube-music-monopoly/,
Okay, yeah, sure, I guess you could persuade me that, in the grand scheme of things, there’s no escaping the algorithm, since algorithms pervade every aspect of contemporary life. Wrong life cannot be lived rightly, et cetera, et cetera.* I know. I know.
(If you are a normal, healthy person you can stop reading here.)
Guy Debord, of Society of the Spectacle fame, advocated, as a revolutionary strategy, just sort of moving around a city with no real goal. He called it the dérive, meaning “drift”: “In a dérive one or more persons during a certain period drop their relations, their work and leisure activities, and all their other usual motives for movement and action, and let themselves be drawn by the attractions of the terrain and the encounters they find there.”** My takeaway from this is that you can at least kind of resist the machine if you use it in a way that it isn’t supposed to be used, at least until the machine finds out about your strategy and makes it into a new means of using the machine.
*Theodor W. Adorno, Minima Moralia: Reflections on a Damaged Life, trans. E.F.N. Jephcott (London and New York: Verso, 2005), p. 39.
**Guy Debord, “Theory of the Dérive,” trans. Ken Knabb, Situationist International Online, accessed December 11, 2021, https://www.cddc.vt.edu/sionline/si/theory.html.
I experienced this at a bar in Paris in 2019. When I walked in, the first song sounded kind of familiar, as did the second and third song. I turned to my brother and said, “I think this is the ‘Housewerk’ playlist on Spotify.” Sure enough, the next song was also on that playlist. Either the bartender had put on that particular playlist or they had relied too much on algorithms to make their own playlists.
The sociologist Zeynep Tufekci documented something very similar on YouTube. See her article “YouTube, the Great Radicalizer” (New York Times, March 10, 2018): “It seems as if you are never ‘hard core’ enough for YouTube’s recommendation algorithm. It promotes, recommends and disseminates videos in a manner that appears to constantly up the stakes. Given its billion or so users, YouTube may be one of the most powerful radicalizing instruments of the 21st century.”
See: Liz Pelly, “Streambait Pop,” The Baffler, December 11, 2018, https://thebaffler.com/downstream/streambait-pop-pelly: “The chill-hits Spotify sound is a product of playlist logic requiring that one song flows seamlessly into the next, a formula that guarantees a greater number of passive streams. It’s music without much risk—it won’t make you change your mind. At times, these whispery, smaller sounds even recall aspects of ASMR, with its performed intimacy and soothing voices. When everyone wants your attention, it makes sense to find reprieve in stuff that requires very little of it, or that might massage your brain a bit. Both traits—its seamlessness and its chillness—reflect music that has become instrumentalized for the platform, whether it resulted from Spotify’s own preferences or the emerging tastes of artists who have developed in its wake.”
Unless, of course, it knew your listening habits well enough, but as I said before, it would still only be showing you what you already like.
Not a real website as of 12/22/21, although hardbass is a real genre.
Compare P.W. Coulson, “On the Question ‘What Should We Read?’” Semi-Public Writings, Substack Inc., October 6, 2021, pwcoulson [dot] substack.com/p/on-the-question-what-should-we-read: “A more up-to-date list [of the references that book reviewers make] can be compiled by regularly reading Bookforum and New York, Los Angeles, and London’s respective Reviews of Books and seeing what authors are most often referenced.”
Just Google “who owns [station].”