Music journalist Liz Pelly has spent most of her professional life working and living in indie music circles, which gives her a distinct and important perspective on the consolidation of the music industry around just a few streaming services, with Spotify foremost among them. In her book Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlist, she explores the birth of Spotify, how it took over the streaming space, and how pernicious its effects have been on indie musicians and labels – as well as listeners who want to hear anything but the most anodyne music available.

Spotify began as an idea to sell ads, not as a way to save the music industry, or even just as a way for its founders, including the military-minded Daniel Ek, to get rich while working in music. Songs were just a vehicle for selling ads and collecting user data to do so more effectively. Ek and other Spotify execs have retconned their history to make Spotify out to be some kind of savior for music and musicians, even as their practices have proven predatory, including bringing back payola, and they’ve cracked down harder on musicians with tiny followings than they have on scammers posting fake music tracks and clips to try to deke Spotify’s algorithms and rack up streams from listeners who aren’t paying much attention to what’s playing.

Pelly spoke to over a hundred current and former Spotify employees, got access to internal Slack logs, and interviewed musicians, label executives, and others to research Mood Machine, and two consistent themes emerged (among many smaller topics). One is that Spotify’s entire structure, built largely around algorithmic playlists, which are made by machine-learning systems and are distinct from curated ones, assembled by humans, like the ones I post every month, is both corrupt and tends to funnel listeners to a small number of artists from major labels, while paying artists from outside of the big 3 labels a pittance (and paying no artists anything close to fair value). The second is that a huge portion of Spotify’s audience isn’t actually listening to the music: One of their main goals is to get listeners to stick on a playlist as some sort of background noise – similar to Netlix’s “second-screen” nonsense – and let it play for hours and hours, and if it runs out let Spotify’s algorithm just keep playing more songs, so they can serve more ads. Playing a song costs Spotify less than half a penny in royalties. Their dream scenario is The Lost Weekend except with some algorithmic chillwave playlist going in the background for 72 hours.

Harper’s excerpted a portion of Mood Machine in 2024 as a story called “The Ghosts in the Machine.” It exposed Spotify’s program to commission fake tracks from outside vendors under its “Perfect Fit Content” program, where those companies would supply songs by ‘ghost artists’ to pad Spotify’s playlists for deeply discounted royalty rates, allowing Spotify to deliver the same quantity of music to listeners for a lower cost. Spotify execs defended the practice, saying that listeners of those playlists were “half-listening anyway,” so why deliver them real music? Pelly spoke to a number of employees dismayed by the practice, saying it was unethical to deceive listeners and harmful to the music industry they were supposedly supporting. The few editors remaining who curated playlists themselves reported increased pressure to use these ghost artists on their playlists, even though that was antithetical to their mission (and, I would presume, their music fandom). Some of the companies Pelly cites as ghost-artist producers are now getting into AI-generated tracks, such as Epidemic Sound, which also sells royalty-free background music to video producers. It’s the most interesting chapter in a book that’s full of them, and the most overt explanation of why Spotify is harmful to the music industry.

Pelly wraps up with a chapter (not the epilogue) about how listeners can support artists better in the streaming era. The obvious answer – and I’m not criticizing Pelly here – is to buy music. You love an album? Pay for it, whether digitally or physically. Buy it direct from the label or from the artist. Going to shows and buying merch helps as well. Streaming an artist’s music barely makes a dent; you’d have to stream a song around 300 times to get them $1. She points out that while Spotify is generally presumed to have the worst payout rates to artists – no one knows because Spotify’s agreements with major labels are confidential and protected by NDAs, which seems like maybe a restraint of trade, but I’m not a lawyer – the other major streaming sites, like Apple and Amazon, aren’t that much better. I don’t believe she ever said not to stream music, but her message is more like “stream responsibly, and buy liberally.”

The book is so thoroughly researched, with citations as well as quotes from relevant sources, that it can bog down a little in the details, but I will gladly accept that tradeoff to get such an academic take on a popular topic. It is easy to find reasons to hate streaming and find people to tell you they hate streaming. It is a much harder task to explain just why streaming is bad for music, and culture, and explain why it’s also going to be hard to fix the problems streaming, and Spotify in particular, has created. Mood Machine does all of this, without once allowing you to forget that the reason you downloaded Spotify in the first place was because you, in fact, like music.

Next up: Susan Choi’s Flashlight. Dun dun, dun dun dun.