The reason artists on these playlists feel forgettable might be because some don’t actually exist. An investigation by Swedish newspaper Dagens Nyheter identified hundreds of “fake artists” that had been featured in Spotify-curated playlists. (When I asked about this, Spotify declined to comment.) Meanwhile, label executives gripe that playlists are pumped with generic tracks because streaming platforms get a higher share of the profits when labels aren’t involved. Right now, every stream is ascribed the same value to be split between the platform and those who provided the track, whether a user listens to rain music or the Rolling Stones.

Record label Universal is fighting the sludge by teaming up with French streaming service Deezer to explore new “artist-centric” business models. Deezer’s CEO, Jeronimo Folgueira, tells me he’s thinking about a tiered system where different types of music generate different value—rain sounds might be cheaper than a top artist—but that’s not an easy thing to do when there are 100 million tracks on your platform. He agrees streaming has overseen a shift away from engaged listening. “We're consuming so much music all the time that music is now becoming kind of background noise,” he tells me. And he thinks AI-generated music is going to make that problem worse—potentially flooding platforms like his with low- quality content. “We cannot pay to store hundreds of millions of tracks that no one really cares about.”

Twenty years ago, the iTunes Music Store reshaped how we listen, but we have now arrived at another turning point blurring the definition of music. AI threatens to turbocharge sludge, making it easier than ever to produce. But if the people want sludge, why not give them sludge? That’s the argument of entrepreneur Oleg Stavitsky, who points out that “functional” sounds—designed for focus or sleeping—always feature among streaming platforms’ most popular playlists. He’s one of six cofounders of Endel, a slick Berlin-based app that uses AI to generate one endless piece of music that adapts to your surroundings. When I use it to soundtrack my daily dog walk, the app uses the accelerometers in my phone to generate music that falls in time with my footsteps. If I start skipping or jogging, the music’s tempo catches up. For Stavitsky, functional sound—he’s careful not to call it music—can co-exist with human-made compositions. He believes people want both.

That sounds harmonious. But realistically, how much time for music does a person have? If Stavitsky’s functional music becomes popular (already he claims 2 million monthly listeners) it will inevitably eat into time people spend listening to actual music. I did try to write this article to Endel’s “focus” mode, which overlaid endless chimes on top of thumping bass. But to me, it felt like an intensified version of the sludge I’ve been rebelling against. So I quickly ended up back on Spotify, ignoring that prominent playlist button and listening to the British electronic artist Burial. I’m not against the idea of functional music, but for me, it still has to be music.

Today, the Pirate Bay reckoning feels like a distant part of tech and music industry history. But in 2009, WIRED special correspondent Oscar Swartz captured the frenzied interest around the founders’ trial in Stockholm, where the prosecutor was forced to grapple with the new frontiers of copyright in front of an enraptured, mostly pro–Pirate Bay, audience.

Today I bought a scalped seat in a court of justice for the first time in my life. The Pirate Bay courtroom was full. But locally famous Swedish author Sigge Eklund whispered in my ear that he had an offer for three seats after the intermission. The price would be about $60 each. We met in a corner of the hallway and finalized the transaction with three young men who were leaving. I was in! This pretty much captures the rock-star quality of The Pirate Bay trial.