The Playlist Middleman Problem: How Algorithms Became the New A&R

The Curator's Disappearing Act

For most of recorded music history, there was a human being at the end of the tastemaking chain: a radio DJ, a MTV programmer, a record-store clerk, or a critic whose opinion could break or bury a song. That person had skin in the game—their reputation rode on whether they picked winners. They also had limits: they could only play so many songs in a day.

Now, Spotify's Discover Weekly reaches 40 million people. Apple Music's Mood playlists are generated by algorithm. YouTube's recommendation engine serves billions of hours of video a month. And while these systems can be audited and their mechanics partly understood, they are not people. They have no reputation to risk, no listening history that began in a record store or a listening booth. They optimize for engagement, retention, and the vague directive to "give users what they want"—which often means what they already like, just slightly adjacent.

The result is a new kind of gatekeeping: one that's distributed across millions of playlists, invisible in its decision-making, and vastly more powerful than any single curator was ever allowed to be.

The New Logistics of a Hit

This shift has rewritten how music gets marketed. A decade ago, a label would push a single to radio programmers, music journalists, and tastemakers. There was a campaign, a strategy, a human ask. Today, the conversation has moved upstream—or rather, sideways—to the platforms themselves.

Artists and labels now obsess over playlist placement, algorithmic seed data, and metrics that Spotify and Apple don't fully disclose. Some labels have hired data analysts whose job is essentially to reverse-engineer platform algorithms: studying which songs get playlisted, at what tempo, with which neighboring genres, in what mood category. It's become a technical discipline, almost scientific in its precision and frustration.

But unlike radio, where payola scandals were at least prosecutable, the rules here are opaque. Spotify pays out royalties based partly on playlist placement and stream count, which means the algorithm indirectly controls artist income. Apple Music has different economics. YouTube has yet another. An artist can be mathematically "heard" by millions and still make no money—or make money that's so fragmented across platforms and weighted toward one or two that it's barely livable.

The platforms argue they're democratizing discovery. Artists can theoretically reach anyone without a label blessing. But in practice, algorithms reward certain patterns: high release velocity, playlist-friendly song structures, audio that matches the acoustic profile of existing hits in a given mood category. It's like if radio had become invisible but amplified a thousand-fold, and the program director was a machine learning model trained on five years of engagement data.

Who's Actually Getting Heard

The effect on artist development has been measurable and strange. Breakout hits now often come from algorithmic playlists rather than traditional marketing. That's genuinely democratic in some cases. But the algorithm also tends to consolidate: songs that get early algorithmic traction get more visibility, which generates more streams, which feeds the algorithm more data confirming they're "good," which leads to more placement.

Emerging artists caught in this machine report a different kind of pressure than previous eras faced. There's no single gatekeeper to convince—there's no specific person to email, call, or build a relationship with. Instead, there's a massive, probabilistic system that's partly transparent (you can see your metrics) and partly black box (you can't know exactly why one song charted and another didn't).

Interestingly, this has also accelerated the rise of genre hybridity and micro-genres. If playlists can be infinitely granular—"Sad Night Driving," "Hyperpop for Gamers," "Indie Bedroom Funk"—then an artist doesn't need to fit neatly into one category to get algorithmic traction. They just need to fit a recognizable mood or sonic profile well enough for the system to slot them in.

The Playlist as Product

For platforms, algorithmic playlists are a retention tool, not a revenue center. Spotify doesn't charge users extra for Discover Weekly. But the more time users spend in Spotify discovering new music, the less likely they are to leave the service. That means the algorithm is built to keep you in the platform, not to maximize fairness or art in any traditional sense.

This creates a peculiar incentive: the platforms benefit from diversity of content (to satisfy varied tastes) but also from homogeneity within mood categories (to keep recommendations coherent and "clean"). The result is a kind of algorithmic balkanization: specific sounds dominate specific moods, and deviations from that profile, no matter how innovative, can struggle to get seen.

For the industry—labels, distributors, promoters, and artists—the question is whether this new gatekeeper is better or worse than the one it replaced. The answer, unsatisfying as it is, seems to be: it's different, and it's in control of more of the listening economy than anyone individual ever was.