Algorithm Failed Music | Edge

This is it Step backa weekly newsletter covering one essential story from the world of technology. To learn more about how to break free from the algorithm and discover things you love, follow along Terence O’Brien. Step back It arrives in our subscribers’ inboxes at 8 a.m. ET. Subscribe to Step back here.

I have been practicing this ritual. Every Tuesday, I got off the train on Eighth Street on my way home from work. I would like to enter Other musicbuy a new CD (or three…) and then walk the rest of the way to the Staten Island Ferry to listen to my new CD. Even if there wasn’t a new record I was looking forward to that week, I would buy something. Often, if I walked in without something specific in mind, I would consult the pyramid-shaped shelf at the end of the aisle that housed staff picks. I would read the index cards taped to the shelf containing handwritten endorsements from someone who worked there and pick out something that looked interesting.

It may seem like ancient times, but until 2010, most people discovered new music in similar ways: browsing in a record store, friends at school, someone’s cool older sibling, CMJ New Music Monthly CD mix.

This began to change in the 2000s with the emergence of the first algorithmic recommendation engines. Pandora was the big pioneer there The musical genome project. The goal was to break songs down into easily measurable traits like “gender of lead singer, level of distortion on electric guitar, type of background vocals,” and the like. It then searches for other songs that share a certain number of common traits and plays that track.

Pandora enjoyed early success because its algorithmic approach to music recommendations was new at the time. But there were also warning signs of trouble to come. Anyone who has used Pandora in the mid-to-late period will no doubt be familiar with its tendency to replay the same ten or so songs over and over again.

This was partly due to his presence in The forefront of flow While there is a small library. when Pandora filed for an IPO In February 2011, it included about 800,000 songs by 80,000 artists. Compare that to today, when smaller players love it Goz It has more than 100 million tracks.

Just a few months later, in July 2011, Spotify has landed in the US With a catalog of 15 million songs, it changed everything. Pretty much from the word go, Spotify has been like that It’s all on the algorithms. In 2015, Spotify launched perhaps its most famous feature, the Find out weekly Playlist, which provided new algorithmic recommendations Every weekas the name suggests.

Discover Weekly is much more than that sophisticated From the Music Genome Project. It starts by pulling songs from playlists created by its users, then matches them to each individual user’s taste profile, using technology from a company called The Echo Nest that it bought in 2014. It then performs additional fine-tuning and filtering, including machine learning analysis of the raw audio data, before creating a unique playlist of 30 songs.

Spotify is the most popular music streaming service in the world. Although its algorithmic recommendations aren’t necessarily the reason, its ubiquity means that hundreds of millions of people receive a steady diet of machine-curated music. Spotify’s goal is to keep you listening no matter what. In her book Mood machineJournalist Liz Bailey recounts a story told to her by a former Spotify employee in which Daniel Ek said: “Our only competitor is silence.”

According to this employee, Spotify’s leadership did not see itself as a music company, but rather as a time-filler. “The vast majority of music listeners, they’re not really interested in listening to music per se. They just need a soundtrack for a moment in their day,” the employee explained.

Simply providing a soundtrack to your day may seem simple enough, but it shows how Spotify’s algorithm works. Her goal Not to help you discover new musicIts goal is simply to keep you listening for as long as possible. It serves Safest songs It may prevent you from pressing stop.

The company even went so far as to partner with music library services and production companies under a program called Perfectly relevant contentor PFC. This saw the creation of fake artists or “ghosts” who flooded Spotify with songs specifically designed to be enjoyable and impossible to ignore. It’s music as content, not art.

Streaming services also provided record labels with an incredible amount of data about what people were listening to. And in kind of Feedback loopbrands are starting to prioritize artists that sound like what people are already listening to. And what people were listening to was what the algorithm suggested.

Artists are already starting out, especially the new ones trying to break through Change how you compose them To play better in the era of algorithm-driven streaming. Songs became shorter, albums longer, and intros disappeared. The hook was pushed to the front of the song to try to immediately grab listeners’ attention, and things like guitar solos disappeared from pop music. the A palette of sounds The artists being pulled became smaller, and the arrangements became smaller More simplified, Flattened pop music.

In a world where most content is presented to us algorithmically, whether it’s on Spotify, YouTube or TikTok, music discovery has suffered. Market research company Media An alarming study published in September said: “The more users rely on algorithms, the less music they hear.” It found that while discovering new music is traditionally associated with young people, “16- to 24-year-olds are less likely than 25- to 34-year-olds to have discovered an artist they like in the past year.” Gen Z may hear a song they love on TikTok, but they rarely look further to hear more music from the artist.

Algorithm fatigue Been building for some time. Apple has made human curation a central selling point for its music services, enlisting big names like Jimmy Iovine and Zane Lowe. But recently, rebellion against the algorithm has been rising.

Bandcamp Daily It has been a cornerstone of music discovery since 2016, and the site was launched Bandcamp clubs in 2025. They offer one album selection per month, artist interviews, and live listening parties for subscribers. Qobuz has an algorithmic recommendation engine, but focuses more on the editorial side of it Al-Qabooz Magazine.

Generation Z may be less likely to discover a new artist they love than some older generations. But they also lead a A revival in university radio. Terrestrial radio once seemed like a dying format, but many schools now say they don’t have enough slots to accommodate all the aspiring DJs.

Even the iPod has… renaissance. iPod classics They fetched hundreds of dollars on eBay, and an entire, albeit small, subculture sprang up around modifying them to extend battery life, increase storage capacity, and add modern conveniences like Bluetooth and USB-C.

At this stage, Anti-algorithm He is itself that complete Type to content. Especially on YouTube, where creators are making videos about getting rid of live streaming, stopping destructive scrolling, and how the algorithm has flattened culture.

Of course, once something becomes a trend, it’s only a matter of time before companies start trying to figure out how to capitalize on it. Features To try and address Complaints About her Algorithmincluding the ability to exclude songs from your taste profile. But it also introduced something new Human treatment Features.

It’s likely that more companies will start offering alternative methods as algorithms become more cumbersome. But eventually, companies will figure out how to create the illusion of serendipitous discovery. They will make algorithmic recommendations, but combine them in a way that feels more natural.

It’s not hard to imagine a future where human-operated playlists are ostensibly algorithmically designed to exclude songs that don’t exactly match your listening history. Or where algorithmic recommendations are subtly placed in easy-to-discover places, making you feel like you’ve found a new record on your own. You will still be manipulated by the algorithm; It will be difficult to detect.