Algorithm and blues: Why Discover Weekly might not be showing you anything new

coloredited_ameenagolding_spotify
Ameena Golding/Staff

Related Posts

A question that nobody really knows how to answer: “What kind of music are you into?” Describing the album that changed your life, trying to explain why a new discovery has hooked you or defending a guilty pleasure off the cuff can feel impossible — the reasons why we like certain kinds of music are elusive, personal and complex. In these situations, it can seem like you’re supposed to explain your life story with the casual brevity of a passing comment on the weather.  

Unfortunately for the sentimental music listener, predicting taste is a lot faster and easier than ever before. Anybody who uses the Discover Weekly feature on Spotify will be familiar with the bizarre feeling of not only having their diary read out loud, but also seeing a few new pages added by a stranger. Robots might not be able to stage a cowboy revolution just yet, but they can definitely make a killer playlist.

There are other music services out there: Pandora, Apple Music, Tidal, to name a few. What sets Spotify apart from the competition is its uniquely accurate recommendation system. The app uses a process called collaborative filtering to pull from a massive amount of data to determine what listeners want to hear. Essentially, Spotify takes the songs that listeners repeatedly play or thumbs-up and matches that information to that of other listeners around the world who have the same tastes. Odds are, you’ll probably like your internet soul mate’s other music selections, and vice versa.

Hit-making and predicting the next Billboard No. 1, formerly the dominion of industry tastemakers and executives, has now been reduced down to a science. Artists who end up under Spotify’s “RapCaviar” and “Today’s Top Hits” have already filtered through multiple other smaller lists. This process is called “playlisting,” and the data collected through essentially test-running songs on smaller, niche playlists all but guarantee that these songs will go viral.

Even stranger are Spotify’s mood playlists — the ones specifically designed for sleep, study or “chill.” In 2016, Music Business Worldwide discovered that Spotify was slipping its own music into “Peaceful Piano” and similar sets; producers are paid a flat fee to make ambient sounds under pseudonyms such as Risto Carto or Hiroshi Yamazaki. It wasn’t immediately obvious to listeners because, as Alexis Petridis wrote for the Guardian, you don’t really hear anybody announcing, “You know what my favourite kind of music is? Piano in the background.” Songs that vaguely sound like “Clair de Lune” might be nice to fall asleep or study to, but they don’t exactly inspire the same kind of engagement as our other favorite artists.

What does it say about the role music plays in our lives that “Deep Sleep” encourages entire segments of listeners to use Spotify to fall asleep? The push to drive constant engagement with the app turns music into vague sounds and notes that simply facilitate other activities. Philosopher Simon Critchley likes that Discover Weekly helped him find new artists but lamented the background role music was starting to play, saying: “It’s wallpaper, though. They’re making it wallpaper.”

Data collection has lifted a veil of mystery from the industry. Music, once all spontaneity and guesswork, dusty crates and chance encounters, has been refined to simple mathematics.

We have unprecedented access to media from all over the world, but music recommendation algorithms only give us songs that fit in with what we already know. With an immeasurably vast musical library at our fingertips, we have the ability to listen to an endless amount of stuff that sounds like the stuff we already like, forever.

There’s music curated to put you to sleep. There’s music engineered to not even be noticed. Don’t think too hard about the fact that the company hired François Pachet, known for developing pop-song-composing artificial intelligence, as director of its Creator Technology Research Lab — the idea of robots writing the next smash hit might make even Kraftwerk nervous.

Leaving the record store basement for the open-plan office might sound like a hipster’s worst nightmare. However, music and AI isn’t so much an unholy matrimony as it is an odd couple.

In an interview at Google, music critic Ben Ratliff recalled explaining to a colleague how before the internet, having to get an obscure album sent to him from some far-flung part of the world somehow made it sound better to him. His colleague responded by asking if that meant he also manually drew water from a well every time he needed to drink.

The record stores are still there. The books, articles, liner notes, cassette tapes, lovingly organized collections are all still there for those curious enthusiasts who have a couple hours to spare. For those who don’t, we have a convenient resource for finding easy listening without the legwork.  

Spotify might provide you hundreds of hours of Chill Relaxing Piano Music to Gently Slip Into A Coma To, but the app’s follow capability can also connect you to other people. Maybe you’ll make some great playlists yourself or have an interesting conversation about a band you saw a friend listening to. Introducing technology into our music doesn’t have to fundamentally change our relationship with the medium. Rather than seeing Spotify’s algorithms as a replacement for the traditional ways of listening and searching, we can see it as a supplement — as long as we remember to occasionally look for new music that pushes us outside of our comfort zones.

Contact Jasmine Garnett at [email protected].