What are the dangers of algorithm-driven music curation?

Earlier this year, Agoria (a.k.a Sebastien Devaud) released Drift, his first album in eight years, and something unlike he’s shared before. Across 10 tracks, the French artist explored a novel musical approach; it’s described as an expression of freedom through music, and a bridge that links territories, all the while creating new frontiers and inventing new codes. It includes collaborations with Phoebe Killdeer, Nicolas Becker (Hollywood sound designer who worked on the Oscar-winning movie “Gravity”), Sacha Rudy (an 18-year-old boy from Paris), STS (the golden voice of L.A.), NOEMIE, for her first track ever, and French hero Jacques. 

The album arrives on the back of a 20-year career, beginning with a series of 12”s on Pias Recordings, which were followed by his debut album, Blossom, a year later. Since then, Agoria has released three more long-players, The Green Armchair (2006), the soundtrack to Go Fast (2008), and Impermanence (2011), all of which showcase his skills in creating deep, stripped-back, melodic techno tracks. Through his labels, Infiné and Sapiens, and releases, which have also appeared on Innervisions, Kompakt, and Hotflush, Agoria likes to challenge his listeners, creating music that refuses to conform to contemporary expectations and labels. 

The backdrop for Drift is a musical landscape that Agoria feels is becoming static. To elaborate on these ideas, and to better explain the inspiration behind the album, Agoria penned an essay for XLR8R’s Real Talk series. 

We live in a world filled with music. Headphones, mobile devices, smart speakers, hotels, subways, airports—we are listening to more music than ever before. But to what, and to whom, are we listening? 

Increasingly, streaming is the musical consumption medium of choice, and these platforms and their mechanized suggestions provide us with an unmatched treasure trove of musical knowledge, not to mention an opportunity to discover new work from artists that we would never have previously found. 

But this also means that our listening habits are increasingly being tracked and then shaped by artificial intelligence, as scary as this may sound. As a result, more and more music is being consigned to boxes or playlist themes, and so the flip side to “discovery” is that we’re being given more of what ticks the same boxes as what we’ve been listening to; more of what we like; or at least more of what the algorithm thinks we like. Is this really discovery at all? 

While some artists win—mainly those who conform to genres—those who fail to conform, whose work is not easy to define or pigeonhole, lose out, because their music is not being presented to new listeners; it’s simply not being discovered. 

This is already having severe consequences on our wider musical landscape: in a fight to survive, some artists are making music with the sole intent of being picked up by an algorithm or playlist rather than because it’s innovative or interesting. Creativity and innovation are in danger of being supplanted by conformity, and new ideas lost to monotony. What we’re seeing is homogenization. 

"...if today’s barometer for success is only quantified in the black and white data of streaming numbers, we risk losing the magical artists that exist in the grey, and music will become dry and, well, boring."

Artists inherently don’t fit in one box. They need and should want to push boundaries and refute definition; and those who refuse to be codified are typically the most rewarding and exciting. Whether its Nicolas Becker’s sound installations for Philippe Parreno, the provocative poise of Arca, or the precocious interpretations of the 18-year-old composer Zer0, we rely on artists like these to push music forward; to present new ideas away from the norm; to blur the line between the boxes. Following a formula and producing a certain sound or style may yield sought-after streaming numbers, but it comes at a considerable cost; if today’s barometer for success is only quantified in the black and white data of streaming numbers, we risk losing the magical artists that exist in the grey, and music will become dry and, well, boring. 

It’s often been said that music would be killed by machines, by the synths and keyboards of electronic music, but electronic music has actually enriched the musical landscape. However, machines are now posing a threat in a very different sense, and so now we must do something about it before it’s too late. 

We need to create an environment or infrastructure that supports these artists. As listeners, we must ensure that they are supported; and as artists, we must continue to fight to make music that challenges; to resist the temptation to make music that abides by formulas. An album might seem indulgent in today’s world, but we have a duty to produce bodies of work that aren’t aimed squarely at certain playlists, but tell stories and ask questions. The music community will be considerably worse off without these bodies of work. 


The good news is that it’s not too late. So as an audience, we have to exercise our power and not allow our listening habits to be lulled by repetition. Remind yourself that we don’t want music to become boring; we don’t want our ears to be filled with the Spotify-friendly drab that’s becoming so common. 

"Curation is a human quality that must not be overlooked in the algorithmic age."

None of this is to say that playlists are the enemy, for they too can be thoughtful and provocative. Much like an eclectic DJ mix, a well-curated playlist can stitch together tracks that don’t appear to belong together and ask questions of its audience. But as listeners we must be wary of simply allowing our music to be programmed by mood or genre; two hours can easily pass without realizing whether two tracks or 20 have been heard. Streaming platforms have a vested interest in plays, and time spent streaming, but they also hold huge curatorial power. We must not be lazy, unconscious listeners, and we must not allow curators to encourage lazy listening. There are humans on both sides of the machine and we must behave as such. Curation is a human quality that must not be overlooked in the algorithmic age.

Exploration and discovery have always been a thrill for music lovers. Whether it was visiting the record store or hearing an amazing new track on your Discover playlist, it is a magical feeling to hear something that grabs hold of you and worms its way inside your head. But today, if we are only listening to what we are being served, does this still count as exploration? If we only hear tracks programmed to match our likes, what are we losing? When I worked at a radio station in my teens, my friends and I tried to find, share, and discover the most obscure records that refused to fit in any boxes. It was not always the song you wanted to hear at that moment, it wasn’t always easy listening, but it made you listen, and you learned as much from a song you hated as one you loved. Listening was not disposable; it was essential and educational, and we must fight to uphold these values today.

I encourage listeners to take advantage of the algorithm, to allow music that fits your taste to be served up to you. But I also suggest that we consciously continue seeking out moments of human curation. Explore a second-hand store bin or market for vinyl that just looks interesting; ask your local record store owner for his or her recommendations; listen to an alternative radio station for a few hours; and spend money going to shows to hear something new and different. All of these will reaffirm how wonderful musical discovery, your own musical discovery, can really be, while nourishing the wider musical landscape. 

The musical opportunity that the algorithm affords us is almost limitless. It can be a tremendous force for good, but only if we use it correctly. If we simply allow our listening habits to be programmed and defined by data, we, in turn, require our artists to conform and risk losing the idiosyncrasy, the challenge, and the humanity of great music. That’s a playlist I never want to listen to.

A human brain requires only 20 watts to fire all its synapses, a computer requires over 20,000,000 watts to yield the same computing power. Whose algorithm do you want to listen to?

Drift LP is available now via Sapiens/Virgin Recordings.