Are Algorithms Ruining Music Discovery?

By Elleana Dumper

Image credit: Edu Grande

Image credit: Edu Grande

The future of music is Artificial Intelligence… and other inaccuracies we’ve been told by Spotify.

Some of my earliest experiences with active music discovery trace back to the peak of YouTube tastemaking channels, namely Majestic Casual.  My obsession with this new music prompted me to peel back the layers until I landed in the world of SoundCloud: the most exciting, messy, and anti-mainstream pool of music that I had ever come across.  After spending hours wading amongst the volumes of casual compositions, throwaway tracks, and raw beats, it was always the ultimate accomplishment to find a track which totally spoke to your soul.

For awhile I avoided Spotify, I did not like the idea of a robot dictating my taste.  Every so often I have made attempts to embrace its “lean-back” experience at the chance of finding a gem, to which my Discover Weekly would suggest music which I liked, but it was not mind-blowing.  It was mostly music which was familiar and in the realm of what I already liked. Not overly impressed, I would revert back to doing the dark-corner digging elsewhere - it seemed to be the only way I would find interesting music was not from an algorithm.

Spotify boasts its ability to cast a wider net than humans, with its access to a diverse scope of music from “the smallest, strangest musician in the world, doing something that only 20 people in the world will dig” (Pasick A., 2015).  Yet a 2016 report from Next Big Sound investigated that half of the artists in the world are undiscovered (Next Big Sound, 2016).  So I am stuck here wondering where all of this underground music is hiding, and why does it cease to breakthrough on these algorithms?

If you are looking to discover the new chart-toppers, sure, Spotify will cater to this.  After all, these playlists are backed by the labels with the big bucks (Pelly L., 2016), and this influence will probably allow them to make make it onto any mainstream media shortly after.  But for the independent labels who do not have strength in numbers, this is far from realistic.

Playlists have increasingly become an aspiration for many artists and an influential marker for success; strategies quickly diverting from “how can I get a spot on that radio show?” to “how can I get a spot on that playlist?”.  This digital progression is clear and natural: playlists are the new form of radio. But do algorithms have the capability to take on such an important gatekeeping role?

Jimmy Iovine, legendary producer and major label insider, acknowledges the inability algorithms have in ruling the meaningfulness and relevance that Apple Music’s curation service needs for its listeners, which continues to be led by music experts (Popper B., 2017).

When we listen to a radio show, we place our trust in a single person to take us on a journey through sounds they have encountered and pieced together.  There is value in the human ability to contextualise music, aiming for a more genuine focus on discovering niche music, rather than being a vessel to deliver the “next big hit” to the masses.  Soulection is a radio show which delivers anything you can imagine would fall under its opening tagline “future beats, eclectic soul, forgotten gems, timeless sounds” recited by its host Joe Kay.  Each song, sound, and transition is purposeful and complementary to another, and I might not have resonated with them if they were played in isolation to the show.

Spotify has attempted to create atmosphere with mood-based playlists.  But often in trying to help you discover new music, algorithms will chuck together tracks that miss your taste by a mile, feature on the same compilation but are extremely dissimilar, or sound plain terrible next to one another.  We often anticipate that it will be hit-and-miss, perhaps due to the platform’s skip-happy nature, rather than engaging in the deep, immersive listening that occurs with a thoughtfully-curated radio show, an art and science which the radio industry has developed over time (Popper B., 2017).

It is not only distance between listeners and artists that algorithms have created, but we need to consider the effects that automated discovery has on inspiring and creating interesting music.  Artists are concerned with creating music which sounds the same in hope of being picked up by an algorithm, so it is a fair concern that this tunnel vision of trending sounds could lead to a flattening of creativity.

A skill which remains unique to humans is that they are able to spotlight interesting music which does not get picked up by an algorithm because it is not “popular”.  So its much-needed to encourage diverse music-making.

However critiquing algorithms purely on discovery, Spotify does do a sufficient job at suggesting music to people who are intimidated by the discovery process. Paul Lumere, CEO of Spotify’s key data partner Echo Nest, has identified these people as “indifferents”, following a three-year study which unpacked different listening habits (Dredge S., 2014).  These indifferents make up 40% of listeners (“savants” being the die-hard musical participators on the other end at 10%) and are receptive to the convenience of automated playlists as it gives them a place to begin their personal discovery.  Spotify is where the majority, if not all, of discovery takes place for many people, which puts them and similar platforms in a position of relevance and power.

For me, it is the friction of discovery, which Spotify continually tries to avoid, that I especially enjoy about the process.  Automated discovery will never quite compare to the organic experiences of human connection when journeying alongside the taste of another, picking something based on its cover art, or finding a diamond among multitudes of tracks.  It gives music the permission to be mysterious, and the discovery of it more meaningful.

While algorithms are extremely innovative for sifting through the volumes of music being uploaded online, it has still got a long way to go with delivering the emotion and cohesion humans are insightful of. It is crucial that streaming platforms continue to utilise this and give listeners the depth that algorithms are not currently reaching. But I also think it is the responsibility of people who are really deep into the music to share and talk about underground artists that we like, what is so great about them or why they should be listening. If we do not do that, algorithms will be telling us what to listen to, which ultimately estranges people from connecting with music and each other, and this is not progressive.

Although, I do not think we should completely ignore any of the different ways people are discovering music; whether it is automated or organic.  Every now and then out of pure convenience, I will chuck on a SoundCloud song radio if I am occupied with other tasks, and I am not going to punish myself for it.  As long as new music is being discovered, tastes are being broadened, and artists are getting due appreciation, I can see myself utilising a variety of methods to optimise my discovery experience.  Next time you catch me on Discover Weekly, I am listening because I want to, not because it told me to.


Dredge S. (2014). What's next for Spotify? Perhaps going 'beyond the Play button'. Retrieved from

Dredge S. (2014). The five types of music discovery. Retrieved from

Next Big Sound. (2016). The Taxonomy for Artists. Retrieved from

Pasick A. (2015). The magic that makes Spotify’s Discover Weekly playlists so damn good. Retrieved from

Pelly L. (2017). The Problem With Muzak. Retrieved from

Popper B. (n.d.) Tastemaker: How Spotify’s Discover Weekly cracked human curation at internet scale. Retrieved from

Sabine N., Price B., Hu C. (Panel). (2018, November 9). The Future of Music Discovery [Audio Podcast]. Retrieved from