Below is the piece that serves as the basis for my thesis:
Meet the New Boss, Same as the Old Boss
What is the number one most important aspect of becoming a successful musician? The instinctual answer to this question might be practice, honing your skill, writing good music, etc. and though that very well may be in a perfect world, that is not the world that we live in. The most important aspect to becoming a successful musician is distribution, because one could be the greatest artist in the world, but with no means to get said art into listeners’ hands, the music is worth nothing more than personal fulfillment. So how does one go about distributing their music, marketing themselves, and getting into the hands of potential fans? Well for most of the history of recorded music this pathway was relatively clear-cut if exclusionary— form a band or go solo, play gigs, get scouted by a record company, get signed, let them market you and pray to God that you have a radio hit before their marketing models tell them that you’re not worth the money. As great of a system that this was for the minority of artists this worked for, for the much larger group of artists that never got signed, they were left by the wayside with no hope for mainstream success.
This all changed in the mid-aughts when the first social media and music streaming services arose, ones where anyone with a computer and an internet connection could share their every thought, picture, and song with the whole world. When this happened was the beginning of the end for the studio monopoly in which artists get to be heard, a new age was heralded in by the YouTubes, Spotifys, and TikToks of the world. Instead of who and what we listened to being decided by some faceless capitalist overseer on the top floor of Capitol Records, it could be decided by us, the consumer, with our clicks and our views. Many thought this was the beginning of true musical democracy. Industry analyst Tatiana Cirisano explains our current moment as “an era where audiences are choosing what they want to hear, and record labels and the rest of the music industry are listening to that," instead of the other way around for the first time ever.
This streaming utopia sounds like a thing of beauty—any bright-eyed teen in their bedroom with the iPhone Garageband app and maybe a midi keyboard if they were feeling really fancy could now make it big. No industry bias, no more outsiders, this was a system based on talent rather than luck or connections. A perfect world, but sadly once more, not the world we live in. For the record labels were not triumphantly replaced by democracy, but rather another institution with the same cold, cutthroat attitude: the algorithm.
Though AI and algorithms are all the buzz right now, these concepts are nothing new. Before ChatGPT was even a little bundle of 0s and 1s in its mothers womb, AI was already taking over the way we interact with technology. Anyone who watched YouTube in the mid-2010s knows that the eternal plight of the YouTuber was making content that hit the enigmatic boxes of the algorithm which would propel it to the front page and thus instant virality. Ten years later and this is what has happened with music. Spotify released the statistic that over one third of new artists discovered by listeners are discovered in the “Made for You” section of the service. A system rooted in AI and neural networks analyzing all music in Spotify in order to quantify its emotion and genre in order to more aptly recommend it to its user base. A similar system is in place on websites such as YouTube (which despite paying the lowest amount per song streamed, is still the number one platform people consume music on. Services like Tidal tried to remedy this and for all 12 users of the platform, if you’re reading this I commend you). One song or video that the algorithm picks up and decides to boost to everyone can make a new artist. This happened with Lil Nas X in the mainstream, and Boy Pablo or Clairo on the indie side.
The algorithms can also rejuvenate old music that wasn’t even popular at its release from artists that are no longer even together. There are fascinating cases such Pavement’s Harness Your Hopes or Galaxie 500’s Strange, both of which were merely unknown B-sides at their release decades ago, but appealed to the Spotify algorithm in just the right way to spur recommendation to a certain brand of listeners, whose listens and subsequent adds to their playlist would spur the algorithm to pedal the song even more to more listeners who continued this positive feedback cycle until both of these songs are now the artists’ top streamed by large margins. This all sounds great, but the million dollar question is how can artists use this to their advantage? And here’s the two million dollar answer: no one really knows.
Algorithms are a black box. Not only are these companies extremely private about the actual software behind their programs (threatening cease-and-desists to those who pry too closely), but due to the fact that these algorithms are based in AI and neural nets—meaning they’re constantly learning and growing based on the new information they receive—even their engineers cannot truly understand their inner workings. They can make guesses—Spotify Data Alchemist Glenn McDonald hypothesized the Pavement case to be an unexpected result of a new autoplay feature and a lucky hyper-specific genre classification, but no one actually knows. Regardless, whatever the true answer may be does not matter as much as the clear fact that it is not really the listener who is determining a hit nowadays, but the algorithm
So what do artists do with this information? They use it to grow. A Google search of how to become a musician no longer details how to acquire a record deal, but rather countless tutorials claiming to know the inner secrets of the algorithms and how to market yourself as efficiently as possible on the apps that matter— TikTok, Youtube, and Spotify—because even though we don't understand how these algorithms work exactly, we understand what often correlates with success. Longer videos, high engagement rates, duets on TikTok, sharing with friends, etc. all seems to boost the content in the algorithm, so what used to be silly little things like getting a song associated with a certain TikTok challenge or dance is now a legitimate marketing tool that can expose your music to millions of people.
The whole business of music marketing now is playing to the algorithm, not even simply for small artists trying to find an audience. Big artists with multi-million dollar record deals such as Charli XCX or Halsey are complaining about being prodded by their labels to make TikToks and other social media content. Even the old bosses are submitting to the new. Which brings up the moral grey area of this entire system—should individual companies, and more specifically their small groups of engineers have sole control over sweeping parts of the music industry?
In addition to this, for as cold and calculated as computer programs are supposed to be, human bias inevitably works its way in. It has already been established that the algorithm is completely capable of making hits, and if there’s a way to monetize that at the cost of the artist, Spotify will find it. In 2020 they rolled out a service for artists wherein the artist could choose a song to be prioritized in the algorithm in return for what they, in their coyly sanitized millennial marketing jargon, named “a promotional recording royalty rate.” This concept which they pitched as egalitarian, allowing artists of all sizes to be able to use this feature without having to pay huge sums, is really just an excuse to lessen the already meager portion of revenue that the musician gets from the streaming service in return for a slightly greater chance at success.
There are many possible dangers in weaponizing the algorithm like this, but before they can even be identified and regulated, we need to evaluate one major question: what truly is the algorithm and what is our relationship to it?
Everyone in the tutorials, industry analysts, and even the musicians themselves seem to regard the algorithm as something that operates independently, choosing the content that it likes the best and spitting it out for us to consume, but the line between us and the algorithm is not so clear cut. Chris Robley from DIY Musician in his Spotify algorithm breakdown in response to the question of if one should make music to appease the algorithm responds with the assertion that “the algorithm is the audience.” The algorithm does not exist in isolation, it is, at its heart, a reflection of what we listen to. It studies our habits and preferences and creates musical models of all of us of which it tries to find music to fit into. We, again through our clicks and streams, create the algorithm. So if the algorithm is only choosing music based on what we listen to, but we’re predominantly listening to the music that the algorithm feeds us, who, if anyone, is really choosing what we consume?