It is 2010 and the music industry is being destroyed. Idealistic music pirates fight for a utopian world where all ideas are free but artists are poor. Greedy music label executives of the big four record labels struggle to keep the industry afloat, squashing the free-loading pirates every chance they get with their version of the Avengers (a team of the world’s most ruthless IP lawyers).This is the plot of The Playlist, a biographical series about Spotify’s righteous battle for existence. The overarching theme in the series is that Spotify acted as a peace broker when tensions between artists, labels, and consumers were reaching a breaking point.
The compromise they depict is good for every party. On Spotify, artists have a platform where they can be paid fairly for digital uploads that are instantly distributed to millions of potential fans, supposedly without needing the support of a record label. Record labels gain a legitimate digital distributor of their product that will expand their reach to a previously unforeseen scale. Users get access to an unimaginably large library of music that makes listening to music as easy as turning on a water faucet and for a good price!
Spoilers: this compromise did not exactly work out as described. Spotify’s payment model compensates rights-holders based on the proportion of all streams on the platform that come from their uploads (explained well here). If a rights-holder’s discography accounts for 0.05% of all streams on the platform for the payment period then they would receive 0.05% of the revenue share pool for that period. How that revenue-share is divided up depends on the structure of the music-rights agreement for the musical work(s) the agreement applies to.
Stream counts as a system, like most count-rank-based systems, is subject to zipf’s law where a small number of types (artists) hold a large majority of the instances (streams). Preferential attachment, referred to colloquially as the phenomena in which "the rich-get-richer", is a common characteristic of these systems where the more instances of a certain type exist, the higher the probability a new instance will be of that type. For streams, this translates to the most popular artists (e.g. Taylor Swift, Drake, and The Weeknd) having stream-shares orders of magnitude larger than their less popular counterparts. Spotify has never denied this. For a long time, they even tried to describe part of their business model as promoting the musical proletariat referred to as "the long tail."
The core message is that by connecting these smaller artists with new listeners via playlists and radios, Spotify is "democratizing" the music industry. With the trusty Spotify Algorithm (which in reality is a myriad of different algorithms), artists don’t need the support of a major label to get promoted. The artist just needs to make good music that resonates with you, the listener, and let the algorithm do the rest.
The reality for the long-tail artist is considerably more bleak than this. Last year, Luminate (a company that has become the industry standard for music streaming metrics, estimated that 86.2% of the 184 million songs they tracked last year received received less than 1,000 streams since they had been uploaded. Some might say that this is simply because there is a lot of bad or even mediocre music out there, that this is simply a "skill issue." But I assure you, this has little to do with an artists musical "skill", and more to do with the economics of streaming.
Let’s say that you had a fantastic 2023 as an artist on Spotify last year bringing in ten million total streams, performing towards the top of the long-tail. You would have made $43,700 from Spotify if you owned your music out-right. This is roughly $30k less than the median household income in 2022. This is how Benn Jordan, a long-time professional musician under the name The Flashbulb, has made more money shorting Spotify on the stock market than he has from his streaming royalties on Spotify.
More surprising about last year is that 45 million of the 184 million tracks recorded had zero plays at all since they had been uploaded. These statistics become even more disheartening in the context of Spotify’s "new feature" added to their payment model. Announced in the Fall of 2023, tracks that have been streamed less than 1,000 times in the last year will be demonetized. Spotify justified this claim by explaining that this feature/choice would help prevent fraud and provide artists with uploads meeting the threshold higher compensation. They also cited the $40 million held up in accounts of smaller rights-holders in which the balances individually do not even meet the transfer minimum required to get money to artists.
Spotify has billed the new feature as a part of an "artist-centric" platform. But the question begs, central to which artists? It certainly does not seem to be the long tail artist. Remember that according to Luminate, 86.2% of music of the 186 million recorded tracks did not meet this threshold. Spotify states that this threshold helps direct funds towards the people that are making the money for the platform - "99.5% of all streams are of tracks that have at least 1,000 annual streams, and each of those tracks will earn more under this policy." And that may be true, considering that Taylor Swift out-streams the entire Jazz and Classical genres.
So let’s say for the sake of simplicity you are a jazz or classical musician who meets the new threshold, the only other artist on Spotify is Taylor Swift, and the payment period is the 24 hours-span known as January 13th, 2024. Taylor has accumulated 79.7M streams today. Let’s generously say that you have been streamed 100,000 times today. If that $40 million gained after the threshold shift is redistributed based on January 13th, Taylor gets $39,950,000, you get $50,000.
For those thinking, ‘$50,000, is still a significant pay increase, I would love that!' Remember, there are only a handful more artists being streamed as much as Taylor. Everyone else is like you. Taylor’s $39,950,000, representing the majority of streams on the platform, gets split by a handful of rights-holders. Your $50,000, representing the portion of streams made up by long-tail artists, gets split by every other rights holder on the platform, nearly a dozen million ways. Spotify’s "new feature" is a pay increase to those who dominate the music industry with major label support and world-wide fame, not those that make up the long tail of the industry. To be clear, no problems with Taylor Swift dominating the streaming charts, you can’t deny greatness.
High counts of low play and no play tracks and a payment model suited for the industry-elite are opposite from what you would expect from a platform in which The Algorithm is prioritizing the promotion of unheard artists. Well, it is most likely because Spotify is not doing that. Instead, Spotify has chosen to monetize it’s high value position in the music distribution chain by allowing artists and labels to pay for promotion on coveted playlists, radios, and personal recommendations.1 And by creating algorithms that maximize platform engagement by recommending you things that are popular and incredibly similar to what you haven chosen to listen to in the past, Spotify can maximize the value of their promotional playlist and recommendation slots.
Spotify is not unique in all of the above behaviors either. Streaming platforms like YouTube (music), Apple Music, Amazon Music, Deezer, and Tidal all follow similar formulas to various degrees of extremity. But that does not mean streaming itself is the problem, though many music enthusiasts see it that way.
All of the people interviewed in this article described how streaming negatively affected their listening experience in a variety of ways. Many of the people interviewed describe a sense that they were simply "using music" when streaming rather than engaging with music. While each person described a different solution, they all seemed to want some sort of connection with the way that they discovered their music.
To say "streaming" is the source of these issues is an over-simplification. Streaming is just an efficient way to deliver the audio. While other mediums add friction which enforces a kind of curation that can be useful to the consumer, this lack of friction is not inherently bad. The real issue is how the platforms offering streaming services have decided to serve music: hyper-personalization.
Hyper-personalization is the idea that your entire experience on a platform is specific to you. It is the core-concept behind the recommendation strategy described earlier to maximize engagement. It is the reason no two users will have the same Spotify homepage. It is the reason your recommended playlists all sound the same, why you get recommended the same twenty-five songs every time a playlist ends. The personalization algorithms are machine learning algorithms with objective functions. While the algorithms may change depending on the consumption mode and context, the abstract objective remains the same - to give you things you like with as high accuracy as possible. Without knowledge of how those recommendations are generated, you become dependent on your music-drip-feed to find what sounds good to you. Hyper-personalization fundamentally disconnects you from the way that you find music.
There is no incentive to show you things you might not like. That would risk you having a bad experience and possibly looking elsewhere, and that would decrease the value of the platform as a distributor of product. You are separated from the contents of your current musical taste and derivative is sold back to you for profit. The more comfortable and reliant you become on these systems, the more valuable the platform becomes. Going back to the analogy Daniel Ek made comparing streaming music to turning on the faucet, you do not expect the water to taste poorly, if it does you filter it or fix the issue.
It ought to be acknowledged that a lot of people for one reason or another find a lot of value in hyper-personalization. It is incredibly convenient for those who just want what they are listening to at the moment to sound good. But for those who have more than a functional relationship with music, it could not be worse. Glenn McDonald, former Spotify Data Alchemist, creator of the beloved everynoise.com, and someone who inspires me, once said "The joy of discovery must be the moment you are in, while you're in it, not a monotonicity of catalogs,". When one logs on to Spotify, all they are met with is a catalog of catalogs, each tailored made to be predictably satisfactory. By providing a hyper-personalized drip-feed with minimal search, no mechanisms for interactive exploration, this joy Glenn describes is removed from the music lovers experience.
For artists of the long-tail, this removes more than just joy from the experience, it has the potential to eliminate income. Without substantial income from streaming revenue, long-tail artists are more reliant on other sources such as concert ticket sales, merchandise, and non-streaming formats. Spotify’s research shows that listeners who feel some sort of connection with an artist are more likely to spend money that flows through those alternative revenue streams via a correlation analysis with artist follows and song likes.
Still, Spotify does not provide fans a meaningful way to engage with the artist or listening community beyond those likes and follows. In fact, 80% of artists on Spotify have less than fifty monthly listeners. Spotify does not facilitate social or community interactions, they have canonically left this to other social-focused platforms. Streaming provides music via faucet, not watering cooler. Without the infrastructure for communities to interact with each other and the artist, communities will struggle to form naturally, and long-tail artists will struggle to maintain alternative sources of income.
As streaming accounts for more and more of the way people consume music, the harder it will become for long-tail artists to cultivate the communities outside of the environment in which they are inherently disadvantaged. The less important maintaining mechanisms for community engagement are to music distributors with the largest market shares, the less viable it becomes to be a long-tail artist.
It may be that the most profitable business model is to build a music distribution system that guarantees functional satisfaction but ignores the inherent social component of music consumption. It might be that rewarding the rights-holders of the products that dominate the platform and selling slots in your playlists and recommendations are great moves for the long term financial health of the company. I hope for the sake of all the employees working at Spotify that they get there one day. However, if this is the structure of the company, it ought not to advertise itself as looking out for the little guy or cultivating "connection." And I certainly don’t think long-tail artists and music listening communities should be forced into extinction simply because they are not as profitable a commodity.
That is not to say people have not tried or are not trying to build alternatives. Glenn McDonald built everynoise.com, providing users with what was essentially a research engine using genres as a sort of taxonomy. The site will no longer be updated with new music following Glenn’s departure from Spotify. Marek Gibney, creator of the Global Network of Discovery, built music-map.com that allows you to clicked through artist-space based on listener preferences collected here. DJs such as Sam at VinylJukies and HouseShoes have created independent radio stations on Twitch where they curate and share music based using their personal vinyl collections they have developed throughout their lives. I have even created a draft of what I think an interactive music search could be like.
I don’t know if any of these alone or in combination are enough infrastructure to foster sustainable musical communities for artists of the long-tail. I don’t know if there ever will be a solution, but I do know people will continue to try and that it wouldn’t kill you to try finding music somewhere other than your streaming service of choice.
Carter is a programmer/writer spending his time thinking about dots, the lines between them, and how to make things that help people forge their own connections. He is a firm believer that music ought to be listened to one album at a time. Here is one album that is currently heavy in his rotation.