The most dependable weekly stack of darkish, vervey, bounce tech-space-prog-house vibes on the internet is Refresh Radio by Solid Stone. I listen every week. One day years ago my eye drifted to an episode’s play count; there were around 1,000. I followed the curious thread to his Twitter profile: just under 5,000 followers and a blank space where the blue check should be.
No surprise in those low counts; my beats fanhood has been finely filtered. It has been a lot of years of hearing mixes roll by in the background, occasionally noticing an interesting new twist of a sound that I trace to a previously unheard-of artist with a small following. But the knowledge lingered like an aroma, that that an artist I listen to every week, whose music to me doesn’t sound faraway and strange, but steady and right, has fewer followers than there are fans at a college hockey game.
Like any self-aware narcissist, I wonder, I feel, I know, that my experience must be significant. There has to be some Vox headline-worthy trend to my narrow taste in kick drums and basslines. Surely my Solid Stone fanhood heralds the Era of Niche, supporting a Middle Class of Creativity.
And, in my defense, an understanding friend might play through the slide reel of anecdotes hinting at an emergence of nichiness: First, the Facebook groups with lists of members longer than can be counted or named without aid of a machine, who feel as I do about Solid Stone and other producers who refine a continuous blend of interstellar space, emotion, and the daily grind. Second, another list, also longer than any human can tabulate with the naked brain, of genre and novel subgenre and other ways of classifying groups of musicians whose sounds are described by their fans in equally preposterous mixed metaphors. Each of which has their own Facebook groups and other like forms of gathering and affirmation. Third, the fact that if this were 2005, I would have no way of learning Solid Stone’s name and following his releases, and if it were 1994, if I was not a card-carrying member of a local club scene, I would have no way of knowing he even existed.
But when it comes to judging whether something matters to America, the global project of growth and exchange, or some other abstraction I am attached to emotionally, I am like Spiros and Eton, the foreign gangsters shipping heroin by the container into Baltimore in Season 2 of The Wire, when they direct dockworker Nick to their subordinate for further reups. “We are wholesalers.” A change in national story can’t rest on a single g-pack’s worth of job creation. I don’t care that it’s easy for Facebook’s multiplicity of groups trick my brain into thinking that music consumption is becoming so well-distributed that there can now be a lot of artists making a moderate amount of money, instead of a few artists making a lot of money.
The tip of the data says my brain is indeed tricked. The absolute number of employed “musicians and singers” (per BLS) sank between 1999 and 2016. Revenue concentration figures for the last few years have evaded me, but in 2014 the top 1 percent of artists accounted for 77 percent of sales. It feels very real that the infiniteness of Youtube and Spotify, which is supposed to facilitate discovery of artists overlooked by major labels and radio personalities, has just taken anything that attempts to carry a melody and called it a song. And that the long tail is comprised of hobby tracks with between 3 and 300 listens each, worth approximately $0.000023 in revenue to the artists. Indeed (again) the Times of London reported that 76% of tracks available for digital download failed to attract a single buyer.
But the library size determines sales concentration. The top-consumed tracks comprise the same percentage of revenue but a much, much smaller percentage of the absolute number of titles when you increase the size of the library exponentially. In fact, relative revenue concentration could skyrocket even if the absolute number of tracks in that top-1 percentile is much larger than the top-10 percentile when library sizes were limited. The sun will still look tiny from Pluto even if you quintuple it in size. If the absolute number of artists responsible for the bulk of revenue increases significantly, then pending a spin of the Inception top to make sure I’m not in a dream of shitty intuitive logic, this could support a solid chunk of new musical livelihoods. Hopefully at least a few hundred thousand more, the range needed to become sufficiently commonplace to be accepted into the sacred scroll of what is considered normal, healthy, dignified American behavior.
Any evidence of that? There are three datapoints suggesting that subscription streaming use increases consumption of non-superstar tracks by in the 10s of percent. But there was little comparability as to what “non-superstar” really means.
First on stage is the Spotification study I flowed about earlier in quarantine. For the faithful readers who don't remember, it found that switching from downloads to subscription streaming increased total music consumption by 49 percent, while increasing the consumption of top 500 artists by 2.4 percent. This means that Spotify shot up listens for lesser known artists by around half.
The second bit of evidence is our familiar Rhapsody bite. Again for forgetters: in a time dominated by digital purchases, users of a zero cost to discovery subscription service listened to the bottom 99 percent of songs 68 percent of the time, but the bottom 90 percent only 22 percent of the time. That is 68 percent of listens going to artists not found at Wal-Mart; if you imagined that, say, 40 percent of listens in the 90s went to these deep tracks nationwide, a hypothetical universal subscription model the shift towards niche would be almost 30 percentage points, and some actual proportion I don’t want to figure out right now.
Finally is a new scrap of data from Professor Laurina Zhang at the University of Western Ontario. Remember back in 2006, when you got a new computer and tried to play one of the four songs you bought on iTunes because you couldn't find them on Limewire? You would have had to enter your iTunes password to prove that you were an authorized listener. Well, in 2007 one of the four record companies controlling virtually all music production freed its library from these digital shackles. In 2009, the other three followed.
Professor Zhang theorized that removal of Digital Rights Management (DRM) controls would make it easier to share songs, reducing the cost of discovery, and boosting consumption of less popular albums. She found that removing DRM increased sales of albums that had sold fewer than 100k copies over the previous three years by 24 percent, and by 30 for albums that had sold fewer than 25k copies in the same time. Sales of moderately and very popular albums didn’t increase or decrease at all.
If as weak a boost to sharability as being able to email songs among friends increases long-tail sales by 30 percent, how much could more powerful discovery mechanisms, like Spotify’s playlists and rec engines, pump up long tail consumption? Other hand: removing DRM is such a weak discovery boost that it makes for a pretty soft comparison to contemporary discovery tools.
And if discoverability does increase in consumption of non-superstar tracks of between 20 and 40 percent, for how many artists does that make the difference between making music full time and working at McDonalds, or McKinsey, or whatever other shitty place a person might work?
The Spotify study applied the increase to artists outside the top 500; the 500th most-streamed artist this week was tapped over 4M times. If the increase went to artists 500-1000, it’s probably not going to make many more musicians with mortgages. But too far down the long tail, and you’re just pumping listens for hobbyists who should keep coding by day. It’s likewise foggy to tell whether the Rhapsody increase is good news for people who aspire to both produce music and own property; it only would be so if the 2nd - 9th percentile of tracks to which a big increase was observed were made by artists right on the edge of middle class life. The under 25k sales in trailing three years benchmark of the Western Ontario study is the only absolute measure we have, and who knows how it aligns with the relative categories of the other two. There’s a “sweet spot” in the long tail and I need to find out where it is.
I need two long tail curves with absolute consumption and revenue marked, one each for before and after Spotify. Not sure where to find them; maybe I’ll conjecture my way to my own.