Turns out I'm not alone in this musical home invasion and Spotify has some explaining to do.
Hundreds of Redditors have had enough of Carpenter's "Please Please Please" popping up everywhere. Same goes for Taylor Swift. Spotify users reported Swift showing up in their "last played" even when they've never given her the time of day.
And let's not forget Drake. The guy is known for sneaking onto people's Spotify Wrapped list for years.
And this isn't just annoying — it's fishy. How can streaming numbers be trusted when the system's clearly pushing some of the “most streamed” artists harder than a used car salesman? I don’t know about you, but my tinfoil hat is so on.
I asked Glenn McDonald, data alchemist, ex-Spotify, and creator of Every Noise at Once, about this algorithmic favoritism. He said: "In a sense most recommendation algorithms are specifically trying to 'promote' something more than it would appear randomly. Basically all of my work at Spotify was like this: trying to find out what some audience is disproportionately fond of, so that collective knowledge can be shared: with that audience or other curious listeners."
Sounds nice in theory. But when your playlist suddenly turns into the Sabrina Carpenter Show, you have to wonder if there’s someone else who's pulling the strings about some particular artists. And it goes beyond Spotify. Let’s look at YouTube. Sabrina Carpenter's "Taste" video is trending #1 for music for days now despite featuring pretty dark witchcraft and gory body dismemberment scenes that definitely violate YouTube's own content policies. If anybody else uploaded something like this on the platform they would be kicked out immediately. Why are we forced to watch it? What is it all about?
Few mainstream media have already branded these suspicions as a big conspiracy (and that’s exactly when you know that the theory actually has legs). In this article, writers, after repeatedly calling it a conspiracy, ended up with industry experts’ comments like “What are we even talking about here? She's a pop star making pop music,” and “really and truly, we're all vibing to it". (I am personally not, so “all” is definitely not truly or really).
Read also: Music Charts in 2024: Industry Dinosaurs or Still Relevant?
Streaming algorithm: curation or paid control?
The most influential music streaming platform, Spotify's been busy. They've axed 2,300 people in three rounds of layoffs to "move faster into AI". Since nothing says "innovation" and “profits” like kicking humans to the curb, Wall Street loved it. (Their stock grew more than 100% within the last 12 months at the moment of writing this article).
While Spotify's trimming the fat, they're also fattening up their wallets. They've hiked prices twice in 12 months. And what listeners are getting for all that extra cash? Apparently, the privilege of having an AI decide what we should listen to next.
But how do these fancy algorithms actually work? According to Glenn McDonald: "They could work any way the engineers want them to, and anything I knew when I worked at Spotify could well have changed since I left. And as with anything, if the people operating a system don't tell you how it works, you can't do anything other than speculate. And if they won't tell you, it's fair to wonder whether it's because they know you'd be upset if you found out."
As for personalization? McDonald drops this truth bomb: "Almost all personalization is an attempt to guess at unexpressed aspects of your taste by comparing the expressed aspects of your taste to the expressed tastes of other people. If 'true' personalization would come from really getting to know you and your background and your emotional needs and the exact shapes of your ears, then hardly any 'personalization' in music-streaming, or indeed in most tech, really counts. It would be more accurate to call it 'targeting'."
Read also: Let’s predict the future of curated playlists in the streaming era. Will AI's efficiency outplay human creativity, or can both coexist in harmony?
Now, let's talk about discovery. Because isn’t that Spotify's big promise? Discover new artists you will like with sophisticated AI driven recommendation algorithms? That sounds so good in theory — you got to easily find new favorite artists and it’s so simple for emerging indie bands to be discovered — just upload your tracks to Spotify and if it’s good you’d be rewarded. Right? Right?
Let’s analyze some actual numbers to get some perspective and compare most streamed of today with most played of the past. Back in the day, Billboard's top 5 singles from 1970 to 1974 had 24 unique artists out of 25 spots. Roberta Flack was the only one to get a sport in the top 5 twice.
In the 80s (from 1980 to 1985) there were 22 new artists out of 25 top 5 spots each year. With Michael Jackson hitting the rating 3 times, and Paul McCartney and Olivia Newton-John 2 times each.
Fast forward to Spotify's most streamed artists from 2019 to 2023, and we've got 13 unique artists hogging those same 25 spots. Bad Bunny showed up 5 years in a row. It's less "discover new music" and more "listen to Bad Bunny on repeat".
Does it sound like you are discovering more music? These numbers tell me the answer is no.
Meanwhile, Spotify's paying songwriters about $150 million less next year. But don't worry, they're on track to meet their 2024 revenue growth goal. Spotify's CEO Daniel Ek is thrilled: "As an adult company we are now consistently profitable, which is great news." Yeah, great news for everyone except the artists and the listeners.
So, what's the bottom line? Spotify's algorithm isn't so much curating as it is controlling. It's less about introducing you to your next favorite band and more about maximizing profits. And while Spotify execs have sold more than $250 million in stock this year, artists are left wondering where their slice of the pie went.
Payola 2.0 in the age of Spotify dominance
Payola… that shady practice where record labels slipped DJs some cash to play their songs? Some speculate it’s back in the shape of those overly aggressive recommendations.
So how exactly does this new-age payola stack up against its predecessor? Glenn McDonald breaks it down: "The idea of payola is substituting money for taste. In the original radio version, you paid a DJ, who was supposedly choosing what to play based on their taste and expertise, to play your song a lot even if they didn't like it, and to carefully not explain why they were. In the modern version, like Spotify's Discovery Mode, you pay by accepting a lower royalty rate, and the algorithm, which is supposedly choosing what to play based on the listener's taste and other people's collective knowledge, plays your song more because you pay, and very carefully does not admit why."
And the best part for modern streamers — it's not even illegal. As McDonald points out, "Legally, it's not payola because the anti-payola laws were only written to apply to broadcasters."
But just because it's legal doesn't mean it's ethical. McDonald lays out the moral quandary:
"The systemic implication, beyond the inherent listener deceit, is that the more artists buy into them, the more pressure it puts on all the other artists to also buy in, just to keep up. Structurally, this is called a 'race to the bottom': eventually everybody feels like they have to do it, at which point it no longer gives any artist an advantage, but all of them are paying the price, and effectively the platform has just lowered their royalty rates under the guise of offering an optional marketing program."
McDonald sums it up perfectly: "Morally, I personally think it's pretty clear that promoted tracks ought to be labeled as such the same way that ads are labeled in search results."
At least with old-school payola, the benefits were spread around. Thousands of DJs got their kickbacks, and some were still nice enough to be selective and have some class (that’s why we had that diversity according to our little study analyzing Billboard most played singles in the 70s and the 80s).
Now? It's just a handful of corporate suits reaping the rewards while artists get less royalties and listeners pay more for a shrinking playlist. I’d be bold and say it - modern streaming is actually way worse than payola.
Artists or audience: who really benefits from modern streaming algorithms?
While streaming platforms (along with their shareholders) seem to be proud with their algorithmic recommendations, a recent study found that listeners feel that algorithmic curation is "impersonal." And it doesn’t seem that AI DJ really cares what users want to hear. While at the same time streamers charge listeners more for a worsening experience.
Do these algorithms do any good for the artists maybe? Well, since artists are getting paid less it doesn’t seem so.
How about impact on emerging bands and music discovery? McDonald believes: "If somebody operating algorithms at you won't tell you how they work, it's probably safe to assume that those algorithms are not designed to serve your interests." He adds, "Emerging artists need community-building. Music listeners need tools that amplify their curiosity and invite them to explore. Passive personalization isn't great for either of those. It's great for cultivating listener dependency. Guess who benefits from that."
The only winners seem to be the corporate suits and their precious algorithms. So here we are, stuck in a musical Matrix where the machines are in control, and they're not even good DJs. The artists are getting squeezed, the listeners are getting swindled, the profits are being prioritized over passion.
Maybe it's time to unplug from the matrix, hit up some local gigs, talk to actual humans about emerging bands? It feels like the future of music depends on it.
🍿🍿 You may also like:
