top of page

Stream Me a River: The Sad Tale of Music Algorithms (or) Why Your Indie Band Isn't Famous Yet

Updated: Jan 19


Ah, the streaming world… A vast ocean of tunes where every track is fighting to be the next big earworm against a veritable worm farm of even bigger earworms. Too far? OK, that’s fair. But seriously(ish), I want to address the million-dollar question: "Do streaming services play favourites with artists, especially the already popular ones?"


The Rich get Richer, The Indie get... Indie-er?


Here’s my take. It's high school all over again. The popular kids get all the attention while the cool indie kids (you know, the ones with the awesome underground band and the shaggy haircuts) linger on the fringes, playing for beer and wings.


But why? Is it the algorithmic overlords wielding their favouritism wands? Or is it something even more sinister? Let's dive into the world of digital music where the beats are plenty, but fairness? Not so much.


Big Streaming’s Discovery Mode: A Twist in the Musical Tale


A while ago Spotify, our most beloved music giant (tongue, meet cheek), introduced something called "Discovery Mode." It's like a backstage pass for artists, but instead of meeting fans, they get a nudge in Spotify’s algorithm. The catch? They pay a commission. It's like paying the DJ to play your song, but instead of a DJ, it's a super-complex algorithm. Transparency? I’m not convinced. It’s pay to play, in a kind of grotty way.


The Algorithmic Bias Boogeyman


But wait, there's more! Bias and discrimination in algorithms aren’t just Spotify’s new hit single. It's a chart-topper in the whole AI influenced music streaming industry, and beyond. Everyone from tech whizzes to academic nerds are jamming about whether these algorithms are fair or if they're just playing favourites.


So, what is "fair"? If you ask Mr. Oxford Dictionary, it's about being impartial and just. In the streaming world, equity is about giving every artist, whether they're The Rolling Stones or your cousin's garage band, equal opportunities to shine. But is this utopian musical dream really achievable in the vast musical catalogues that exist in the global music streaming machine, and is it, really, fair?


Measuring Fairness: The Musician's Scale


In truth, I don’t believe there’s an ideal way to measure fairness in music streaming – because people’s taste is people’s taste, and while it can always be influenced, it’s a matter of personal choice, and policing choice is akin to stifling freedom of expression. At The Pack, we don't hold with that.


But I do believe that we can influence the equity of the system by being more intentional about how we, as music consumers, allow the systems to influence us. We can also create some space for those artists who are routinely marginalised, perhaps by considering diversity metrics. This is akin to measuring the variety in your music diet. Are you only munching on a single sound or savouring a smorgasbord of genres? Like a diet, the more diverse people’s musical consumption, the healthier the industry.


And guess what? Research shows that the more diverse digital music munchers do have broader taste generally (and, interestingly, above average intelligence… just saying…). But diversity? Still a tough nut to crack in a system where the four major labels have a chokehold on the streaming services, and therefore on your listening choices.

 

The Incentive Conundrum


And of course, streaming platforms don’t really want to play all the music cards evenly. More choice means more listeners sticking around, and this is why they have catalogues of many millions of artists, and even more millions of tracks. It's like offering a buffet instead of just pizza – sure, everyone loves pizza, but sometimes you crave sushi, right? But by offering listeners unlimited choice, there is a perception that people are more diverse in their listening… and that is certainly not a given.


The Diversity-Recommendation Tango



And for the algorithms, it’s a careful dance between serving listeners what they love and throwing in a surprise track every now and then to shake up the sound. Play it too safe, and listeners get bored. But throw in too many wild cards, and they might just start skipping like stones. The goal? Keep the punters listening, but also throw in a few unexpected tunes so as to maintain a pretense of fairness.


But algorithms, like humans, can be biased. They might love pop and ignore hair metal, not because they have bad taste (giggle), but because that's what they've been taught. AI, like a child, is entirely influenced by its parenting. The problem starts with the data we feed these hungry algorithms – if they're mainly munching on mainstream music (how’s the alliteration?), they'll keep recommending it like a broken record (oh, the puns!). Welcome to the world of feedback loops, where popularity breeds more popularity, and indie tracks get lost in the shuffle.


The Feedback Loop: A Musical Echo Chamber


Imagine a world where only the hits play on repeat. That's the feedback loop. The more a song is played, the more it's recommended, turning streaming services into an echo chamber of Top 40 hits. It's like being stuck at a party where the DJ only knows 10 songs.


Gender Bias: The Unsung Tune


Oh, and let's not forget about the gender bias. It's the music industry's guilty secret. Taylor and Beyonce aside, studies show that algorithms often sideline female artists and listeners. It's like a boys' club, but with playlists instead of secret handshakes and overt misogyny.


The Filter Bubble Blues


So, with all of these influences compounded, you enter the "filter bubble" - your own personal musical echo chamber. Here, you're fed a diet of what you already like, never discovering the wild, wacky, and wonderful world outside (oh my, the alliterative muscle is flexing today!). And that’s fine, if you have no intention of evolving your musical palate… but, at the risk of sounding preachy, it’s time to break free from the bubble, or risk missing out on that obscure jazz-fusion band you never knew you loved (OK… that was a stretch).


Spotify's Diversity Dilemma


And hey, don’t just take my word for it, even Spotify admits it: their algorithms can lead to a less diverse musical platter. But when users branch out, they can step away from algorithmic suggestions and into a world of musical exploration. It's like choosing the road less travelled, for your ears (hey, I’m a Surrealist, just go with it).

 

 The Great Algorithmic Balancing Act


So, how do we make these recommendation engines more inclusive? The answer lies in tweaking the algorithmic recipe. Add a pinch of diversity, a dash of fairness, and voilà! You've got a more balanced musical meal. Sounds simple, huh? Yeah… Nah…


TikTok's Trendsetting Tactics


But TikTok's doing something right. They're mixing up their algorithm to favour diversity over predictability. It's like a DJ who's not afraid to play something off the beaten track, shaking up the dance floor with new beats. And while I’ll admit I’ve not spent much time indulging in TikTok because… well… I am old and there’s very little to be done about that… I’m impressed that something so new, is taking a different view.


Debiasing (it's a word if I say it's a word...k...) the Musical Mind



So, how can we make algorithms less biased? Well, it's like fixing a mildly stuck, old-school jukebox. You shuffle the playlists (re-ranking), change the records (rebalancing), tweak the machine's settings (regularisation), and sometimes, you just need to unplug it and start over (adversarial training). Then you give it a good kick for good measure.


Editorial Influence: Just a Little of that Human Touch


But here’s the real kicker, and the one that the big tech giants don’t want us to think about. The bottom line is that it’s actually important to include a human element. Editors curating playlists without undue influence from labels and AI could steer the ship towards more diverse waters. But the cost reduction models of most of the major tech players eschew the addition of human curators in favour of frictionless systems. But in truth, isn’t our difference and our humanity what makes music truly meaningful?


The Bottom Line: A Symbiotic Symphony of Algorithms and Humans


In the end, creating a fair and diverse musical world is like conducting an orchestra and to some degree, it's actually up to us. It takes a mix of algorithmic genius and human touch, of data and intuition, to create a symphony that resonates with everyone. And while we're figuring it out, let's not forget to enjoy the music – whether it's from a chart-topping star or your neighbour’s garage band. After all, every song has a story, and every melody deserves its chance to become your new favourite tune.


So, next time you hit play on your preferred streaming service, remember the complex symphony of influence and choice that brought that track to your ears. Maybe give that random recommendation a chance – who knows, it might just be the next big thing in your personal playlist. Or even better, go searching for something you’ve never heard before and support an independent artist, who, let’s face it, is more likely to appreciate having you as a new fan.


In the world of streaming, the tune of fairness and diversity is an ongoing composition, a creative collab, a melody that we're writing together. From data scientists to playlist curators, from indie bands to pop icons, and right through to the consumers of music - you, and me - each of us plays a note in this grand musical adventure.


And remember, in the end, it's not just about the songs that reach the top of the charts. It's about the hidden gems waiting to be discovered, the underdog anthems, and the niche beats that speak to your soul. It's about breaking free from the algorithm fabricated echo chamber, owning your choice, stepping out of the filter bubble, and embracing the full spectrum of sound.


So, go ahead, press play on something unexpected. Dance to a different drumbeat. Who knows? You might just find your new anthem in the most unlikely of places.

223 views1 comment

Recent Posts

See All
bottom of page