How music discovery became predictable

Mint

20-06-2025

If I could, I'd pay serious money to travel 20-something years back in time to experience Nirvana's ground-breaking album, Nevermind, for the very first time again. Having borrowed a cassette from a schoolfriend, I found an opportune time to go to my parents' room and use the two-in-one music system—a 'deck". This wasn't a parent-friendly record; on Smells Like Teen Spirit, the main guy, Kurt Cobain, screams about his libido repeatedly. While I'd heard one Nirvana song—Come As You Are, via a stray MP3 on someone's CD—I had little idea what was coming next: a sonic thunderstorm that would blow my teenage brain right out of my ears.
All of this today sounds like gibberish. 'Two-in-one"? 'Cassette"? 'MP3"? In the early-to-mid-2000s, these were essential terms in the cultural lexicon. Music consumption and discovery, as with every generation prior and since, was for millennials too dictated by the prevailing technology of the time, and indeed its limitations. Only, that particular period is the most tumultuous in recent music history. It was an era of upheaval, transformation, and chaos, as the world shifted from the physical to the digital: cassettes were commonplace and affordable (a standard ₹125), but they were being phased out. CDs were a popular if rather more expensive format. These were found, neatly arranged by name and genre, in brick-and-mortar shops, imagine.
MP3s, available for download online, became a convenient and free alternative, existing in a lawless, peer-to-peer digital jungle via file-sharing software Napster and the clones that followed. While no longer a complete novelty, digital music wasn't yet pervasive either. But it was gaining traction, leaving the industry in turmoil as bands lost significant revenue and labels' bottomline got wrecked. Everything was illegal, pirated by amoral music nerds and spread widely by internet anarchists. Starved as we were of a lot of current music that just wouldn't release in India via conventional routes, we hit the download button.
These trends defined how young people discovered their music. You could go to a Planet M to window-shop, and you'd end up finding a random band or artist that could ruin your week or change your life. Grey market spaces like Palika Bazaar—an underground and 100% illegal market in the heart of Delhi—became a source for complete (pirated) discographies, sorted into digestible MP3 folders. Cable TV, pre-streaming, was another place to find music. MTV and Channel V and, later, a channel called VH1, would play music videos all day long.
NO MORE BARRIER TO ENTRY
Musical tastes, for millennials and those preceding them, were shaped by a range of eclectic factors. The most exciting among these were the happy discoveries. The life-changing accidents. A random untitled mix-CD from a friend's friend's friend. A mislabelled song on the pirate software Limewire. Something you stumble upon on VH1 while channel-surfing. The songs you've never heard before, that catch you by surprise. It's this feeling where a greater force takes over your being, and compels you to dig deeper, and find out everything about that band. You have no choice but to start a new obsession immediately.
Much of these tools of discovery have now, for reasons good and bad, been rendered obsolete. And while it's tempting to romanticise the past, it was also genuinely exhausting to hunt for music. Nothing ever released here on time; they played the same 50 songs on TV; MP3s were mislabelled and impossible to sift through; downloads took hours, days, weeks; tapes were dying, CDs were pricey.
Today, for the price of a single cassette, a hundred-and-bit rupees, I have access to Spotify's entire library of over 100 million songs. (A relevant counterpoint here is that you're only renting this music; it could disappear tomorrow.) There was a prolonged battle for the soul of music but, by the mid-2010s, streaming had won out, becoming the preferred mode of listening globally. The barrier to entry was decimated. There are dozens of streaming platforms—the chief ones being Spotify, YouTube Music and Apple Music—each one offering (to Indians) affordable prices for their premium versions and free versions with ads.
A quick sidenote: streaming platforms are a net evil to society; they've done untold damage to artists by offering them literal peanuts and devaluing art, while training listeners to never pay for what they consume. It's legalised theft. The P2P MP3 era that pioneered digitisation, while not without its problems, had a sense of reckless freedom and idealism to it. That chaos and anarchism has been replaced by a cold-blooded capitalism where the artist gets shafted while the guy above him lines his pockets. Indeed, Spotify—the loudest player in the market—faces regular criticism and has been the subject of high-profile boycotts and walkouts. (They've all returned, hat in hand, as bands are left in a no-win situation, having to pick between fans and principles.) And yet, at the same time the tech has liberated the listener by opening up access in this way. It's all very messy.
NEEDLE DROPS
We'll come back to streaming since it's such an omnipresent force in the world of discovery. But the olden methods—cable TV, physical stores, and such—have either withered away or been re-interpreted in modern settings. Instead of Channel V late-night broadcasts curated by Luke Kenny, people are discovering music accidentally through 'needle drops" on TV/web series they're watching on second screens. This is a curious inversion; previously, shows would use popular, recognisable songs as a cheat code to signal a pre-determined mood to the viewer. Like how no medical drama could resist throwing in the awful How to Save a Life by the Fray for a while. Now, that arrangement has flipped. Songs on shows—which are experienced differently as the viewer has an existing emotional relationship with the show's characters, as well as visual cues for context—take on new meaning and serve as introduction to an artist. Excited, the viewers rush to YouTube to comment in solidarity. They search online for more needle drops. SEO-driven aggregator websites and click-hungry publications rush to compile a list of all the songs featured on a show, which is duly converted into playlists by fans.
There's also the rather more controversial method of discovery: Instagram Reels. There can be an inauthenticity and, if I may, a dishonesty about music written expressly for the purpose of going viral on social media in 30-second teasers. But it works because we all spend an inordinate amount of time on social media. Often, these songs have inescapable hooks. The format of social media short-form videos is such that the same template is reused, recycled, and rejigged during its window of relevance. Just by repeat exposure, these songs can get stuck in one's head and lead the listener on to a path of greater discovery.
A lot of music listening, thus, seems to have shifted online. And to the ever-present cellphone. While previously there were different avenues—computers, music systems, Walkman or Discman, iPods—a streamlining of technology has meant that the phone is the primary device now. By way of outliers, we do have vinyl fetishists with record players; audiophiles; music nerds going deep on centralised forums like Reddit or Discord, using the Bandcamp/SoundCloud online catalogues, even buying CDs directly from young, independent bands. But mainstream habits revolve very much around streaming.
In physical spaces, too, you may—at a restaurant or a bar—come across someone pointing their phone at the speaker playing music. They're 'Shazaming" a song so they can hear it again. Shazam, an app that processes a song being played and provides all details related to it, has been around forever. But it really caught fire over the past decade, and was acquired by Apple in 2018. Previously, you'd have to memorise the lyrics to look up later, or hope to hear the song again. It's a nice reminder, again, that technology can be such a valuable asset in the process of discovery. And just as often a hindrance.
LIMITS OF THE ALGORITHM
Which brings us to the elephant. The algorithm. Recently, I discovered something called Spotify Blend. Users can 'blend" your profile with that of a friend's, and Spotify will do its algorithm witchcraft to create a custom, shared playlist incorporating both people's musical preferences. It even offers a 'match score" to see if your music tastes align, a quick and foolproof way to tell if the relationship is going to last. You can add up to 10 friends in a Group Blend, each with their unique taste profile coming together to create one giant khichdi playlist for everyone to parse through. This is a modern retelling of community exchange; people have forever shared their music with friends in group settings. Except that we have an additional friend in the mix here: the algorithm.
Streaming services offer a series of playlist options, from user playlists to 'algatorial" ones. The ones driven by the algorithm are of particular interest here. On Spotify, you get Time Capsules, Discover playlists, homepage recommendations, autoplay options—the algo never sleeps. Multiple AI and machine-learning processes work simultaneously to create this entity. Based on research, theories, and information available, the technology analyses songs via content filtering—looking at a song in isolation, studying its metadata and such—and collaborative filtering, where it's placed within a larger context. User behaviour, search history, lyrical themes, compositional structures—they're all factored in to craft personalised recommendations. I've even noticed the algorithm sometimes picking up the key in which a song is composed, and playing a series of songs that all start in that same key. Regardless of one's principled opposition to streaming, these features aid the process of discovery and make it so much easier. The algorithm is sharp, well-informed, intuitive, and will instantly gauge a listener's interest, guiding them to new places.
But it raises a couple of semi-philosophical questions. For one, why should I allow the machine to tell me what to listen to? There's a volatility attached to discovery—repeat trial-and-errors driven by human emotions and external variance. Streaming, with its robotic efficiency, can flatten that unpredictability into a horizontal structure leaving little room for experiments. It knows what I like, and it'll keep feeding me.
More importantly, what about the music that even I don't know I like? At a time when my music habits exclusively comprised alt-rocker misanthropes, I stumbled, on MTV, upon a song called Surfing on a Rocket by French electronica/dream pop duo Air. It led me down a circuitous path of cool electronic music I'd never have found otherwise. Algorithms—on Netflix, on social media, on music streaming—can create bubbles and echo chambers. They keep feeding you versions of things you already like and engage with. And they hide you from a world of discovery that you don't even realise exists. Those avenues for happy accidents, while very much still around, can get constricted by the self-limiting nature of algorithmic excellence.
It's a complex subject, riddled with questions around access, ethics, tech manipulation, listener behaviour, maybe some moral panic—as such, all discussions around art and consumption do eventually circulate within these idealism labyrinths. And conversations around the algorithm deserve critical examination without being tainted by generational bias. But what remains steady is that new generations find novel ways to access and consume music; it can feel alienating—even existentially distressing—to those on the outside. Maybe we're losing some recipes. But the music landscape is forever fluid and evolving. And people within it will always find systems that work for them.
Akhil Sood is a Delhi-based writer.