There was a time when Friday meant Release Radar.

I’d open Spotify and there it was: a little digital treasure chest of new music.

Sometimes it was brilliant. Sometimes it was shit. Sometimes it was something I’d never heard of from a band I’d never heard of and suddenly I’d spent the next three weeks listening to their back catalogue.

That was the magic.

Spotify didn’t just give me music. It gave me the possibility of discovering something.

I’m not sure when that disappeared. But my love affair with Spotify is over.

Spotify won.

That’s the funny part.

It has mass acceptance. It’s become infrastructure. Music streaming isn’t some revolutionary new technology anymore. It’s plumbing.

And once something becomes plumbing, you stop asking whether it’s magical. You just turn the tap on. And the water is there.

Which raises an uncomfortable question: What happens when the thing you once loved becomes merely good enough?

Spotify keeps raising prices. The product keeps evolving. There are more features, more data, more playlists, more recommendations, more algorithms and more ways to consume music than ever before.

And yet I’m increasingly convinced that the actual experience of discovering music has become worse. That’s quite an achievement.

We’ve gone from having too little music to having virtually all of it.

And somehow I’m listening to the same records I bought thirty years ago.

The Discover playlist used to be good. Like, really good.

Eight years ago it could occasionally make me stop what I was doing.

“What the fuck is this?”

That’s the best feeling a music service can give you.

Now?

Not so much.

The recommendations increasingly feel like they’ve been generated by somebody who looked at my listening history and decided that the safest possible interpretation of it was the best one.

I listen to Thin Lizzy. Here’s some Thin Lizzy. I listen to Whitesnake. Here’s Whitesnake. I listen to Iron Maiden. Here’s Iron Maiden.

I’ve saved these songs.

So apparently Spotify thinks I want to hear them again tomorrow.

And the day after that. And perhaps in a different playlist on Thursday.

Saving a song does not mean I want to hear it every fucking day. It means I like the song. There is a difference.

The Daily Mixes are probably the perfect example. They’re not really discovery. They’re comfort food. They’re Spotify saying:

“We’ve studied your behaviour and concluded that you’d probably like the things you already like.”

Well, thanks. I knew that. That’s why I saved them.

The whole reason I use a streaming service is because I don’t want to have to manually construct my listening experience every day.

I want to be surprised. I want the obscure album. The forgotten B-side. The guitarist I somehow missed. The band from 1987 that released one great album and disappeared. The song that isn’t popular enough to have been algorithmically blessed.

I want the rabbit hole. And the rabbit hole is where Spotify increasingly feels lost.

Then there’s Release Radar.

This one hurts because Release Radar was once my favourite thing Spotify did.

Every Friday I’d go looking for new music. And now I get:

A new XYZ release from a Hindu Yogi. A new Loverboy release from a young rapper. A new Sykes release from someone who definitely isn’t John Sykes. A new Ratt release that isn’t from Ratt.

Somewhere between the artist, distributor, metadata database, record label and algorithm, something has gone spectacularly fucking wrong.

And I understand how this happens. There are thousands of artists. There are multiple artists with the same names.

There are distributors.

Labels.

Rights holders.

Metadata.

Automated ingestion.

Millions of releases.

It’s complicated.

But that’s the problem.

The system is complicated enough to explain the mistake and still not good enough to prevent it.

And every Friday, instead of opening a treasure chest, I’m checking whether Spotify has accidentally delivered somebody else’s mail.

And the music itself can disappear.

There’s another bizarre problem with the streaming age. We were promised access to everything. And for the most part, we got it.

But streaming isn’t ownership. It’s access. The music is sitting there until somebody decides it shouldn’t be.

Licensing changes. Rights change hands. Distributors disappear. Labels change. Artists change their minds. Regional agreements expire.

Albums appear. Albums disappear. Albums come back. Then disappear again.

Maybe some of the master tapes don’t even exist anymore. Maybe they burned down in one of those mysterious fires that seem to have consumed half the history of recorded music.

But here’s the strange thing: We were told digital music would solve scarcity.

Instead, we’ve created a different kind of scarcity.

The music isn’t necessarily hard to find. It’s hard to know whether it’ll still be there tomorrow.

The cloud looks permanent.

It isn’t.

Here’s where I really started to understand the problem. Ask Spotify for something simple:

“Best emotive guitar solos.”

Go on.

Do it.

You don’t get emotion. You get metadata. You get lists compiled by magazines.

Lists compiled by websites. Lists compiled from other lists. Wikipedia. “The 50 Greatest Guitar Solos Of All Time.”

And there’s the problem.

Spotify knows what an emotive guitar solo is supposed to look like in data. It doesn’t necessarily know what makes one matter.

Because how do you quantify Michael Schenker playing “Try Me”?

How do you quantify David Gilmour playing “Comfortably Numb”?

How do you explain Zakk Wylde on “Angel of Mercy”?

Mick Mars on “Home Sweet Home”?

C.C. DeVille on “Life Goes On”?

John Sykes on “Straight For The Heart”?

Warren DeMartini on “Giving Yourself Away”?

George Lynch on “Alone Again”?

John Petrucci on “Learning To Live”?

Al Pitrelli on “Blue For You”?

Gary Moore on “Empty Rooms”?

Eddie Ojeda on “The Price”?

Vivian Campbell on whatever particular piece of magic he happened to be responsible for that week during his stint with Dio.

Yngwie Malmsteen doing what Yngwie does.

Vinnie Vincent on “I Still Love You.”

Steve Rothery on “Easter.”

Bruce Kulick on “Tears Are Falling.”

Bob Kulick on “The Idol.”

Randy Piper on “Sleeping in the Fire.”

Richie Sambora on “Dry County.”

Dave Meniketti on almost fucking anything.

Adrian Smith on “Stranger in a Strange Land.”

James Hetfield on “Nothing Else Matters.”

Kirk Hammett on the outro to “Bleeding Me.”

Randy Rhoads on “Goodbye To Romance”. There are many others that I can’t think of right now.

These aren’t simply guitar solos. They’re moments. They’re performances. They’re context. They’re tone. They’re restraint. They’re excess. They’re the note that arrives half a second later than you expect.

They’re the bend that hangs in the air. They’re the player knowing when not to play. They’re everything that happens between the notes. Try putting that into a database.

And this is where I don’t buy the lazy argument that Spotify’s engineers are idiots.

They’re not.

They’re probably fucking brilliant.

They’ve built an extraordinary machine. Spotify knows what I’ve listened to. It knows what I’ve skipped. It knows what I’ve saved. It knows what people who listen to the bands I listen to also listen to.

It knows tempo. Genre. Popularity. Release dates. Listening patterns.

It can probably predict with frightening accuracy that if you listen to Judas Priest at 10:47pm on a Wednesday, there’s a statistically meaningful chance you’ll listen to Iron Maiden next.

It’s incredible.

And yet I ask it for an emotive guitar solo and it gives me an SEO list.

That’s the paradox.

The machine understands the data surrounding music extraordinarily well.

I’m not convinced it understands music.

This is where Spotify’s problem becomes everybody’s problem.

We’re increasingly trying to turn taste into data. But taste isn’t simply a collection of preferences.

Taste is memory. Context. Association. Experience. History.

You don’t just like David Gilmour.

Maybe you heard “Comfortably Numb” when you were fifteen.

Maybe you heard it when you were thirty-five.

Maybe the guitar solo reminds you of someone who isn’t in your life anymore.

Maybe you heard Gary Moore’s “Empty Rooms” at exactly the wrong moment in your life and now the opening notes carry twenty years of emotional baggage.

Spotify doesn’t know that.

And it can’t.

Because the emotional meaning of music isn’t contained inside the audio file.

It’s contained inside you.

That’s why two people can listen to the same song and hear completely different things.

And that’s why the idea that an algorithm can eventually know your taste better than you do is worth questioning.

Maybe it can predict your behaviour.

That’s not the same thing.

We lost the obsessive music nerd.

There used to be people who knew music. The guy in the record shop who’d hear you mention Van Halen and say: “You need to hear this.”

The radio DJ who’d play something you’d never heard. The bloke at school who had an album nobody else had. The friend who’d make you a mixtape.

The magazine writer who’d disappear down some obscure musical rabbit hole and drag you along with them.

They were often wrong.

And that’s what made them useful.

Algorithms want evidence.

Humans have hunches.

Humans make mistakes.

Humans have obsessions.

And sometimes an obsession is exactly what you need.

Spotify has become safe. It got too good at being Spotify.

It knows what I like. It knows what people like me like. It knows what songs are popular. It knows what songs are associated with other songs.

It knows what gets skipped. It knows what gets saved. It knows everything except the one thing I’m actually asking: Surprise me.

Not randomly. Not stupidly.

Not with another artist whose name happens to match the artist I’m looking for.

Surprise me intelligently.

Take a risk. Connect two things that don’t obviously belong together.

Tell me why. That’s what a great music obsessive does.

And then there’s YouTube, sitting there with an enormous music catalogue as well.

And YouTube has something Spotify doesn’t have in quite the same way: the internet’s collective memory.

Live performances. Interviews. Bootlegs. Old television appearances. Fan uploads. Rare tracks. Alternate mixes. Gear breakdowns. Concert footage. Someone filming a guitarist playing one obscure solo in 1989.

The ecosystem is chaotic. But chaos has advantages.

Spotify is beautifully organised. YouTube is a fucking junk drawer.

And sometimes the junk drawer is exactly where you find the thing you were looking for.

So what am I actually paying Spotify for?

That’s the question I keep coming back to. Because Spotify isn’t bad. That’s almost the problem. It’s convenient. It works.

The catalogue is enormous. The interface is good. The sound is good. The integration is good. It’s easy. (In saying all of this, I’m currently devouring the Bernie Torme history).

And that’s exactly why I’m frustrated.

I’m paying for convenience and getting increasingly less discovery.

The service is becoming more expensive while the thing I personally valued most about it seems to be deteriorating. Maybe that’s not a universal experience.

Maybe millions of people couldn’t care less about Release Radar.

Maybe most people just want to press play and hear songs they already like.

That’s perfectly rational. But I’m not most people.

And I don’t want Spotify to know me so well that it stops introducing me to things I don’t know.

We’ve never had more music. Millions of albums. Decades of recorded history.

Every obscure corner of rock, metal, jazz, blues, punk, pop, electronic music and everything else sitting inside a phone.

Humanity has created an almost unimaginable archive of sound.

And yet the technological solution to infinite choice seems increasingly to be: Here’s the stuff you’ve already heard.

That’s not discovery. That’s a loop. An echo chamber.

And the larger the database becomes, the more important curation becomes.

Not less.

Because abundance creates its own scarcity.

There is too much music. There are too many releases. Too many artists. Too many playlists. Too much content.

The scarce resource isn’t music anymore.

It’s attention.

And the service that wins isn’t necessarily the one with the most music.

It’s the one that can make me care about something I didn’t know I cared about.

I don’t want better algorithms.

Maybe that’s the wrong conclusion. Maybe I don’t need another algorithm. Maybe I need better curators.

Or weird curators. Or human curators assisted by algorithms.

Let the machines do what they’re brilliant at. Find patterns. Search billions of tracks. Surface connections.

Handle the impossible scale of the catalogue.

But let humans inject the thing machines are still terrible at: taste.

Give me the obsessive metal nerd.

Give me the person who knows every forgotten 1980s hard-rock record.

Give me the person who can tell you which pressing of an album sounds better and why.

Give me someone who has spent thirty years listening to guitar players. Someone who knows why that solo matters.

Maybe the problem is that Spotify stopped feeling like a record shop.

That might be what I’m really missing. The record shop wasn’t efficient. It wasn’t personalized.

It didn’t know my listening history.

It couldn’t calculate my predicted enjoyment of an album.

It didn’t have machine learning.

It had a bloke behind the counter who had listened to 10,000 records and had strong opinions.

And sometimes he’d hand you something and say:

“You’ll hate this.”

Then you’d buy it anyway.

And thirty years later you’d still be listening to it.

That’s discovery.

That’s the thing Spotify used to make me feel.

And somewhere along the way, the algorithm got better and the experience got worse.

So no, Spotify isn’t dead.

Far from it. It’s enormous. It’s successful. It’s going to be around for a very long time.

But my love affair with it is over.

Because Spotify has access to more music than any human being in history.

And somehow, I’m finding myself listening to the same records I bought thirty years ago.

Maybe the problem isn’t that Spotify doesn’t have enough music.

Maybe it has too much.

Maybe the problem is that somewhere inside that enormous machine, nobody knows what makes one fucking note matter.