Somewhere in the American South, there's a retired steelworker who can tell you the difference between a good and great production of La Bohème. He's never been to Milan. Nobody in his family listens to opera. He found a recording on a whim at a library book sale in 1987 and something in Puccini's second act broke him open. He's been chasing that feeling for almost forty years.
In Tokyo, a twenty-three-year-old office worker listens to nothing but outlaw country. Waylon Jennings. Merle Haggard. Townes Van Zandt. She doesn't speak English fluently. She found a Spotify playlist during the pandemic and heard something in the steel guitar that felt like homesickness for a home she'd never had.
In a suburb outside London, a thirteen-year-old boy is deep into Bollywood — not ironically, not as a phase, not because of a school assignment. He's studying Hindi on Duolingo so he can stop reading subtitles.
None of these people are the target audience. The market research never imagined them. The algorithms would never have suggested these things to these people. And yet.
These are the most interesting stories on any shelf.
The Demographic Cage
Here's something the recommendation engines don't want you to think about: they are, at their core, demographic machines.
Not in the crude sense. They don't literally sort by age and gender anymore — that's too obvious, too legally uncomfortable. But they achieve the same result through a thousand softer signals. Your listening history. Your watch time. Your click patterns. The behavior of people who behave like you. And "people who behave like you" is, more often than not, a polite way of saying "people who are like you."
The result is a kind of invisible fencing. You get served things that people in your demographic cluster already like. The recommendations feel personal — made for you, the interfaces promise — but they're actually tribal. You're seeing what your demographic sees. You're hearing what your demographic hears. You're reading what your demographic reads.
This is efficient. This is optimized. This is — and I mean this literally — the opposite of discovery.
Because discovery, real discovery, is the moment you stumble across something that nobody in your cluster has found. Something that wasn't on any playlist generated from your history. Something that violates the prediction. The algorithm looks at you and says, This doesn't make sense. And you look at the thing you've just found and think, I know. But I love it anyway.
The Passport Theory of Taste
I've started thinking about taste in terms of passports.
Most of the things you love live in your home country. They're from your era, your culture, your language, your demographic. This isn't a failure — it's natural. You grew up swimming in a particular cultural ocean and of course you absorbed what was in the water.
But the things that make your shelf yours — that make it different from every other shelf held by someone your age, in your city, with your education — are the stamps. The entries in the passport. The things you loved that required crossing a border.
Not a literal border, usually. A border of genre. Of language. Of era. Of expected identity. The hip-hop head who loves Joni Mitchell. The literary fiction reader who can't stop rereading Dune. The person who grew up on punk and now listens almost exclusively to classical piano.
Each of these crossings is a small act of disobedience against the person you're supposed to be. And each one makes your shelf more honestly, irreducibly you.
The problem is that the entire infrastructure of modern media consumption is designed to keep you in your country.
How Borders Used to Be Open
This is the part that stings: it used to be easier.
Not because people were more open-minded — that's nostalgia talking. But because the delivery mechanisms were dumber, and dumb delivery mechanisms leave room for accidents.
The radio DJ who played jazz between rock sets. The video store clerk who put a Hong Kong action film in the staff picks next to Die Hard. The library's "new arrivals" shelf that didn't know you only read mysteries. The friend of a friend's CD case in the back seat. The channel you landed on at 2 AM because you were too tired to change it.
None of this was optimized. None of it was efficient. And that inefficiency was a feature so powerful that we didn't even recognize it until it was gone.
Those systems had friction. They had randomness. They had the equivalent of wrong turns on a road trip — the kind of wrong turns that lead you to the best diner you've ever eaten at. You encountered things you weren't looking for because the systems weren't smart enough to prevent it.
Now the systems are very smart. And you encounter almost nothing you aren't looking for.
The Wander Problem
Here's the thing about wandering: you can't do it efficiently. That's the whole point.
You can't wander with a destination. You can't wander on a schedule. You can't wander inside a system that's constantly trying to guide you toward something it already knows you'll like.
And yet wandering is how most people find the thing that changes them. Not the thing they enjoyed — the thing that rewired them. The thing that opened a door to a room they didn't know existed in their own house.
That steelworker at the library book sale wasn't looking for opera. He was looking for nothing in particular. He was in a state of open attention — receptive, undirected, available to surprise. And Puccini walked through the opening.
This state is almost impossible to achieve inside a recommendation feed. Not because feeds are bad, exactly, but because they're always doing something. Always suggesting. Always interpreting your behavior as a signal. Every pause, every scroll, every half-second of attention is being read and fed back into the system. You can't be undirected inside a system that's constantly directing you.
The feeds want to help. I believe that. The engineers who build them genuinely want you to find things you'll love. But there's a category of love that can only be found through aimlessness, and aimlessness is the one thing an optimization engine cannot offer.
What Your Unexpected Loves Tell You
Pay attention to the things you love that surprise you. They're doing something important.
When you love something that wasn't made for you, you're discovering a part of yourself that your environment didn't predict. You're learning that your taste has rooms you haven't visited. That you contain sympathies and resonances and aesthetic appetites that your biography didn't account for.
The person who grew up in the suburbs and falls in love with rural folk music — that's not appropriation. That's recognition. Something in the loneliness of those songs matched something in the loneliness they already carried. The form was unfamiliar. The feeling was not.
The person who speaks only English and becomes obsessed with French cinema — they aren't performing sophistication. They're discovering that they think in images more than words, that pacing matters to them in ways they'd never had language for, that a certain quality of light on a Parisian street speaks to something they've felt but never named.
These unexpected loves are dispatches from the parts of yourself that the algorithm doesn't know about. Because the algorithm only knows the parts of you that you've already expressed. It models the you that exists in data. The you that clicked, watched, listened, and bought. But you're bigger than your data trail. You contain multitudes that haven't been captured yet.
The unexplored room is always the most interesting one.
The Shelf as a Map
Look at your shelf — your real one, the total picture of everything you love. Books, albums, films, shows, games, whatever your medium is.
Now look at the things that don't fit. The outliers. The ones that don't connect to the others in any obvious way. The ones you can't explain with a single story about your upbringing or your peer group or your education.
Those are the stamps in your passport. Those are the evidence that you've traveled. Not physically, necessarily, but aesthetically. Emotionally. You went somewhere you didn't have to go, and you brought something back.
A shelf with no stamps is a shelf that stayed home. It might be beautiful. It might be full of excellent things. But it's a self-portrait with only one expression.
The most interesting shelves are the ones with gaps in the logic. The ones where you look at two items side by side and think, How does this person contain both of these things? That confusion is the fingerprint. That's where the person is.
How to Cross a Border
I'm not going to give you a listicle of ways to broaden your taste. That would defeat the point.
But I'll say this: the next time you're choosing something to watch or listen to or read, notice the moment of filtering. The moment where your hand reaches for the familiar. The moment where the algorithm suggests something and you think, That's not really my thing.
Sit in that moment for a second. Ask yourself: How do I know?
Because sometimes "that's not my thing" is genuine self-knowledge. You've tried it. It didn't resonate. Fair enough.
But sometimes "that's not my thing" is just "that's not my demographic's thing." And you've never actually tried it. You've just absorbed, through a thousand invisible signals, that people like you don't like things like that. And you've let that absorption pass for preference.
The difference between those two — between genuine preference and inherited demographic assumption — is the difference between a shelf that knows itself and a shelf that's been told who it is.
You won't know which one you have until you wander.
Stacks is a place where your shelf is yours — the expected parts and the surprising parts. The things you love that everyone loves, and the things you love that nobody saw coming. All of it counts. All of it's real. Build your shelf →