You think math is about numbers? Wrong. It's about taste. It's about choosing which problem to even look at, and knowing when a proof is not just correct but beautiful. That's what I picked up from a decade of talking to mathematicians who actually love their work — and it's exactly what's being squeezed out of the field today.
The piece that got my gears turning — posted on Hacker News, of all places — asks a simple question: Is mathematics about to enter the conservatory? Not a literal conservatory, of course. The metaphor: math is turning into classical music. A discipline where you practice endlessly, follow strict rules, and perform for a tiny audience of connoisseurs. Innovation? Daring? That's for the jazz age, not the philharmonic.
The creeping conservatism of mathematics
Here's what's happening. Over the past few decades, mathematics has become increasingly formalized. Papers are dense, proofs are long, and the emphasis is on rigor over insight. You can't just say, "This seems right, and here's a sketch." You need to dot every i, cross every t, and make sure no referee in Stockholm can find a gap.
That's not inherently bad. Rigor is the backbone of the discipline. But when it becomes the only thing that matters, you lose the creative spark. Mathematicians are supposed to be explorers, not auditors. They're supposed to follow a hunch, chase a pattern, and occasionally be wrong — but productively wrong.
"The conservatory approach rewards those who can play the existing repertoire flawlessly. But it punishes those who want to compose something new."
And that's exactly what's happening. Young mathematicians are trained to solve problems within established frameworks, not to question the frameworks themselves. Tenure committees look at citation counts and publication numbers, not at whether someone tackled a bold problem and failed spectacularly.
The AI factor — a new layer of polish
Now throw AI into the mix. With tools like Lean and Coq, you can formalize proofs in ways that were impossible a decade ago. That's a boon for verification. But it's also a trap.
Already, I've seen papers that read like they were generated by an algorithm — because they were. Not the ideas, necessarily, but the presentation. Every proof is a smooth, frictionless march from assumptions to conclusion. No dead ends. No intuition. No personality.
That's the conservatory effect on steroids. You get perfect performance, but you lose the human element. The music is technically flawless, but it doesn't move you.
What we lose when math becomes a recital
Think about the great mathematicians of history — Gauss, Riemann, Grothendieck. They weren't just technicians. They were visionaries who had an almost artistic sense of what was important. Gauss allegedly said that a mathematician "who is also something of a poet will never be a perfect mathematician." But that was about the perfection, not the poetry.
Today, we're producing a lot of perfect mathematicians. And they're doing important work, no doubt — cryptography, optimization, data science. But the field is losing its soul. The number of truly unexpected breakthroughs — the kind that make you gasp and rethink everything — has dwindled.
Take the Langlands program. It's been called the "grand unified theory of mathematics," and it's the kind of far-reaching connection that should make every researcher's heart race. But how many young mathematicians are actually working on it? Too few, because it requires deep intuition across multiple fields. It's not flashy. It doesn't produce quick papers.
The cost of the conservatory model
The conservatory model is driving people away. I've talked to grad students who love math but hate the publish-or-perish treadmill. They're expected to churn out incremental results in obscure subfields, not to take risks. The message is clear: don't try anything new unless you're sure it will work.
That's the opposite of what makes math exciting. The best moments in the field come from someone daring to ask a silly question — and then finding out it's not silly at all.
And let's be honest: the conservatory model also creates a culture of intellectual conformity. If everyone is playing the same game, you get groupthink. You get schools of thought that dominate for decades, not because they're right, but because they're safe.
There's a reason the Fields Medal has gone to people like Maryam Mirzakhani, who worked on moduli spaces and dynamics — fields that were considered fringe when she started. She didn't follow the crowd. She made her own path. But how many Maryams are we losing because they're told to stick to the canonical problems?
Can we change the tune?
Here's the thing: mathematics has been through this before. In the early 20th century, the formalists — led by David Hilbert — pushed for total rigor. They wanted to reduce math to a set of axioms and rules. And they succeeded, in a way. But they also sparked a reaction: the intuitionists, who insisted that math was a human activity, not a formal game.
That tension is healthy. Too much formalism, and you lose the creative spark. Too much intuitionism, and you get sloppy. The field needs both — the rigor to check whether an idea is correct, and the intuition to come up with the idea in the first place.
But right now, the pendulum has swung too far toward the formal side. The AI tools that are supposed to help us verify proofs are making it worse. Instead of freeing mathematicians to think bigger, they're enabling them to work on smaller, more technical problems and call it a day.
"We need to teach math as an art, not a martial art."
There are signs of pushback. Some journals have started to encourage "experimental" math — papers that include computer experiments and conjectures, not just proofs. Some conferences are featuring talks on failed approaches, because they can be just as instructive as successes. But it's not enough.
What we need — and what we can't have
We need to change how we train mathematicians. That means less focus on technical skill and more on taste. It means encouraging students to read the works of the masters — not to imitate them, but to understand how they thought. It means giving researchers the freedom to follow their curiosity, even if it doesn't lead to a publication.
We need to change how we evaluate mathematicians. That's harder. In a world of scarce resources, universities and funding agencies want measurable output. But citations and publication counts are a poor proxy for intellectual impact.
We need to accept that some of the most important mathematical work is invisible. It's the kind of insight that doesn't lead to a paper but that shapes how other people think. It's the kind of idea that you can't put in a formal proof because it's a feeling, a hunch, a direction.
But here's the uncomfortable truth: we probably won't get any of that. The conservatory is too entrenched. The incentives are too strong. And now, with AI, we have a way to make the whole thing even more mechanical.
So yes, mathematics is about to enter the conservatory. And if it does, it will survive — for a while. The concert halls will still be full, the performers will be virtuosos, and the audience will applaud politely. But the music will be dead. The spark that made it a human endeavor, a form of intellectual play, will be gone.
The question is: are we OK with that? Do we want a field that's pristine but passionless? Or do we want to risk something — to let a few wrong notes ring out, because they might just lead to a new symphony?
I know which I prefer. I just don't know if it matters anymore.
