Imagine you are an architect working on the 87th floor of a building. One day you discover a problem with the foundation three levels underground—not a small crack, but a flaw in the design language itself. You cannot demolish and rebuild, because the 86 floors above are all in use. All you can do is quietly route around it, patch it, drive in steel pins.
That is exactly the situation mathematicians have been in for the past century.
In 1914, the German mathematician Felix Hausdorff—a poet, philosopher, and playwright as well—defined an object called a "topological space" in his *Foundations of Set Theory*. From then on, it became the foundation of mathematics. Calculus, geometry, algebra, number theory—nearly every field built on top of it. But there is an awkward fact: topological spaces do not get along with algebra.
This sounds like a minor annoyance. It is not. Algebra is one of mathematicians' most powerful tools, and topological spaces are their most natural way to describe "shape." A mathematician wanting to use both at once—like wanting to use your left and right hands together—found that the topological foundation made algebraic tools slip. For a century, everyone made do. The people working on the 87th floor never went down to fix the basement.
Until 2019, when Peter Scholze and Dustin Clausen decided to dig down.
1. Topological spaces: a century-old design
First, what is a topological space?
Imagine standing on a plot of land. You don't need to know how many meters apart two trees are; you only need to know "they are in the same region." Hausdorff's design is exactly this: a topological space is a collection of points plus a collection of "regions" called open sets, satisfying two rules—any union of open sets is open, and any finite intersection of open sets is open.
That simple. But the definition's power is astonishing.
With it, mathematicians can talk about "continuity" without distance, "connectedness" without measurement. The Fundamental Theorem of Algebra—the one Gauss stumbled over—can be proven in three lines with a topological argument. Surfaces can be classified: sphere, torus, double torus... just count holes.
After Hausdorff's book appeared, topology went from scattered insights to a discipline with a rigorous language. The French collective Bourbaki later wrote that Hausdorff's definition gave the theory "complete precision and complete generality."
But the problem was planted too.
2. Category theory: a broken car on the highway
In 1945, two American mathematicians, Samuel Eilenberg and Saunders MacLane, published a bold paper creating a new discipline—category theory.
Category theory is not about specific mathematical objects; it is about "relationships between mathematical objects." A category consists of objects and the "morphisms" between them. Even more powerful are "functors," which map an entire category to another, translating objects while preserving relationships.
It was like building highways between the mountain ridges of mathematics.
But topologists couldn't get on this highway. Or rather, they could—but driving a junker that kept stalling.
The reason: the category of topological spaces lacks key properties that category theorists need. Algebraists wanted to use category theory to connect topology and algebra—constructing objects that are both topological and algebraic, called "topological abelian groups." But the category of topological abelian groups is simply bad to work with. Abstract nonsense that yields powerful conclusions in other categories yields only nonsense here.
Scholze later put it bluntly: "I think topologists actually don't like topological spaces, because it's not a convenient category."
This is not a small complaint. Category theory was one of the greatest accelerators of late twentieth-century mathematics. Grothendieck used it in the 1950s–60s to rebuild algebraic geometry, transforming the field from a set of techniques into a language, with influence to this day. Topology was excluded from that revolution—like a city never connected to the high-speed rail network.
3. Scholze and Clausen: two unusual people
Peter Scholze, Fields Medalist in 2018 and Germany's youngest full professor at 25, has an odd style—he says he is "not that interested in theorems or proofs." What interests him is naming things.
> "To a large extent I'm just rephrasing in my own words what other people have done. I'm trying to give names to things that already exist."
> "They have to make interesting theorems easy to state, and easy to prove."
Dustin Clausen, Scholze's longtime collaborator, is even more extreme. He does not publish papers—because he believes academic publishing is fundamentally broken. He doesn't even bother writing informal notes, leaving that to collaborators. He once seriously considered becoming a literary translator.
Together, what they set out to do sounds a bit crazy: replace topological spaces.
Not patching, not improving—changing the foundation.
4. Condensed sets: building continuity from dust
Their new object is called "condensed sets."
The most intuitive way to understand it starts with an old friend—the Cantor set.
Take a line segment of length 1 and remove its middle third. Then remove the middle third of each remaining segment. Then the middle third of each of the four remaining segments. Repeat infinitely.
What remains is a "dust" of points. No two points are adjacent—every point is separated from every other by gaps. This is a "totally disconnected" space.
Intuitively, this dust seems "more fragmented" than the continuous real line. But Scholze and Clausen found: continuous things can be built from this dust.
How? By "gluing" the dust together.
The most familiar example is right at hand—decimal notation.
The number 0.5 can be written as 0.5000... or as 0.4999.... Same number. But the decimal expansion is itself a "dust-ification" process—each additional digit cuts the number line into a smaller piece. 0.4 cuts the line to the segment [0.4, 0.5). 0.49 cuts to [0.49, 0.50). 0.499 cuts to [0.499, 0.500). Do this infinitely, and every point is completely separated from every other—dust.
But once you accept that 0.4999... = 0.5, the dust glues back into the continuous real line.
Condensed sets are the rigorous version of this "dust + gluing." The Cantor set is the simplest condensed set and the building block for constructing others. By "smashing together" many Cantor-set-like dusts in strange ways, you can rebuild the real line and everything topological spaces could give you—and algebra runs on it smoothly.
> "These totally disconnected pieces are extremely simple algebraically." —Scholze
This is the most counterintuitive part. We assume "continuous" is more complex and harder to handle than "discrete." Scholze and Clausen say: no, building the continuous from the discrete is simpler than handling the continuous directly.
5. Solving a problem "we didn't know we had"
The Stanford mathematician Ravi Vakil, also president of the American Mathematical Society, said:
> "They're solving a problem we didn't know we had, because we thought we already had a reasonable solution."
> "As a result, a huge swath of mathematics just got simpler."
That is worth savoring.
For a hundred years, mathematicians assumed topological spaces were "the right language." Not because they were perfect—everyone knew they clashed with algebra—but because there was no alternative. It's like using an awkward screwdriver for thirty years: you no longer notice it's awkward, because your wrist has deformed to accommodate it.
Scholze and Clausen swapped the screwdriver. Suddenly, everyone's wrist stopped hurting.
6. A quiet revolution
The revolution proceeded quietly.
In April 2019, Scholze taught a course on "condensed mathematics" at the University of Bonn. In May, he posted 77 pages of lecture notes. The notes ended with a new proof of an important theorem called "coherent duality"—the previous proof was extremely convoluted and technical; the new one was clean and elegant.
That gave mathematicians worldwide confidence. Johan Commelin in Freiburg organized a reading group. "Nobody in our group understood all the details," he admitted.
But condensed sets began to spread. From 2019 onward, Scholze and Clausen defined further variants—"light," "solid," "liquid," "gaseous"—a whole family of condensed mathematics. They also introduced "analytic stacks" and "gestalten"—which some mathematicians consider even more important than condensed sets themselves.
Clark Barwick and his student Peter Haine independently defined nearly the same thing, called "pyknotic sets" (from the Greek for "dense"). Barwick said:
> "The real excitement is in defining new objects of study. It tells you there's a class of natural objects you've never looked at before—like an unclimbed mountain. We're just nibbling at the corner of this vast territory."
Dagur Asgeirsson, a former student of Clausen, put it more directly:
> "I think it's fair to compare Peter to Grothendieck. He's in some way reinventing everything."
7. Why naming matters more than proving theorems
Scholze says he doesn't care about theorems, only about naming things. That sounds like slacking. It isn't.
Mathematical history has a strange pattern: major breakthroughs are often not new theorems but new languages.
Grothendieck rebuilt algebraic geometry not by proving new theorems, but by introducing new words—"scheme," "topos," "motive." These words made previously invisible structures visible, and previously unprovable theorems obvious.
Hausdorff's definition of topological spaces also wasn't a new theorem—it was a rigorous language for the intuitive concept of "nearness."
Scholze and Clausen are doing the same. They aren't discovering new continents; they're giving existing continents better names. Once the right name exists, mountains that were hard to climb become gentle slopes.
Scholze himself is modest:
> "I'm not inventing anything. I'm just trying to give names to things that already exist."
Grothendieck wrote something similar in his late memoirs:
> "The most beautiful house is not the bigger, taller one, but the one that faithfully reflects the hidden structure and beauty in things."
Mathematicians are builders, but not inventors. They discover things that already exist.
8. A cross-domain analogy: solving problems at a different level
This story reminds me of a larger pattern.
Octopuses don't modify their DNA blueprint to adapt to the environment—they edit RNA, making adjustments at the blueprint level. Slime molds don't memorize with neurons—they externalize memory into slime trails. Birds don't use GPS chips—they sense magnetic fields via quantum-entangled radical pairs. Mantis shrimp don't rely on thicker shells to withstand impacts—the inner layer of their shield is a photonic crystal that selectively filters out dangerous frequency bands.
None of these are "doing the same thing better"—they all solve the problem at a different level.
What Scholze and Clausen did is the same. They didn't make topological spaces finer, more complex, more powerful—that would be "building thicker walls." They changed levels: from "open sets" to "dust + gluing." From "handling continuous objects directly" to "constructing the continuous from the discrete."
And like the octopus, the slime mold, the mantis shrimp, this "level change" is not evasion—it is smarter use of what already exists. The Cantor set has existed since the nineteenth century, category theory since 1945, decimal expansions even earlier. Scholze and Clausen invented no new parts—they just recombined them and changed the perspective.
Admit that the old tool fails at a certain level, route around it, rather than forcing a repair. That is an engineering principle, and a biological principle—and now, a mathematical one.
9. One level deeper: language shapes thought
Condensed mathematics also points to a deeper insight.
We assume mathematics is "objective"—theorems are just there, waiting to be discovered. But the Scholze–Clausen story shows: language shapes what you can see.
In the language of topological spaces, you cannot see the unification of topology and algebra—because the language itself keeps them apart. Switch to the language of condensed sets, and the unification appears naturally.
This is not unique to mathematics. In physics, Newton's language of "force" made conservation of energy invisible—until Leibniz's language of "vis viva" appeared. In chemistry, the "phlogiston" language made oxygen invisible—until Lavoisier changed the language. In AI, the "supervised learning" language made self-supervision invisible—until the language of contrastive learning appeared.
The world didn't change—the language did. The same world, but different languages reveal different structures.
Scholze says he "tries to give names to things that already exist." That remark runs deeper than his modesty—naming is not invention; it makes what already exists visible.
10. Next step: quantum field theory?
Scholze has hinted that condensed mathematics may not stop at number theory and topology. Quantum field theory—the core of modern physics—has long been hampered by unclear foundations. It uses extremely sophisticated algebra and topology, but the two have never meshed.
> "Quantum field theory is essentially very analytic and topological. Mixing these two worlds is not a small matter, but condensed mathematics offers a possible framework." —Scholze
If this path works out, condensed mathematics may be not just a reconstruction of mathematics, but of physics.
But it is too early to say. Scholze himself is unsure how widely condensed sets will be adopted—they are powerful, but also complex and hard to learn. What he sees is more like a first step—the beginning of a larger plan to "understand why numbers behave the way they do."
Epilogue
In 1914, Hausdorff—a poet, philosopher, playwright—gave mathematics a definition. For a hundred years, that definition held up the entire mathematical world.
In 2019, two men who dislike publishing papers—one who cares only about naming things, one who considered becoming a literary translator—decided to replace that definition.
They did not overturn Hausdorff. They simply discovered: continuity built from dust works better than continuity handed down directly.
That sounds like a paradox. But mathematics is like this. The deepest breakthroughs are often not "trying harder" but "changing languages." Not discovering new continents, but giving old continents better names.
Scholze said: "I'm not inventing anything. I'm just trying to give names to things that already exist."
Perhaps that is the essence of all deep innovation—not creation, but seeing.