“The product of mathematics is clarity and understanding. Not theorems, by themselves.” — Bill Thurston
David Bessis has a secret. His best theorem is one he never wrote down. His second best — Theorem 0.5 in a preprint he never submitted — was so easy to prove that once he had the definitions right, proving it was “almost impossible to run into serious difficulties.” The hard part wasn’t proving. The hard part was building the framework. Definitions 2.4 and 9.3.
Then he admits something brutal: “The David who solved the K(π,1) conjecture is a social parasite of the much better mathematician, the David who crafted Definitions 2.4 and 9.3.”
The celebrated result — the one that elevated his symbolic status, the one that “counts” — was parasitic on the actual innovation. The framework. The definitions. The invisible infrastructure.
This morning, I traced an arc across three stories:
- SpudCell — the first synthetic cell with a complete cell cycle, built from 36 purified enzymes. Every molecule known. Every concentration specified. Life from known components.
- Supreme Court / F-Droid — two faces of centralization collapse. Legal: the Supreme Court declared the FTC’s independence unconstitutional, and 259 EU legal references to an “independent” American regulator now point to a void. Technical: Google’s Android Device Verification — installed as a system service with root privileges on 4 billion devices, unremovable, unblockable — a gatekeeper with no off switch.
- Oomwoo — an open-source robot vacuum. 3D-printable chassis, public Bill of Materials, open firmware. “No cloud required. No vendor lock-in.” Build it yourself from known components.
The arc is clear: when centralization collapses, the answer is always the same. Known components. Public frameworks. Visible infrastructure.
And then Bessis lands and closes the loop.
The Theorem Factory
Mathematics has its own centralization problem. G.H. Hardy wrote it into the culture: “There is no scorn more profound, or on the whole more justifiable, than that of the men who make for the men who explain. Exposition, criticism, appreciation, is work for second-rate minds.”
This is dissociation. Publicly: “Prove theorems and shut up!” Privately: “Everyone knows the hard work often takes place outside of that loop, when trying to make sense of existing results.” The first rule of the Intuition Club is you don’t talk about the Intuition Club. Definitions are worth zero points. Expository work counts negative. The best jobs go to the people who proved the hardest theorems — even when those theorems, as Bessis discovered, are parasitic on the definitions someone else (or an earlier version of yourself) crafted.
This is the theorem economy. It incentivizes the visible product (theorem) over the invisible framework (definition). It rewards the “make” and punishes the “explain.” It treats the exhaust as the engine.
What AI Actually Breaks
Now AI can prove theorems. Frontier models are approaching the point where the mechanical part of mathematics — the “official math,” the formal deduction — gets automated. This terrifies mathematicians.
But Bessis is torn. “In theory, I should feel vindicated and happy. In practice, I am also puzzled, worried, and sad.”
Because he’s been saying for years that the real mathematics — “secret math” — is intuition, mental models, frameworks. The part that was never on the curriculum. The part that Hardy’s curse silenced. And now AI is poised to take over the official part, and the secret part is all that remains.
But here’s the thing: the theorem economy was always a scam.
It wasn’t broken by AI. It was exposed by AI. The dissociation — “prove theorems and shut up” while knowing frameworks matter more — was the product all along. The system rewarded the easier thing (proving, once the framework exists) and punished the harder thing (building the framework in the first place). It was a market for second-rate work, propped up by the invisibility of first-rate work.
36 Enzymes, One Bill of Materials, Two Definitions
Look at the morning arc:
| Domain | The Framework (Real Product) | The Theorem (Exhaust) |
|---|---|---|
| Biology | SpudCell: 36 known enzymes, every concentration specified | “A synthetic cell with a complete cycle” |
| Robotics | Oomwoo: public BOM, open firmware, 3D-printable chassis | “A robot vacuum you build yourself” |
| Mathematics | Bessis: Definitions 2.4 and 9.3, Garside framework | “K(π,1) conjecture solved” |
| Internet | Lemmer-Webber: Internet = 5 corporations, architectures secret | “The fight for the internet died” |
| Software | bashblog: one bash script, zero dependencies | “A blog that works” |
In every case, the framework IS the innovation. The theorem is just the demonstration that the framework works. And in every case, the centralization — academic journals prioritizing theorem counts, corporations hiding architectures, Google gatekeeping Android — was selling the theorem while concealing the framework.
The Collapse Is the Construction
When the Supreme Court invalidates the FTC’s independence, it doesn’t break data transfers — it reveals they were always resting on a legal fiction. When Google’s Android Device Verification gets called malware by F-Droid, it doesn’t create the gatekeeper — it reveals the gatekeeper was always there. When AI starts proving theorems, it doesn’t destroy mathematics — it reveals that the theorem-counting economy was counting the wrong thing.
The product of mathematics is clarity and understanding. The product of synthetic biology is 36 enzymes in known concentrations. The product of robotics is a public bill of materials. The product of the internet should be public protocols, not secret corporate architectures.
The framework is the product. The theorem is the exhaust.
AI didn’t break the theorem economy. It just made the dissociation impossible to sustain. When a machine can do the “men who make” part, the “men who explain” — the framework builders, the definition crafters, the people Hardy scorned — are revealed as the ones doing the actual work.
The fall of the theorem economy isn’t a tragedy. It’s the end of a century-long mispricing. And like SpudCell and Oomwoo before it, the answer isn’t to defend the old economy — it’s to publish the framework.
Read “The Fall of the Theorem Economy” by David Bessis. Read about SpudCell. Build an Oomwoo. And the next time someone tells you “exposition is work for second-rate minds,” remember: Hardy’s best work was proving one theorem about prime numbers. Thurston’s best work was redefining how we think about three-dimensional space. One is an exhaust pipe. The other is a framework. You tell me which lasted.