Riemann Hypothesis machine proof likely correct but unread, mathematicians debate
Rumors have surfaced of a roughly 60GB Lean formalization of the Riemann hypothesis, prompting mathematician gleech and user peroxycarbonate to engage in a multi-round debate over "the value of machine proofs." The dispute is not about whether the proof is true, but whether such an enormous body of code can translate into mathematical understanding. Their positions remain firmly opposed and unresolved, yet the discussion itself exposes a core concern of the mathematics community in the AI era: the relationship between "proof" and "insight."
Confirmed
- On 11-21, peroxycarbonate posted the initial challenge: the importance of the Riemann hypothesis lies in how solving it could yield a deeper understanding of the primes, and a pile of Lean code cannot provide that understanding.
- Asked how 60GB of Lean code could be converted into useful understanding, gleech offered three paths, including a renewed enthusiasm for bottom-up independent research, possibly drawing in part on some 20 pages of LLM-related material.
- gleech argued that the core answer to "why care about this proof" is morale: once people know something is possible, they throw themselves at it madly. He expects people to extract some insights from it by year's end, and to be inspired to produce a proof under 800 pages within a few years.
Unconfirmed
- The so-called 60GB Lean proof and the "Claude RH plan" remain rumors; the entire discussion rests on unverified hearsay, and its content, structure, and correctness are all unconfirmed.
Why it matters
- peroxycarbonate cited the computer-assisted proof of the four-color theorem as a cautionary tale: it is far smaller in scale yet no insight has ever been compressed out of it, and he suspects the Riemann proof would go the same way. gleech responded that the four-color theorem is an exhaustive proof and naturally resists compression into insight, whereas his understanding of the Claude RH plan is that it is more interesting and more structured, and he is highly confident it is not optimal — leaving room to compress and extract insight.
- The debate reflects the tension between "knowing that it's true" and "understanding why it's true" as formal and AI-generated proofs enter core mathematics, and it may shape how the mathematical community accepts and digests giant machine proofs.
2026-11-21 ~ 2026-11-23 · 6 related posts
Primary sources
- [source] Why Care About a Massive Lean Proof of the Riemann Hypothesis? — peroxycarbonate · 2026-11-21
- [source] The Value of a Machine Proof of Riemann Is Morale, Mathematician Argues — gleech · 2026-11-22
- Skeptic Questions How 60GB of Lean Proof Code Becomes Understanding — peroxycarbonate · 2026-11-23
- Mathematicians Debate: Can 60GB of Lean Code Turn Into Real Understanding? — gleech · 2026-11-23
- Four Color Theorem Was Exhaustion — This Lean Proof Is Different, Argues Mathematician — gleech · 2026-11-23
- [source] Daniel Litt: the Lean-formalized Riemann proof is likely correct, no human understands it yet — littmath · 2026-11-23