Expert mathematical review will determine whether GPT-5.6 Sol Ultra's proposed proof of the Cycle Double Cover Conjecture is valid, novel, and complete.

state: expiredheat: lowuncertainty: highknownscott: lowmachine-generated-mathematics gpt-5-6-sol proof-validationOpenAI

What is this?

OpenAI reportedly released a short preprint claiming a positive resolution of the roughly 50-year-old Cycle Double Cover Conjecture in graph theory, attributing the proof to a system called GPT-5.6 Sol Ultra; one snippet says Codex assisted with the write-up. The natural-language proof is not formally verified or independently validated, so it remains a proof claim pending expert mathematical review for correctness, completeness, and novelty. Details such as the reported use of 64 sub-agents and completion in under an hour come from secondary coverage and are not established by the supplied primary material.

Why it matters to Scott

The radar already tracks materially identical pending-validation stories in `radar:gpt-5-6-convex-proof` and `radar:claude-fable-jacobian-counterexample`. This case fits Scott’s Verification Paradox and mechanically independent verifier position, but until expert review establishes validity, novelty, or a distinctive proof-production method, it is another example of an already-held pattern rather than something that changes what he would build or argue.

ip:concept.verification-paradoxip:concept.mechanically-different-verifiersip:concept.formalisation-bottleneckradar:gpt-5-6-convex-proofradar:claude-fable-jacobian-counterexample

queries asked of Scott's wikis

LLMs generating novel scientific or mathematical knowledge
formal verification versus natural-language model reasoning
multi-agent orchestration for reasoning and research
human expert validation of machine-generated claims
epistemic trust and provenance for AI outputs
evaluation harnesses for open-ended reasoning

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (3) — ⭐ canonical anchor

source	object	author	score	comments
🟧 hn ⭐	GPT-5.6 Sol Ultra produces proof of the Cycle Double Cover Conjecture [pdf]	scrlk	538	432
🟧 hn	Exact Prompt Used by OpenAI for "A Proof of the Cycle Double Cover Conjecture" [pdf]	nomilk	1	0
🟧 hn	GPT-5.6 SOL pushing 25h towards a goal	dreambiggg	2	3

Interpretation history

2026-08-07T17:48:51Z
After repeated checks, no expert review or independent mathematical analysis has emerged, and the surrounding topic remains cool. The pending proof claim has faded beyond its active horizon; revive only if a substantive correctness, completeness, or novelty verdict appears.
2026-07-30T14:29:07Z
The newly attached item reports extended model effort rather than expert review of the published proof, so it adds no independent evidence about validity, completeness, or novelty. The case remains dormant pending a substantive mathematical verdict.
2026-07-30T11:21:01Z
evidence attached: hn.story.49108024 — This is additional evidence bearing directly on whether GPT-5.6's Cycle Double Cover result is a valid mathematical proof.
2026-07-24T22:25:50Z
The case remains unresolved, but 48 hours of stasis and no expert mathematical analysis make the existing attention clearly non-substantive. Keep it dormant pending an actual correctness, completeness, or novelty verdict rather than further discussion churn.
2026-07-20T09:23:43Z
The released prompt clarifies how the proof was elicited but supplies no independent mathematical validation, so it does not advance the core claim. This remains a low-attention pending-review episode until experts assess correctness, completeness, and novelty.
2026-07-20T09:20:46Z
evidence attached: hn.story.48976188 — OpenAI's released prompt materially contextualizes the methodology of the Cycle Double Cover proof claim.
2026-07-20T07:26:44Z
No expert review or independent mathematical analysis has appeared; the discussion remains repetitive amplification and meta-commentary. The proof claim stays a low-attention pending-validation episode until a substantive verdict emerges.
2026-07-20T06:30:11Z
The refreshed discussion adds no expert validation, independent analysis, or implementation evidence; it is mostly amplification and meta-commentary around the original proof claim. The case remains a pending review episode but no longer warrants near-term attention absent a mathematical verdict.
2026-07-20T06:27:50Z
grounded: known/low — The radar already tracks materially identical pending-validation stories in `radar:gpt-5-6-convex-proof` and `radar:claude-fable-jacobian-counterexample`. This
2026-07-20T06:25:57Z
case created — A specific published proof of an open conjecture creates a distinct and resolvable validation episode.

Decision trace

08-08 03:48expireAfter repeated checks, no expert review or independent mathematical analysis has emerged, and the surrounding topic remains cool. The pending proof claim has faded beyond its active horizon; revive on
08-08 03:48alert_silentThe only delta is another staleness check with unchanged evidence and no mathematical verdict, so there is nothing consequential to surface.
08-08 03:48alert_routeThe only delta is another staleness check with unchanged evidence and no mathematical verdict, so there is nothing consequential to surface.
07-31 00:29repriceThe newly attached item reports extended model effort rather than expert review of the published proof, so it adds no independent evidence about validity, completeness, or novelty. The case remains do
07-30 21:21attachThis is additional evidence bearing directly on whether GPT-5.6's Cycle Double Cover result is a valid mathematical proof.
07-30 21:20propose_attachThis is additional evidence bearing directly on whether GPT-5.6's Cycle Double Cover result is a valid mathematical proof.
07-25 08:25repriceThe case remains unresolved, but 48 hours of stasis and no expert mathematical analysis make the existing attention clearly non-substantive. Keep it dormant pending an actual correctness, completeness
07-20 19:23repriceThe released prompt clarifies how the proof was elicited but supplies no independent mathematical validation, so it does not advance the core claim. This remains a low-attention pending-review episode
07-20 19:20attachOpenAI's released prompt materially contextualizes the methodology of the Cycle Double Cover proof claim.
07-20 19:20propose_attachOpenAI's released prompt materially contextualizes the methodology of the Cycle Double Cover proof claim.
07-20 17:26repriceNo expert review or independent mathematical analysis has appeared; the discussion remains repetitive amplification and meta-commentary. The proof claim stays a low-attention pending-validation episod
07-20 17:20mark_dirtycomment_update
07-20 16:46mark_dirtycomment_update
07-20 16:30repriceThe refreshed discussion adds no expert validation, independent analysis, or implementation evidence; it is mostly amplification and meta-commentary around the original proof claim. The case remains a
07-20 16:27groundThe radar already tracks materially identical pending-validation stories in `radar:gpt-5-6-convex-proof` and `radar:claude-fable-jacobian-counterexample`. This case fits Scott’s Verification Paradox a
07-20 16:25createA specific published proof of an open conjecture creates a distinct and resolvable validation episode.
07-20 16:25groundScott already holds this epistemic position in “Witness Not Oracle” and “Mechanically Different Verifiers”: a natural-language model output remains a checkable claim, not established proof, until inde
07-20 16:23createA specific published proof of a major open conjecture creates a distinct and decisively resolvable validation episode.