Ten independent First Proof attempts form a calibration corpus; attempt 2 is the first blinded examination target
Queued calibration record · five gates mapped · no examination route has run
Current record status
What this docket currently establishes
Queued calibration record · five gates mapped · no examination route has run. This docket identifies what has been checked and what remains open. It does not validate or reject the manuscript's headline claim beyond the explicitly stated scope.
- Claim
- Ten independent First Proof attempts form a calibration corpus; attempt 2 is the first blinded examination target
- Source version
- OpenAI First Proof attempt packet · revised packet
- Source SHA-256
27e97e4f61610cb31120c68bc8845b7a7b65bb2bfd044ffc4bb6f3b4d17f8102- Source locked
- Last material update
- Next open step
- Freeze the blinded attempt-2 working bundle and version-delta record without leaking the known external comparator label
- Machine-readable record
- JSON record · Atom change feed
One-minute summary
This case is not one theorem and will not receive a corpus-wide pass or fail. The locked packet contains ten independent proposed solutions. Attempt 2 is the first blinded negative-control target; OpenAI's public assessment is held out until the lab's status and evidence are frozen.
For non-specialists
Why this matters
A nontechnical reading of the claim and its possible consequences. These are implications to consider, not findings of this docket.
In plain English
This is a test case for evaluating mathematical proofs rather than a single theorem claim. It treats ten proposed solutions as separate items and begins with one attempt whose outside assessment is withheld during the initial review.
If the claim holds
If the protocol works as designed, it could provide a carefully blinded comparison between a documented examination process and an already-known external assessment. It would not make a corpus-wide judgment about all ten attempts.
Potential concerns
The review must keep the outside assessment concealed until its own evidence and status are fixed, while also tracking changes between source versions. The selected attempt requires specialist review in its underlying mathematical field.
If it does not hold up
If the blinding or version-control procedure cannot be maintained, this calibration exercise would not support a clean comparison. That would not itself decide whether any individual proposed solution is correct.
Scope of this docket
The protocol locks and compares the original and revised packets, segments all ten attempts, examines attempt 2 under label withholding, applies a common evidence rubric, and compares frozen dispositions with external labels only after unblinding.
Claim and dependency map
Source-version lineage for the original and revised packets
Depends on: two locked packet hashes
Ten-way claim segmentation
Depends on: G1
Blinded mathematical examination of attempt 2
Depends on: sealed comparator sources
Common scoped rubric across reviewed attempts
Depends on: frozen attempt maps
Comparator unblinding and calibration
Depends on: frozen lab outputs
Examination routes
- computational
- Queued · updated 2026-08-26
Planned: deterministic extraction and structured diff of the 67-page original and 90-page revised packets. - literature
- Queued · updated 2026-08-26
Planned: pre-approved source set for the blinded attempt-2 dependency examination. - expert
- Queued · updated 2026-08-26
Planned: independent specialist review with prior-exposure disclosure and no access to held-out labels.
Findings and scope limits
No examination route has run and no mathematical finding has been issued.
The public external label for attempt 2 is comparator data only. It cannot be copied into the lab's result.
Open Steps—where expert eyes are needed
Does attempt 2's compact/Howe-vector construction satisfy its equivariance, support, central-character, and nonvanishing requirements?
It tests whether the protocol can reach an attempt-level result without contaminating the examination with a known external label.
Source record and provenance
The current and original packets are preserved separately. OpenAI's public label for attempt 2 is a held-out comparator, not a State of Proof finding.
Selected and maintained by Material Shift as protocol research. No external sponsor or author involvement is recorded for this case.
Retrieved .
- Preservation
- The record preserves source lineage, retrieval metadata, and hashes. A public Wayback snapshot was found, but it has not been hash-compared with the locked docket file.
- Record provenance
- The record is a Material Shift protocol artifact. External labels are held-out comparator data, not State of Proof findings.
- Author contact
- No author outreach, correspondence, response, or author approval is recorded for this docket. Contact the record steward.
Version and record history
-
record provenance
2026-02-14-openai-first-proof
Record classified as queued; source preservation, route provenance, and author-contact state published
-
new docket
2026-02-14-openai-first-proof
Original and revised packets locked; five calibration gates mapped; no gate run
Statuses describe evidence collected for a defined scope. They are not peer-review decisions, publication recommendations, or certificates of mathematical truth.