OpenAI reports a 21 move Maker strategy for Snaky

In the initially empty infinite square-board weak Maker–Breaker game, Maker can complete a translated, rotated or reflected Snaky hexomino within 21 legal Maker claims against every legal Breaker play.

Sources for this record

What was already known

Earlier planar results gave a Maker win with a one-stone handicap, while the ordinary empty-board planar case remained unresolved in the cited literature.

AI involvement AI-led

OpenAI says its internal frontier model produced this mathematics release and credits OpenAI as manuscript author. The source does not detail the individual human mathematical input for this result.

OpenAI says its internal frontier model produced this release and credits OpenAI as the paper author. The source does not establish a more precise individual autonomy claim.

Importance Notable

An explicit strategy settles the ordinary planar Snaky achievement question and gives a 21 move upper bound.

Limitations

The 21 move theorem does not claim the minimum move count or a strategy for the strong game. The Lean scope note checks the infinite-board strategy but not the paper’s separate 17×17-board restriction. No named human reconstruction or independent proof review is documented.

Technical verification Formal (Lean 4)

The pinned repository note states that its Lean theorem covers the selected statement, with a named Comparator theorem and source file. This is source-reported machine checking; repository review is unchecked and no independent run is recorded here.

Formal checking

The pinned scope note identifies a Lean proof of the selected statement and links the comparator theorem.

Checker
OpenAI-reported Lean 4 / Comparator
Scope
In the initially empty infinite square-board weak Maker–Breaker game, Maker can complete a translated, rotated or reflected Snaky hexomino within 21 legal Maker claims against every legal Breaker play.
Basis
Reported by the linked checker
Evidence source

Checking a formal statement does not establish that it matches the original problem.

Human understanding Few

Few is the default when evidence for a higher level is insufficient. The manuscript explains the proof, but OpenAI is its credited author and no named human reconstruction of the central argument is documented. Few is the evidence default, not a finding that people cannot understand it.

Correction and publication history

2026-10-18 · publication

Primary manuscript, prior-work and pinned formalization review of one OpenAI-authored contribution; historical non-notifying inclusion.

Read this published revision