Looks 2D, isn’t?
An AI preprint claims a counterexample to Wall’s D(2) problem
Covers result 321 in OpenAI’s catalog: its one paper. See its connections to other results and the work it cites.
3:18. English captions available. The video loads from YouTube when you press play.
Chapters
The claim, in plain words
A finite cell complex of dimension 3 can sometimes be squashed, up to homotopy, to dimension 2. In 1965 C. T. C. Wall asked when. If the complex already looks 2-dimensional to homology, with nothing above degree 2 in its universal cover and no third cohomology for any twisted coefficients, must it be homotopy equivalent to a finite 2-complex? That is the D(2) problem.
It sits next to the Eilenberg–Ganea and Whitehead conjectures, and most work on it has been for finite fundamental groups, where it ties to deep questions about group presentations. In 2021 Mannan and Popiel wrote that confirming one main kind of counterexample, an Euler characteristic gap, would require a fundamentally novel obstruction. Before October 6, recent papers still listed the problem as open.
Result 321 is one OpenAI paper, 10 pages, dated and made public October 6, 2026. It claims a finite connected 3-dimensional complex that passes Wall’s homological tests, but has no finite 2-dimensional model. Its fundamental group is infinite: by our reading, two finite groups of orders 30 and 78 glued together, with an extra loop.
What it doesn’t claim
Not claimed: anything for finite fundamental groups, where the problem stays open, or anything about the Eilenberg–Ganea or Whitehead conjectures. The stabilized version, adding enough spheres, was already known to work.
Where it stands
- Posted as a preprint
- OpenAI, 10 pages, dated October 6, 2026
- Peer reviewed
- Not yet
- Checked in Lean
- No formalization for this item
Last checked October 19, 2026. The paper is unrefereed and unchanged since release, with no Lean formalization. The tower obstruction, which carries the proof, has not been publicly checked by anyone we know of. In the first 12 days we found no public comment from any mathematician, no reported error and no competing claim. OSVerify has not reviewed it.
How the proof goes
- A twisted coefficient system. The complex is built by a finite plus construction, which kills a perfect subgroup; perfection comes from the coprime numbers 15 and 26. Then the paper chooses a 1-dimensional twisted coefficient system, sending one element to −1 and another to a cube root of unity. Our computer recomputes the paper’s 4 × 5 Fox matrix entry by entry, confirms its key minor, −4(1 − ζ²), and finds that the obstruction needs both the sign and the cube root: either alone does nothing.
- A tower that cannot stop. The new ingredient is an obstruction, proved by a tower argument in the style of Howie. Any finite 2-complex with this fundamental group would have twisted second homology that cannot vanish, because some element of finite order sees the coefficient twist. But a 2-dimensional model would inherit the complex’s own vanishing twisted homology. Our computer checks this pattern on small complexes, from the projective plane up to presentations of the two finite groups: whenever the twist sees torsion, the twisted second homology is at least 1.
The numbers in the video
- Ours: the paper’s 4 × 5 Fox matrix, recomputed entry by entry, and its key minor −4(1 − ζ²).
- Ours: the obstruction needs both the sign −1 and a cube root of unity; either alone does nothing.
- Ours: on small complexes, from the projective plane to presentations of the two finite groups, the twisted second homology is at least 1 whenever the twist sees torsion.
Read the source
- A Counterexample to Wall’s D(2) Problem (PDF), 10 pages.
- The collection’s catalog, where this is result 321.
The paper
OpenAI groups related papers into one result; this one has a single paper.
Transcript
Read the full script
0:00The question
A finite cell complex of dimension 3 can sometimes be squashed, up to homotopy, to dimension 2. In 1965 C. T. C. Wall asked when. If the complex already looks 2-dimensional to homology, with nothing above degree 2 in its universal cover and no third cohomology for any twisted coefficients, must it be homotopy equivalent to a finite 2-complex? That is the D(2) problem. It sits next to the Eilenberg–Ganea and Whitehead conjectures, and most work on it has been for finite fundamental groups, where it ties to deep questions about group presentations. In 2021 Mannan and Popiel wrote that confirming one main kind of counterexample, an Euler characteristic gap, would require a fundamentally novel obstruction. Before October 6, recent papers still listed the problem as open.
0:52The claim
Result 321 is one OpenAI paper, 10 pages, dated and made public October 6, 2026. It claims a finite connected 3-dimensional complex that passes Wall’s homological tests, but has no finite 2-dimensional model. Its fundamental group is infinite: by our reading, two finite groups of orders 30 and 78 glued together, with an extra loop. Not claimed: anything for finite fundamental groups, where the problem stays open, or anything about the Eilenberg–Ganea or Whitehead conjectures. The stabilized version, adding enough spheres, was already known to work.
1:32How the proof goes
The complex is built by a finite plus construction, which kills a perfect subgroup; perfection comes from the coprime numbers 15 and 26. Then the paper chooses a 1-dimensional twisted coefficient system, sending one element to −1 and another to a cube root of unity. Our computer recomputes the paper’s 4 × 5 Fox matrix entry by entry, confirms its key minor, −4(1 − ζ²), and finds that the obstruction needs both the sign and the cube root: either alone does nothing. The new ingredient is an obstruction, proved by a tower argument in the style of Howie. Any finite 2-complex with this fundamental group would have twisted second homology that cannot vanish, because some element of finite order sees the coefficient twist. But a 2-dimensional model would inherit the complex’s own vanishing twisted homology. Our computer checks this pattern on small complexes, from the projective plane up to presentations of the two finite groups: whenever the twist sees torsion, the twisted second homology is at least 1.
2:42Where it stands
The paper is unrefereed and unchanged since release, with no Lean formalization. The tower obstruction, which carries the proof, has not been publicly checked by anyone we know of. In the first 12 days we found no public comment from any mathematician, no reported error and no competing claim. If it holds, a complex can pass Wall’s homological tests for dimension 2 and still not be 2-dimensional, and Wall’s question has a negative answer, shown with an infinite fundamental group.
Spot a mistake?
This episode simplifies on purpose, but it shouldn’t be wrong. If something is, open an issue on GitHub or leave a comment on YouTube. Corrections are noted here with a date.