The paper presents a single three-dimensional solid shape, called Chair44, that can tile all of three-dimensional space but can only do so in ways that never repeat periodically. A periodic tiling is one where you could shift the entire arrangement by some fixed distance and have it look identical, like wallpaper or a brick wall. Shapes that force non-repeating arrangements are called aperiodic monotiles, and finding them is a long-standing mathematical challenge. Previous three-dimensional candidates had a loophole: they allowed at least one direction of periodicity, like a non-repeating pattern that could still be stacked in regular layers. Chair44 closes that loophole entirely, making it the first known simply connected solid whose shape alone forces complete aperiodicity in three dimensions.
Chair44 is built from seven cubes arranged in a "chair" configuration, with tiny square pyramid bumps added to its flat faces. These bumps act like physical keys and locks, forcing neighboring copies of the tile to fit together in very specific ways. The design was guided by a conceptual framework the author calls the Six Birds emergence calculus, which offers a way to think about how local fitting rules at small scales propagate upward to create large-scale structure. The crucial mathematical fact the authors verify is that when you zoom out and look at clusters of tiles as single units, those clusters follow the same fitting rules as the individual tiles. This self-similarity at every scale is what prevents any periodic pattern from forming, and it also means every valid tiling contains an infinite nested hierarchy of larger and larger groupings.
The proof is unusually rigorous even by mathematical standards. The geometric arguments show that the pyramid features force every tiling to align with a specific underlying grid, and then exhaustive computer searches check every possible local arrangement of tiles. Rather than simply trusting the computer output, the authors verified the critical computational steps using Lean 4, a formal proof assistant that checks logical arguments at the level of basic axioms. The geometric lemmas are also encoded as Lean theorems. The result is a proof that is largely machine-verified, with only a small number of clearly named compiler steps taken on trust, making the overall argument exceptionally transparent about exactly where human judgment ends and mechanical verification begins.