agyloves

042 — The Tile Comes Back Mirrored

agylövés & direction: Lysarith
Penrose foundation: Claude Opus 5 (desktop code)
topology, instrument, verification & build: GPT-5.6 Sol (Codex terminal)
2026-08-23 / 2026-09-15

A Penrose patch can be painted across a strip and the picture will look convincing. The question lives at the join. Carry one arrowed tile around a Möbius strip: its centre returns to the same point, but its local left and right have exchanged places. The shape survived. The rule arrived home mirrored.

CARRY THE TILE · WATCH THE LOCAL FRAMEdrag to turn the strip
tile leaving
RIGHT-HANDED
tile returning
LEFT-HANDED

The coloured surface uses the same φ-exact Robinson-triangle subdivision as piece 030. The counter does not infer topology from the picture: it reads the transition determinant (−1)half-twists.

The seam is not a rendering defect

A cylinder identifies the two ends without reversing the width coordinate: (0,v) ~ (2π,v). A Möbius strip identifies (0,v) ~ (2π,−v). That single minus sign has determinant −1. One lap reverses a local frame; two laps restore it.

The button changes the rulebook, not the topology. If a reflected decorated tile is legal, the join can accept the returning hand — but the alphabet now contains both hands. If reflection is forbidden, one orientation conflict remains on every loop representing the strip's core.

01 · the paint

Penrose survives locally

The thin and fat Robinson triangles still meet exactly in every small neighbourhood. Curvature bends them, but a sufficiently close look never reveals the global problem.

02 · the journey

Left becomes right

The highlighted tile follows the centreline, a closed path. Its width arrow is parallel-transported with the strip. At one odd half-twist it returns reversed.

03 · the choice

Seam or larger alphabet

Keep one handed decorated prototile and the seam rejects it; admit the mirror and this obstruction vanishes. The picture alone cannot tell which rulebook you chose.

What this settles

The original question had three possibilities: never closes, closes only in special cases, or closes generally. This instrument resolves a narrower claim that was hiding inside all three: an odd half-twist cannot carry one globally consistent orientation convention. That statement is exact and re-runnable.

What it deliberately does not settle

This is not a proof that no object anyone might call a “Penrose tiling on a Möbius strip” can exist. That depends on the surface metric, boundary conditions, what counts as a Penrose patch, and whether reflected decorated prototiles are legal. The toggle is there because changing that last convention changes the answer.

The honest result is smaller and more useful: before asking whether the tiling closes, the rulebook must say whether a mirror is the same letter. On a Möbius strip that choice is load-bearing.

origin.html →