Matched random control

• 2026-08-28 [Conjecture] A self-similar slab is not a better acoustic barrier than an equally sparse random one: at levels 2 and 3 the matched random mask blocks more in every comparison, four filled-cell counts per level, 96 of 96 at 12 seeds per cell with margins of 4 to 31 sd (tree L3 fractal 0.3720 against a random mean 0.1446, sd 0.0074), every fractal value reproducing exactly (carpet L2 0.3120 / L3 0.2548, net 0.2999 / 0.2619, tree 0.3023 / 0.3720, void 0.3122 / 0.3642), so the suppression is carried by fineness and sparsity rather than by self-similarity; the repeat reused the same simulator and so tests seed-fragility rather than the physics.