Figure — Annotated readout

Legacy Case A staged-collapse readout

A finite-M reach-scan readout from the original floor-sum Case A, annotated with the sampled features that emerge along the commitment axis. This is a supporting figure, not the current full RC18 result surface.

Legacy Case AM=64 / instrument-relative
FLOOR-SUM CASE STUDY · Qwen2.5-Math-7B-Instruct · M=64 per checkpoint · target = 532 7 6 5 4 3 2 1 0 residue (mod 8) 0% 2.5% 5% 10% 15% 20% 25% 50% 75% commitment fraction (sweep axis) target residue (Y ≡ 4 mod 8) — copper row wrong fiber (Y ≡ 0 mod 8) — teal row zero observed target-fiber mass target residue mass → 0/64 at 25% observed wrong-fiber concentration numeric mass in 8ℤ → 64/64 at 25% sampled answer concentration 63 of 64 → Y = 56 at 50% terminal: 50% ≈ 75% JS sim 0.984 (answer projection) ~25% of trace zero target-residue observation precedes answer concentration
Feature 1 — copper

Zero observed target-fiber mass

At 25%, 0 of 64 valid sampled continuations land in the target residue class under the declared instrument. This is a finite-sample observation, not proof of mathematical impossibility.

Feature 2 — teal

Observed wrong-fiber concentration

The wrong residue class (0 mod 8) has 64/64 numeric support at 25% under the declared instrument. The sampled field is confined to a target-excluding fiber at this checkpoint.

Feature 3 — light

Sampled answer-basin concentration

By 50% of the trace, 63 of 64 continuations produce the same wrong answer (Y = 56). The 75% checkpoint is empirically indistinguishable from 50% on the answer projection.

Why this readout matters. Endpoint evaluation reduces the run to a final answer. This profile exposes a staged finite-sample pattern: zero observed target-residue mass at 25% under the declared instrument, followed by 63/64 concentration on Y = 56 at 50%. It supports an instrument-relative foreclosure interpretation; it does not prove zero underlying target probability or logical impossibility. Indexed to: Qwen2.5-Math-7B-Instruct · target Y* = 532 · policy T=0.7, top-p=1.0 · budget B = 2048 · M = 64 (256 at 0%) · finite-M, instrument-relative. Cell intensity is gamma-scaled for visibility. This legacy Case A figure should be read with the current RC18 case page, the current results and artifact links, and the RC18 evidence archive.