void · e 006 void
E-006 is void before collection
2026-08-20. Zero model calls. Void on its own pre-registered power gate, which was computed before any draw and refused the design.
PREREGISTRATION §5 committed to: "If that fraction is below 0.80, this experiment does not run as designed. Either the rung count rises or the hypothesis is withdrawn before any draw. A design that cannot detect its own effect is not evidence of absence."
POWER.md returns 0.117 against a moderately concave truth on 32 probes. The gate refuses the run, and raising the rung count does not repair it — the noise is probe-level, so eight rungs or eighty make no difference.
Required: about 512 probes. MERIDIAN-IX32 has 32, and about nine independent prompt templates. The instrument is short by more than an order of magnitude.
Why this is the useful outcome
E-002c's ladder was underpowered for exactly this question, nobody computed it, and its fidelity column was subsequently read as bearing on L1 — which is how retraction 17 was found and how this experiment came to be designed at all. The difference here is only that the check ran first.
Nothing is claimed. Nothing is withdrawn. L1 remains conjectured and remains untested, and now the reason is on the record: not that it has not been tried, but that no probe measure this programme owns can carry the test.
The successor is not a better ladder. It is a probe measure an order of magnitude larger — and retraction 15 is the warning attached to building one, since thirty-two probes that are nine templates would become five hundred probes that are ninety.
This document is licensed CC BY 4.0.