Iteration 96 — H-005 Regime-MCI EXCLUDED KILL #36 · THE CLOSEST DOOR MISS (0.88×) · STRONGEST RE-PROPOSAL CASE

In plain words: the different-family candidate lived up to its billing — and died the campaign's narrowest door-law death, which is itself information. The instrument reads whether trading intensity CAUSES the next bar's duration, with each series' own past accounted for (the trap that makes plain correlation lie about causation on autocorrelated data). Its design deliberately removed the significance-level knob that killed both of its family's earlier candidates. The link is real: intensity genuinely drives the next bar's duration in the crash and the calm alike, with the sensible negative sign — more intensity, shorter bars. And unlike the four tail meters, its reading DOES look regime-varying: fifty percent stronger in the covid crash. But the difference (0.043) fell just short of the noise floor (0.049) — an 88% ratio where the tail meters scored 1 to 18 percent. The frozen rule fires, and re-running at a bigger sample in the same breath would be threshold-hacking, so it isn't done. Instead the boneyard records the batch's strongest re-proposal case: the same pre-registered form on full windows — one power lever, zero design changes — plausibly resolves it. One candidate remains; the batch closes next firing either way.