From fb60217b98c97ab354d4f73a94805f32a785539f Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Thu, 3 Sep 2026 08:24:48 +0200 Subject: [PATCH] =?UTF-8?q?docs(EXPERIMENTS):=20CE=20curves=20for=20E5/E6?= =?UTF-8?q?=20=E2=80=94=20the=20confound=20quantified,=20E6=20clean?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Control curve unrecoverable (recorded absent); E6 shows no anomaly across 2,550-11,050; E5 fresh-head segment decays 0.84 -> 0.035 over ~2,000 steps, confirming the diagnosis. Neither lever shows a convergence-speed effect; DER gates remain the verdicts. --- docs/EXPERIMENTS.md | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/docs/EXPERIMENTS.md b/docs/EXPERIMENTS.md index 19bc162..4435f3b 100644 --- a/docs/EXPERIMENTS.md +++ b/docs/EXPERIMENTS.md @@ -285,6 +285,17 @@ All rows passed the CER parity gate at release. Readings: units move the rung; register swaps at constant budget do not. Clean negative; direction shifts to G2b (48k add). Labels sha256 39b4213b5ab4c9f7273da90a9a1954f7701320b62870a954e6e2c02e810617a4. +- **CE curves (2026-09-03, registered checkbox):** control run-006's + per-step curve is not recoverable (client died pre-compaction; + steps never persisted) — recorded as absent, not reconstructed. + E6 (local log, steps 2,550-11,050): CE oscillates 0.003-0.023 with + no anomaly at either resume; both preemptions invisible in CE. + E5's captured segment IS the confound quantified: CE 0.8444 at step + 8,550 (fresh aux head) decaying through 0.16 @ 9,000 and 0.069 @ + 9,500 to ~0.035 by 10,500 — ~2,000 steps to return to trend, + independently confirming the fresh-head diagnosis. Neither lever + shows a convergence-speed effect distinguishable from noise on the + surviving curves; the DER gates remain the only verdicts. ## Parked