Counterfactual Echo Gain (CEG): Future-Aware Metriplectic Assistance Yields Gate-Certified Echo Improvement

Summary: This now includes the preregistered CEG outcome and gate statuses for the metriplectic assisted‑echo experiment. It makes no novelty claims about echo mechanisms; the claim is strictly that a model‑aware assistance micro‑sequence yields a statistically positive echo gain under instrument gates and an energy‑match constraint. Numerical scheme is treated as the measuring instrument per VDM canon. Under preregistered conditions and strict gates, the metriplectic assisted‑echo exhibits a statistically positive CEG (median\ₘax 0. 054552 at λ=0. 5) while preserving J‑Noether and M‑monotonicity and passing Strang defect QC. This supports the preregistered claim that internal J/M knowledge can be operationally exploited to improve echo recovery without cheating via energy injection. Original proposal description: This is a T4 preregistration for an experiment that tests whether a metriplectic, model-aware assisted time-reversal can improve echo fidelity in the Void Dynamics Model (VDM). We define the preregistered metric Counterfactual Echo Gain (CEG) and a set of strict physics gates (J-Noether drift, M-monotonicity / H-theorem, and energy-matching) that must be satisfied. The primary hypothesis is that the median CEG across preregistered seeds will be ≥ 0. 05 under the gates and controls. No results or artifacts are included here — this deposit documents the exact predictions, machine-readable specs, and provenance (commit SHA) prior to running the preregistered experiments. Original results description: A structure‑preserving, metriplectic Strang‑composed KG⊕RD solver is paired with a preregistered, equal‑work assisted‑echo meter (CEG). Under strict J‑Noether, M‑Lyapunov, and Strang order gates, the instrument exhibits a statistically positive counterfactual echo gain (medianλ_λ CEG =0. 054552=0. 054552=0. 054552 at λ=0. 5=0. 5λ=0. 5, n=12n=12n=12 seeds), with energy parity and invariants within machine‑precision tolerances. Artifacts (CSV/JSON/PNG) and seeds are pinned for reproduction.