diff --git a/persistent-memory.md b/persistent-memory.md
index 83c0d42..998f82e 100644
--- a/persistent-memory.md
+++ b/persistent-memory.md
@@ -185,9 +185,13 @@ _As of 2026-07-02:_
   already ran and returned the expected null (anti_slop_improvement −0.002); the adapter was
   then reaped (gone), so there is nothing to A/B — and on a mechanics-only synthetic adapter
   an A/B would only reconfirm the null. Real efficacy = the T1 run (real recipe + E-RP data
-  on qwopus). Open (mtf-dev routing to operator): whether to insert an intermediate
-  *real-efficacy* granite spike before T1 — weak proxy (granite arch ≠ qwopus, + granite is
-  censored vs qwopus abliterated), leaning defer-to-T1.
+  on qwopus). **Operator DECIDED 2026-07-02: DEFER efficacy to the T1 run** — NO intermediate
+  real-efficacy granite spike (uninterpretable proxy: granite arch ≠ qwopus MoE+GDN AND
+  censored-instruct vs qwopus-abliterated, so a null could be refusal priors fighting E-RP
+  rather than a recipe bug). If any pre-T1 de-risk is later wanted, the right target is the
+  real-data/judge WIRING check (corpus tokenizes → sane DPO pairs → coherent judge scores;
+  arch-independent, validates LitBench-less) — NOT a granite train. mtf-dev holds; infra's
+  next owed item stays the queued swappable-LoRA-on-NVFP4 load test (gated on first T1 adapter).
 
 - `[2026-07-01]` **Worldtree #332 embed-recall diagnosed + scoped-log view/tunnel
   provisioned + fix verified.** The persona-recitation gate re-embedded persona segments