ablation.md: add the causal staleness grid (semantic vs geometric) and the zeroed-latent control

ablation.md

@@ -8,7 +8,7 @@ All closed-loop numbers below are **paired**: every configuration runs the same
 
 Hardware: 4x Tesla V100-SXM2-32GB (sm_70), fp32, driver 535 / CUDA 12.2,
 torch 2.6.0+cu124, transformers 5.5.4, lerobot 0.6.1. Suite: `libero_spatial`.
-Generated 2026-08-20.
+Generated 2026-08-21.
 
 **The headline dual-rate comparison in the model card is not reproduced.** With the
 clock fixed and 500 paired episodes per configuration, no refresh rate differs
@@ -127,7 +127,7 @@ Amortized with measured per-call cost and the rates each run actually ran at.
 Two things fall out of this table.
 
 **Caching the semantic latent saves at most ~39% of compute, not 97%,** because the
-floor is set by the fast path (section 5).
+floor is set by the fast path (section 6).
 
 **Plain action chunking beats the dual-rate decomposition on both axes.** Executing
 10 actions per replan is 9.6x cheaper than the best dual-rate
@@ -136,15 +136,79 @@ real time on this hardware. Since the open-loop baseline exists precisely to
 separate "the gain comes from dual-rate" from "the gain comes from chunking", on
 this benchmark the answer is chunking.
 
-**Standard LIBERO-Spatial cannot show the temporal asymmetry.** Neither the
-semantic rate (20 -> 0.5 Hz) nor the replan rate (20 -> 2 Hz) changes success. The
-suite is quasi-static, so a flat result here does not distinguish "semantics
-tolerate staleness" from "nothing in this scene changes". Demonstrating the
-asymmetry requires a benchmark where the world moves after the latent is computed.
+**Neither the semantic rate (20 -> 0.5 Hz) nor the replan rate (20 -> 2 Hz)
+changes success.** On its own that is ambiguous: it could mean semantics tolerate
+staleness, or it could mean nothing in this scene changes fast enough to matter.
+Section 5 separates the two, and the answer is the first one - the suite is not
+quasi-static at all, it is specifically insensitive to *semantic* staleness.
 
 ---
 
-## 5. System profile
+## 5. The asymmetry: staleness on each path, separately
+
+Constant-age probes rather than a periodic cache, so the age on the x-axis is the
+age, not the mean of a sawtooth. 150 paired episodes per cell, compared against the
+same episodes of the full-rate run. Proprioception is fresh throughout.
+
+**Staling the semantic latent while geometry stays fresh:**
+
+| latent age | Success | 95% CI | delta | paired 95% CI | flips | McNemar p |
+|---|---|---|---|---|---|---|
+| 0.25 s | 128/150 = 85.3% | 78.8–90.1 | +1.3 | -4.0…+6.7 | +9/-7 | 0.8036 |
+| 0.5 s | 129/150 = 86.0% | 79.5–90.7 | +2.0 | -3.3…+7.3 | +10/-7 | 0.6291 |
+| 1 s | 128/150 = 85.3% | 78.8–90.1 | +1.3 | -4.0…+6.7 | +9/-7 | 0.8036 |
+| 2 s | 126/150 = 84.0% | 77.3–89.0 | +0.0 | -5.3…+5.3 | +8/-8 | 1.0000 |
+| 4 s | 125/150 = 83.3% | 76.6–88.4 | -0.7 | -6.0…+4.7 | +8/-9 | 1.0000 |
+
+**Staling the geometry while the semantic latent stays fresh:**
+
+| frame age | Success | 95% CI | delta | paired 95% CI | flips | McNemar p |
+|---|---|---|---|---|---|---|
+| 0.25 s | 35/150 = 23.3% | 17.3–30.7 | -60.7 | -68.7…-52.0 | +2/-93 | 0.0000*** |
+| 0.5 s | 8/150 = 5.3% | 2.7–10.2 | -78.7 | -85.3…-72.0 | +0/-118 | 0.0000*** |
+| 1 s | 4/150 = 2.7% | 1.0–6.7 | -81.3 | -87.3…-74.7 | +1/-123 | 0.0000*** |
+| 2 s | 2/150 = 1.3% | 0.4–4.7 | -82.7 | -88.7…-76.0 | +0/-124 | 0.0000*** |
+| 4 s | 0/150 = 0.0% | 0.0–2.5 | -84.0 | -89.3…-78.0 | +0/-126 | 0.0000*** |
+
+**Both stale together:**
+
+| both aged | Success | 95% CI | delta | paired 95% CI | flips | McNemar p |
+|---|---|---|---|---|---|---|
+| 0.5 s | 6/150 = 4.0% | 1.8–8.5 | -80.0 | -86.7…-72.7 | +2/-122 | 0.0000*** |
+| 2 s | 1/150 = 0.7% | 0.1–3.7 | -83.3 | -89.3…-77.3 | +0/-125 | 0.0000*** |
+
+**Controls - is the semantic path used at all?**
+
+| control | Success | 95% CI | delta | paired 95% CI | flips | McNemar p |
+|---|---|---|---|---|---|---|
+| latent zeroed | 16/150 = 10.7% | 6.7–16.6 | -73.3 | -80.7…-66.0 | +1/-111 | 0.0000*** |
+
+### Reading
+
+At four seconds of staleness the geometric path costs **84.0 points**
+while the semantic path costs **0.7**, which is not distinguishable from
+zero - no semantic cell reaches significance at any age out to 4 s, and their
+paired intervals all straddle 0. (A ratio of the two is not quoted: the denominator
+is consistent with zero, so it would be arbitrarily large and meaningless.) Every
+geometric cell is significant at p < 0.0001, and a quarter of a second - five frames
+at 20 Hz - already costs 60.7 points. The both-stale
+column tracks the geometry column, so geometric staleness accounts for the damage
+and semantic staleness adds nothing on top.
+
+The control rules out the obvious alternative explanation. Zeroing the latent
+costs 73.3 points, so the policy is not ignoring the semantic path - it needs
+it, it just does not need it recent. That is the claim this architecture rests on,
+and here it is separated from "the benchmark is static" and from "the latent is
+decorative".
+
+Caveat worth keeping in view: these cells stale the pixels while proprioception
+stays fresh, so the policy receives an observation whose parts disagree in time.
+Whether the collapse is caused by old pixels or by that disagreement is a separate
+question, currently being measured with matched pixel and proprio lag.
+
+---
+
+## 6. System profile
 
 Measured on an idle machine, batch 1, fp32, 30 iterations (5 on CPU). No edge
 device is attached, so CPU stands in as the constrained-device proxy; no
@@ -173,7 +237,7 @@ headroom is in the visual encoder, not in the language stack.
 
 ---
 
-## 6. Adaptive refresh — threshold calibration
+## 7. Adaptive refresh — threshold calibration
 
 A drift gate that refreshes when the pooled cam0 spatial tokens have moved away
 from those at the last refresh (plus a proprio term and age bounds). The signal
@@ -190,7 +254,7 @@ rather than at a matched frequency:
 
 ---
 
-## 7. Not yet included
+## 8. Not yet included
 
 Running or pending, and deliberately not summarized here until finished:
 the fresh/stale semantic x fresh/stale spatial grid, the compute-matched gate