Clean-boundary synthetic-pair control: ratio 1.311 (CI [1.258,1.402]) vs 1.471 — corruption inflated by ~12%, structure signal survives at 0% word-slicing; blog section 7 + results.md resolved

1 files changed, 16 insertions, 7 deletions

diff --git a/results.md b/results.md
index 4225e39..c9e0c23 100644
--- a/results.md
+++ b/results.md
@@ -126,13 +126,22 @@ Middle-layer ratio across seeds: 1.47 +/- 0.09 (SD), bootstrap 95% CI
 [1.37, 1.53]. The frequency anti-correlation holds, but the lens also carries
 genuine conditional-predictability signal.
 
-CAVEAT (from adversarial review): '#' was inserted at uniform random character
-positions, which slices through the middle of a word 58% of the time (letter
-on both sides: th#e, ki#ng); '@' always sits at a clean word boundary after
-"the ". Predictability is therefore not perfectly isolated from n-gram
-corruption. A clean-boundary control (noise token after random word
-boundaries, 0% word-slicing) is running; the numbers above should be read with
-that caveat until it lands.
+CAVEAT + CONTROL (resolved): '#' was originally inserted at uniform random
+character positions, which slices through the middle of a word 58% of the time
+(letter on both sides: th#e, ki#ng); '@' always sits at a clean word boundary
+after "the ". A clean-boundary control (noise token after random word
+boundaries, 0% word-slicing, same frequency) was run with three fresh seeds:
+
+```
+  placement of '#'      middle-layer ratio @/#   bootstrap 95% CI
+  random (58% slicing)  1.471 +/- 0.090          [1.368, 1.529]
+  clean boundary (0%)    1.311 +/- 0.079          [1.258, 1.402]
+```
+
+Reading: the corruption confound was real (inflated the ratio by ~12%) but not
+the whole story — the structure signal survives at clean boundaries, CI
+entirely above 1 in every seed. Original run: outputs/synth_pair/seed{s}/;
+control run: outputs/synth_pair_clean/seed{s}/.
 
 ## 3. Loss-reweighting causal test (`src/loss_reweight.py`)