Weksa

map: which R rate the rule is held to is undecided

Soul finding r-draw-owns-every-misread, failure scenario.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

74edaadfbf

2026-11-13 13:50:39 +00:00

Weksa

map: growth-scene re-run with the R split matches the Soul figures

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

cca0664c17

2026-11-13 13:49:00 +00:00

Weksa

calibrate: pin R roles and line columns in the unit test; doc: false rejects can have R as candidate

Soul rinv s1 findings r-roles-unpinned and reject-in-r-candidate.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

5a681b2c3a

2026-11-13 13:48:54 +00:00

Weksa

map: the R split is now printed by calibrate, not yet re-run on growth-scene

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

6c263f0398

2026-11-13 13:24:19 +00:00

Weksa

calibrate: split prompted trials by the recorded lowering R's role

Prints prompted-r-candidate, prompted-r-base and prompted-r-free beside the
prompted total, so the false accepts and rejects that involve R and the
R-free trials are visible from the harness, not from scratch scripts.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

3b72f9b2be

2026-11-13 13:24:09 +00:00

calibrate: split prompted trials by R's role #81

2026-11-13