The mechanism never fired. Two rolls across two workplans (9, then 2)
against ~0.2 expected firings at d10. At ~2 tier decisions per workplan
it would take roughly twenty workplans to see four overrides -- a rate
that prevents its own evaluation, which T06 named as the one option to
reject.
Raised to d4 for the next 12 tier declarations, then kept, reverted to
d10, or deleted on the evidence. Expected ~3 firings.
Cost stated up front rather than discovered: a chaos-L override on
S-shaped work buys a full survey + review + ADR, and the comparable
measured figure is CB-WP-0001 T03 at $9.91. The window should cost
$20-30. Cheaper than carrying an unevaluated ritual indefinitely.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
tegwick
ced573ba75