NC Money & Votes
In progress — no holdout look has been spent. The panel is built and its committee-mapping defect corrected; the referral-specific test that carries the hypothesis stays blocked until bill referral data lands.
Studies with a stated result. Each one names what was tested, what came back, and where the code lives.
In progress — no holdout look has been spent. The panel is built and its committee-mapping defect corrected; the referral-specific test that carries the hypothesis stays blocked until bill referral data lands.
The two mechanisms beat the frozen baseline on neither primary outcome; the single positive among three declared outcome tests is the rate chance alone produces, and rests on an interval a post-hoc diagnostic shows is too narrow.
A readiness audit found the detector was a ghost — fields consumed as if live while returning hardcoded fallbacks, influencing no decision. It now runs in observe-only mode, and a separate harness measures what each rejection reason actually costs.