# n173 — THE TEST CAMPAIGN: all 32 predictions under systematic test
# (2026-07-11, all day, loop-driven. Batch protocol declared here.)

## The protocol
Every standing prediction gets one of three test types:
  R (regression)   - deterministic in-toy re-verification of its
                     load-bearing exact structure (tests_regression.py)
  D (data)         - confrontation with the current state of the
                     relevant experiments/observations (web sweep,
                     the n110 discipline: only existing data)
  S (structural)   - consistency re-derivation against the Action
Verdicts: PASS / LIVE (data pending) / TENSION (quantified) /
ACTION (registry must change). All verdicts land in this note.

## BATCH 1 — REGRESSION (done): 18/18 PASS
The suite (tests_regression.py, committed, re-runnable in ~20 s)
covers: the Nordtvedt identity (2.000000000000, both channels), the
dark states (rotation via signed continuity, shear via unsigned
incidence, winding via divergence - all machine zeros), the fermion
sector (T^l = +-I by twist parity, exact), the radiative kernel
({+,x} exactly, scalars at 0), the photon kernel (2 transverse, no
scalar), the two-channel law (zeta kappa-independent), Alice
holonomy, the knee algebra, the frame margin.
Two harness bugs found and fixed in the writing - one of them a
useful clarification now recorded in the suite itself: THE AUDIT
READS UNSIGNED INCIDENCE ON COUPLING-BUDGETS AND SIGNED CONTINUITY
ON FLOWS (the n166 rows say so; the suite now tests each object
with its own operator).

## BATCH 2 (queued) — cosmology group: 11 (DESI/Gdot), 15 (JADES
## dipole), 16 (birefringence), 24 (r/B-modes)
## BATCH 3 (queued) — GW group: 12 (c_gw), 17 ({+,x}, d_L), 18-20
## BATCH 4 (queued) — particle group: 25 (proton), 26 (Pauli/VIP),
## 27 (nu mass), 28 (DM: LZ/XENON, decay lines, SIDM fits)
## BATCH 5 (queued) — lab group: 3-4 (DP floor/BMV status), 19
## (Penning), 29 (alpha_2), 30-32 (charge-blind, SQUID, inertia)
Each batch: current-data sweep, verdict per prediction, stamp
updates on the predictions page if anything moves.