{
 "name": "QAG Replication Lab \u2014 independent in-house replication of SPARC v3.0",
 "version": "2026-11-06",
 "provider": "R&R Droids",
 "note": "Every figure below was recomputed from the raw SPARC release (Rotmod_LTG). The lab exists to stress the headline result, not to promote it; failed tests are published with the same prominence as passed ones.",
 "phase0_reproduction": {
  "status": "EXACT",
  "detail": "v3.0 headlines reproduced byte-for-byte: QAG red-chi2 5.549148, MOND 56.553877, Newton 616.741738; zero ingest drift vs the original SPARC release."
 },
 "phase1_equal_freedom": {
  "protocol": "2-parameter NFW halo granted the same fitting freedom as AVI (2 params/galaxy), same data, same errors",
  "total_chi2": {
   "qag": 16875,
   "nfw": 17897
  },
  "median_galaxy_red_chi2": {
   "qag": 1.424,
   "nfw": 1.331
  },
  "verdict": "QAG wins total chi2; NFW wins the median galaxy. Both halves of that sentence are the result."
 },
 "phase2_prediction": {
  "held_out_fitting": {
   "status": "PASS",
   "test_red_chi2": 5.595,
   "train_red_chi2": 5.539
  },
  "zero_parameter_blind_test": {
   "status": "FAIL",
   "blind_red_chi2": 63.08,
   "beat_mond_fraction": 0.268,
   "predictability_R2": "0.04-0.17",
   "verdict": "(v_inf, r_aff) are essentially unpredictable from baryonic properties in the tested feature set."
  }
 },
 "phase3_robustness": {
  "status": "ORDERING ROBUST",
  "detail": "Inclination and distance Monte Carlos move all models together; the QAG/MOND/Newton ordering survives, but no model approaches chi2 ~ 1 on published SPARC errors."
 },
 "derivation_hunt": {
  "systems_tested": 73,
  "verdict": "PARTIAL",
  "v_inf": "A zero-constant candidate exists: the a0-BTFR identity (M_bar = v_inf^4 / (G a0); M_bar = 9.36e9 Msun from the BTFR calibration puts NGC3198-class v_inf on the fitted scale without fitted constants).",
  "r_aff": "No system predicts r_aff. This is the framework's open theory gap and its central next test."
 },
 "canonical_summary": "As a two-parameter fitting law, AVI leads its class on total chi2. As a predictive theory it is unfinished: the derivation of (v_inf, r_aff) from baryons is the open problem."
}