# dubai-02 wave 1 (dubai-02-edition-1) after re-extraction under R6 and R7: WAVE1_RERUN, one run, pre-registered basis
before re-extraction (wave2-decision.md, 23 September 2026): 37 of 42 pairs below the split-half mean (88.1%), median ratio 0.685
HEAD: c359103d71a613b84eeecb707e2d1319abfa2ebb
ledger sha256: 4d7aa18f3586cf6c60b472b093c68846ae6961e5dd2dc8deecbdf5f60203ba57
gate.py start (UTC): 2026-10-06T08:45:41Z
command: python analysis/dubai-02-w0/gate.py --ledger data/dubai-02/ledger.db --edition dubai-02-edition-1 --naming preregistered --out analysis/dubai-02-edition-1
  (no --compare: the 21 September outputs carried a drift check against dubai-02-w0, which is not re-extracted)
gate.py end (UTC): 2026-10-06T08:49:32Z
q1_surface_overlap.csv sha256: e89b50db8c4de1d64781b939448d44a345890a77405125dd681957e562430c3f
headline.py run (UTC): 2026-10-06T08:49:32Z
--- headline.py output ---
analysis/dubai-02-edition-1/q1_surface_overlap.csv: 15 questions, 42 surface pairs compared, 3 excluded (G4 chatgpt-gemini, G4 chatgpt-ai_mode, G4 gemini-ai_mode)
below the mean of their two split-halves: 37 of 42 = 88.1%
median ratio of cross-surface overlap to that mean: 0.676
threshold: replicates at share >= 70% and ratio <= 0.80; fails at share < 50% or ratio >= 0.90
verdict: REPLICATES
