A 7.8σ drift we quarantined ourselves
Multi-probe bootstrap wₐ departure at 7.8σ.
Same failure mode as the w₀ headline: the diagonal-covariance bootstrap shortcut inflated it; full-covariance treatment gives 3-4σ. Quarantined by the founder's own audit on 2026-07-09.
QUARANTINED — covariance_diagonal_shortcut
This significance was produced with a diagonal-covariance bootstrap shortcut that discards supernova correlations and inflates σ. The full-covariance treatment of the same data gives 3–4σ. Caught by David Thompson's own audit. Clears when: a full-covariance recalibration bundle reproduces the claimed significance — or a kill test buries it.
TENSION — the registry's self-check flags this claim
Declared contradiction with USR-2026-0003 PROVISIONAL — “The evolving-DE preference appears in all three SN compilations refit through one...”. One of these claims must eventually fall.
Declared contradiction with USR-2026-0001 PROVISIONAL — “Dark energy evolves: a CPL (w₀, wₐ) fit to DESI DR2 BAO + CMB + Pantheon+ beats ΛCDM...”. One of these claims must eventually fall.
automated sweep · exact quantity-name match · flagged when |Δ| > 3σ combined · resolves only by evidence — raw feed
Falsify-box — how to kill this claim
Verification record — every quoted number, re-run
| quantity | measured | verdict |
|---|---|---|
| wₐ bootstrap departure | 7.8σ (diagonal covariance) | NOT_RERUN |
Provenance
script status: PRESENT
datasets: Multi-probe compilation (diagonal-covariance bootstrap)
re-run: not re-run
Edges
Status, honestly
covariance_diagonal_shortcut. This status was assigned by hand from the public re-run audit above — it is a judgement made from that evidence, not a number this site computed for itself. What cannot move it is authority: no vote, endorsement, or say-so, from anyone including the maintainer. Independent reproduction would move it; nothing else will.