Auditing Independence

This Methods paper specifies an agreement-curve diagnostic that distinguishes genuine independent redundancy from correlated pseudo-redundancy. When multiple channels measure a shared latent quantity, the protocol compares the observed rate at which agreement improves with channel count against the rate predicted by an independently specified channel model. That comparison is the sharp model-conditional falsifier — sharp conditional on the channel model's adequacy.

The diagnostic operates in two regimes — classification, where consensus improves exponentially at the per-channel Chernoff-information rate, and continuous, where variance reduces toward σ²/k_eff — and pairs every comparison with an effective-sample-size audit that detects when high pairwise agreement reflects shared upstream structure rather than independent redundancy. Its formal pair, TN-S₁, proves the underlying conditional bound; real-data execution against the Planck PR3 component-separated CMB maps is carried by a separate Tier 1.6 demonstration.

What acceptance commits you to. Acceptance does not require adopting Universal Collapse Theory. It requires only that the conditional lemma is correct and that the audit improves discrimination over standard pairwise agreement statistics.

Keywords: independence audit; multi-channel measurement; effective sample size; agreement curve; S₁ signature.