# Change notes

Prepared and checked by the same AI assistant; no blinded comparison or independent human review is claimed.

| Draft problem | Revision | Brief anchor |
|---|---|---|
| “prove” converted association to causation | Opens with the unresolved causal question; retains three independent cultures | Known fact 1 |
| p = 0.20 treated as equivalence | Preserves the null result and states that equivalence is not established | Known fact 2 |
| Proposed perturbation presented as guaranteed validation | Describes the experiment as a planned test against matched control | Known fact 3 |
| Approved access invented | Restores unresolved access and ethics requirements | Known facts 4–5 |

The revised paragraph preserves all supplied quantities and does not add an effect size, power claim, approval or award. Quality is demonstrated on this one synthetic example; no success rate is inferred.