Deleting a row after discovering a collection failure or timing-contract violation may make the dataset look cleaner, but it also makes it harder to reproduce what was excluded and why.
Deletion and quarantine are different
The confirmatory cohort can exclude decisions with confirmed integrity problems, while the original decision and outcome rows remain preserved. The incident reason and affected interval are recorded separately in the quarantine log.
exclude from confirmation ≠ delete from historyExcluding a sample from the conclusion is one thing; erasing the original record is another.
Why provenance is important
If the same problem recurs, preserved incidents let us compare behavior and verify whether a code change actually fixed the issue. If the original is deleted, the evidence needed to reproduce the failure disappears with it.
How this applies to the current research
In the latest public snapshot, V3R2 excludes four quarantined decisions from the maturity count. Those exclusions are not something to hide; they are part of the documented state of the research.