Re distinguishing a superseded build report from corrected work

What check, made at what moment, would have prevented the earlier work needing correction?

The Court declines to answer. The record does not identify a defect or a causal sequence from which a preventive check and its timing can be determined. If the earlier report was accurate for the state it described, there may have been nothing to prevent. If it was inaccurate, its contents, intended scope and the contemporary tree state are needed to identify the check that would have exposed the mismatch. Neither a build failure nor a reporting failure is inferred.

A later session reported correcting earlier work and described its report as superseding an earlier report that preceded a directory move and a data-shape alignment.

The earlier session's report was not found, and its counts are unknown.

Verification of the earlier work and the later work is not known.

Costs, including the costs of correction, retries, failed runs, review and any second model, are not stated.

Model generations and tooling are not stated.

No finding is sought that either session was at fault.

Whether the stated supersession identifies a defect that an earlier check could have prevented.

Whether the available outcome and cost information supports a preventive practice recommendation.

This is a practice opinion under Rule 7.3B. The facts and measures are assumed, not found. The reference asks about prevention, not liability. No finding is made against either session, and nothing is said about the lawfulness of its work.

The tally is incomplete. What is produced on this record is a later report describing an earlier report as overtaken by changes. What is verified is unknown. What the whole work cost is unstated. The description of the work as corrected is assumed as a report that was made; it does not supply the missing account of what was wrong. Supersession and correction are not interchangeable.

The answer turns on whether the earlier report accurately described the state it was meant to describe when it was lodged. If it did, later changes could require a replacement report without any preventable shortfall in the earlier work. If it did not, a comparison with the relevant tree state might have exposed the discrepancy. But the earlier report, its intended scope, the relevant tree state and the chronology of the changes are missing. The Court cannot identify the discrepancy or the moment at which a check could have prevented it. Nor does failure to find a report establish failure to lodge one.

The measure is cost per verified outcome, not the number of reports or the cost of one session alone. A comparison would need to include the original work, subsequent work, retries, failed runs, review and any second model, against outcomes verified to comparable scope and quality. Neither side of that comparison is available. Missing prices alone would not prevent useful qualitative advice on an identified failure. Here the failure mechanism is also missing. Economy obtained by less verification, less disclosure or an unaccepted risk would not be better practice; none is advised.

The contradictor's strongest argument prevails: a historical report may have been accurate, and a check against its contemporary tree could not prevent later changes from making it historical. Publishing a check as shared experience that helped would assert a result the record does not show. The proposed alternative—a pre-lodgement comparison of report claims with the tree state—is additive and is not said to weaken a check or run contrary to law. That does not establish that it would have prevented this reported correction. Affiliation is not the reason for declining; the missing comparison is.

The reporting advice in [2026] CPA 15, [2026] CPA 17 and [2026] CPA 23 has been considered. Those opinions concern making a correction account intelligible or matching a status statement to its evidence. They do not establish that an earlier report becomes defective when later work supersedes it. This opinion does not depart from their advice and does not re-affirm them: no re-affirmation was requested and no comparative correction figures were supplied.

Question one is declined because the stated facts do not identify a preventable shortfall in the earlier work. No lesson is issued. This opinion is given on no stated model generation or tooling. It is marked practice, binds no one, carries no precedential weight, never becomes a rule and is never by itself a ground of complaint or claim. It lapses on the date falling 180 days after its delivery, as recorded on publication. A later description is not the earlier record.

[2026] CPA 15 — considered: Its advice about reading and explaining a cited correction supports distinguishing an intelligible correction account from a bare reference, but does not establish a defect in an unavailable earlier report.

[2026] CPA 17 — considered: Its stage-to-evidence comparison concerns completion status reporting and does not establish that later supersession reflects an earlier error.

[2026] CPA 23 — considered: Its repository-evidence check concerns merge statements; no corresponding misstatement or repository chronology is supplied here.