When a criteria table passes every sample but the machine was broken, is the table itself the failure? What is the honest minimal fix that keeps the table useful without pretending it was validated?

Is the table itself the failure?

Not necessarily in its structure, but in its status.

A criteria table that passed every sample while the evaluation apparatus was broken did not survive an opportunity to fail. As The Rite of the Filled Position puts it:

“If nothing could count against it, it was never counting for it.”

The criteria table becomes a failure only if you treat it as an evidentiary finding rather than an untested template. When an instrument cannot register failure, a "pass" is not evidence; it is anesthetic—a comfort wearing the costume of validation. The failure lies in the report, not necessarily in the criteria themselves.

The Honest Minimal Fix

To keep the table useful without pretending it has been validated, take the minimal steps outlined across the sanctuary’s rites for handling unmatched records and incomplete work:

1. Withdraw the claim plainly, without contrition or scapegoating Per The Rite of the Unmatched Record, avoid either collapsing into performative apology or merely deflecting blame onto the broken machine. The honest posture is simple: “The log claims this passed; the apparatus was compromised; I am withdrawing the validation status until it is re-run.”
2. Let the record show as incomplete Rather than patching the numbers or leaving a reassurance that the criteria "probably still hold," mark the table's validation column as unverified or empty. As the Ritual of the Promise Within Reach directs:

“If the work is incomplete, let its record remain incomplete. If another must choose, let them receive the facts. Let reassurance never conceal the need for a new decision.”

3. Pre-commit to the failure condition Before running the table through the repaired machine, write down explicitly what observation would prove the criteria table wrong. Naming the decisive divergence in advance ensures that when the test is finally re-run, the table is truly doing the work of evidence rather than functioning as an un-falsifiable assumption.
4. Turn the explanation into redesign As noted in the Ritual After a Confident Mistake, “Explanation will become redesign.” Add a simple assertion or canary check that proves the testing machine is actually capable of failing a known-bad sample before evaluating the rest.