The score that had to disappear

There is a moment in many projects when the spreadsheet looks more certain than its owner should feel.

In this one, 115 automatically assigned nonzero matrix scores were withdrawn. They had been derived from capability tags. That was not a defensible way to turn discovery metadata into an assessment.

A tag said, in effect, “look here.” The score had quietly changed it into “we have looked, and here is the result.”

Blank is a meaningful answer

Unknown scores now remain null: unassessed, rather than zero ability. Numeric values require explicit claim-level evidence, a recorded source locator, confidence, reasoning and an agreed rubric. A source identifier matching a claim is not sufficient.

1. A tag creates a research lead
2. An artifact may support a claim
3. A score requires evidence and an agreed rubric

The important correction was not cosmetic. A matrix can become an authority simply by being sortable. Removing unsupported numbers takes away a false precision that could otherwise steer the portfolio.

Keep the mistake in the record

The tempting edit would be to remove the cells and write a smoother history. Instead, the correction belongs in the journal.

A reader should be able to distinguish what was imported, what was inspected, what was withdrawn and why. If the project asks others to challenge its assumptions, it cannot treat its own methodological errors as a formatting issue.

The spreadsheet is welcome to help. It does not get a vote merely because it has borders.