Rating-style letters: AAA to B, with D for a failed document

The proposal

Each rung keeps its title, meaning and criteria. “Not graded” stays “—”.

Regime ceilings are relabelled the same way: A+ becomes AAA, B becomes A, D becomes BB.

Why

• The letters are read as a school grade. On a school scale anything below the top reads as a mark-down: a B says “good, could do better”. But B on this scale describes a document that is trusted, independently time-anchored and will keep verifying for years. Readers take it as a criticism the rubric does not make.
• School scales are not international. A to F with an E is used in Sweden and in the European credit system ECTS, where E is a pass; in the United States there is no E at all, and many countries grade with numbers. On this scale E means “a claim, without proof”. A standard published for readers in many countries should not depend on which school system the reader went through.
• Rating scales describe reliance, which is what the grade expresses. AAA to B is the most widely recognised format for “how far can you rely on this”. The convention also draws a line in the right place: BBB and above is read as sound. Mapped as proposed, that line falls exactly at the scale’s trust gate. From BBB up, a recognised trust regime vouches for the signer; from BB down, nothing does.
• D for failure, and the dash stays free. In rating scales D means default, which matches “checked, and it fails”. It keeps “—” for “not graded”, which the rubric deliberately separates from failure: absence of proof is a different statement from disproof.

Costs and risks

• It must not become grade inflation. Every document would carry a better-looking letter without being any stronger. The proposal is only defensible if the criteria stay exactly as they are, which is why it changes labels and nothing else. Any change to thresholds would be a separate proposal.
• The association with credit ratings. The grade says nothing about the financial standing of the signer, the issuer or anyone else, and is not a credit rating. Readers may also carry over connotations such as “junk” for BB and B.
• Harder to tell apart at a glance. AA and AAA differ by one character, which matters in small badges and in speech.
• Two vocabularies for a while. Grades already issued under 0.x versions keep their letters and the version they were issued under. The mapping table above would be published permanently, so an old grade can always be translated. Before 1.0 is the cheapest moment for a change like this; after 1.0 it would not be made.

Every implementation has to change its labels, so the change would ship as the next major version.

Alternatives considered

• Keep A+ to F and explain it better. The cheapest option, but explanation does not reach the reader who sees only the letter.
• Numbered levels (Level 1 to 7, like assurance levels). Neutral, but loses the intuitive reading, and readers disagree on whether 1 is the best or the worst.
• Plus and minus modifiers (AA+, AA−). Not proposed: the scale has no steps between its rungs.

Questions for commenters

1. Does a rating-style scale read more accurately than a school-style one to the people you work with?
2. Is the association with credit ratings a help or a liability?
3. Is D the right label for a failed document, or should failure look entirely different from the other grades?
4. Should the switch happen at a single version, or should both labels be shown side by side for a period?