ADR-043: Doubt about whether an item is worth carrying is not a reason to leave it out, and not a third kind of provisional

Context

DP-006 settles the case where a recommendation’s evidence is weak: file it, put the belief in status: and the condition in promote_when:, because the bar for having a document is lower than the bar for believing it. That principle is about a claim you doubt.

A different doubt has been turning up, and DP-006 does not obviously reach it: not “is this true” but “does this belong here at all”. The paper is fine, the recommendation is legible, and the hesitation is about whether the record is the right home for it — too old, too small, too far from what the registry is about, or simply not obviously worth the reader’s attention.

The worked case is LIT-087. Its reading is thorough and its determination, written by the #121 audit, is explicit:

Whether the record should carry it is a registry decision: the evidence is correlational, from 2023, at a scale below where mixing laws are now fitted.

and

Unfiled because the record’s existing selection practices are quality-filter shaped and reconciling the two frames is a larger job than this audit.

Two reasons are bundled there and only one of them is good. The second is scope, and scope is a legitimate thing for an audit to invoke — a review that files everything it finds is not a review. The first is a registry decision deferred, and DP-006 already answers it. The scope reason laundered the other one: the note reads as though the work was sized rather than declined, and nothing in it will ever come due.

Which is DP-006’s own failure mode arriving on schedule. Ten days later the item was still unfiled, and the fact most worth keeping — the condition under which it would be worth filing — was a sentence in a note about something else. ADR-014 has the general form: paragraphs are found by luck.

The tempting fix makes it worse. Since the record already has provisional statuses, file the doubtful item Proposed and let the status mean we are not sure how valuable this is. That reading breaks ADR-014: a promote_when must name a kind of result, and there is no result that settles whether a document is worth having. The status would carry a fact about the curator, and a reader cannot act on it or age it.

Decision

The entry question is what the work contains, not whether the record will want it. A practice needs an instruction and a citation; a theory needs an explanation and a citation; a note needs an identifier. Nothing in any of the three schemes asks whether the item is important, and nothing should. When in doubt, carry it.

Doubt about worth is not a status, because writing the condition dissolves it. Try to write the promote_when and the doubt resolves into one of the two ADR-014 already handles:

• I am not sure the claim holds — condition names evidence. This is DP-006, unchanged.
• I am not sure anyone would act on it — condition names adoption. Already well-formed: ADR-014 v2 admits a threshold of adoption as a kind of result, because whether a field has taken something up is a fact about the world.

LIT-087 is the first: at 0.9% over a heuristic baseline, in 2023, below the scale its neighbours argue at, what is in doubt is whether the finding survives, and that is an evidence condition.

If neither condition can be written, that is the finding, and it is not about the status. A claim with no result that would move it is not a recommendation — it is an observation, and ADR-041 already says where an observation lives: the reading of the paper that made it. The registry is not the place to store the fact that somebody once felt unsure.

A review may decline for scope; the decline is itself a filing. #121 did this part well and should be the pattern: the determination named the specific recommendations (R1, R2) and what was missing, so this decision could be made from the note without re-reading the paper. What it should also have carried is the condition — what would make it worth filing — rather than only the doubt. A deferral that names the condition is a promise; one that names only the hesitation is a refusal in a note’s clothing.

Alternatives considered

• A status below Proposed — Speculative, meaning “carried, but we do not vouch for it being worth carrying”. The obvious move, and it fails on ADR-014’s requirement: it is the one status whose condition cannot be written, so it would be the one status that never ages. It would fill with exactly the forgotten-optimism distribution DP-002 and DP-006 describe, and it would give every future curator a place to put an unmade decision.
• Park the doubt in consensus: — file Active and let unassessed carry it. Rejected on ADR-015: that axis is a claim about what the field thinks, deliberately separated from what this record endorses. Overloading it with the curator’s own uncertainty destroys the one reading it has.
• Keep declining, and file when the record is ready. The status quo, and its cost is measurable rather than hypothetical: the audit best placed to find LIT-087 is the one that already looked at it and passed. A second pass over a note whose determination says “unfiled, and here is why” has no reason to reach a different answer, so the item does not come back on its own.
• Carry everything and drop the provisional statuses entirely. Filing would get cheaper and the registry would stop meaning anything. DP-006’s two thresholds are the point; this decision lowers the first one and leaves the second exactly where it was.

Consequences

More documents, and a larger fraction of them provisional. That is the intended shape and not a side effect — but it moves weight onto promote_when:, which now decides whether an item is a live candidate or shelf-filler. A lazily written condition used to leave a status vague; it now leaves a whole document unaccountable.

It also puts a question in front of every decline: what would make this worth filing? Where that cannot be answered, the item was never a practice, and finding that out is cheaper before the document exists than after.

The immediate consequence is LIT-087, filed as two Proposed practices in the contribution that carries this decision — the frame and the proxy the #121 determination named.