ADR-046: A TERM may be Proposed, and rival, complement or narrow another

Context

ADR-032 defined a TERM as “a concept or distinction. No credence: it has a definition, and a changed definition is a new version or a superseding TERM”. The only relation between terms was supersedes. That covers a definition replaced in sequence. It does not cover the other cases the record has hit:

• A candidate that is not adopted yet. In nucleation#162 a replacement for TERM-060 (frustration) was drafted, proposed as a supersession, and withdrawn before merge. The open disagreement ended up in a QUESTION and a dated note on TERM-060. Each round had to choose between adopting the new definition and leaving it in prose. The owner, afterwards: “I think we might currently lack a way to express proposed and/or competing definitions of a given term, and if that’s the case it might explain the tug-of-war we’ve been experiencing trying to finalize this PR.”
• Two definitions in force for different dimensions. TERM-050 (intrapersonal synchronization) and TERM-061 (practical synchronization) are both Active. TERM-061 says that it does not supersede TERM-050 and that the two are independent dimensions. Only prose said so.
• A special case that stays in force. TERM-043 supersedes TERM-010, but TERM-010 is still Active. Its section “Why it is still Active” says the weighted sum “survives in that role, as the scalar a task-fixed weighting gives”. supersedes was standing in for narrowing, and the term-revisions chain showed a retirement that never happened.

The configuration was already half there. TERM takes the shared statuses vocabulary, which includes Proposed. But the TERM template listed only Active | Superseded | Rejected, and none of the 63 TERMs was Proposed.

The record has solved the same problem for other schemes. THEORY has rivals (ADR-022), symmetric, for accounts “that cannot also be right, where neither yet corrects the other”. CLAIM has complements (ADR-032), symmetric, added because three accounts “were wrongly set up as rivals”. QUESTION has refines / refined_by.

Decision

A TERM’s status is the standing of its definition, not a credence. ADR-032’s “no credence” holds: whether a definition is adopted is a separate matter from whether any claim stated in it is true.

Status	On a TERM
Proposed	a candidate definition, recorded and not adopted. Its body says what it would replace or add, and what would settle it
Active	adopted: the argument may be stated in it
Deferred	a candidate whose adoption is put off
Superseded	adopted once, then replaced. The successor is named in superseded_by
Rejected	a candidate declined, or an adopted term retired with nothing in its place

TERM gains three relations, each borrowed from the scheme that already has it:

• rivals (symmetric). Another definition of the same concept that cannot be adopted beside this one for the same use, where neither has yet replaced the other. A Proposed candidate rivals the Active term it would replace. When the question is settled, the winner supersedes the loser. The loser goes Superseded if it had been adopted, or Rejected if it never was. This is ADR-022’s lifecycle, with supersedes in the role corrects plays for THEORY.
• complements (symmetric). Another definition in force beside this one, for a different sense or dimension of the same subject.
• refines / refined_by. This term is a special case of a broader one, fixed by further conditions. Both stay in force.

supersedes means an adoption that retires the earlier term. The superseded term’s status is Superseded. A term that stays Active beside its successor is not superseded, so the relation it needs is complements or refines.

Applied now:

• TERM-050 and TERM-061 complements each other, as TERM-061’s own text says.
• TERM-010 refines TERM-043, in place of TERM-043 supersedes TERM-010. This removes a supersession the record’s own text says did not retire anything. TERM-010 stays Active. TERM-043’s prose is unchanged: it still says that it replaced the weighted scalar “as the foundation”, which is true.

Nothing is filed for frustration. #162 merged with TERM-060 Active and the open relation to thwarting held in a QUESTION. The next candidate definition of frustration should be filed as a Proposed TERM that rivals TERM-060.

Alternatives considered

• New relation names: proposes_for/proposals (one-to-one) and alternatives_to. These were drafted in the owner’s discussion before this decision. They say what rivals, complements and refines already say elsewhere in the record. A reader who knows THEORY’s rivals would have to learn a second word for the same thing. proposes_for also adds nothing to “Proposed, and rivals the Active term”, and making it one-to-one rules out a candidate that would replace two terms.
• A separate CONCEPT scheme, with definitions pointing at the stable concept they define. That is the clean split by luria’s own test, when a concept’s identity needs a standing of its own apart from every definition of it. Nothing in the record needs that yet: line and the titles already group definitions, and every claim uses a particular definition. It would mean migrating every uses edge. Revisit if a concept needs its own status or relations.
• A TERM-only status vocabulary. The shared statuses blurbs already read correctly for a definition. A separate vocabulary would add a closed list to maintain and nothing that the table above does not give.
• Status quo: candidates in prose, a QUESTION or the workbench. That is what #162 did, and it is what made every proposal a choice between adopting and erasing.

Consequences

• A competing definition can be filed, cited and compared without retiring anything. A CLAIM stated in a Proposed TERM is citing something not in force. Lint warns on it (retired-citations) where the term is cited in the body or in a ref::uses directive, as the record’s claims do. It does not warn on the frontmatter uses: edge alone. This was checked on 2026-10-18 by marking TERM-050 Proposed for one scan: 31 sites in 13 files were reported, among them CLAIM-156’s ref::uses-file line and not its uses: field. That is the intended signal: a claim stated in an unadopted definition says so with an inactive-ok note, or waits.
• complements on TERM and CLAIM is a symmetric relation within one scheme, which luria reads correctly. Its extension across schemes still waits on dmarx/luria#358, as before.
• The term-revisions chain no longer shows TERM-010 → TERM-043, because that was never a revision. Narrowing is visible on the term pages as refines / refined_by.
• ADR-032’s text is unchanged. Its history notes that this decision refines its TERM paragraph.