ADR-029: presupposes joins the THEORY relations

Context

On 2026-10-05 the owner noticed that most theories relate to papers but not to other theories. A count bore it out:

• 129 of the 153 theories name another theory in their bodies;
• 58 declare any theory-to-theory relation;
• 319 named pairs are undeclared;
• 35 bodies say outright “no relation is declared”.

The scheme had three relations:

• extends is a refinement or widening of the same mechanism;
• rivals is another account of the same phenomenon that cannot also be right (ADR-022);
• corrects replaces the other’s reasoning. It has never been used.

The undeclared mentions are mostly contrasts and parallels, but a common kind is a real commitment none of the three can say: one account takes another’s conclusion as a premise for a claim about a different phenomenon. Pettit’s corporate responsibility (THEORY-139) rests on List and Pettit’s result that a rational group must form its judgments holistically (THEORY-127). It argues from that result to a corporation having a mind of its own, and from there to responsibility. It was filed as extends because extends was the nearest relation available, and that is false: responsibility is not a refinement of the aggregation mechanism.

Decision

Add presupposes to the THEORY scheme, with the converse presupposed_by, written by luria link --fix.

• Meaning. This account takes the other’s conclusion as a premise for a claim about something else.
• Consequence for status. If the presupposed account is Rejected, the presupposing one loses a premise and needs another route to its conclusion. It is not thereby wrong. That is the difference from extends, where the extension usually falls with what it extends.
• Not in the chain. The theory chain draws lines of explanation from extends and corrects, with the invariant that linked accounts share a topic. Presupposition often crosses topics (social ontology to ethics, ecology to method), and those are the links a reader most needs, so it joins no chain and the topic invariant does not apply to it.
• Explained where held. As with every relation, the body says in prose what is taken from the other account.

First uses:

• THEORY-139 presupposes THEORY-127, replacing its extends.
• THEORY-136 presupposes THEORY-132. Scheffer and Carpenter’s standard of evidence for alternative stable states is built on Holling’s picture of basins of attraction and resilience as basin size. It is a claim about how to test for alternative states, not a refinement of Holling’s claim that stability and resilience come apart.

The remaining undeclared pairs are left for an audit, which should look for misfiled extends as well as missing relations.

Alternatives considered

• Prose only. This is what filings did before this decision. It is true but invisible to the views and to the older entry, which never gets a back-reference.
• Stretching extends. That is what happened with THEORY-139, and it misstates the claim. It would also put cross-topic presuppositions into the lines of explanation, where the topic invariant flags them as unbound.
• A generic related or see_also. That would let a link stand without saying what it commits to, which defeats the point of declared relations. Contrasts and parallels stay in prose because they commit to nothing that could fail.
• Adding it to LIT as well. Papers already have extends, rivals, corrects and compared_against. No case has yet needed presupposition between papers, so the LIT scheme is unchanged.