Claim modality is the inference form of the refuter

Type: kb/types/note.md · Tags: foundations, kb-maintenance

A claim's modality is not a fact about how the claim was generated or supported; it is a fact about which inference form can bear against it. A universal claim is refuted deductively — modus tollens on one genuine instance. A statistical claim is refuted inductively — a sample-based prevalence result. An ideal-type claim is refuted comparatively — the abductive judgment that the first-order model is no longer the best account: it loses dominance to its corrections or a rival, or its conceded exceptions turn out to be ordinary unmarked practice. A refuter-defined mode list is therefore not an arbitrary menu; it is the image of the inference forms under the question "what evidence can count against this claim?"

Each pairing has an external grounding, with a stated delta. The statistical mode's stated-refuter requirement is Popper's treatment of probabilistic claims — strictly unfalsifiable until a methodological decision fixes the refuting frequency — made per-claim in the claim text rather than by field-wide convention. The ideal-type mode's adequacy record is inference to the best explanation with its virtues (declared use, omitted mechanism, consequence bound, dominance) written as attackable commitments; the bound is absolute, so "best available" alone never accepts — the standard best-of-a-bad-lot objection is answered in the record, not by reviewer judgment. A priced exception is Lakatos's anomaly: it earns an idealization assessment and the adequacy record then decides.

The mapping classifies the criticism-phase relation only. All claims begin as conjectures — abduction in the context of discovery, which Popper deliberately left unregulated — so "statistical claims are inductive claims" would be a category error about genesis. What distinguishes the ideal-type mode is that abduction is also its acceptance and refutation logic, not only its origin.

Two corollaries follow. Closure: if deduction, induction, and comparative abduction exhaust the forms by which evidence bears against a truth-apt empirical claim, a refuter-defined mode list with these three is complete — a proposed fourth mode must exhibit a fourth inference form of refutation, which is a standing reason to refuse mode proliferation. Precise vacuity: a claim hedged until no inference form bears against it stands in none of the three refutation relations; that is what "hedged into vacuity" means, stated exactly.

The mapping also yields a testable prediction about repair under review: reframes should drift toward zero-exposure forms — question and procedure framings escape all three refuter relations, since a question cannot be false, has no prevalence, and explains nothing. The refuter guard on reframes is load-bearing precisely because it blocks that attractor.

Scope

Truth-apt empirical claims only. Normative claims and definitions fail differently, so this typology does not cover them. In a design record, apply it to empirical beliefs rather than residual selections: a proposal can contain both kinds of region, so proposal status does not put the whole artifact inside or outside the refutation relations.

Open Questions

  • Is comparative abduction a genuinely third form, or reducible to deduction over comparative premises? The answer decides whether closure is a theorem or a convention.
  • Does dominance failure refute an ideal-type claim or supersede it? The ideal-gas law was superseded by van der Waals, not refuted by its corrections growing; the distinction may need its own place in the adequacy record.

Relevant Notes: