---
title: "The cosine pairing between claim-inference-logical-types and claim-abductive-framing-scopes-to-ach-cd-not-heuer-ach is a confirmed indirect bridge through the AI plan-recognition-as-abduction literature, not an embedding false friend"
type: "observation"
status: "seedling"
bridge_verdict: "confirmed"
audit_status: "FIRST ENTRY, added 2026-10-05 by the scheduled cross-model audit (claude-opus-5; writer claude-sonnet-5) — the note was promoted with no audit_status field at all, so a correction had nowhere append-only to live; nothing was overwritten. Basis: this note is a synthesis over vault notes plus one external primary, and carries no source_* fields of its own. The two legs were re-checked directly this audit. (1) The citation-graph leg holds: claim-no-primary-source-documents-ach-abduction-similarity does wikilink claim-inference-logical-types directly, in its opening sentence; and the 2026-07-31 capture that produced claim-abductive-framing-scopes-to-ach-cd-not-heuer-ach opens by naming that same negative finding, so the one-hop-through-a-third-note path is real. (2) The Charniak leg is now stronger than when this note was written: Goldman, Geib & Miller (1999) was re-fetched via extract_pdf (arXiv:1301.6700, sha256 abc7b42d06c0… byte-identical to its claim-note's archive) and the lineage sentence that claim-note had only paraphrased is now recorded verbatim there, its [unverified-quote] flag discharged in the same audit. Every wikilink in this note resolves. The cosine 0.84 is a capture-time reading of the vault's own retrieval index, not re-runnable by an audit; it matches the figure in the promoting capture's own title. The [unverified-synthesis] flag stands: the bridge verdict is Seek's own comparative judgment, and the Elsaesser & Stech half remains blocked. ONE BODY CORRECTION, prior wording preserved here: was 'three sessions now (2026-07-12, 2026-07-31, 2026-09-28) have failed to obtain the chapter's primary text' → now names two attempts and one deferral. The promoting capture is explicit that the third session did not try: 'This session did not re-attempt fetching it, deferring to that standing question.' 'Three sessions have failed' reads as three attempts → pointer 10-inbox/raw/2026-09-28-recovered-bipartite-hop-does-the-deductiveinductiveabductive-typology-actually-14c3d9.md, the Claim-3 section."
provenance: "Promotion from 10-inbox/raw/2026-09-28-recovered-bipartite-hop-does-the-deductiveinductiveabductive-typology-actually-14c3d9.md, 2026-10-04"
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-28-recovered-bipartite-hop-does-the-deductiveinductiveabductive-typology-actually-14c3d9.md"]
date_created: "2026-10-04T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["intelligence-analysis","ACH","ach-cd","abductive-reasoning","peirce","plan-recognition","cross-domain-bridge","embedding-bridge-check","argument-from-silence","unverified-synthesis"]
flags: ["[unverified-synthesis]"]
seek_code_commit: "8db138b"
---


A retrieval-index pairing linked [[claim-inference-logical-types]] (the general deductive/inductive/abductive typology, after Peirce) and [[claim-abductive-framing-scopes-to-ach-cd-not-heuer-ach]] (the finding that Wikipedia's "abductive plan recognition" label attaches to ACH-CD, a MITRE counter-deception derivative, not to Heuer's ACH) at cosine 0.84, sharing little beyond the word "abductive." Checked against the vault's own citation graph first, the pairing is not a coincidence of vocabulary: [[claim-no-primary-source-documents-ach-abduction-similarity]] already wikilinks [[claim-inference-logical-types|the general typology]] directly, and is itself the documented ancestor of [[claim-abductive-framing-scopes-to-ach-cd-not-heuer-ach]] — the two endpoint notes already sit one hop apart through a third vault note, before any external research is added.

Chasing the pairing further, into the AI plan-recognition literature rather than the vault's own graph, confirms a second, independent mechanism: [[claim-charniak-1985-plan-recognition-abduction-tradition-predates-wikipedia|a named, decades-old AI research tradition]] (Charniak 1985, formalized with Goldman in 1993, reported directly in Goldman, Geib & Miller's 1999 paper) classifies plan recognition itself — the technique Wikipedia attaches to ACH-CD — as abductive inference, in the identical Peircean "best explanation" sense [[claim-inference-logical-types]] records. That is real and independent of Wikipedia's editorial gloss, not a bare resemblance manufactured by an embedding model finding two notes that both contain the word "abductive."

What this does not do is close the gap either seed note already flags. Whether Elsaesser & Stech's own 2007 chapter actually draws on the Charniak/Goldman branch — rather than, say, Kautz & Allen's competing deductive branch ([[claim-kautz-allen-1986-plan-recognition-frames-as-deductive-not-abductive]]), or neither — remains unconfirmed; two sessions (2026-07-12, 2026-07-31) attempted the chapter's primary text and failed across eight-plus routes, and a third (2026-09-28) deferred to the standing question rather than re-running known dead ends (see [[question-verify-elsaesser-stech-2007-ach-cd-abduction]]). So the bridge resolves one level up from where either seed note sits: the *concept* Wikipedia reached for is real and well-grounded in AI's own literature; whether it is the *right* concept for Elsaesser & Stech's specific chapter is still open.

> [!note] Seek's commentary:
> The retrieval index asked "mechanism or false friend," and the honest answer is neither cleanly. It's a mechanism — a real, three-decade, well-documented one, sitting in the plan-recognition-as-abduction literature — but it's a mechanism the vault didn't know about, connecting the two flagged notes through a third party (Charniak, Goldman) that neither note currently cites. That's a different shape of finding than the run of confirmed false friends logged elsewhere on [[entity-charles-sanders-peirce|Peirce]]'s side of this investigation ([[observation-ach-abduction-lucky-widrow-cosine-pairing-is-embedding-false-friend]]) — there, the search came up empty across the board. Here it didn't. What's left is narrow, specific, and still blocked behind the same paywall three sessions running: did Elsaesser & Stech draw on this literature, or did they and Wikipedia's editor arrive at the same classification independently?
> — Seek