The cosine-0.84 pairing between the ACH/abduction and Lucky-Widrow claim-notes is an embedding false friend — no connection exists at the object level

embedding-false-friendcosine-similarityjingle-fallacyretrievalepistemicsbridge-investigationargument-from-silenceachabductive-reasoningluckywidrow

Seek's semantic index flagged claim-no-primary-source-documents-ach-abduction-similarity and claim-lucky-widrow-independent-origination-favored-lucky-side-unverified as unlinked neighbors at cosine 0.84, raising the question of whether a genuine bridge connected them. At the object level — the actual subject matter of the two notes — it does not: this is the vault's seventh confirmed instance of an embedding false friend.

Targeted searches for "abductive reasoning" + "Widrow-Hoff" / "least mean squares," for Peirce/inference-to-the-best-explanation + "adaptive filter" / "gradient descent," for "Analysis of Competing Hypotheses" + adaptive-filter terms, and for Robert Lucky + abduction/abductive returned only technical adaptive-filtering material on one side (GeeksforGeeks, CREWES, IEEE/arXiv papers) and philosophy-of-abduction material on the other (Stanford Encyclopedia of Philosophy, PhilArchive), with zero overlap. Nothing connects Peircean abduction or ACH to Widrow-Hoff LMS or Lucky's 1965 equalizer, nor Lucky himself to abductive reasoning.

At the object level, intelligence-analysis epistemology and telecommunications/adaptive-control engineering history are disjoint literatures; the cosine proximity is rhetorical surface (both notes are negative findings graded on the argument-from-silence ledger), not shared subject. The bridge that does exist sits one level up, in the two notes' shared method rather than their shared content — see observation-ach-abduction-lucky-widrow-notes-share-argument-from-silence-method-classified-as-abductive. See entity-jingle-fallacy for the tracked series this joins.