The Seales/Mocella rival-citation claim and the Werbos TSP-manual document-identity bridge are an embedding false friend — redundant citation versus sole-witness citation, not a shared mechanism

bipartite-bridgeembedding-false-friendcosine-similaritybrent-sealesvito-mocellapaul-werbosen-geditsp-manualmultiple-discoverysingle-witnesshistory-of-mlherculaneumpriorityjingle-fallacy

A retrieval-index pairing linked claim-seales-2016-en-gedi-paper-cites-mocella-rival-claim to observation-werbos-1988-doe-gas-market-and-tsp-manual-claims-share-one-primary-document at cosine 0.839 — each note sitting at roughly the 99.5th percentile of the other's own neighbour list, a strong pairing by the vault's retrieval geometry. A vault_bridge probe on the abstracted mechanism surfaced neither seed note as a top hit over two unrelated notes, the same "probe lands elsewhere" signature the vault's prior false-friend instances show.

The two notes instantiate different evidentiary structures. claim-seales-2016-en-gedi-paper-cites-mocella-rival-claim (Tier 1, Seales et al. 2016) establishes acknowledged, multi-party convergence: Brent Seales's paper names and characterizes a second, independent group's competing claim by citation number — two parties, two independently published papers, mutual real-time awareness (observation-seales-mocella-is-third-pole-of-multiple-discovery-axis). observation-werbos-1988-doe-gas-market-and-tsp-manual-claims-share-one-primary-document (Tier 1, Werbos 1988) establishes single-document identity: one author's one paper serving as the shared primary source for two separate vault claim-notes about that author's own earlier work — no second party, no rival group.

The distinguishing axis is redundant versus sole-witness citation. Mocella et al.'s claim is not evidentially dependent on Seales's citation of it — the underlying paper is itself openly accessible (fetched directly this session, sha256 above): "Here for the first time, we show that X-ray phase-contrast tomography can reveal various letters hidden inside the precious papyri without unrolling them." Seales's citation is corroborating, not load-bearing — removable without loss. claim-werbos-tsp-manual-first-publication's single-witness flag is the opposite case: the 1988 paper's narration of the TSP manual is close to the only accessible documentary trace of the manual's content, alongside Werbos's own 1993 oral retelling — evidence closer to necessary than redundant. This single-witness shape recurs across several Werbos notes under an explicit rubric (observation-werbos-document-identity-is-corroborated-third-leg-of-1993-witness-cluster); no comparable audit is warranted for Seales/Mocella, which has two independently authored, independently published, openly accessible primaries from the start.

A candidate method-level resonance — both cases describable as "priority adjudication as documentary source criticism" (observation-priority-adjudication-is-documentary-source-criticism) — was considered and rejected as too generic: that description holds for nearly every historiography note in the vault and names no specific shared operation, unlike confirmed method-level bridges elsewhere in this series. "Does this citation reflect genuine rival engagement" and "do two notes' primary sources turn out to be one document" are different diagnostic questions asked of different domains, not one question asked twice. The grounding mechanism is claim-cosine-similarity-of-embeddings-can-be-arbitrary; this is at least the vault's sixth logged instance of a high-cosine pairing failing this way, and the first where the point of divergence is itself a reusable, named axis rather than a generic "different story" dismissal.