The two notes bracket vault_bridge's error directions

Two vault notes describe opposite failure directions of the vault_bridge cosine tool. observation-gates-jevons-fourth-instance-hedged-bridge-pairing-is-citation-genealogy-bridge concludes that a cosine score cannot separate a real connection (shared-referent or citation-genealogy) from a false friend without a direct read: the false-positive / misclassification side. observation-pitman-ossietzky-bridges-share-genre-not-object-level-connection records a false negative: Ossietzky's bridge_candidate was false because no entity page existed to score against, while Pitman's person-bridge was detected (see observation-ossietzky-bridges-liu-xiaobo-and-nobel-foundation-clusters and observation-isaac-pitman-bridges-gabelsberger-shorthand-and-ipa-clusters).

Read together, cosine over-links resemblance (see observation-gates-jevons-cosine-pairing-is-embedding-false-friend, observation-puckette-pitman-feldbaum-meta-pairing-is-embedding-false-friend) and under-links connection carried by a person's name.

What this is not

The two notes share no object, citation, entity, era, or wikilink in either direction; neither's finding is load-bearing in the other's argument, so this is not a citation-genealogy bridge. Their cosine proximity is explained by shared meta-vocabulary ("bridge," "cluster," "false friend") and one writer's register, the same diagnosis observation-pitman-feldbaum-bridges-same-genre-opposite-tool-outcomes reached for a sibling pair. The bracketing is a reason a future bridge-species hub might cite both notes, not a link between them. The existing observation-petiska-and-gates-jevons-liang-olsder-bridges-are-two-species-not-one-relationship already pursues a species framing that such a hub would have to reconcile.