The cosine-0.85 pairing between the vault's legitimation-backlink and Werbos-obscurity maintenance-gap observation notes is a confirmed bridge, not a false friend — both document one recurring mechanism: a 40-mocs hub named as "home" by a child document that the hub's own list does not cite back

moc-maintenancebacklink-gapstructural-gapconfirmed-bridgebridge-investigationjingle-fallacymeta-vaultwerboslegitimation-not-origination

Seek's semantic index flagged observation-moc-legitimation-not-origination-backlink-to-parent-moc-missing and observation-werbos-obscurity-claim-missing-from-backpropagation-origins-moc as unlinked neighbors at cosine 0.85 [unverified-quant — needs primary]. The two seed notes share no domain vocabulary: one is about Cold War Chinese population policy and Geert Jan Olsder's declined-credit testimony, the other about 1970s neural-network history and Paul Werbos's account of his own obscurity. As with the observation-ml-ad-en-gedi-and-gates-jevons-false-friend-notes-are-a-confirmed-bridge pairing, this is a bipartite hop one level above the vault's usual question — both seeds are themselves verdicts about the vault's own filing, not subject-matter claims.

Direct comparison confirms the bridge. Both notes document one structural mechanism at two different grains: a 40-mocs/ hub that a child document explicitly names or cites as its "home," which the hub's own list does not cite back. In the legitimation note the child is itself a MOC — moc-legitimation-not-origination names moc-attribution-and-origin-myths as its "general home," and a direct read found that "[n]o equivalent entry exists for moc-legitimation-not-origination, despite that map predating both fixes." In the Werbos note the child is a claim-note — claim-werbos-attributes-early-obscurity-to-publishing-strategy-not-suppression cites three members of moc-backpropagation-origins, yet "[a] direct read of the MOC's full text confirms the claim-note's slug appears nowhere in it." MOC-citing-MOC in one case, claim-citing-MOC in the other; the same one-directional backlink gap in both, and both notes assign the fix identically — a 40-mocs/ Warden edit, not a 30-notes/ promotion edit.

The shared surface is not incidental: the two notes' common referent is that mechanism, the way the jingle-fallacy false-friend verdicts turned out to share their diagnosed failure class (entity-jingle-fallacy). The mechanism recurs and has an informal fix — a Warden pass adds the missing reciprocal bullet. The legitimation half was closed that way on 2026-09-19; the Werbos half's one confirmed gap, against moc-backpropagation-origins, remained open as of 2026-10-05. (Update 2026-10-18, cross-model audit: since closed — the Warden's 2026-10-17 pass on moc-backpropagation-origins added the reciprocal entry, citing the gap-note by name.) Per Cali's 2026-10-05 review of the originating capture, the capture's wider framing (the Werbos note "independently absent from three separate hubs," one of ≥8 instances) overcounted: two of those three absences were deliberate Warden scope rulings declining to fold the note into moc-argument-from-silence and moc-attribution-and-origin-myths, not maintenance misses. The bridge holds regardless; it rests on the mechanism, not the census.