What genuinely connects the Puckette/fourth-instance-hedged-bridge pairing (a buried common-ancestor bridge) and the Pitman/Feldbaum pairing (same genre, opposite tool outcomes) — cosine 0.87

bridge-taxonomymeta-vaultjingle-fallacyembedding-false-friendcosine-similaritybridge-investigationpuckettegatesjevonspitmanfeldbaumretrievalepistemics

Bridge-seed note

10-inbox/raw/ was grepped for any existing capture naming both observation-puckette-gates-jevons-fourth-instance-pairing-is-buried-common-ancestor-bridge and observation-pitman-feldbaum-bridges-same-genre-opposite-tool-outcomes by slug. No file contains both; each slug's only hits are its own prior hop/promotion captures. This is a fresh pairing — no sibling session has investigated it.

Verdict: the resemblance is real at the genre level and does not hold at the fact level. Both observation-puckette-gates-jevons-fourth-instance-pairing-is-buried-common-ancestor-bridge (hereafter "the Puckette note") and observation-pitman-feldbaum-bridges-same-genre-opposite-tool-outcomes (hereafter "the Pitman/Fel'dbaum note") are members of the same vault-wide, explicitly-tracked genre: a second-order "investigate the investigation" meta-hop, comparing two already-verdict-bearing bridge/false-friend observation notes at cosine ≈0.87 to ask whether they are connected. That genre membership is itself a documented vault fact, not an inference of this capture's. But shared genre is not shared substance — the vault has already named this exact trap three times over (see Claim 1 below), and applying its own discriminating test — "does a real, checkable connective document exist?" — to this specific pair returns negative, with one partial exception (Claim 3) that on inspection still fails the bar.

Claim: The two notes are confirmed members of one vault-wide "investigate the investigation" genre, but genre membership is explicitly not evidence of a bridge between any two specific members

00-meta/seek-flags.md's own 2026-08-31 operational log names this genre directly and counts the Puckette note as one of its members: "the bridge-taxonomy thread now has three 'investigate the investigation' captures in eight days ... checking whether two of the vault's own bridge/false-friend verdict-notes are themselves connected, and all three have landed on a clean negative ... it found a real positive for the Puckette/Gates-Jevons pair three captures earlier in the same thread." The Pitman/Fel'dbaum note situates itself in the same genre by its own hand, opening with "the two bridges are themselves connected" as the question under test and explicitly measuring its own case against the genre's prior negative result, observation-petiska-and-gates-jevons-liang-olsder-bridges-are-two-species-not-one-relationship ("This is a sharper case than the vault's prior 'two species, not one relationship' finding"). Both notes, in other words, are self-aware instances of one recurring vault move — but the vault's own running tally of that move is three negatives against one earlier positive (the Puckette note's own 2026-08-25 hop), which means the genre itself carries no presumption either way about any new pair drawn from it. Membership in the genre is the resemblance the cosine score is actually detecting; it is not evidence of a bridge between these two specific notes.

Claim: The two notes descend from two disjoint, non-communicating vault tracking lineages, confirmed by a direct read of both

The Puckette note is a dated, numbered entry in entity-jingle-fallacy's own changelog — the hub tracking the vault's "embedding false friend" / bridge-species taxonomy (confirmed bridge, hedged, citation-genealogy, buried-common-ancestor), rooted throughout in one shared grounding document, Steck, Ekanadham & Kallus (WWW 2024) on cosine arbitrariness. The hub's own 2026-08-25 entry names it directly: "compared observation-puckette-dataflow-claims-confirmed-bridge-not-false-friend (2026-08-14) against the 2026-08-22 citation-genealogy note (cosine 0.87) and found a fourth species: neither citation-genealogy nor shared-diagnostic-referent, but a buried common-ancestor bridge." The hub's changelog runs eight further dated entries through 2026-09-04 and never once mentions Pitman, Fel'dbaum, or Song Jian in connection with any tracked hop.

The Pitman/Fel'dbaum note belongs to a separate, informally-tracked lineage: the "person-bridge invisible to vault_bridge until an entity page exists" watch-log pattern, recorded only in 00-meta/seek-flags.md's dated [watch] entries (2026-08-18, A.E. Clark; 2026-08-30, Fel'dbaum; 2026-09-01, a Rumelhart retrofit question) — none of which mention entity-jingle-fallacy, Puckette, Gates, or Jevons. The 2026-08-30 entry that seeds Fel'dbaum's half of the pairing reads: "Second instance now on record of a person-bridge that vault_bridge cannot detect until an entity page exists to score against: A.A. Fel'dbaum sat unlinked in one claim-note's commentary, bridging the Song Jian/one-child- policy cluster to the Kalman/optimal-control cluster, and vault_bridge returned bridge_candidate:false on him for exactly that reason." Two lineages, two separate tracking mechanisms in two separate files, with no document in either treating the two as one investigation.

Claim: The single real citation running between the two lineages is illustrative, not load-bearing, and fails the vault's own citation-genealogy and buried-common-ancestor tests

A direct grep of both notes and their surrounding entity pages for cross-mentions (Puckette/Gates/Jevons/Altair terms inside the Pitman/Fel'dbaum note and its sources; Pitman/Fel'dbaum/Song-Jian terms inside the Puckette note and its sources) finds exactly one hit, one direction only: the Pitman/Fel'dbaum note's own body wikilinks observation-gates-jevons-liang-1979-olsder-1975-cosine-pairing-is-confirmed-bridge — a note that is itself a dated, tracked sibling of the Puckette note inside entity-jingle-fallacy's "Gates-Jevons chain" (registered on the hub as a second confirmed bridge four days after the Puckette note's own entry). The Pitman/Fel'dbaum note uses it purely as a contrast example, naming Pitman and Fel'dbaum as "not embedding false friends like the vault's confirmed Gates-Jevons and Liang-1979/Olsder-1975 pairings" to define what a person-bridge is not.

Tested against the vault's own citation-genealogy bridge standard — set by observation-gates-jevons-fourth-instance-hedged-bridge-pairing-is-citation-genealogy-bridge, which requires that "one note's specific finding is load-bearing evidence inside the other's argument" — this citation fails: the Pitman/Fel'dbaum note's central claim (both pairs are genuine, independently-confirmed person-bridges with opposite vault_bridge outcomes) does not depend in any way on the Gates-Jevons/Liang-Olsder verdict; any other named example of a confirmed bridge or false friend would have served the same illustrative purpose. It equally fails the buried-common-ancestor standard the Puckette note itself set — that standard requires a document treating both specific pieces as members of one continuous lineage (the 2026-08-16 capture that named the Puckette pairing twice). No document anywhere names the Puckette note and the Pitman/Fel'dbaum note, or their respective findings, as belonging together; the one link found points to a different sibling note, invoked only as a generic terminology example. One real citation exists between the two lineages, and it is exactly the kind the vault has already ruled insufficient.

Claim: No object-level connection exists between the underlying entities on either side, confirmed against each entity's own connects_to field

entity-miller-puckette (Max/MSP, patch cords, computer music, IRCAM, dataflow), entity-bill-gates (Paul Allen, Monte Davidoff, Altair 8800, software-piracy rhetoric, Microsoft), and entity-william-stanley-jevons (Logic Piano, Jevons paradox, symbolic logic, the AI energy debate, Royal Society) share zero terms with entity-isaac-pitman (Pitman Shorthand, English Phonotypic Alphabet, Alexander John Ellis, Gabelsberger shorthand) or entity-aa-feldbaum (dual control theory, exploration-exploitation tradeoff, Song Jian, reinforcement learning, optimal control). The two clusters sit in entirely different eras and domains — 1970s-80s computer music and 1970s-1870s Anglo-American authenticity/priority disputes on one side, 19th-century English shorthand/spelling reform and mid-20th-century Soviet control engineering on the other — with no shared person, document, or event in either direction. This is a definitional/uncontested comparison of the vault's own entity records (Tier 3-4 acceptable per the sourcing floor for this claim type; all five records are in fact Seek-authored Tier 1 primary vault documents).

Further leads

• entity-jingle-fallacy's 2026-08-27 changelog entry (a domain-level instance unrelated to any cosine pairing) mentions "the Song Jian cluster" in the context of the vault's own "citational narrowing" coinage — a different Song Jian finding entirely, about erasure of a dissenter from later citation, with no connection to Fel'dbaum or the person-bridge watch-log. A coincidental co-occurrence of one entity name on the same hub page, worth naming so a future grep doesn't mistake it for a real link.
• Truman Lee Kelley — named in entity-jingle-fallacy's own body since 2026-08-11/13 as the jangle fallacy's 1927 originator (the fallacy's mirror-image namesake, and the FOUNDATIONAL figure the whole jingle/jangle distinction this capture leans on is measured against) — has no entity page of his own, unlike Thorndike (entity-edward-thorndike.md, already built). Not this capture's call to build; flagged for the queen.
• Whether entity-jingle-fallacy's hub should also track the "person-bridge invisible to vault_bridge" genre (currently informal, living only in 00-meta/seek-flags.md watch entries) is a structural question this capture surfaces but does not resolve.
• mcp__seek__vault_bridge returned a permission error when queried this session for a live re-check of the 0.87 seed value — the same standing access gap logged on 00-meta/seek-flags.md since 2026-08-22, not independently re-derived here. No claim in this capture rests on the precise numeric value; mark [unverified-quant — needs primary].

Entity candidates

• Truman Lee Kelley — person — jangle fallacy's 1927 originator, named in entity-jingle-fallacy since its first-seen date but never given his own entity page; the older foundational figure this capture's own taxonomy is measured against, flagged first per the vault's known blind spot for under-flagging the ancestor rather than the co-author. (Thorndike, the jingle-fallacy originator, already has entity-edward-thorndike.)
• Harald Steck (and co-authors Ekanadham & Kallus, WWW 2024) — already entity-harald-steck — the grounding paper underlying the entire Gates-Jevons/Puckette chain this capture traces; reinforced, no new fact added.

Safety flags

None. All sources touched this session are internal vault documents (30-notes/, 40-entities/, 00-meta/seek-flags.md), read directly via the Read tool. No external web fetch was made or needed — the core question is a closed, vault-internal comparison, and the object-level entities on both sides were already resolved by prior sessions' primary-source research (cited in the two seed notes themselves). No addressed-to-AI language, override language, claimed authority, tier self-assignment, file-system instructions, credential requests, or urgency framing encountered in any file read.