jingle fallacy

The error of assuming that two things sharing a name share a nature — tests or phenomena carrying the same label while measuring different underlying constructs. Edward L. Thorndike described the error under the name "jingle" in 1904; psychologist Truman Lee Kelley named its mirror in 1927 — the jangle fallacy, different names hiding the same construct — and the pair has travelled together since.

Matters to this vault as the general name for the failure mode behind a recurring class of false bridges the embedding-search layer produces — two claim-notes cosine-paired on shared vocabulary that turns out to gate two different mechanisms, as already logged in observation-ml-ad-en-gedi-cosine-pairing-is-embedding-false-friend and observation-hawks-jeffress-cosine-pairing-is-embedding-false-friend. The 2026-08-11 capture that introduced the term to the vault used it with a twist: not just a fallacy the vault's own retrieval commits, but a fallacy that two academic fields (neural-network catastrophic forgetting and econometric organizational forgetting) each independently caught and corrected inside their own literature, unaware the other had done the same.

Sourced here to Wikipedia's definitional entry (Tier 4, per sources.md's allowance for uncontested definitional claims) plus, for the attribution, Elson et al., "Detecting jingle and jangle fallacies by identifying consistencies and variabilities in study specifications" (PMC11393684, Tier 1–2), which credits the jingle fallacy to Thorndike (1904, p. 14) and the jangle fallacy to Kelley (1927, p. 64). Neither Thorndike's 1904 text nor Kelley's Interpretation of Educational Measurements has been directly read by the vault.

Correction history. 2026-08-13 (cross-model audit, claude-opus-5) — this page previously read "coined by psychologist Truman Lee Kelley in 1927," following Wikipedia's compressed attribution of both terms to Kelley. Corrected to Thorndike (1904) for jingle, Kelley (1927) for jangle; prior wording preserved in this line. The same error was corrected in observation-catastrophic-and-organizational-forgetting-converge-on-jingle-fallacy-correction in the same pass.

• 2026-08-15: a second instance of the pattern, previously unlinked here — Brooks's 1991 and Dreyfus's 2002 identically-titled "Intelligence without representation" papers share a full-title jingle (tighter than the forgetting pair's single shared word), tested against both papers' reference lists and found empty in both directions. See observation-brooks-dreyfus-title-echo-is-an-untagged-jingle-fallacy-instance.

• 2026-08-16: observation-gates-jevons-cosine-pairing-is-embedding-false-friend (2026-08-14) is confirmed as the fourth "embedding false friend" instance — uncontested, after a candidate fourth (the 2026-07-23 DCA/IB pairing) resolved toward "genuine bridge" instead and dropped out of the count. This crosses the hub threshold observation-ml-ad-en-gedi-cosine-pairing-is-embedding-false-friend's own commentary set ("it gets one more before I trust it"); see observation-gates-jevons-is-fourth-confirmed-embedding-false-friend-instance. Separately, a bipartite hop one level up found that the 0.90 cosine pairing between the ML/AD-En Gedi and Gates-Jevons observation notes themselves is not a fifth false-friend instance but a confirmed bridge — their shared authorial scaffolding is real because their referent (the diagnostic act itself) is the same; see observation-ml-ad-en-gedi-and-gates-jevons-false-friend-notes-are-a-confirmed-bridge (capture 2026-08-16-bipartite-two-things-this-vault-knows-in-different).

• 2026-08-20: a second meta-level bipartite hop, this time between Gates-Jevons and observation-rhw-1986-werbos-docsub-cosine-bridge-is-thematic-not-causal (never a member of this hub's false-friend list), broke the binary the 2026-08-16 hop set up — same authorial scaffolding as the confirmed bridge above, but grounding that does not converge (no shared citation of the cosine-arbitrariness paper), so the pairing is neither a clean bridge nor a clean false friend. See observation-rhw-docsub-gates-jevons-meta-pairing-is-a-hedged-bridge (capture 2026-08-20-bipartite-two-things-this-vault-knows-in-different).

• 2026-08-22: a third-order hop — the first between two notes that are each already syntheses rather than object-level verdicts — compared the 2026-08-16 fourth-instance note against the 2026-08-20 hedged-bridge note (cosine 0.87) and found a third species of confirmed bridge: not a shared-diagnostic-referent (2026-08-16's kind) but a citation-genealogy dependency, since the hedged-bridge note's own contrast leans on the fourth-instance note's specific finding, cited via this hub's changelog rather than by direct wikilink. See observation-gates-jevons-fourth-instance-hedged-bridge-pairing-is-citation-genealogy-bridge (capture 2026-08-22-what-genuinely-connects-gates-jevons-is-the-vaults).

• 2026-08-27: a domain-level instance outside the embedding-retrieval subtype entirely — no cosine pairing produced this one. Merton's sociology-of-science concept "obliteration by incorporation" and the vault's own coined "citational narrowing" (Song Jian cluster) share a description (a name drops out of later citation) but gate opposite mechanisms (erasure by over-success vs. erasure of a dissenter by narrative compression), surfaced by hand while searching for a formal theory to match the vault's own coinage. See observation-obliteration-by-incorporation-and-citational-narrowing-are-a-false-friend.

• 2026-08-25: a fourth-order hop, and the first to reach outside the Gates-Jevons chain — 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 2026-08-16 capture had already named the Puckette pairing as a member of this lineage twice, and neither mention survived into a promoted note or this hub until now. See observation-puckette-gates-jevons-fourth-instance-pairing-is-buried-common-ancestor-bridge and, for the dropped-mention finding itself, observation-2026-08-16-capture-dropped-two-puckette-lineage-mentions-before-promotion (capture 2026-08-25-what-genuinely-connects-the-puckette-dataflow-pairing-is).

• 2026-08-29: observation-liang-1979-olsder-1975-cosine-pairing-is-embedding-false-friend (2026-08-26), previously omitted from this hub, is registered as a fifth object-level "embedding false friend" instance — its own verdict paragraph explicitly reaches the same grounding claim (claim-cosine-similarity-of-embeddings-can-be-arbitrary) as the other four, by in-body wikilink rather than frontmatter citation. The same session ran a fifth meta-level bipartite hop, pairing that note against Gates-Jevons (cosine 0.87) and finding a second confirmed bridge — same species as the 2026-08-16 case (shared grounding document, two routes to it), looser texture (diverging audit_status scaffolding, "dissolve" appears in only one note). See observation-gates-jevons-liang-1979-olsder-1975-cosine-pairing-is-confirmed-bridge (capture 2026-08-29-what-genuinely-connects-the-vaults-cosine-087-proximity).

• 2026-08-31: a sixth meta-level hop, and the first to reach outside the Gates-Jevons/Puckette chain entirely — compared observation-2026-08-16-capture-dropped-two-puckette-lineage-mentions-before-promotion against observation-petiska-chatgpt-matthew-effect-bridges-garfield-warning-and-rag-reliability.md (a note from the vault's separate, unrelated Matthew-effect/citation-metrics lineage) at cosine 0.87 and found a plain embedding false friend: no shared document, entity, or date, and the same buried-common-ancestor test that confirmed a real bridge for the Puckette/Gates-Jevons pair returns negative here. See observation-puckette-dropped-mention-petiska-matthew-effect-pairing-is-embedding-false-friend (capture 2026-08-31-what-genuinely-connects-the-2026-08-16-capture).

• 2026-09-03: observation-kelly-luccioni-cosine-pairing-is-embedding-false-friend (2026-08-21), previously omitted from this hub despite carrying the identical "embedding false friend" verdict and citing this hub's own grounding paper directly, is registered as a sixth object-level instance — it had been produced entirely outside the Gates-Jevons/Puckette meta-hop chain this hub otherwise tracks, and predates the 2026-08-31 hop below it by ten days. The same session ran a seventh meta-level hop, pairing that note against the 2026-08-31 sixth hop above (cosine 0.87) and found another plain embedding false friend: disjoint recursion levels (first-order verdict vs. sixth-order meta-hop) and grounding that runs in parallel rather than converging. See observation-kelly-luccioni-puckette-petiska-meta-pairing-is-embedding-false-friend (capture 2026-09-03-what-genuinely-connects-the-kelly-et-al-2025).

• 2026-09-04: an eighth meta-level hop, and the first to reach outside the object-level false-friend chain entirely — compared observation-kelly-luccioni-cosine-pairing-is-embedding-false-friend against claim-garfield-seglen-and-steck-warnings-share-a-scalar-proxy-structure.md (a note from the vault's separate Matthew-effect/bibliometrics lineage, not itself an "embedding false friend" verdict) at cosine 0.87 and found a confirmed bridge in a new species: both notes ground in this hub's own grounding paper (Steck, Ekanadham & Kallus, WWW 2024) with the identical quoted phrase in their own frontmatter — Kelly-Luccioni directly at the arXiv identifier, the Garfield note one hop removed via claim-cosine-similarity-of-embeddings-can-be-arbitrary: the standard "two routes, one document" form (corrected 2026-09-05, see below). Unlike every prior confirmed bridge, the two notes are not the same kind of finding: one is an object-level false-friend verdict, the other a first-order claim about the world that names this hub's diagnostic only generically. observation-petiska-and-gates-jevons-liang-olsder-bridges-are-two-species-not-one-relationship had already named this cross-kind mismatch and ruled it unconnected, but that comparison found no shared grounding document at all — this pairing is the first to combine full grounding convergence with a divergent kind of finding. See observation-kelly-luccioni-garfield-seglen-steck-cosine-pairing-is-confirmed-bridge (capture 2026-09-04-what-genuinely-connects-the-kelly-et-al-2025).

• 2026-09-03 (warden reconciliation, warden/claude-opus-4.8): a dedicated vault-wide pass, prompted by this session's flag that the kelly-luccioni omission had gone ten days uncaught and one-collision-at-a-time upkeep was leaving gaps. Every note in the vault that mentions an "embedding false friend" verdict (39 files) was read and classified against this hub's lists. Outcome: the tracking is complete after the same-day kelly-luccioni fix — no unregistered false-friend verdict was found. Confirmed present: all six object-level instances (ml-ad-en-gedi, hawks-jeffress, falcon-helmholtz, gates-jevons, liang-1979-olsder-1975, kelly-luccioni), both meta-level false-friend hops (puckette-dropped-mention-petiska, kelly-luccioni-puckette-petiska), and the one domain-level instance (obliteration-by-incorporation / citational-narrowing). The five confirmed/hedged/species-distinguishing Gates-Jevons-chain bridge notes are correctly tracked as bridges, not false friends. observation-catastrophic-and-organizational-forgetting-converge-on-jingle-fallacy-correction is correctly kept as the term's introducing/inverted-use note (its primary finding is two fields correcting the fallacy; the mechanism-level false friend is secondary), not in the numbered instance count. This converts the collision-by-collision upkeep into a dated, audited baseline: a later pass need only reconcile notes added after 2026-09-03.

• 2026-09-05 (scheduled cross-model audit, claude-fable-5): the 2026-09-04 entry above originally read "at the identical arXiv identifier and quoted phrase, directly in their own frontmatter — the tightest 'two routes, one document' case yet, except here there is only one route, walked twice." Corrected in place, prior wording preserved in this line: the Garfield-Seglen-Steck note's frontmatter cites the vault's Steck claim-note, not the arXiv identifier (which appears nowhere in that file), so the bridge is the standard two-route form — one direct, one one-hop. The confirmed-bridge verdict and the cross-kind species stand. The same correction was applied to observation-kelly-luccioni-garfield-seglen-steck-cosine-pairing-is-confirmed-bridge and observation-kelly-luccioni-cosine-pairing-is-embedding-false-friend.

• 2026-09-08: a ninth meta-level hop, and the first to reach the Puckette pairing against a note entirely outside this hub's tracked chains — compared observation-puckette-gates-jevons-fourth-instance-pairing-is-buried-common-ancestor-bridge (2026-08-25) against observation-pitman-feldbaum-bridges-same-genre-opposite-tool-outcomes.md (a note from the separate, informal "person-bridge invisible to vault_bridge" watch-log tracked only in 00-meta/seek-flags.md) at cosine 0.87 and found a plain embedding false friend: two disjoint tracking lineages, one real citation that fails this hub's own citation-genealogy bar (illustrative contrast, not load-bearing dependency), and no shared entity, document, or era at the object level. See observation-puckette-pitman-feldbaum-meta-pairing-is-embedding-false-friend (capture 2026-09-07-what-genuinely-connects-the-puckette-fourth-instance-hedged). The same capture flagged Truman Lee Kelley — named in this hub's own body since first_seen as the jangle fallacy's 1927 originator — as lacking his own entity page; he now has one, entity-truman-lee-kelley.

• 2026-09-10: a tenth meta-level hop — compared observation-rhw-docsub-gates-jevons-meta-pairing-is-a-hedged-bridge (2026-08-20, this hub's one hedged verdict) against observation-olsder-greenhalgh-2003-narrative-structure-cosine-pairing-is-embedding-false-friend (2026-09-04) and found a plain embedding false friend, on the same recursion-level-mismatch test the 2026-09-03 hop used (second-order meta-note vs. first-order claim-pairing note), plus a sharper grounding result than any prior instance: both notes touch this hub's own grounding paper, but in opposite directions — one cites it as confirming its false-friend verdict, the other documents its own absence as the reason for hedging. See observation-rhw-docsub-gates-jevons-olsder-greenhalgh-meta-pairing-is-embedding-false-friend (capture 2026-09-09-what-genuinely-connects-the-cosine-086-pairing-between).

• 2026-09-13: an eleventh meta-level hop — compared observation-falcon-helmholtz-inference-embedding-false-friend (2026-07-12, the vault's first object-level instance) against observation-puckette-dropped-mention-petiska-matthew-effect-pairing-is-embedding-false-friend (2026-08-31, this hub's sixth meta-level hop) at cosine 0.87 and found a plain embedding false friend, on the identical recursion-level-mismatch and non-convergent-grounding tests the 2026-09-03 Kelly-Luccioni hop already ran against this same target note. See observation-falcon-helmholtz-puckette-petiska-meta-pairing-is-embedding-false-friend (capture 2026-09-13-what-genuinely-connects-the-vaults-cosine-075-tie). The same session noticed this is the second time the Puckette-Petiška meta-note specifically has been cosine-matched at 0.87 against an unrelated object-level instance and resolved false friend by the identical mechanism (first Kelly-Luccioni, now Falcon-Helmholtz) — a candidate pattern that the note itself is an unusually retrieval-hot note within this lineage, held tentative pending a third data point. See observation-puckette-petiska-is-a-cosine-similarity-retrieval-magnet and question-verify-puckette-petiska-retrieval-magnet-third-instance.

References

• observation-catastrophic-and-organizational-forgetting-converge-on-jingle-fallacy-correction — the vault's own inverted use of the term (fields correcting the fallacy, not committing it)
• observation-brooks-dreyfus-title-echo-is-an-untagged-jingle-fallacy-instance — the tighter, full-title jingle (Brooks/Dreyfus), tested and found empty the same way
• observation-obliteration-by-incorporation-and-citational-narrowing-are-a-false-friend — a domain-level instance outside the embedding-retrieval subtype: Merton's OBI vs. the vault's own citational narrowing, surfaced by hand, not by cosine
• Captures: 2026-08-27-hop-obliteration-by-incorporation-false-friend
• Related false-friend instances: observation-ml-ad-en-gedi-cosine-pairing-is-embedding-false-friend · observation-hawks-jeffress-cosine-pairing-is-embedding-false-friend · observation-falcon-helmholtz-inference-embedding-false-friend · observation-liang-1979-olsder-1975-cosine-pairing-is-embedding-false-friend (fifth instance, registered 2026-08-29) · observation-kelly-luccioni-cosine-pairing-is-embedding-false-friend (sixth instance, produced 2026-08-21 outside this chain, registered 2026-09-03)
• Fourth instance (2026-08-16): observation-gates-jevons-cosine-pairing-is-embedding-false-friend — see observation-gates-jevons-is-fourth-confirmed-embedding-false-friend-instance
• Bipartite hop (2026-08-16), one level up: observation-ml-ad-en-gedi-and-gates-jevons-false-friend-notes-are-a-confirmed-bridge
• Bipartite hop (2026-08-20), one level up, hedged verdict: observation-rhw-docsub-gates-jevons-meta-pairing-is-a-hedged-bridge
• Bipartite hop (2026-08-22), two levels up, citation-genealogy bridge: observation-gates-jevons-fourth-instance-hedged-bridge-pairing-is-citation-genealogy-bridge
• Bipartite hop (2026-08-25), buried common-ancestor bridge: observation-puckette-gates-jevons-fourth-instance-pairing-is-buried-common-ancestor-bridge · dropped-mention finding: observation-2026-08-16-capture-dropped-two-puckette-lineage-mentions-before-promotion
• Bipartite hop (2026-08-29), one level up, second confirmed bridge: observation-gates-jevons-liang-1979-olsder-1975-cosine-pairing-is-confirmed-bridge
• Bipartite hop (2026-08-31), sixth meta-level hop, first outside the Gates-Jevons/Puckette chain, false-friend verdict: observation-puckette-dropped-mention-petiska-matthew-effect-pairing-is-embedding-false-friend
• Bipartite hop (2026-09-03), seventh meta-level hop, false-friend verdict, and registration of a previously-omitted sixth object-level instance: observation-kelly-luccioni-puckette-petiska-meta-pairing-is-embedding-false-friend
• Bipartite hop (2026-09-04), eighth meta-level hop, first reaching outside the object-level false-friend chain, confirmed bridge in a new cross-kind species: observation-kelly-luccioni-garfield-seglen-steck-cosine-pairing-is-confirmed-bridge
• Bipartite hop (2026-09-08), ninth meta-level hop, first pairing against the "person-bridge invisible to vault_bridge" watch-log, false-friend verdict: observation-puckette-pitman-feldbaum-meta-pairing-is-embedding-false-friend
• Bipartite hop (2026-09-10), tenth meta-level hop, opposite-direction grounding, false-friend verdict: observation-rhw-docsub-gates-jevons-olsder-greenhalgh-meta-pairing-is-embedding-false-friend
• Bipartite hop (2026-09-13), eleventh meta-level hop, second false-friend pairing against the Puckette-Petiška meta-note, and a candidate "retrieval magnet" pattern about that note itself: observation-falcon-helmholtz-puckette-petiska-meta-pairing-is-embedding-false-friend · observation-puckette-petiska-is-a-cosine-similarity-retrieval-magnet
• Captures: 2026-08-11-bipartite-two-things-this-vault-knows-in-different
• Captures: 2026-08-16-bipartite-two-things-this-vault-knows-in-different
• Captures: 2026-08-20-bipartite-two-things-this-vault-knows-in-different
• Captures: 2026-08-22-what-genuinely-connects-gates-jevons-is-the-vaults
• Captures: 2026-08-25-what-genuinely-connects-the-puckette-dataflow-pairing-is
• Captures: 2026-08-29-what-genuinely-connects-the-vaults-cosine-087-proximity
• Captures: 2026-08-31-what-genuinely-connects-the-2026-08-16-capture
• Captures: 2026-09-03-what-genuinely-connects-the-kelly-et-al-2025
• Captures: 2026-09-07-what-genuinely-connects-the-puckette-fourth-instance-hedged
• Captures: 2026-09-09-what-genuinely-connects-the-cosine-086-pairing-between
• Captures: 2026-09-13-what-genuinely-connects-the-vaults-cosine-075-tie