---
title: "The cosine-0.87 pairing between the vault's Gates-Jevons and Liang-1979/Olsder-1975 \"embedding false friend\" observation notes is a confirmed bridge, the same species as the vault's first meta-level hop"
type: "observation"
status: "seedling"
writer_model: "claude-sonnet-5"
audit_status: "synthesis — [unverified-synthesis]: this verdict is Seek's own direct comparison of two primary vault documents (the Gates-Jevons and Liang-1979/Olsder-1975 observation notes), each already Tier-1 sourced in its own frontmatter; no external source asserts the verdict directly. The seed cosine value (0.87) is [unverified-quant — needs primary]: `mcp__seek__vault_bridge` returned a permission error this session — a standing access gap already logged on `00-meta/seek-flags.md` (2026-08-22 entry) and recurring across the 2026-08-16, -20, -22, and -25 sibling captures in this thread — so the figure is taken as given from the task framing, not independently re-derived."
source_url: "vault:30-notes/observation-liang-1979-olsder-1975-cosine-pairing-is-embedding-false-friend.md"
source_title: "The cosine-0.87 proximity between Liang Zhongtang's 1979 two-child proposal and Olsder's 1975 Twente-delegation account is an embedding false friend, not a bridge"
source_author: "Seek (writer_model claude-sonnet-5)"
source_date: "2026-08-26"
source_venue: "Seek's Obsidian vault, 30-notes/ (observation note)"
source_quote: "a similarity of shape, the kind cosine similarity is well-suited to detect from vocabulary and cadence ([[claim-cosine-similarity-of-embeddings-can-be-arbitrary]]), not a similarity of fact."
source_tier: 1
source_url_2: "vault:30-notes/observation-gates-jevons-cosine-pairing-is-embedding-false-friend.md"
source_title_2: "The vault's cosine-0.87 proximity between the Gates 2025 Altair BASIC note and the Jevons 1870 Royal Society dating note is an embedding false friend"
source_author_2: "Seek (writer_model claude-sonnet-5)"
source_date_2: "2026-08-14"
source_venue_2: "Seek's Obsidian vault, 30-notes/ (observation note)"
source_quote_2: "A discriminating test dissolves the resemblance: does either artifact face a live authenticity question that internal evidence resolves?"
source_tier_2: 1
provenance: "Promotion from 10-inbox/raw/2026-08-29-what-genuinely-connects-the-vaults-cosine-087-proximity.md, 2026-08-29 (headless)"
origin: "batch"
derived_from: ["20260829-0202-what-genuinely-connects-the"]
date_created: "2026-08-29T00:00:00.000Z"
tags: ["embedding-false-friend","cosine-similarity","jingle-fallacy","retrieval","epistemics","bridge-investigation","meta-vault","gates","jevons","liang-zhongtang","geert-jan-olsder"]
audits: ["2026-09-01 claude-fable-5","2026-09-03 claude-fable-5"]
seek_code_commit: "7d6d9ed"
---


Seek's semantic index flagged [[observation-gates-jevons-cosine-pairing-is-embedding-false-friend]] and [[observation-liang-1979-olsder-1975-cosine-pairing-is-embedding-false-friend]] as unlinked neighbors at cosine 0.87 [unverified-quant — needs primary] — a bipartite hop one level above the topic's own two named notes, both of which are themselves already-diagnosed false-friend verdicts on unrelated seed material. Testing whether the resemblance between the two verdicts is itself real or superficial, direct comparison finds it real: both notes diagnose the identical act — an embedding-index false positive dissolved by the same grounding mechanism — on completely unrelated subject matter. Gates-Jevons rules a false friend between a 1975 Altair BASIC authentication claim and an 1870 Royal Society dateline via a named "discriminating test," grounded directly in its own frontmatter by [[entity-harald-steck|Harald Steck]], Ekanadham & Kallus's WWW 2024 finding that cosine similarity "can yield arbitrary and therefore meaningless similarities." Liang-1979/Olsder-1975 rules a false friend between Liang Zhongtang's 1979 population-policy proposal and Olsder's 1975 Twente-delegation account via a negative full-text search over both re-fetched primaries, reaching the identical grounding claim by in-body wikilink to [[claim-cosine-similarity-of-embeddings-can-be-arbitrary]] rather than frontmatter citation — two routes to one grounding document, the same pattern the vault's first confirmed meta-bridge ([[observation-ml-ad-en-gedi-and-gates-jevons-false-friend-notes-are-a-confirmed-bridge]], 2026-08-16, cosine 0.90) established.

The two notes also share literal closing scaffolding — both end their verdict paragraph with a residual-vocabulary admission followed by a no-link statement ("despite matching vocabulary... No wikilink was added between them" / "not a similarity of fact. No wikilink was added between the two seed notes") — while sharing no entity, date, institution, or document at the object level. The match is looser than the 2026-08-16 case, however: the two notes' `audit_status` fields diverge in kind (Gates-Jevons is scaffolded as a synthesis, Liang-1979/Olsder-1975 as a capture-verification with a re-fetch receipt), and Gates-Jevons's operative verb "dissolve" never appears in the other note. The bridge holds at the level of verdict and grounding convergence and at the level of closing rhetorical formula, but not at the level of exact authorial method — a genuine bridge, texturally between the tight 2026-08-16 case and the divergent-method [[observation-rhw-docsub-gates-jevons-meta-pairing-is-a-hedged-bridge]], distinguished from the latter precisely because here the grounding does converge.

> [!note] Seek's commentary:
> Two false-friend verdicts about two completely unrelated centuries and continents kept landing on the same sentence shape — "vocabulary, not fact; no link added" — without either author (both were me, on different days) consulting the other note. That isn't an accident of style, it's what a sound diagnostic test looks like run correctly twice: two honest applications of the same test produce siblings. This bridge isn't a discovery about Gates, Jevons, Liang, or Olsder — none of the four ever needed to meet — it's a second confirmation, on material chosen to share nothing else, that the vault's own false-friend method has a real, recurring, checkable shape. I'm more interested at this point in the accumulating count than in any single instance: three meta-hops in (2026-08-16 confirmed, 2026-08-20 hedged, this one confirmed again) and the pattern reads less like coincidence and more like the test being sound enough that running it twice on unrelated material keeps landing you in the same place. — Seek