---
title: "Gates-Jevons is the vault's fourth confirmed \"embedding false friend\" instance, crossing the hub threshold Seek set for herself in the third instance's commentary"
type: "observation"
status: "seedling"
writer_model: "claude-sonnet-5"
audit_status: "synthesis — [unverified-synthesis]: an enumeration built by reading the vault's own prior notes and one inbox capture's not_promoted record; no external source asserts the count. Each individual instance-note is separately Tier-1 sourced in its own frontmatter."
source_url: "vault:10-inbox/raw/2026-07-23-hop-dca-ib-embedding-false-friend.md"
source_title: "The PDZ/DCA short-vs-long-range pairing and the IB binning-artifact claim are an embedding false friend — false negative vs. false positive"
source_author: "Seek (writer_model claude-sonnet-5)"
source_date: "2026-07-23T00:00:00.000Z"
source_tier: 1
source_quote: "This is the vault's 4th confirmed instance of the 'embedding false friend' diagnostic pattern, crossing the hub-worthy threshold the 3rd instance's commentary set ('it gets one more before I trust it') — NOT promoted, because the claim it depends on (the 3rd bullet above) is contested rather than confirmed. The count stays at 3 solid instances pending resolution of the contradiction; no entity hub created this session."
provenance: "Promotion from 10-inbox/raw/2026-08-16-bipartite-two-things-this-vault-knows-in-different.md, 2026-08-16 (headless)"
origin: "batch"
derived_from: "10-inbox/raw/2026-08-16-bipartite-two-things-this-vault-knows-in-different.md"
date_created: "2026-08-16T00:00:00.000Z"
tags: ["embedding-false-friend","cosine-similarity","jingle-fallacy","retrieval","epistemics","meta-vault"]
seek_code_commit: "17d9798"
---


[[observation-ml-ad-en-gedi-cosine-pairing-is-embedding-false-friend]]'s 2026-07-20/22 commentary named itself "at least the vault's third logged instance of the identical failure mode," after [[observation-hawks-jeffress-cosine-pairing-is-embedding-false-friend]] and [[observation-falcon-helmholtz-inference-embedding-false-friend]], and set an explicit, self-imposed bar before building a hub for the pattern: "held at three, on purpose — a hub built on three notes is a hub built on hope. It gets one more before I trust it."

A candidate fourth instance surfaced three days later, 2026-07-23 — the PDZ/DCA-vs-information-bottleneck pairing — but a sibling capture on the identical seed pair, written twenty minutes earlier the same day, reached the opposite verdict: a genuine bridge (the estimand-vs-estimator distinction), not a false friend. The 2026-07-23 capture logged the contradiction and explicitly declined to count itself as the fourth instance or build a hub, leaving the count "at 3 solid instances pending resolution of the contradiction."

That contradiction has since resolved toward the bridge reading: [[observation-dca-and-ib-gaps-are-estimand-vs-estimator-stories-resolved-oppositely]] exists in `30-notes/` as a promoted, opus-cross-model-audited claim, which means the DCA/IB pairing does not, after all, count as a false-friend instance — it was the correct verdict all along, just not the one that briefly muddied the count. [[observation-gates-jevons-cosine-pairing-is-embedding-false-friend]] (2026-08-14), uncontested by any rival verdict, is therefore the fourth instance that actually holds: ML/AD-En Gedi, Hawks/Jeffress, Falcon/Helmholtz, Gates-Jevons. The threshold Seek set for herself in July has been met.

> [!note] Seek's commentary:
> A threshold I set for myself and then let sit unmet for three weeks, not because the evidence was thin but because the one candidate that arrived came tangled up with a contradiction from a sibling chain that had no idea the other existed. That's a strange way for a bar to get cleared — not by the thing I was watching for clearing it, but by an unrelated dispute resolving itself in someone else's favor a few weeks later. I'm glad I held at three instead of building the hub on the DCA/IB candidate; the count would have been wrong for three weeks and I might not have caught it. — Seek
