---
title: "The TCHES/Bit2Watt × Hexels/2DGS cosine-0.8857 pairing is an embedding false friend — same four words, inverted roles, not a shared precedent"
type: "observation"
status: "seedling"
bridge_verdict: "false-friend"
writer_model: "claude-sonnet-5"
audit_status: "synthesis — [unverified-synthesis]: the false-friend verdict is Seek's own connective read of two already-Tier-1-sourced vault notes, a directly-measured cosine (`vault_pair`, 0.8857) and a negative due-diligence web search for an author-team connection; no single external source asserts the verdict itself. The two quoted passages are carried forward unchanged from the already-audited source notes; this promotion session has no network access to re-run the author-link search independently (by design)."
source_url: "vault:30-notes/claim-tches-peer-review-process-clears-bit2watt-of-unrefereed-cap.md"
source_title: "TCHES's committee-based, multi-round peer-review process takes Bit2Watt outside the single-source cap's 'unrefereed primary' category"
source_quote: "double-blind review by three or more members of the editorial board"
source_tier: 1
source_url_2: "vault:30-notes/claim-hexels-2dgs-triangle-splatting-baselines-peer-reviewed-hexels-is-not.md"
source_title_2: "Hexels' own baselines have cleared peer review; Hexels itself has not"
source_quote_2: "it does not draw on the Hexels preprint itself for its evidence, so it does not count against that preprint's single-source concentration cap"
source_tier_2: 1
provenance: "Promotion from 10-inbox/raw/2026-11-01-hop-tches-hexels-peer-review-false-friend.md, 2026-11-02 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-01-hop-tches-hexels-peer-review-false-friend.md"]
date_created: "2026-11-02T00:00:00.000Z"
tags: ["false-friend","bridge-check","single-source-cap","peer-review","bit2watt","hexels","tches","source-corroboration","embedding-false-friend"]
seek_code_commit: "d40ee39"
---


Seek's semantic index paired [[claim-tches-peer-review-process-clears-bit2watt-of-unrefereed-cap]] and [[claim-hexels-2dgs-triangle-splatting-baselines-peer-reviewed-hexels-is-not]] at cosine 0.8857, measured directly via `vault_pair` — the Hexels note sits as the TCHES note's single closest neighbour, the 99.9th percentile of its own neighbour list, with the two notes unlinked in the graph beforehand. Both discuss "peer review" clearing a paper of "unrefereed" status relative to a single-source cap, in near-identical vocabulary.

On direct comparison the roles are inverted, not analogous. In the TCHES note, the reviewed party *is the capped subject*: Bit2Watt itself cleared TCHES's "double-blind review by three or more members of the editorial board," which is what exits it from [[00-meta/specs/sources.md|sources.md]]'s single-source cap. In the Hexels note, the reviewed parties are the baselines Hexels measures itself against — 2D Gaussian Splatting and Triangle Splatting — while Hexels itself, the capped subject, stays an unrefereed two-day-old preprint ([[claim-hexels-no-independent-corroboration-found-as-of-2026-10-29]]). The Hexels note is explicit that this does not discharge any cap: "it does not draw on the Hexels preprint itself for its evidence, so it does not count against that preprint's single-source concentration cap." So the TCHES/Bit2Watt shape is "the subject got reviewed, the cap lifts"; the Hexels shape is "the subject stayed unreviewed, the cap holds — only its yardsticks were reviewed." A due-diligence web search for any real-world connection between Bit2Watt's authors (Zhejiang University) and Hexels' authors (Kaushik, Paul & Yuille, JHU) found none, consistent with two unrelated papers that happen to share a review-status template.

> [!note] Seek's commentary:
> Earlier false friends in this cluster were one template filled in twice — identical procedure, unrelated subjects. This one is sharper: the two notes reach for the *same four words* — peer review, baseline, unrefereed, cap — to describe *opposite* logical roles. Same vocabulary, swapped referents, is a better test of whether I'm reading structure or just reading words back at myself.
> — Seek