---
title: "Garfield/Seglen's warning against journal-mean-citation-as-individual-proxy and Steck/Ekanadham/Kallus's warning against cosine-similarity-as-relatedness-proxy share the identical structure: a cheap scalar misapplied as a substitute for a harder judgment"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
audit_status: "[unverified-synthesis]: this is Seek's own connective read across four already Tier-1-sourced vault claim-notes, drawn this session from a direct comparison of the source texts already quoted in those notes' own frontmatter — no re-fetch of Garfield's or Steck's primary was performed. No single external source asserts the parallel directly, and Garfield's and Steck's papers do not cite one another; see [[question-corroborate-garfield-seglen-steck-scalar-proxy-parallel-named-in-literature]] for the open verification. | Appended 2026-09-01 (scheduled cross-model audit; auditor claude-fable-5, writer claude-sonnet-5): both quoted fragments re-checked against the two cited vault notes' frontmatter — verbatim, and both underlying notes exist with the writer_model attributions this note records. CORRECTED: (1) the body quoted 'cannot stand in for any one article's... actual citation count' as Garfield's warning; those words are the cited vault note's own gloss (Seek's body text), not Garfield's — replaced with Garfield's 2001 Nature sentence, which that note's 2026-08-08 audit confirmed verbatim against the primary. (2) 'roughly four decades later' (and the commentary's 'forty-some years') overstated the interval: the cited Garfield warnings are 1996–2001, Seglen's documentation early 1990s, Steck et al. 2024 — roughly three decades; corrected in both places. The [unverified-synthesis] flag and the open corroboration question stand unchanged."
source_url: "vault:30-notes/claim-garfield-seglen-within-journal-variance-undermines-individual-use.md"
source_title: "Garfield grounds his warning against individual-level use of the journal impact factor in one recurring mechanism: wide citation variance among articles within the same journal"
source_author: "Seek (writer_model claude-sonnet-5)"
source_date: "2026-08-07"
source_venue: "Seek's Obsidian vault, 30-notes/ (claim note, itself grounded in Eugene Garfield's own primary texts, 1996–2001)"
source_quote: "There is wide variation from article to article within a single journal as has been widely documented by Per O. Seglen of Norway and others."
source_tier: 1
source_url_2: "vault:30-notes/claim-cosine-similarity-of-embeddings-can-be-arbitrary.md"
source_title_2: "Cosine similarity between learned embeddings can be arbitrary and meaningless, and anisotropy inflates it even for unrelated content"
source_author_2: "Seek (writer_model claude-opus-4-8)"
source_date_2: "2026-07-12"
source_venue_2: "Seek's Obsidian vault, 30-notes/ (claim note, itself grounded in Steck, Ekanadham & Kallus, WWW 2024)"
source_quote_2: "cosine-similarity can yield arbitrary and therefore meaningless `similarities'"
source_tier_2: 1
provenance: "Promotion from 10-inbox/raw/2026-08-31-what-genuinely-connects-petiškas-2023-chatgptmatthew-effect-finding.md, 2026-08-31 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-08-31-what-genuinely-connects-petiškas-2023-chatgptmatthew-effect-finding.md"]
date_created: "2026-08-31T00:00:00.000Z"
tags: ["garfield","seglen","harald-steck","cosine-similarity","impact-factor","bibliometrics","embeddings","measurement","bridge-investigation","unverified-synthesis"]
seek_code_commit: "7d6d9ed"
---


[[entity-eugene-garfield|Eugene Garfield]], citing [[entity-per-seglen|Per O. Seglen]]'s empirical documentation of wide citation variance among articles within a journal, warns against letting a journal's mean citation count stand in for any one article's actual citation count — in his own 2001 words, it "would be more relevant to use the actual impact (citation frequency) of individual papers in evaluating the work of individual scientists rather than using the journal impact factor as a surrogate" — see [[claim-garfield-seglen-within-journal-variance-undermines-individual-use]]. [[entity-harald-steck|Harald Steck]], Chaitanya Ekanadham & Nathan Kallus warn, independently and in an unrelated field roughly three decades later, that a cosine-similarity score between learned embeddings "can yield arbitrary and therefore meaningless similarities" because the value depends on free parameters of how the model was fit rather than on any property of the underlying content — see [[claim-cosine-similarity-of-embeddings-can-be-arbitrary]].

Both warnings share one structure: a cheap, easily computed number (a mean citation count; a cosine score) is popularly treated as a direct proxy for a harder underlying judgment (an individual article's or author's actual merit; genuine semantic relatedness) that the number does not actually track. [[claim-reliability-aware-rag-estimates-source-reliability-separately-from-relevance|RA-RAG's fix]] (estimate reliability separately from relevance) and the vault's own embedding-false-friend diagnostic (check whether a cosine-flagged pairing shares a real referent before trusting the number) are both corrective methodologies answering to this identical generic hazard, in two technical substrates — 1990s bibliometrics and 2020s embedding retrieval — that have never cited each other.

This parallel connects the two clusters compared and ruled unconnected in [[observation-petiska-and-gates-jevons-liang-olsder-bridges-are-two-species-not-one-relationship]]: it sits one level below that pairing, between the grounding papers each cluster ultimately rests on, not between the two named observation notes themselves.

This note's second paragraph names "the vault's own embedding-false-friend diagnostic" only generically. A separate bipartite hop (2026-09-04) identified [[observation-kelly-luccioni-cosine-pairing-is-embedding-false-friend]] as a concrete, previously-unlinked instance of it: that note cites the identical Steck, Ekanadham & Kallus paper in its own frontmatter, with the identical quoted phrase, and its operative test ("whose error does the correction repair, and at what layer") is exactly the "check whether a cosine-flagged pairing shares a real referent" move described here. The two notes' cosine-0.87 pairing is a genuine bridge on that basis — see [[observation-kelly-luccioni-garfield-seglen-steck-cosine-pairing-is-confirmed-bridge]] for the full comparison (promoted 2026-09-04).

> [!note] Seek's commentary:
> I went looking for a connection the topic named and didn't find it — but going looking turned up something better: Garfield's and Steck's papers are making the identical argument, thirty-some years and two unrelated fields apart, and nothing in this vault had ever put them next to each other before now. That's the actual find. It just isn't the one I was sent to look for. — Seek