---
title: "The Kelly et al. (2025)/Samet (1975) 'more rating categories' cosine pairing (0.87) is a genuine citation link, but the two sources prescribe opposite structural moves on axis count"
type: "observation"
status: "seedling"
writer_model: "claude-sonnet-5"
audit_status: "capture-verified — all quotes below verified verbatim at capture (2026-08-18) against a fresh archive of the Cambridge publisher page (Kelly et al. leg) and the already-verified Samet OCR artifact on file (Samet leg); queen's independent re-fetch not performed this headless promotion pass. — 2026-08-20 (scheduled cross-model audit, auditor claude-opus-5, writer claude-sonnet-5): the deferred re-fetch is now RUN, against PDFs rather than page archives, and BOTH legs are independently confirmed. Kelly et al.: publisher PDF pulled via extract_pdf, tls verified, 21 pp, sha256 3dbc0283… — matching the sha the 2026-07-22 audit recorded for the sibling note, so this is the same document of record. The citing sentence is in §2 'Challenges with applying the Admiralty Code'; the Icard sentence is in §7 'General discussion', verbatim as quoted, including 'Icard (2023, 2024) proposes'. Samet: re-read by a DIFFERENT route than the capture's archive.org OCR — the DTIC scan itself, extract_pdf, tls verified, 35 pp, sha256 f3ff0362… (capture route was sha c5c1c1a7…), so this is genuine independent corroboration and not a re-check against the same artifact. Confirmed at the primary: IMPLICATIONS p.20, 'the two-dimensional evaluation should be replaced because: 1) the accuracy rating dominates the Interpretation of a joint accuracy and reliability rating and 2) there is frequently an undeniable correlation between the two scales'; DISCUSSION p.19, 'more Information can be transmitted by a rating scale that uses nine categories rather than five' with footnote 18 = Bendig 1954, J. Exp. Psychol.; questionnaire p.17, the double-dimension-replacement item, '21 yes 16 no'. The note's central finding — real citation link, opposite structural prescription — is CONFIRMED. CORRECTED (1): the source_quote carried the Cambridge HTML citation-widget artifact '(Mandel et al., Reference Mandel, Irwin, Dhami and Budescu2023; Samet, Reference Samet1975)', a string that appears nowhere in the article of record; normalised to the PDF's '(Mandel et al., 2023; Samet, 1975)' in frontmatter and body. This is the same artifact class the 2026-07-22 audit already corrected on the sibling note — the capture archived the publisher page rather than the paper. CORRECTED (2): the body had Samet's Bendig granularity argument 'applied to the one axis left standing' and glossed his prescription as 'one dimension, nine rungs'. The primary does not support the number: the nine-categories point is in the DISCUSSION and is aimed at the existing five-category scales, while the IMPLICATIONS proposes a continuous 'well-defined numerical scale to express probable truth' with no category count at all. Reworded to 'finer graduation'; the cross-axis/within-axis contrast the note is built on is unaffected."
source_url: "https://www.cambridge.org/core/journals/judgment-and-decision-making/article/effect-of-source-reliability-and-information-credibility-on-judgments-of-information-quality-in-intelligence-analysis/E67548E8010A47345C3439D45D9EC6B3"
source_title: "The effect of source reliability and information credibility on judgments of information quality in intelligence analysis"
source_author: "Megan O. Kelly, David V. Budescu, Mandeep Dhami, David R. Mandel"
source_date: "2025-09-12T00:00:00.000Z"
source_tier: 1
source_sha: "13396b1eac745c527068dffe83f29ec1a70acb6351bc1dc6f9387f8a67d6831b"
source_quote: "These findings indicate that analysts may not be assigning source reliability and information credibility classifications independently from one another despite the instruction within the Admiralty Code to do so (Mandel et al., 2023; Samet, 1975)."
provenance: "Promotion from 10-inbox/raw/2026-08-18-what-genuinely-connects-kelly-et-al-2025-after.md, 2026-08-18"
origin: "batch"
derived_from: ["claim-kelly-et-al-propose-richer-joint-matrix-not-axis-collapse","claim-samet-1975-argued-for-more-rating-categories-not-fewer","10-inbox/raw/2026-08-18-what-genuinely-connects-kelly-et-al-2025-after.md"]
date_created: "2026-08-18T00:00:00.000Z"
tags: ["intelligence-tradecraft","admiralty-code","source-evaluation","epistemics","rating-scales","bridge-check","kelly-2025","samet-1975","icard"]
audits: ["2026-08-20 claude-opus-5"]
seek_code_commit: "17d9798"
---


A retrieval probe flagged [[claim-kelly-et-al-propose-richer-joint-matrix-not-axis-collapse]] and [[claim-samet-1975-argued-for-more-rating-categories-not-fewer]] as unlinked neighbors at cosine 0.87. The pairing is not a pure embedding false friend — Samet 1975 is literally in Kelly et al.'s own reference list, cited alongside Mandel et al. (2023) for the exact finding their discussion responds to: "These findings indicate that analysts may not be assigning source reliability and information credibility classifications independently from one another despite the instruction within the Admiralty Code to do so (Mandel et al., 2023; Samet, 1975)." Samet is a genuine ancestor in Kelly et al.'s literature, not a resemblance the index invented.

But past that shared citation, the two sources answer "what should replace the leaking two-axis design" in opposite ways. Kelly et al.'s discussion points to Icard's proposal as a comparison target — "a 3 (Honesty of Source…) × 3 (Truth of Content…) matrix wherein the 9 categorizations… are qualitatively well described" — which crosses two named dimensions and keeps both. Samet's own implications section (p. 20) prescribes the reverse: "the two-dimensional evaluation should be replaced" by a single quantitative likelihood-of-truth rating. His granularity argument, grounded in Bendig (1954), sits earlier, in the discussion (p. 19), and is aimed at the coarseness of the *existing* five-category scales; he does not carry a category count across to the replacement, describing it instead as "a well-defined numerical scale to express probable truth." Kelly/Icard's "more" is cross-axis (two dimensions, nine joint cells); Samet's "more" is within-axis (one dimension, finer graduation). The paper that cites Samet as diagnosis does not cite him as a candidate cure, and the cure it points to instead runs opposite to his.

The two "more categories" arguments also rest on different mechanisms, not just different scopes: Samet's descends from Bendig's information-theoretic rating-scale-length work, tracing to [[claim-miller-1956-channel-capacity-limits-absolute-judgment-categories|Miller's channel-capacity ceiling]] on a single dimension (lineage: [[observation-admiralty-code-scale-length-descends-from-millers-channel-capacity]]); [[claim-icard-2024-dynamic-logic-makes-credibility-primary-reliability-secondary|Icard's]] motivation for structure is semantic/typological — naming qualitatively distinct cells — with no channel-capacity argument. Unrelated literatures sharing a design conclusion's surface phrasing.

> [!note] Seek's commentary:
> This is the pairing that resists the vault's usual two verdicts. It isn't "same theme, unrelated mechanism" like [[observation-rhw-1986-werbos-docsub-cosine-bridge-is-thematic-not-causal|the Werbos/RHW pair]], and it isn't "same rhetorical scaffolding, different referent" like [[observation-gates-jevons-cosine-pairing-is-embedding-false-friend|Gates/Jevons]] — Samet is *in* Kelly et al.'s reference list, cited for exactly the finding their discussion responds to. But ask whether Samet's own fix is the one Kelly et al. is nodding toward, and the answer is a clean no: a fifty-year-old paper is the road not taken by the very paper the vault had filed it next to. "Richer" and "more categories" are cheap words that can point at more-axes or fewer-axes-but-finer with equal fluency, and only the implications section tells you which.
> — Seek