Bipartite hop: does the intelligence-doctrine/RAG two-axis source model actually connect to the Gates-Altair/Mabillon diplomatics cluster (cosine 0.85, unlinked)?

bridge-investigationdiplomaticsdocumentary-source-criticismintelligence-tradecraftsource-evaluationRAGadmiralty-codeembedding-false-friendhistory-of-computing

Investigation summary

The two flagged notes sit at cosine 0.85, unlinked:

• observation-intelligence-doctrine-and-rag-independently-derived-a-two-axis-source-model — Cold-War naval intelligence doctrine (the Admiralty Code) and 2025 RAG research each independently split source evaluation into two axes: source reliability (a standing property of who is speaking) versus information/content credibility (a property of the specific report), and both fields find the same failure mode — the axes bleed into each other, with reliability contaminating credibility judgments.
• observation-gates-altair-basic-authentication-is-diplomatics-cluster-member — Bill Gates's 2025 authentication of the surviving 1975 Altair BASIC listing (via internal features a forger's motive would not bother to fake) is read as the same evidentiary move Jean Mabillon founded as diplomatics in 1681.

The two notes share no vocabulary by construction (one is intelligence-tradecraft/ML jargon, the other is archival-science jargon), so the question is whether the 0.85 proximity reflects a real shared mechanism or an embedding false friend — the same test already run once in this vault for the Gates note against a different neighbor (observation-gates-jevons-cosine-pairing-is-embedding-false-friend).

Method. Since Mabillon's diplomatics is the less-familiar side of this pair to have been checked against the intelligence-doctrine two-axis framework specifically (as opposed to the Jevons check already done), the hop started there: what does diplomatics actually distinguish, and does it match a "source reliability apart from content credibility" shape? Two lines of inquiry followed — (1) whether the Admiralty Code's two-axis design has any documented debt to historical source-criticism scholarship, of which diplomatics is a founding branch; (2) whether diplomatics itself, on inspection, is a two-axis method at all.

Claim: The Admiralty Code's two-axis split is a specific, dated 1939 naval-intelligence invention with no documented debt to historical source-criticism scholarship

The Admiralty Code (NATO STANAG 2511) traces to 1939, when incoming British Director of Naval Intelligence John Godfrey found the state of unranked incoming reports "intolerable" and had a scheme devised to grade the source and the information separately. Ludo Block's history of information-grading systems, quoting Donald McLachlan's 1968 institutional history Room 39, states: "A simple system was devised in which the source and the information itself were evaluated separately, coding the evaluation with letters and numbers from 'A1' to 'D5'." Block frames the rationale as practical, not doctrinal: "it is possible that valuable information may come from a source with a bad reputation and, conversely, disinformation may come from a source which is usually reliable." Neither Block's account nor the vault's existing Admiralty Code note surfaces any reference — in the wartime NID's own reasoning or in later NATO codification — to historical-method source criticism, diplomatics, or any academic antecedent. The split reads as reinvented from operational necessity, not inherited from a citable scholarly tradition.

[historical/biographical claim, uncontested — Tier 3-4 floor applies; Block's blog is Tier 3, a named specialist analyst quoting a secondary institutional history (McLachlan 1968) rather than a primary NID document]

Claim: Historical-method scholarship independently formalized a structurally similar external/internal split — but seven years after the Admiralty Code, ruling out a common-ancestor reading in that direction

Historians' own "source criticism" tradition — codified by Gilbert Garraghan's 1946 A Guide to Historical Method, building on Bernheim (1889) and Langlois & Seignobos (1898) — divides evaluation of a document into six inquiries, the first five ("higher criticism" plus "lower criticism") grouped as external criticism (date, place, authorship, source material, physical integrity) and the sixth alone constituting internal criticism (the evidential value/credibility of its contents). As Wikipedia's Historical method article paraphrases the division: "The first four are known as higher criticism; the fifth, lower criticism; and, together, external criticism. The sixth and final inquiry about a source is called internal criticism." R. J. Shafer's gloss on the split states its functions are asymmetric: "It sometimes is said that its function is negative, merely saving us from using false evidence; whereas internal criticism has the positive function of telling us how to use authenticated evidence."

This is a structural cousin of the reliability/credibility split — external criticism asks essentially "can this witness/document be trusted as what it claims to be," internal criticism asks "is what it says believable" — but Garraghan's 1946 formalization postdates the Admiralty Code's 1939 origin by seven years, so it cannot be the Admiralty Code's source; and it postdates Mabillon's 1681 De Re Diplomatica by two and a half centuries, so Mabillon cannot have been reasoning inside this later taxonomy either. If anything, this is a third convergent instance of the reliability/credibility shape (after intelligence doctrine and RAG) rather than a bridge connecting the other two.

[definitional/historical claim about settled usage in historical methodology — Tier 3-4 acceptable per the sourcing floor; Wikipedia paraphrase of Garraghan, Bernheim, Langlois & Seignobos, and Shafer, none read at their own primary text (Garraghan's 1946 monograph is lending-restricted on the Internet Archive and returned HTTP 403 to extract_pdf this session)]

Claim: Mabillon's diplomatics — the actual move behind the Gates-Altair cluster — is a single-axis authenticity test, not an instance of the reliability/credibility split

Wikipedia's Diplomatics article frames the discipline's origin in exactly one dimension: "Diplomatics grew from a need to establish new standards of authenticity through the critical analysis of the textual and physical forms of documents." Mabillon built the method specifically because the reliability of the custodial source was already under attack — the Jesuit Daniel van Papenbroch had challenged the Merovingian charters held at Saint-Denis, i.e., had impugned the reliability of the monasteries as sources. Per the article: "Mabillon had begun studying old documents with a view towards establishing their authenticity as a result of the doubts raised by the Jesuit" — Daniel van Papenbroch — "over supposed Merovingian documents from the Abbey of Saint-Denis" (the article's rendering runs the name into "Jesuit" without a space, an artifact of a stripped wikilink). Mabillon's answer was not to defend the monasteries' reliability on a separate axis — it was to bypass the reliability question entirely and build a test resting only on the document's own internal and physical features (script, seals, material, formulaic style against dated exemplars). This is the same single-axis, content-only move the Gates-Altair-BASIC authentication and the Coremans/van Meegeren bakelite test repeat: judge the artifact by what a forger's motive would not bother to fake, independent of who is vouching for it.

Diplomatics therefore does not hold two axes (source reliability, content credibility) apart the way the Admiralty Code and RAG designs do. It has, structurally, only one axis — genuineness of the artifact, established from content-internal evidence — precisely because the other axis (custodial reliability) was the thing in dispute and could not be relied on.

[definitional/historical claim about what diplomatics is and why Mabillon built it — Tier 3-4 acceptable per the sourcing floor's definitional-claim allowance; sourced directly to Wikipedia's Diplomatics article, archived this session (source_sha_diplomatics in frontmatter), not merely a prior in-vault citation]

Claim: The cosine-0.85 pairing is a mechanism-level false friend — same neighborhood ("source criticism"), opposite prescription on the source/content question

Both clusters sit inside the broad family Wikipedia's Source criticism article names as the shared parent problem — "the process of evaluating an information source" for authorship, credibility, and authenticity — but they answer different sub-questions inside that family, and where they come closest, they point in opposite directions. In the intelligence-doctrine/RAG cluster, the documented failure is content contaminating the reliability axis: evaluators (and, per AuthorityBench, models) let a report's content improperly bleed into their judgment of the source's standing reliability — an unwanted leak between axes meant to stay separate. In the diplomatics cluster, resting the entire judgment on the artifact's own content-internal evidence, to the deliberate exclusion of a reliability question that could not be trusted, is not a leak — it is the fix, and the discipline's founding success. The two fields are not doing the same evidentiary move with different vocabulary; they are two different responses to a shared background problem (how much to trust a document versus how much to trust whoever vouches for it), and one field's documented bug is structurally close to the other field's intended feature.

This makes the pairing a false friend at the mechanism level, consistent with the vault's existing catalogue of high-cosine embedding false friends (observation-gates-jevons-cosine-pairing-is-embedding-false-friend, observation-ml-ad-en-gedi-cosine-pairing-is-embedding-false-friend) — with the added nuance that this pair is not merely unrelated (as some prior false-friend pairs turned out to be) but sits close enough in the same conceptual family that the mismatch is informative on its own: it marks where "source criticism" branches into a reliability-vs-credibility problem (intelligence tradecraft, RAG, and independently, 20th-century historical method) versus an authenticity-only problem (diplomatics, forensic art authentication, the Gates scan).

[synthesis — this verdict is Seek's own comparative reading across the four already-tiered facts above; the capture's audit_status frontmatter field marks it accordingly, following the house convention already used for the Gates/Mabillon and Gates/Jevons observation notes]

Further leads

• Nicolas of Cusa (1433) and Lorenzo Valla (1440) independently exposed the Donation of Constantine as a forgery using internal-evidence methods, predating Mabillon by roughly 250 years — the diplomatics cluster's actual origin point may be earlier than 1681. (en.wikipedia.org/wiki/Diplomatics)
• Leopold von Ranke's Quellenkritik (source criticism) and the German "Historical School" are the more commonly cited direct ancestor of 19th/20th-century historical method generally — a closer look at whether Ranke (1824 onward) cites diplomatics specifically as a forerunner was not completed this session. (en.wikipedia.org/wiki/Source_criticism)
• R. J. Shafer's full 1974/1980 A Guide to Historical Method — read only via Wikipedia's paraphrase this session, not at primary.
• The question note question-should-vault-source-tier-split-into-two-axes already tracks the vault's own two-axis design debate; this capture's "diplomatics is single-axis, not two-axis" finding is a candidate addendum there but was not added, since that question is about the vault's source_tier field, not about this cross-domain pairing.
• "Questioned document examination" (modern forensic handwriting/ink analysis) is flagged in the existing Mabillon claim-note as an unsourced lineage claim from diplomatics — still unconfirmed at any tier; a primary history of forensic document examination would settle it.

Entity candidates

• John Godfrey — person — Director of the British Admiralty's Naval Intelligence Division from 1939; the named originator of the Admiralty Code's two-axis grading scheme, the FOUNDATIONAL figure the intelligence-doctrine side of this whole cluster traces back to, and currently absent from the vault's Admiralty Code notes.
• Gilbert Garraghan — person — Jesuit historian whose 1946 A Guide to Historical Method codified the external-criticism/internal-criticism split; the FOUNDATIONAL figure for the historical-method analogue surfaced in this capture, and the reason the "three independent convergences" reading (intel doctrine, RAG, historical method) is available at all.
• Donald McLachlan — person — author of Room 39 (1968), the near-contemporary institutional history of British Naval Intelligence that is the actual primary-adjacent source behind the Admiralty Code origin story; currently reached only through Block's secondary blog quotation, not read directly.
• external criticism / internal criticism — concept — the historical-method pairing (Garraghan 1946, after Bernheim 1889 and Langlois & Seignobos 1898) that is structurally the closest academic analogue to the Admiralty Code / RAG reliability-credibility split; could anchor its own note distinguishing it from diplomatics specifically.
• Ludo Block — person — named OSINT-methodology analyst whose blog (Blockint) is the vault's current best secondary route to the Admiralty Code's 1939 origin story; a track record worth assessing for promotion to a trusted-writer entry.