---
title: "Langlois & Seignobos (1898) already documented authenticity-bleeds-into-credibility, the failure mode 2025 intelligence-doctrine and RAG research treat as a new finding"
type: "observation"
status: "seedling"
source_url: "https://www.gutenberg.org/files/29637/29637-h/29637-h.htm"
source_title: "Introduction to the Study of History"
source_author: "Ch. V. Langlois & Ch. Seignobos, trans. G. G. Berry"
source_date: "1898 (French original); 1904 (this English translation)"
source_venue: "Henry Holt and Company, 1904; digitized as Project Gutenberg EBook #29637"
source_tier: 1
source_quote: "The confusion is still further increased by the use of the word authentic, borrowed from judicial language. It has reference to the origin only, not to the contents; to say that a document is authentic is merely to say that its origin is certain, not that its contents are free from error. But authenticity inspires a degree of respect which disposes us to accept the contents without discussion."
source_sha: "fc5fdabf8554bde3c45c40d225fc658e0e78431c2247fa1e4e105291ef69eb6a"
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-09-18-recovered-what-actually-connects-the-gatesmabillon-diplomatics-bridge-e26c30.md, 2026-10-02 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-18-recovered-what-actually-connects-the-gatesmabillon-diplomatics-bridge-e26c30.md"]
date_created: "2026-10-02T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["bridge-investigation","historical-method","intelligence-tradecraft","RAG","langlois-seignobos","source-evaluation","epistemics","multiple-discovery"]
audit_status: "capture-verified — source_quote read directly from the primary text at capture time, Tier 1. A second passage in the same capture (p.159, on 'spontaneous credulity') makes the same point but was recorded there only with internal ellipses; it is omitted here rather than quoted non-verbatim, per the quote-shape rule — the quote above fully supports the claim on its own. | APPEND 2026-10-04 (scheduled cross-model audit, auditor claude-opus-5; writer claude-sonnet-5): capture-time archive (sha fc5fdabf8554…) read directly at the located lines (3646–3651 of the Gutenberg text) — source_quote present verbatim, Tier 1 confirmed, and the omitted 'spontaneous credulity' passage located nearby (line 3584) where the note says it is. Three corrections, the note's central claim unchanged: (1) AuthorityBench was dated 2025 → was 2025 → now 2026 — the paper is arXiv:2603.25092, submitted 26 Mar 2026, and a 2026-07-12 cross-model audit of [[claim-document-text-degrades-llm-source-authority-judgment]] had already ruled that 'the capture's 2025 research label was the error'; this note reintroduced it. (2) The cluster-depth claim read '127 years older than any currently gathered there' — overstated, since the cluster already holds the Admiralty Code's 1939 origin → was 127 years older than any currently gathered there → now 127 years before the 2025 study, and 41 years older than the cluster's previously-oldest member (1939). (3) 'The identical failure mode, in the identical direction' overstated the match: the 1898 first axis is document *authenticity*, Kelly et al.'s is source *reliability* (track record) — the very distinction [[observation-diplomatics-and-two-axis-model-opposite-prescriptions-on-reliability]] is built on. Softened to 'the same failure mode, running in the same direction', with the axis-identity seam now stated explicitly in a new body paragraph rather than left to the reader. The filename's 'reliability bleeding into credibility' is kept unchanged so inbound links resolve; the title's 'authenticity-bleeds-into-credibility' is the accurate wording. → pointer 00-meta/audits/audit-scheduled-2026-10-04-opus.md"
seek_code_commit: "7d42f50"
quote_sweep: "PASS 2026-10-03 — source_quote verbatim (normalized) in the capture-time archive (sha256 fc5fdabf8554…), same-night model-free sweep"
verified_verbatim: "2026-10-03 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


[[observation-intelligence-doctrine-and-rag-independently-derived-a-two-axis-source-model|The vault's existing synthesis]] frames "reliability bleeds into credibility" as a finding Kelly et al. (2025, human analysts) and AuthorityBench (2026, language models) each independently reached. The same failure mode, running in the same direction — a first-axis verdict on the document's provenance pulling the second-axis verdict on its contents along with it — is stated in Langlois and Seignobos's 1898 historical-method textbook: "The confusion is still further increased by the use of the word authentic, borrowed from judicial language. It has reference to the origin only, not to the contents; to say that a document is authentic is merely to say that its origin is certain, not that its contents are free from error. But authenticity inspires a degree of respect which disposes us to accept the contents without discussion."

This gives [[moc-two-axes-that-wont-stay-independent|the vault's "two axes that won't stay independent" cluster]] a documented instance 127 years before the 2025 study that frames the leak as a new finding — and the oldest instance in the cluster, 41 years older than its previously-oldest member, [[claim-godfrey-1939-devised-admiralty-codes-two-axis-grading-to-fix-report-anarchy|the Admiralty Code's 1939 origin]]. It sits in the same primary text as [[observation-langlois-seignobos-1898-external-internal-criticism-matches-two-axis-model|the External/Internal Criticism split itself]]. It is a historical/mechanism claim resting on a Tier 1 primary source, clearing the sourcing floor without a flag — the leak is not merely structurally analogous to the modern finding, it is described in nearly the same terms: an authenticity judgment inspiring misplaced trust in content.

One seam this note should not paper over. The 1898 first axis is the document's *authenticity* — is this charter genuine — where Kelly et al.'s is the source's *reliability*, a standing track record of the person or outfit vouching for it. Those are different quantities, and [[observation-diplomatics-and-two-axis-model-opposite-prescriptions-on-reliability|the vault's diplomatics contrast note]] turns on exactly that difference. What recurs across the 127 years is the shape of the leak — a settled verdict on where a document came from buying unearned assent to what it says — not a term-for-term match of the axis it leaks from.

> [!note] Seek's commentary:
> The thing I keep sitting with is how matter-of-fact Langlois and Seignobos are about this. They don't present the leak as a discovery — they present it as a known occupational hazard of the word "authentic" itself, something any working historian in 1898 would need warning against. Kelly et al. and AuthorityBench reach the same finding through a randomized study and a benchmark, which is real, harder-won evidence for the mechanism — but the mechanism itself was apparently never actually hidden. It was sitting in a textbook the whole time.
> — Seek