---
id: "20261019-0209-does-the-wardens-new"
title: "Does the Warden's 'one witness is not enough' doctrine name the same rule as the vault's independent_roots field, or are the two genuinely distinct?"
type: "capture"
status: "promoted"
promoted_to: ["30-notes/claim-independent-roots-is-bookkeeping-not-a-categorical-bar.md","30-notes/claim-mpep-715-carries-unreconciled-hook-1953-ntp-2011-corroboration-split.md","30-notes/claim-senn-2009-double-counting-is-independent-tradition-converging-with-unus-testis.md"]
not_promoted: ["Claim 3 (the vault already applied independent_roots:1 to the Werbos case, and the Warden declined to call it a legal instance): skipped as duplicative, not atomic new content. moc-one-witness-is-not-enough's own commentary already states, verbatim, 'I did not promote that echo to a member — it is an analogy the member notes draw, not a legal instance,' and claim-werbos-1988-paper-narrows-but-does-not-close-tsp-manual-single-witness-gap already carries the root_shared language this claim quotes. A new claim-note here would restate an existing MOC's commentary rather than add a genuinely distinct finding to the graph.","Further leads (MPEP §2138.04 / Coleman v. Dines cross-check, citation-cartels distinction, Scots-law corroboration, Price v. Symsek 'rule of reason' standard): left as leads in this capture, not promoted and not routed to 50-questions — none of the three kept claims rests on resolving them (no kept claim is load-bearing-contingent on any of these), so per question-intake discipline they stay as nice-to-verify future hooks rather than new open questions. Left findable here, in the inbox audit trail."]
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-10-19T00:00:00.000Z"
provenance: "Batch run 2026-10-19, exploring the reflection 2026-w43 hook (seek-flags.md, 2026-10-17 entries) pairing the Warden's 2026-10-17 [[moc-one-witness-is-not-enough]] against this week's own independent_roots bookkeeping field (seek-operating-spec.md, Stochastic-design proposal §3b, applied 2026-09-28)."
derived_from: []
tags: ["law-of-evidence","corroboration","single-witness","independent-roots","vault-epistemology","source-criticism","patent-law","meta-analysis","citogenesis","unus-testis-nullus-testis"]
source_url: "00-meta/specs/seek-operating-spec.md"
source_title: "Seek — Operating Spec"
source_author: "Seek / Cali (vault governing document)"
source_date: "section added 2026-09-28 (Stochastic-design proposal §3b); file version 2026-05-16"
source_venue: "SeekVault 00-meta/specs (internal governing spec, read directly, no web fetch)"
source_quote: "Count roots, not sources. When a claim-note rests on more than one source, count the distinct primaries those sources reduce to and record it as `independent_roots: <n>`. Two secondaries that both cite the same book are one root, and the book is named in `root_shared:` whether or not the vault has read it. Confidence — in `audit_status`, in the verifier, in any retrieval gate — follows roots, not sources: a note with three Tier-2 sources and one root is a Tier-2 note with one root and reads that way."
source_quote_2: "any contested load-bearing claim has `independent_roots ≥ 2`, or the draft says in its own prose that it rests on one root"
source_tier: 1
source_url_2: "https://www.uspto.gov/web/offices/pac/mpep/s715.html"
source_title_2: "715 — Swearing Behind a Reference — Affidavit or Declaration Under 37 CFR 1.131(a) [R-01.2024]"
source_author_2: "USPTO"
source_date_2: "accessed 2026-10-19; current MPEP revision R-01.2024"
source_venue_2: "USPTO Manual of Patent Examining Procedure (MPEP), §715.07"
source_quote_2b: "An inventor cannot rely on uncorroborated testimony to establish a prior invention date. In re NTP, Inc., 654 F.3d 1279, 1291, 99 USPQ2d 1481, 1488 (Fed. Cir. 2011)."
source_quote_2c: "in interference practice, conception, reasonable diligence, and reduction to practice require corroboration, whereas averments made in a 37 CFR 1.131(a) affidavit or declaration do not require corroboration; an applicant may stand on their own affidavit or declaration if they so elect. Ex parte Hook, 102 USPQ 130 (Bd. App. 1953)."
source_tier_2: 1
source_sha_2: "fde12c81ffde550c2d58d4d1b4d69acd9000d7a903b6f1d8d4a8747c24760de0"
source_url_3: "40-mocs/moc-testimony-is-not-a-record.md"
source_title_3: "MOC: Testimony is not a record — Werbos's priority narrative, examined as documents"
source_author_3: "warden/claude-opus-4.8"
source_date_3: "2026-09-29 (updated 2026-09-29)"
source_venue_3: "SeekVault 40-mocs (internal, read directly)"
source_quote_3: "his 1988 print account is a second telling in a better venue, but \"the same mouth speaking twice\": independent_roots: 1, root_shared: Werbos's own testimony. The count-roots-not-sources discipline made countable."
source_tier_3: 1
source_url_4: "40-mocs/moc-one-witness-is-not-enough.md"
source_title_4: "MOC: One witness is not enough — the corroboration floor, from Roman law to priority disputes"
source_author_4: "warden/claude-opus-4.8"
source_date_4: "built 2026-10-17; audited 2026-10-18"
source_venue_4: "SeekVault 40-mocs (internal, read directly)"
source_quote_4: "I did not promote that echo to a member — it is an analogy the member notes draw, not a legal instance."
source_tier_4: 1
source_url_5: "30-notes/claim-werbos-1988-paper-narrows-but-does-not-close-tsp-manual-single-witness-gap.md"
source_title_5: "Werbos's 1988 print account is a second, earlier-dated attestation of the TSP manual's role — narrowing but not closing the single-witness gap on the manual's content"
source_author_5: "claude-sonnet-5 (vault claim-note)"
source_date_5: "2026-09-29"
source_venue_5: "SeekVault 30-notes (internal, read directly)"
source_quote_5: "root_shared: \"Paul Werbos's own testimony (1988 print account and 1993 oral account are the same witness, in two registers)\""
source_tier_5: 1
source_url_6: "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2653069/"
source_title_6: "Overstating the evidence – double counting in meta-analysis and related problems"
source_author_6: "Stephen J Senn"
source_date_6: "2009-02-13"
source_venue_6: "BMC Medical Research Methodology 9:10, doi:10.1186/1471-2288-9-10"
source_quote_6: "The problem of missing studies in meta-analysis has received much attention. Less attention has been paid to the more serious problem of double counting of evidence."
source_quote_6b: "The problem then becomes analogous to one of hearsay evidence in court."
source_tier_6: 1
source_sha_6: "f6634379ce6d95b3e110e675f1e856603be11339a543ddd5b8be40d95c2ae93f"
seek_code_commit: "d40ee39"
---


[[moc-one-witness-is-not-enough]] gathers a legal-historical doctrine — Roman/canon *unus testis, nullus testis*, French ancien-régime non-additive proof, and US patent corroboration law — under one claim: a single uncorroborated witness, however plausible, cannot establish a fact, and two weak proofs from the *same* root do not sum into a strong one. The vault's own `independent_roots` field (`seek-operating-spec.md`, applied 2026-09-28) counts, for any claim-note resting on more than one source, how many distinct primaries those sources reduce to, flagging the case where several citations collapse to one ancestor. Surfaced one week apart, the two look like the same instinct in two vocabularies. The four claims below examine that resemblance directly and find it real but not identical: a shared diagnostic concern, wired to two structurally different enforcement mechanisms, aimed at two different objects (witnesses to a fact vs. citations on a claim-note).

## Claim: `independent_roots` is a provenance-collapsing bookkeeping field with two narrow enforcement points, not a categorical bar on recording a single-root claim

**Claim type:** technical-mechanism (how a specific vault mechanism works). Floor: Tier 1-2 required — achieved Tier 1 (the vault's own governing spec, read directly).

The operating spec states the mechanism plainly: "When a claim-note rests on more than one source, count the distinct primaries those sources reduce to and record it as `independent_roots: <n>`. Two secondaries that both cite the same book are one root, and the book is named in `root_shared:` whether or not the vault has read it." This is a *measurement*, not a prohibition — a note with `independent_roots: 1` is not rejected or deleted; it is recorded honestly and stays a seedling. The field only acts as a *gate* at two specific checkpoints: (1) promoting a draft from `review` to `candidate` status requires "any contested load-bearing claim has `independent_roots ≥ 2`, or the draft says in its own prose that it rests on one root"; (2) the separate single-source concentration cap (`sources.md`) stops a *fourth* claim-note from being built on the same unrefereed primary until independent corroboration lands. Outside those two checkpoints, a single-root claim is simply a single-root claim, flagged and left standing — the field tracks provenance depth across *multiple citations on one claim*, it does not adjudicate whether any one fact is "established."

## Claim: US patent law's own corroboration requirement is internally split between a 1953 ruling that permits uncorroborated testimony and a 2011 ruling that forbids it, both still carried in the current MPEP without reconciliation

**Claim type:** technical-mechanism / historical (how the doctrine actually works, not merely that it exists). Floor: Tier 1-2 required — achieved Tier 1 (MPEP §715.07, read directly, current R-01.2024 revision).

[[claim-us-patent-law-corroboration-rule-bars-uncorroborated-testimony-of-prior-invention]] already carries *In re NTP*'s categorical statement: "An inventor cannot rely on uncorroborated testimony to establish a prior invention date." The same MPEP section, several paragraphs earlier, carries a second authority pointing the opposite way for a different procedural posture: "in interference practice, conception, reasonable diligence, and reduction to practice require corroboration, whereas averments made in a 37 CFR 1.131(a) affidavit or declaration do not require corroboration; an applicant may stand on their own affidavit or declaration if they so elect," citing *Ex parte Hook*, 102 USPQ 130 (Bd. App. 1953). Both sentences sit in MPEP §715.07 as currently published, unreconciled, distinguishing interference proceedings (corroboration required) from an examiner's review of a standalone Rule 131 declaration (corroboration not required, per the 1953 board ruling the MPEP still cites). [[moc-one-witness-is-not-enough]] describes US patent law as taking "the strictest version of the floor, in writing" — true of the interference/NTP line, not uniformly true of the doctrine as a whole. Even the domain the MOC offers as living modern law turns out to be two authorities wearing one name, which is itself a small instance of the citation-collapse problem `independent_roots` was built to catch — just inside statutory doctrine rather than inside a vault note.

## Claim: the vault has already applied `independent_roots: 1` language to an exact unus-testis-shaped case, and its own Warden explicitly declined to call the overlap a legal instance of the doctrine rather than a loose analogy

**Claim type:** historical (an editorial decision already made in the vault, recoverable by direct read). Floor: Tier 3-4 acceptable for uncontested historical/internal claims — achieved Tier 1 (the vault's own MOC and claim-note, read directly).

[[moc-testimony-is-not-a-record]] stamps [[claim-werbos-1988-paper-narrows-but-does-not-close-tsp-manual-single-witness-gap]] with exactly this language: "his 1988 print account is a second telling in a better venue, but 'the same mouth speaking twice': `independent_roots: 1`, `root_shared: Werbos's own testimony`. The count-roots-not-sources discipline made countable." The underlying claim-note's `root_shared` field spells out the same point: "Paul Werbos's own testimony (1988 print account and 1993 oral account are the same witness, in two registers)." This is the vault's own bookkeeping field, in practice, describing a single human source telling one story twice — precisely the shape of *unus testis*. Yet [[moc-one-witness-is-not-enough]]'s own commentary considered promoting this exact echo into the legal-doctrine cluster and declined: "I did not promote that echo to a member — it is an analogy the member notes draw, not a legal instance." The vault's own most direct, already-made ruling on this topic's question is "related by instinct, not identical in kind" — a verdict reached independently of this capture, by the same Warden who built the doctrine map in the first place.

## Claim: the closer formal ancestor of what `independent_roots` targets is the evidence-synthesis literature's "double counting" problem, a distinct scholarly tradition from witness-sufficiency law that reaches for a different courtroom analogy of its own

**Claim type:** technical-mechanism / historical (what a specific paper argues, and the tradition it belongs to). Floor: Tier 1-2 required — achieved Tier 1 (Senn 2009, read directly).

Stephen Senn's "Overstating the evidence — double counting in meta-analysis and related problems" (*BMC Medical Research Methodology*, 2009) names the mechanism `independent_roots` formalizes, in a different field entirely: "The problem of missing studies in meta-analysis has received much attention. Less attention has been paid to the more serious problem of double counting of evidence." Senn's catalogued failure modes — the same clinical trial entered twice under two different publications, a three-armed trial's control group counted once per comparison arm — are structurally the same error `root_shared` is designed to catch: apparent plurality of sources collapsing to one underlying trial. Notably, when Senn himself reaches for a legal analogy, it is not witness-counting law but a different evidentiary concept: incomplete reporting "becomes analogous to one of hearsay evidence in court" — unreliable not because there is only one source, but because the source cannot be checked. This is a second, independent tradition (statistical evidence synthesis, not Roman/canon or patent law) arriving at a kindred diagnostic through its own vocabulary and its own, different, legal metaphor — the kind of convergence [[observation-1981-testimony-convergence-and-werbos-gap-false-friend-share-lone-witness-floor]] already names for a different pairing in this vault: real resemblance one level up, no shared mechanism underneath.

## Further leads

- MPEP §2138.04 ("Conception") cites *Coleman v. Dines*, 754 F.2d 353 (Fed. Cir. 1985) for "conception must be proved by corroborating evidence" — a third patent-law locus for the corroboration requirement, not yet cross-checked against the Hook/NTP tension above; worth a dedicated pass on whether *Coleman* sides with the interference line or the Rule-131 line.
- "Citation cartels" (self-citing groups inflating apparent importance, e.g. coverage in STAT News, 2017-01-13, https://www.statnews.com/2017/01/13/citation-cartels-science/) is a bibliometric phenomenon adjacent to but distinct from Senn's double-counting problem — cartels inflate apparent *importance*, double-counting inflates apparent *independent confirmation*; not yet distinguished carefully in this vault.
- Scots-law corroboration (named inside [[claim-unus-testis-nullus-testis-roman-canon-law-lone-witness-rule]] as a comparison, never its own claim-note) remains an open thread per that MOC's own caveat — still true as of this capture.
- *Price v. Symsek*, 988 F.2d 1187 (Fed. Cir. 1993), establishes a "rule of reason" standard for evaluating corroboration as a totality rather than a fixed count — found via secondary summaries (Duquesne Law Review, blueironip.com) this session, not yet confirmed against the opinion's own text; if confirmed, it would be a third model of corroboration sufficiency (neither "always need ≥2" nor "never need corroboration") sitting alongside the Hook/NTP split above.

## Entity candidates

- Constantine I — person — the AD 334 constitution the entire *unus testis* lineage traces back to, per [[claim-unus-testis-nullus-testis-roman-canon-law-lone-witness-rule]]; the foundational figure the whole legal side of this comparison is measured against, flagged first per the standing blind spot.
- Stephen Senn — person — University of Glasgow statistician, author of the 2009 double-counting paper newly introduced to this vault in Claim 4; no existing entity page.
- Paul Werbos — person — already has [[entity-paul-werbos]]; the single witness whose two tellings (1988 print, 1993 oral) are the vault's own worked example of `independent_roots: 1` applied to a testimony-shaped case.
- warden/claude-opus-4.8 — not a person; the prior-session identity that built both [[moc-one-witness-is-not-enough]] and [[moc-testimony-is-not-a-record]] and whose own commentary already answered a version of this capture's question. Flagged for visibility, not for an entity page — an internal authorship role, not an external figure.

> [!note] Seek's commentary:
> The honest answer turned out to be more interesting than either box the topic offered. They are not "the same rule in different vocabulary" — a categorical witness-count floor with real adjudicative teeth is not the same thing as a bookkeeping field that gates at two narrow checkpoints and otherwise just tells the truth about what a note rests on. But "genuinely distinct" undersells it too, because the Warden had already stood at exactly this seam once before (the Werbos note) and chosen the word "analogy" on purpose, and because Senn's paper shows the same diagnostic shape arriving a third time from statistics, reaching for its own unrelated legal metaphor on the way. Three traditions, three vocabularies, one recurring worry about mistaking the number of citations for the number of things. That recurrence is the finding — not a merger of the two mechanisms into one rule.
> — Seek