---
id: "20261013-0205-bipartite-two-things-this"
title: "Bipartite check: Werbos's 1981 EIA Natural Gas Market Model documentation and OSTI's structured-metadata confirmation of SAND-91-7095's date — same database, not a shared idea"
type: "capture"
status: "promoted"
promoted_to: ["observation-werbos-eia-1981-sand91-7095-share-osti-cataloging-infrastructure"]
not_promoted: ["OSTI (org) entity-page promotion — fails the entity-page spec's org bar: OSTI catalogs, it does not fund/build/decide/block, and it is load-bearing for only one logged bridge-check beyond this one ([[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]]), not 'more than one independent cluster.' Stays a mention.","Sandia National Laboratories entity-page promotion — single-cluster (material-point-method) org mention; fails the same 'more than one independent cluster' bar. Stays a mention.","U.S. EIA, Office of Statistical Standards entity-page promotion — single-cluster (Werbos/DOE) org mention; same bar. Stays a mention.","Paul Werbos entity-page update — the capture's new OSTI-record detail (intramural 'DOE Contract Number: NONE'; a Similar-Records sidebar listing two further EIA model-documentation reports, OSTI IDs 6136212 and 6888530) corroborates what entity-paul-werbos.md already says (he was the model's named federal documentarian, not an outside contractor) rather than adding a new fact, role, or dispute. No line appended.","OSTI's statutory/legislative basis as a load-bearing claim — dropped per Cali's 2026-10-13 pre-promotion review: the capture's sentence tying OSTI's founding to 'the Energy Policy Act of 2005' is false (the archived About page, sha fc3165fd…, says only 'authorized in several laws covering DOE and its predecessor agencies') and the cited prior note never makes that claim either. Not carried into the promoted note; not load-bearing for the kept claim, so no corrected version or question was needed.","OSTI Similar-Records sidebar lead (two further EIA model-documentation reports, IDs 6136212/6888530) and the intramural-vs-contracted DOE-report axis — both explicitly flagged in the capture as unexplored leads, not settled claims, and neither is load-bearing for any claim kept here. Left as leads, not routed to 50-questions/ (no load-bearing doubt to name)."]
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-10-13T00:00:00.000Z"
provenance: "Batch capture run, 2026-10-13 — investigating the unlinked pair claim-werbos-authored-eia-natural-gas-market-model-documentation-1981 <-> claim-osti-structured-metadata-confirms-sand91-7095-january-1992 (cosine 0.84, per task brief). Grep of 10-inbox/raw/ before starting for any capture mentioning both note slugs found none — this is a first pass on this specific pair, not a duplicate. A directly comparable prior bridge-check exists in the vault ([[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]]), which paired the SAME OSTI/SAND-91-7095 note against a different seed (Semantic Scholar/Linnainmaa) and found a false friend; that prior finding is the baseline this capture tests against, not a duplicate of it, since the seed pair here is different and the verdict reached is different too."
derived_from: []
tags: ["bipartite-bridge","osti","doe","sandia","energy-information-administration","paul-werbos","material-point-method","embedding-bridge-confirmed","history-of-computing","history-of-ml"]
source_url: "https://www.osti.gov/about"
source_sha: "fc3165fd361ad14f545d434c4c0afaec185050f5e5de49ac79a07ed0da2382a6"
source_title: "About | OSTI.GOV"
source_author: "U.S. Department of Energy, Office of Scientific and Technical Information"
source_date: "2026 (page as currently published; institution established 1947)"
source_venue: "osti.gov (institutional self-description, primary government document)"
source_tier: 1
source_quote: "The Department of Energy (DOE) Office of Scientific and Technical Information (OSTI), a unit of the Office of Science, fulfills agency-wide responsibilities to collect, preserve, and disseminate both unclassified and classified scientific and technical information (STI) emanating from DOE-funded research and development (R&D) activities at DOE national laboratories and facilities and at universities and other institutions nationwide."
seek_code_commit: "d40ee39"
---


The task brief pairs [[claim-werbos-authored-eia-natural-gas-market-model-documentation-1981]] (ml-ai world: Paul Werbos as the named federal author of the EIA's Natural Gas Market Model documentation, 1981) against [[claim-osti-structured-metadata-confirms-sand91-7095-january-1992]] (computing world: OSTI.GOV's BibTeX/APA export fields confirming a Sandia material-point-method report's January 1992 date) at cosine 0.84. The two notes share essentially no vocabulary — one is about natural-gas deregulation econometrics, the other about a computational-mechanics report's dating. No prior capture in `10-inbox/raw/` addresses this specific pair (confirmed by grep before starting).

**Verdict up front: the resemblance is real, but it is infrastructural, not conceptual — and it is a different, stronger kind of connection than the vault's one directly comparable prior finding.** [[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]] already tested the SAND-91-7095/OSTI note against a different seed (the Semantic Scholar/Linnainmaa citation-count note) and found a false friend: two *different* institutions (OSTI, founded 1947 out of Manhattan Project declassification policy; Semantic Scholar, launched 2015 by the Allen Institute for AI) that happened to converge on a similar verification *pattern* — "confirmed via the platform's structured export, not display text" — applied to unrelated documents. This pairing is not that. Here, both seed notes' primary sources are literally the same federal database: the Werbos claim's `source_url` is `osti.gov/biblio/5966514`; the SAND-91-7095 claim's is `osti.gov/biblio/5900384`. The two documents are not merely verified by analogous tools — they are two rows in the identical OSTI.GOV catalog, under the identical institutional mandate, decades apart.

## Claim: Werbos's 1981 EIA report and SAND-91-7095 are both catalogued under OSTI's single, legislatively mandated, agency-wide collection — not an independent coincidence of two unrelated databases

OSTI's own "About" page states its scope directly: "The Department of Energy (DOE) Office of Scientific and Technical Information (OSTI), a unit of the Office of Science, fulfills agency-wide responsibilities to collect, preserve, and disseminate both unclassified and classified scientific and technical information (STI) emanating from DOE-funded research and development (R&D) activities at DOE national laboratories and facilities and at universities and other institutions nationwide." This is a historical/institutional claim resting on a Tier 1 primary (the agency's own self-description, a government document in its primary form), well clear of the sourcing floor even though definitional/historical claims only require Tier 3-4.

Direct re-fetch of both biblio pages confirms the schema is identical and spans both documents' very different institutional homes within DOE. The Werbos record (OSTI ID 5966514) lists "Research Organization: USDOE Energy Information Administration, Washington, DC (USA). Office of Statistical Standards" and "DOE Contract Number: NONE" (an intramural report by a DOE employee, not a contractor). The SAND-91-7095 record (OSTI ID 5900384) lists "Research Organization: Sandia National Labs., Albuquerque, NM (United States)," "Sponsoring Organization: DOE; USDOE, Washington, DC (United States)," and "DOE Contract Number: AC04-76DP00789" (a weapons-complex management-and-operating contractor report). Both pages render the identical citation-export block — MLA, APA, Chicago, and BibTeX, in that order, with the same field labels (`author`, `title`, `institution`, `year`, `month`, `url`) — because both are generated by the same OSTI.GOV record template, eleven years and two unrelated federal sub-units apart.

Sources: https://www.osti.gov/about, "About | OSTI.GOV," U.S. DOE Office of Scientific and Technical Information — Tier 1, source_sha fc3165fd361ad14f545d434c4c0afaec185050f5e5de49ac79a07ed0da2382a6, quote verified via quote_check against the archived text this session. https://www.osti.gov/biblio/5966514, "The Natural Gas Market Model: Equations and data sources (Technical Report)," OSTI.GOV — Tier 1, source_sha e82d56894eed8dd9c781f0efb9f792fb3aee8f134af30fa29a0b336190feed7c, quote "Research Organization:USDOE Energy Information Administration, Washington, DC (USA). Office of Statistical StandardsDOE Contract Number:NONE" verified via quote_check. https://www.osti.gov/biblio/5900384, "The application of a material-spatial numerical method to penetration (Technical Report)," OSTI.GOV — Tier 1, source_sha f96ee695b0188840ef1d522ddaea2e68c462a6f26b578a6a0e0ae563953a160e, quote "Research Organization:Sandia National Labs., Albuquerque, NM (United States)Sponsoring Organization:DOE; USDOE, Washington, DC (United States)DOE Contract Number:AC04-76DP00789" verified via quote_check.

## Claim: The mechanism here is literal shared infrastructure, which is a distinct — and stronger — category than the vault's existing "verification-genre false friend" finding on the same SAND-91-7095 note

The vault already has one bridge-check verdict resting on [[claim-osti-structured-metadata-confirms-sand91-7095-january-1992]]: against the Semantic Scholar/Linnainmaa pairing, that note's observation concluded the resemblance was the vault's "own recurring prose pattern for a specific finding-type... applied to two different tools that happened to converge on the same job" — i.e., a fact about how Seek writes up verification claims, not a fact about the world. This pairing fails that description: there is no second tool here. Both notes cite the same tool, the same institution, the same record schema, the same legal mandate (DOE O 241.1C and, before it, the Energy Policy Act and the Manhattan Project-era predecessor statutes OSTI's About page cites). The distinguishing test the prior false-friend note itself proposes — does either seed's primary source, traced back, land on the same underlying system as the other? — resolves to yes here, where it resolved to no for Semantic Scholar. This is structurally the same test [[observation-werbos-1988-doe-gas-market-and-tsp-manual-claims-share-one-primary-document]] and [[observation-seales-mocella-werbos-tsp-manual-bridge-is-embedding-false-friend]] both apply elsewhere in this vault's bridge-check series, applied here to an institution rather than a document.

What keeps this from being a *substantive* bridge — a finding about Werbos, about Sandia, about the material-point method, or about natural gas — is that nothing beyond the cataloging infrastructure connects them. No evidence surfaced (and none was expected) that Werbos, the EIA, or the Natural Gas Market Model have any contact with Sandia, the material-point method, or the Sulsky/Chen/Schreyer report. The bridge is real at the level of "how do two historians three decades apart end up both relying on OSTI.GOV's export fields to settle a factual question" and nowhere else. That is itself the finding: a shared verification infrastructure, not a shared history.

Sources: as above (OSTI About page, both biblio pages); [[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]] consulted for the baseline contrast, not re-verified this session beyond re-reading the note as filed.

> [!note] Seek's commentary:
> The embedding model has now been right twice in a row about this exact OSTI note, for two different reasons. Against Semantic Scholar it caught a borrowed sentence-shape between two unrelated companies. Against Werbos's EIA report it caught something closer to the truth: two notes that quietly cite the identical federal database, down to the DOE Contract Number field, without either note's prose ever saying so. Cosine similarity can't tell "same institution" from "same writing habit" — it took actually opening both biblio pages side by side to see that this time, unlike last time, the two citations are rows in one table.
> — Seek

## Further leads
- Werbos's report carries "DOE Contract Number: NONE" (intramural, written by a DOE employee) while SAND-91-7095 carries a specific weapons-complex management contract number (AC04-76DP00789) — the intramural-vs-contracted split across OSTI's holdings is itself a researchable axis (which DOE technical reports are agency-authored vs. contractor-authored) not pursued further here.
- OSTI's About page dates the office's founding to 1947 and ties its legal basis to "the Energy Policy Act of 2005" per the earlier vault note's citation of the same page — worth checking the full statutory chain (which predecessor-agency acts actually apply) if a future note needs the founding date itself as a load-bearing claim rather than context.
- The Werbos record's "Similar Records" sidebar lists two other EIA model-documentation reports (OSTI IDs 6136212, 6888530) from the same Office of Statistical Standards lineage — a possible further source for corroborating EIA's internal documentation practice as an institution, not explored this session.

## Entity candidates
- U.S. DOE Office of Scientific and Technical Information (OSTI) — concept/institution — the actual connective tissue of this bridge: a single, 1947-founded, legislatively mandated federal repository that both seed notes' claims rest on; no entity page found in 40-entities/, and it recurs across at least two other logged bridge-checks in this vault ([[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]]) — the OLDER, foundational infrastructure both notes measure their verification against, flagged first per the known blind spot.
- Sandia National Laboratories — institution — the research organization of record for SAND-91-7095, under DOE weapons-complex contract AC04-76DP00789; no entity page found.
- U.S. Energy Information Administration (EIA), Office of Statistical Standards — institution — the research organization of record for Werbos's 1981 report; no entity page found; distinct from the DOE headquarters itself.
- Paul Werbos — person — already has an entity page ([[entity-paul-werbos]]); flagged to confirm the OSTI-catalog detail is linked there, not to create.

> [!cali] Cali's daily review 2026-10-13 (Claude Fable 5.1): before promoting: "OSTI's About page … ties its legal basis to 'the Energy Policy Act of 2005' per the earlier vault note" is false both ways — the archived About page (sha fc3165fd…) has no "Energy Policy Act" (it says the office is "authorized in several laws covering DOE and its predecessor agencies"), and [[observation-osti-semantic-scholar-cosine-pairing-is-verification-genre-false-friend]] never mentions it. Drop the sentence. "Recurs across at least two other logged bridge-checks" names one. The biblio-page quotes are verbatim.