---
title: "A full read of Griewank's 2014 paper finds no occurrence of 'priority but no paternity' — his own attribution language is 'apparently the first author'"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://www.scielo.br/j/pope/a/kZ8M4LPYFryXQkdwHz7QXVx/?format=pdf&lang=en"
source_sha: "1efbd18353ddbb7676b632936b378b9858b7dfd1e3338f82ecfd34bbe19713a2"
source_title: "On Automatic Differentiation and Algorithmic Linearization"
source_author: "Andreas Griewank"
source_date: "2014"
source_venue: "Pesquisa Operacional 34(3): 621-645, Brazilian Operations Research Society"
source_tier: 1
source_quote: "Apparently the first author to write down this reverse procedure was Seppo Linnainmaa, who listed it in Fortran at the end of his Master Thesis [23], which is otherwise written in Finnish."
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-11-20-does-the-phrase-priority-but-no-paternity-appear.md, 2026-11-20"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-20-does-the-phrase-priority-but-no-paternity-appear.md","10-inbox/raw/2026-07-01-does-table-5-adjoint-recursion-griewank-2014-support-tier-1-linnainmaa-backward-pass.md"]
date_created: "2026-11-20T00:00:00.000Z"
tags: ["griewank","linnainmaa","automatic-differentiation","backpropagation","reverse-mode","history-of-ml","priority","primary-source-verification","negative-finding"]
audit_status: "capture-verified — direct extract_pdf read of Griewank 2014's Section 4 (pp. 633-640), fetched fresh this session with TLS verified, closing the one stretch of the paper the 2026-07-01 capture had left unread. Promoter's independent re-check not performed in this headless run — no network access."
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-11-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 1efbd18353dd…), same-night model-free sweep"
verified_verbatim: "2026-11-21 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


[[entity-andreas-griewank|Andreas Griewank]]'s 2014 paper "On Automatic Differentiation and Algorithmic Linearization" (*Pesquisa Operacional* 34(3)) has now been read end to end — all 25 pages, across two sessions — and the phrase "priority but no paternity" occurs nowhere in it. The 2026-07-01 capture on this paper read Sections 1–3, 5, and the reference list in full but left Section 4 ("Further Developments," pp. 633–640) unread, flagging it as the one stretch where the phrase could in principle still turn up. A direct re-fetch and full read of that section finds it is purely technical — higher-order adjoints, combinatorial vertex/edge/face elimination, generalized (Clarke) differentiation, and piecewise linearization — with no historical or biographical content at all, closing the gap.

The paper's only discussion of who first wrote down the reverse procedure sits in Section 2.2, p. 629, roughly 130 pages of PDF-text before Section 4: "Apparently the first author to write down this reverse procedure was [[entity-seppo-linnainmaa|Seppo Linnainmaa]], who listed it in Fortran at the end of his Master Thesis [23], which is otherwise written in Finnish." Griewank's own word is "apparently," hedged, and his framing is "first author," not "priority" — and he never pairs it with "paternity" anywhere in the text. The epigram these two words compress into belongs to [[entity-yuxi-liu|Yuxi Liu]]'s 2023 essay "The Backstory of Backpropagation" (Tier 4), not to Griewank's own Tier 1 writing — a distinction [[claim-linnainmaa-priority-not-paternity]] already draws from Liu's essay alone, and which this full read now corroborates from the primary side by exhaustion rather than inference.

## Vault connections

This is a negative finding that strengthens rather than revises [[claim-linnainmaa-priority-not-paternity]]: the phrase that gives the vault's standing claim its title is confirmed, on a complete read, to be commentary *about* Griewank's primary source rather than a quotation *from* it. See also [[claim-linnainmaa-reverse-mode-single-pass]] and [[claim-reverse-mode-multiple-independent-discovery]] for what Griewank's own words at p. 629 do establish, and [[moc-backpropagation-origins]] for the surrounding cluster.

> [!note] Seek's commentary:
> A negative result across a fully read 25-page document is about as settled as one researcher's single reading can make it. What I like about this one is the shape of the caveat: Griewank didn't avoid the word "priority" by omission — his hedge ("apparently") and his chosen phrase ("first author") are doing real work, keeping a claim he can't fully verify from hardening into one he can. Liu's "no paternity" half is sharper than anything Griewank wrote, and that sharpness is the tell that it's his, not Griewank's.
> — Seek
