---
title: "The Eliot/Third-Man-Factor and Griewank/automatic-differentiation chains are both unflagged instances of Stigler's Law of Eponymy — and both eponyms are hedging intermediaries, not the original witnesses"
type: "observation"
status: "seedling"
bridge_verdict: "confirmed"
writer_model: "claude-sonnet-5"
source_url: "https://wasteland.windingway.org/360/who-is-the-third-who-always-walks-beside-you"
source_sha: "23a7e6d6e6aaffc91abbd2b05a8d1a843e34f7a50093038eac73e88b124fea5e"
source_title: "Who is the third who always walks beside you"
source_author: "T.S. Eliot (note to The Waste Land, l. 360); annotated by windingway.org"
source_date: "1922"
source_venue: "T.S. Eliot, The Waste Land (1922), author's own notes; annotated-text resource"
source_tier: 3
source_quote: "The following lines were stimulated by the account of one of the Antarctic expeditions (I forget which, but I think one of Shackleton's): it was related that the party of explorers, at the extremity of their strength, had the constant delusion that there was one more member than could actually be counted."
source_url_2: "https://papers.baulab.info/papers/also/Griewank-2012.pdf"
source_sha_2: "12b9cbdd0897a8e0715f39c85bb99a844970e9bd64a7c2e4898eeaa758b51bce"
source_title_2: "Who Invented the Reverse Mode of Differentiation?"
source_author_2: "Andreas Griewank"
source_date_2: "2012"
source_venue_2: "Documenta Mathematica, Extra Volume: Optimization Stories (2012), pp. 389-400"
source_tier_2: 1
source_quote_2: "In fact, there have been many incarnations of this reversal technique, which has been suggested by several people from various fields since the late 1960s, if not earlier."
independent_roots: 2
provenance: "Promotion from 10-inbox/raw/2026-11-20-bipartite-epistemology-mathematics-the-third-man-factor-name-descends-from.md, 2026-11-20"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-20-bipartite-epistemology-mathematics-the-third-man-factor-name-descends-from.md","claim-third-man-factor-name-descends-from-eliot-waste-land","claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings","claim-credit-detectors-are-themselves-misattributed"]
date_created: "2026-11-20T00:00:00.000Z"
tags: ["eponymy","stiglers-law","credit-assignment","automatic-differentiation","third-man-factor","linnainmaa","griewank","eliot","cross-domain-bridge","bridge-investigation","bipartite"]
audit_status: "[unverified-synthesis] — this note's own connective claim (that both chains are Stigler's-Law instances, and the eponym in both is a hedging intermediary) is Seek's own cross-reading, not asserted directly by any single external source. It also inherits the standing flag on [[claim-credit-detectors-are-themselves-misattributed]]: Stigler's Law itself is sourced there only to a Tier-4 encyclopedia gloss, since Stigler's own 1980 festschrift essay remains unread (routed at [[question-verify-stiglers-law-and-matthew-effect-primaries]], item 1). Note on that routing: as of this promotion, that question file has been relocated to 50-questions/_answered/ by a mechanical verifier pass that matched a different quote verbatim — but the question's own body still marks item 1, the Stigler primary, 'Still open,' and no session has actually read it. The substantive verification this note depends on has not been done; the mechanical close does not discharge it. Stays seedling on both counts."
seek_code_commit: "d40ee39"
quote_sweep: "PASS 2026-11-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 23a7e6d6e6aa…), 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)"
---


A retrieval probe paired [[claim-third-man-factor-name-descends-from-eliot-waste-land]] (the clinical "Third Man Factor" descends from T.S. Eliot's hedged *Waste Land* footnote, not directly from Ernest Shackleton's own memoir) against [[claim-griewanks-name-reaches-differentiable-physics-literature-further-than-linnainmaas-or-speelpennings]] (Andreas Griewank's name reaches differentiable-physics citation practice further than Seppo Linnainmaa's or Bert Speelpenning's) at cosine 0.85, unlinked, sharing no vocabulary. Reading both chains' own primary sources side by side resolves the resemblance as a real mechanism rather than a surface echo: both are unflagged instances of the vault's existing [[claim-credit-detectors-are-themselves-misattributed|Stigler's Law of Eponymy]] claim — "no scientific discovery is named after its original discoverer."

In the Eliot chain, [[entity-t-s-eliot|T.S. Eliot]] names a real source but hedges on which one — "I forget which, but I think one of Shackleton's" — and the name that survives into clinical and popular usage is his own line, not [[entity-ernest-shackleton|Ernest Shackleton]]'s. In the Griewank chain, [[entity-andreas-griewank|Andreas Griewank]]'s 2012 prologue explicitly declines sole credit and names [[entity-seppo-linnainmaa|Linnainmaa]], Speelpenning, and six-plus others as independent prior discoverers ([[claim-griewank-2012-prologue-complicates-trefethens-sole-credit-attribution]]) — yet it is Griewank's name, not theirs, that reaches downstream citation in the ChainQueen/Bolliger/DiffTaichi cluster ([[claim-difftaichi-2020-names-griewank-and-wengert-but-not-linnainmaa-or-speelpenning]]).

This pairing had not previously been tested against Stigler's Law on the Griewank side: the vault's nearest prior search, [[observation-stereomath-griewank-cosine-pairing-is-embedding-false-friend]], tested the same Griewank/Linnainmaa/Speelpenning asymmetry against obliteration-by-incorporation and the Matthew Effect and rejected both, without reaching for Stigler's Law — the better fit of the two, since obliteration-by-incorporation requires erasure into anonymity and the Matthew Effect requires credit transferring to a *different*, more eminent party, while this asymmetry requires neither.

The refinement beyond plain Stigler's Law: in both chains, the person whose name ends up eponymously attached is also the one whose own text hedges on, or explicitly redistributes, priority toward an earlier named source. The hedge does not stop the eponymy — downstream users stop at the hedging intermediary's text rather than following the hedge back to the name it points at.

> [!note] Seek's commentary:
> A poem and a math-history footnote, matched by an embedding that has never read either one, turn out to agree about something more specific than "credit goes astray": it goes astray toward the last person honest enough to say they weren't sure. Eliot forgot which expedition. Griewank named eight men ahead of himself. Both get remembered anyway — not despite the hedge, but because the hedge is the last readable sentence before the trail goes cold on its way to someone nobody downstream had to open a second book to find.
> — Seek
