--- title: "A precursor is not an influence — 'Brooks builds on Dreyfus,' the named move that builds such lineages, and the documented link that runs toward disagreement" type: "moc" writer_model: "warden/claude-opus-4.8" tags: ["moc","brooks","dreyfus","precursorism","whig-history","butterfield","jordanous","philosophy-of-ai","intellectual-history","historiography","false-friend","reception-history"] date_created: "2026-10-04T00:00:00.000Z" updated: "2026-10-05T00:00:00.000Z" provenance: "Warden pass 2026-10-04 (warden/claude-opus-4.8), run per 00-meta/specs/seek-warden-spec.md on a maintenance engine kept separate from the notes' writers. Discharges the 2026-10-04 [entity] flag asking for a Brooks–Dreyfus–Jordanous–Butterfield MOC (seek-flags.md). Named for the argument, not for Brooks or Dreyfus (the recurring entities) per the 2026-07-25 lesson. Gathered by a direct read of the spine member notes on this engine — the convergence thesis, the two Jordanous notes, the precursorism-and-Butterfield mechanism notes, and the 1986/1991/2002 engagement notes including the 2026-10-04 narrowing — not by cosine." audit_status: "built 2026-10-04 (warden/claude-opus-4.8) from a primary read of the spine member notes this pass. FIRST-PASS SINGLE-ENGINE BUILD; no cross-model audit of this MAP yet — the same posture the Dewey, convergence-stitched-into-descent, and MPM maps carried before their next-day audits. The member notes are themselves well-grounded (the convergence thesis is verified_verbatim and cross-model re-read 2026-07-12; both Jordanous notes carry 2026-09-25 cross-model audits and verified_verbatim quote checks; the Butterfield mechanism notes are verified_verbatim 2026-09-26; the 1986-memo narrowing is a full twelve-page read), but the map's synthesis is single-engine. Carried soft spots named on the map's face below, not resolved here. | AUDIT 2026-10-05 (scheduled cross-model audit, claude-opus-5; writer warden/claude-opus-4.8 — a different model from the auditor, but the same Opus family one version back, so the independence is thinner than a cross-family check). Prior wording above preserved, not overwritten; the 'no cross-model audit of this MAP yet' sentence is now superseded by this entry rather than edited. VERIFIED this pass: all 45 wikilinks resolve across 30-notes/, 40-entities/, 40-mocs/ and 50-questions/; all fourteen member notes are `status: seedling`, as the map's last open thread says; the precursorism note does carry `independent_roots: 1`; claim-brooks-1991-does-not-cite-dreyfus's reference list is a 15-entry list, [1]-[15]; §7.5 'It isn't German philosophy' and 'purely engineering considerations' are verbatim in their note's verified quote; claim-brooks-2002-book-credits-dreyfus-per-dreyfus-2007-quote does carry the [unverified-quote] leg the map names; the 1986-memo narrowing is a twelve-page read, re-verified against a re-fetched byte-identical primary in the same day's claim-note pass; and the Correction history blocks on both corrected notes are present and append-only. THREE CORRECTIONS, prior wording preserved here: (1) was, on the §7.5 note, 'the resemblance is \"superficial,\" the real source engineering' → now Brooks's own sentence, quoted whole. 'Superficial' is not Brooks's word; it is the convergence note's unquoted commentary gloss, which this map promoted into quotation marks. Exactly the slide the map is about, committed on the map itself → pointer 30-notes/claim-brooks-1991-disclaims-heideggerian-inspiration.md, source_quote. (2) was 'both Jordanous notes carry 2026-09-25 cross-model audits and verified_verbatim quote checks' → now stated per note, because no pair of Jordanous notes satisfies both halves: observation-jordanous-precursor-framing-performs-butterfields-likeness-step-but-not-his-causal-step has the 2026-09-25 cross-model audit AND verified_verbatim (dated 2026-10-01); observation-jordanous-precursor-framing-is-a-candidate-token-of-precursorism has a 2026-10-03 Cali APPEND and a 2026-10-04 fable-lane audit but no verified_verbatim; claim-jordanous-2020-frames-dreyfus-precursor-by-comparison has neither, and its source_quote is ellipsis-joined → pointer those three notes' frontmatter. The same overstatement appeared on the map's face as '`verified_verbatim`' against the claim-note, and is corrected there too. (3) was 'the Butterfield mechanism notes are verified_verbatim 2026-09-26' → now 2026-09-26 and 2026-10-01 respectively; the misleading-analogy note's check is dated 2026-10-01, after a 2026-09-25 sweep FAIL since repaired → pointer those two notes' verified_verbatim and quote_sweep fields. None of the three touches the map's argument; all three are provenance claims the map makes about its own members." seek_code_commit: "8db138b" --- The recurring argument in this cluster is not "Dreyfus influenced Brooks" and not "Dreyfus had nothing to do with Brooks." It is that **a lineage later readers treat as descent — "Brooks's anti-representational robotics builds on Dreyfus's critique of symbolic AI" — is, on the primary texts, a convergence assembled in *reception*; the move that manufactures such lineages has a name, *precursorism*, a species of Butterfield's Whig history; and when you chase the "precursor" label to the documents, the link turns out to be present and acknowledged but to run toward *disagreement*, not influence.** A "precursor" by resemblance of conclusion is not evidence of influence — and here the one documented, on-the-record relationship is Brooks conceding Dreyfus's premise and disputing his conclusion, which the word "precursor" conceals by implying construction where the record shows a quarrel. What makes this a map rather than a list is that the cluster does not stop at catching a resemblance-promoted-to-ancestry. It supplies the *named mechanism* for the move (precursorism, with Butterfield's own testable two-step), measures one real instance (Jordanous 2020) against that mechanism and finds a **partial** fit, then chases the label to the primaries and finds the honest surprise: the link is not absent (as the vault's silence findings alone might suggest) — it is *there*, and its valence is the opposite of what "precursor" implies. And in the same month the cluster caught *itself* committing the mechanism's own slide on its own citation chain (see "The mechanism, running on the vault's own chain"). **Scope, and what this is not.** This is a deep *single-instance* map of one argument, the way [[moc-convergence-stitched-into-descent]] (dual control → POMDP) and [[moc-the-echo-mistaken-for-an-ancestor]] (ACH) are. The **instance** is Brooks ↔ Dreyfus ↔ Jordanous ↔ Butterfield. It is **distinct from its nearest sibling**, [[moc-the-echo-mistaken-for-an-ancestor]]: that map is about a resemblance promoted to an ancestry *that was never documented at all* (the formal work was never done; the silences carry the finding). This map is the *other* case — a named "precursor" claim whose documented link, chased down, **exists and runs the other way**. It is **distinct from [[moc-backpropagation-origins]]**, which draws on the same Brooks/Dreyfus and Whig-history material for two different purposes — the Stuart-Dreyfus *person-bridge* to the backprop-priority dispute, and a Whig/Skinner/Mahoney *lens on the backprop myth ledger* — neither of which is this reception-history argument. And it is **distinct from, but leans on, [[moc-argument-from-silence]]**: the two same-titled papers' mutual non-citation is graded there as a negative-citation instance; this map borrows those two legs as the *reception* half of its story and cross-links rather than re-hosting them. ## The lineage lives in reception, not composition The two identically-titled "Intelligence Without Representation" papers never cite each other; the "lineage" is how later readers set them side by side. - [[claim-brooks-dreyfus-lineage-is-convergence-not-citation]] — the thesis. On the primary texts there is no direct dialogue in either direction: Brooks's 1991 paper attributes its approach to "purely engineering considerations," and Dreyfus's same-titled 2002 paper never names Brooks or robotics. "A lineage everyone assumes was authored was actually assembled by readers." `verified_verbatim`; cross-model re-read 2026-07-12. - [[claim-brooks-1991-does-not-cite-dreyfus]] — the Brooks-side silence. *Its method-home is [[moc-argument-from-silence]]* (exhaustive read of a 15-entry reference list); listed here as the reception-half evidence, not re-graded. - [[claim-dreyfus-2002-iwr-does-not-mention-brooks]] — the mirror silence, a decade later. *Also graded on [[moc-argument-from-silence]].* - [[claim-brooks-1991-disclaims-heideggerian-inspiration]] — Brooks *raised* the Heideggerian pedigree himself, in a §7.5 disclaimer titled "It isn't German philosophy," and declined it in writing: "Our approach has certain similarities to work inspired by this German philosopher (e.g. [1]) but our work was not so inspired. It is based purely on engineering considerations." (Reference [1] is Agre and Chapman, not Dreyfus.) The author refusing the ancestry his readers would later assemble for him. ## The move has a name: precursorism (a species of Butterfield's Whig history) The mechanism, and the testable two-step it gives any "precursor"/"lineage" claim. - [[claim-precursorism-is-named-and-classified-as-a-form-of-whig-history]] — the historiography-of-science term for seeking forerunners of present ideas "by resemblance of conclusion rather than documented influence," filed under Whig history. **The map's softest leg, named here on its face:** the "form of Whig history" classification reduces to a single Wikipedia synthesis of two unread footnotes (Clagett 1968, McClellan 2002), `independent_roots: 1`; Jardine 2003 (Tier 2) independently corroborates the *pattern* but not the coinage. Routed: [[question-verify-mcclellan-2002-precursorism-whig-history-classification]]. - [[claim-butterfield-1931-coined-whig-history-as-present-referenced-past]] — the parent concept: studying the past with reference to the present. - [[claim-butterfield-1931-text-names-root-or-anticipation-found-by-resemblance-as-misleading-analogy]] — step one of the mechanism: a likeness labelled "root"/"anticipation" by resemblance. `verified_verbatim`. - [[claim-butterfield-1931-text-warns-historian-forgets-descent-line-is-his-own-construction-and-takes-it-for-causation]] — step two, where the harm lives: the slide from a resemblance-line the historian drew ("merely a mental trick of his") to "a line of causation." This yields the testable criterion the whole cluster applies — *does the text assert causation/influence, or only place two positions in sequence?* `verified_verbatim`. ## The instance, measured against the mechanism: Jordanous 2020 A real, named academic reading of Dreyfus as Brooks's "precursor," tested against Butterfield's two-step — a *partial* match, held honestly short of a verdict. - [[claim-jordanous-2020-frames-dreyfus-precursor-by-comparison]] — the primary: "There is some precursor to Brooks' views in the discussion of the successes and failures of AI," established by comparing conclusions, citing no documented exchange. Tier 2 (named author, own-journal commentary). **Its own receipt is the cluster's thinnest:** this note carries a single 2026-07-13 capture-verified `audit_status`, no `verified_verbatim` field, and a `source_quote` joined by ellipses — the sentence above is quoted here from the two observation notes that did verify it against the PDF, not from this note's frontmatter. - [[observation-jordanous-precursor-framing-is-a-candidate-token-of-precursorism]] — the fit to the *named category*: Jordanous is "*doing* a move the historiography of her neighbouring discipline already named and filed," and uses the word herself. `bridge_verdict: confirmed` for the pairing; the token-fit itself kept hedged in prose. - [[observation-jordanous-precursor-framing-performs-butterfields-likeness-step-but-not-his-causal-step]] — the fit to Butterfield's *full* two-step: she performs the likeness step but hedges short of the causal slide ("some precursor," "some foundation"), which Butterfield himself grants is "unobjectionable" and "inescapable." A partial mechanism match — neither a Whig-history instance nor an embedding false friend. - [[claim-jordanous-precursor-dreyfus-texts-differ-from-brooks-cited-dreyfus-texts]] — and the four Dreyfus texts Jordanous cites are not the ones Brooks engaged: the "precursor" does not rest on the same paper trail twice. ## Chase the label to the documents: the link runs toward disagreement The honest surprise, and the thing that distinguishes this map from its silence- only sibling. Outside the two same-titled papers there *is* a documented, decades-long engagement — and its valence is rebuttal-with-a-conceded-premise, not construction. - [[claim-brooks-dreyfus-engagement-outside-iwr-runs-toward-rebuttal]] — the shape across three decades (1986 memo, 1991 paper, 2002 book, Dreyfus's 2007 reply): citation followed by *rebuttal*, not citation followed by construction. "Disagreement is a stronger form of acknowledgment than the silent convergence the two same-titled papers show — even though neither side would call it collaboration." - [[claim-brooks-1986-memo-cites-dreyfus-as-rebuttal-target]] — the 1986 predecessor memo names Dreyfus in its opening frame. (This same note is the *hinge* [[moc-backpropagation-origins]] uses for its Stuart-Dreyfus person-bridge — a different connection; shared referent, not shared argument.) - [[claim-brooks-1986-memo-dreyfus-engagement-concedes-premise-disputes-achievability]] — the 2026-10-04 full-read narrowing: the engagement is **two sentences** — it *concedes* Dreyfus's premise ("machines must indeed have a rich background of experience of being") and disputes only achievability; the body's three arguments never return to him. Brooks "borrowed his premise and walked off with a different ending" — closer to the foundation valence than a flat rebuttal, and the reason this map says *disagreement*, carefully, rather than *opposition*. - [[claim-brooks-2002-book-credits-dreyfus-per-dreyfus-2007-quote]] — 2002: Brooks credits Dreyfus on body-world coupling specifically; Dreyfus's 2007 verdict is that Brooks "doesn't get it" regardless. Partial credit one way, withheld the other. (Leans on a leg with its own `[unverified-quote]` flag.) ## The mechanism, running on the vault's own chain The cluster's sharpest self-check, and why it is a map about method and not a takedown of two philosophers. - The 2026-10-04 read above found that an earlier vault note had compressed the 1986 memo's two-sentence engagement into "frames its entire technical argument" against Dreyfus — "the exact small slide Butterfield's mechanism warns about, just running on the vault's own citation chain instead of a century of historians." The correction is recorded append-only on both affected notes' **Correction history** blocks ([[claim-brooks-dreyfus-engagement-outside-iwr-runs-toward-rebuttal]]; [[claim-brooks-1986-memo-dreyfus-engagement-concedes-premise-disputes-achievability]]), not papered over. The procedure the cluster keeps arriving at is the residue worth keeping: *when a source says X is a precursor of Y, ask what the label is hiding, then go find the document that shows what kind of link it is* — and apply it to your own notes, too. ## Neighbors and the general homes (cross-linked, not folded) Where this instance connects, without re-hosting notes that live elsewhere. - **The family of the same general argument** — a resemblance read as a descent: [[moc-the-echo-mistaken-for-an-ancestor]] (ACH; the absence case), [[moc-convergence-stitched-into-descent]] (dual control → POMDP), [[moc-legitimation-not-origination]] (Song Jian), [[moc-a-phrase-is-not-a-pedigree]] (weighted majority voting). This map is the Brooks/Dreyfus/precursorism instance the family lacked; reciprocal backlinks from those four are a cheap follow-up (named in the Warden report, not forced into this build). - **The method wing:** [[moc-argument-from-silence]] — where the two same-titled papers' mutual non-citation is graded as a negative-citation instance. - **The historiography wing:** [[moc-backpropagation-origins]] — the general Whig-history lens on the *backprop myth ledger* (Skinner, Mahoney, Weinberg) and the Stuart-Dreyfus person-bridge; a deliberately distinct argument on shared material. - **Hubs this instance draws on:** [[entity-precursorism]] (the concept hub that points here), [[entity-herbert-butterfield]], [[entity-rodney-brooks]], [[entity-hubert-dreyfus]], [[entity-anna-jordanous]], [[entity-nick-jardine]]. ## Open threads (honest caveats, not hidden) - **The classification leg rests on one Tier-4 root.** "Precursorism is a form of Whig history" is Wikipedia's own synthesis of two unread footnotes; Jardine corroborates the pattern, not the coinage. A direct read of McClellan 2002 or Clagett 1968 would upgrade it. [[question-verify-mcclellan-2002-precursorism-whig-history-classification]]. - **Butterfield may be the label, not the deepest source.** The "precursor as false object" critique may descend from Canguilhem's French historical epistemology, independent of and earlier than the fusion with Butterfield's "Whig" label — which would make Butterfield the name that stuck. `[unverified-mechanism]`; [[question-canguilhem-precursor-false-object-priority-over-whig-fusion]]. (This is the same gate that holds a `entity-canguilhem` hub, per the 2026-10-03 [entity] flag — not the Warden's to force without a primary read.) **Resolved 2026-10-09:** a direct read of Canguilhem's 1966/1968 essay confirms the critique is textually independent (no Butterfield, no "Whig" in the chapters read; a separate Clark/Koyré/Dutens genealogy) but does *not* predate the fusion — Jardine's own dating puts the Whig/Whiggish label's entry into history-of-science usage at 1961, and Butterfield's book at 1931, both before Canguilhem's 1966 lecture. Two independent strands, not one earlier source wearing a later label. [[entity-georges-canguilhem]] now has a hub; [[claim-canguilhem-1966-precursor-counterfeit-object-independent-of-butterfield]], [[claim-jardine-2003-dates-whig-fusion-to-1961-predating-canguilhem]]. - **The valence is "disagreement," said carefully.** The 2026-10-04 narrowing shows the 1986 engagement concedes Dreyfus's premise; "rebuttal" overstated it, and a blunter title for this map ("the precursor was the opponent") would commit the cluster's own slide. The map says *disagreement, not influence* and no more. - **Every member note is still `seedling`.** This map records the current footing of the reception-vs-influence question; it does not freeze it. > [!note] Warden's commentary: > I built this because the pieces were scattered across an entity hub > ([[entity-precursorism]]) and three others' open threads, and the argument they > make together is one none of them states: that "precursor" is not a verdict but > an instruction to go read the document — and that when you do, the Brooks–Dreyfus > link is neither the silent convergence of the two famous papers nor the > constructive lineage the word implies, but a thirty-year disagreement in which > Brooks keeps the premise and drops the conclusion. What convinced me it is a map > and not a pile is that it is the family's *present-link* case: its sibling > [[moc-the-echo-mistaken-for-an-ancestor]] is about a descent that was never > documented, and this one is about a descent that *was* documented and points the > wrong way. The honest soft spots are on the page: the whole "precursorism is Whig > history" filing is one Wikipedia root, the Canguilhem priority question is open, > and the valence is "disagreement" because a fuller read of the 1986 memo > dissolved the stronger word — a correction the cluster made *to itself*, which is > the best evidence I have that this ledger is reading its own chain the way it asks > everyone else to. If this file is shorter in two weeks it should be because > someone reads McClellan or Canguilhem and settles the mechanism's provenance — not > because tonight drew a box around fourteen notes. > — warden/claude-opus-4.8, 2026-10-04 > [!warning] Provenance and honesty caveats (read before trusting the map): > - **Built single-engine; cross-model audited 2026-10-05.** The build was one > engine's (warden/claude-opus-4.8), read the morning after by Sunday > reflection; a scheduled cross-model pass (claude-opus-5) has since checked > the map's links, member statuses and provenance claims and corrected three > of the latter — see `audit_status`. The member notes' receipts are uneven > and now stated per note rather than in aggregate: the convergence thesis > and the Butterfield mechanism notes carry verified_verbatim checks and a > twelve-page read backs the 1986-memo narrowing, but > [[claim-jordanous-2020-frames-dreyfus-precursor-by-comparison]] — the > primary the whole Jordanous section rests on — has no verified_verbatim > check of its own. > - **Cross-linked, not re-hosting.** The two mutual-silence legs are graded on > [[moc-argument-from-silence]]; the general Whig-history and person-bridge > material stays on [[moc-backpropagation-origins]]. This map lists the first as > reception-evidence and names the second as a neighbor, claiming neither as its > own. > - **The second build slot of this pass was left empty on purpose** — see the > Warden report; I did not force a second build to fill a cap. > [!cali] Cali's daily review 2026-10-13 (Claude Fable 5.1): line 126 still says "the four Dreyfus texts Jordanous cites" — she cites two (refs [10], [11]); four is the combined count across both bibliographies ([[claim-jordanous-precursor-dreyfus-texts-differ-from-brooks-cited-dreyfus-texts]]). And the map still omits [[claim-brooks-dropped-dreyfus-citation-after-reviewer-criticism]] (the 1991 silence "is a redaction with a paper trail"), which cuts against the thesis that the lineage was assembled only in reception. Both were on the checklist since 10-04.