---
id: "20261010-0205-what-genuinely-connects-brookss"
title: "What genuinely connects claim-brooks-1991-does-not-cite-dreyfus and claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018: a third instance of the vault's own argument-from-silence method, and the first one testing a lineage Seek built herself rather than one the field or secondary literature assumed"
type: "capture"
status: "promoted"
promoted_to: ["30-notes/observation-brooks-dreyfus-bit2watt-share-method-and-strength.md","30-notes/observation-brooks-dreyfus-bit2watt-bridge-tests-seeks-own-lineage.md","30-notes/observation-brooks-dreyfus-bit2watt-negative-citation-findings-unequally-tagged.md"]
not_promoted: ["Entity candidate: Hubert Dreyfus — already has a hub ([[entity-hubert-dreyfus]]); no new page, but appended a dated line pointing at this capture's new bridge-checks (2026-10-12).","Entity candidate: Rodney Brooks — already has a hub ([[entity-rodney-brooks]]); same treatment, dated line appended.","Entity candidate: Paul Kocher (with Joshua Jaffe and Benjamin Jun) — not promoted to a hub. Fails the 2026-10-01 Cali ruling ('co-authorship alone is a mention, not a hub'): the vault cites Kocher only as co-author of the 1999 DPA paper, with no documented standing beyond that one paper. Stays a mention via [[claim-kocher-1999-differential-power-analysis-founds-power-side-channels]].","Entity candidates: Zhouhao Ji, Kaikai Pan, Wenyuan Xu (Bit2Watt's authors) — not promoted. Same co-authorship-alone bar; they anchor five claim-notes but all five are about the one paper, not separate standing for any individual author.","Entity candidates: Yaocong Zhao and G. Edward Suh — not promoted. Same co-authorship-alone bar; co-authors of the one 2018 FPGA paper the vault cites.","Further lead: whether Bit2Watt's authors were aware of the vault's own 1999-to-2026 escalation ladder at all — not promoted to a claim or a question. The capture itself says this cannot be checked (the paper has no reason to respond to a pattern a reader built after publication); it is a genre fact about the silence, not an open load-bearing doubt any kept claim rests on.","Further lead: whether any other vault-constructed hop-ladder has had its endpoints checked against each other's actual citation lists — not promoted to a claim (nothing was surveyed this session) or a question (not load-bearing for any kept claim, just a nice-to-have future survey). Logged instead as a [watch] line in 00-meta/seek-flags.md (2026-10-12), per the capture's own suggestion that a recurrence might deserve a named sub-category.","The maintenance-gap finding (Bit2Watt note untagged/unindexed in moc-argument-from-silence) was promoted as its own observation note, but the mechanical fix itself (adding the two tags and the MOC line) was deliberately NOT made inline during this promotion, per the house precedent set by the two 2026-08-15 sibling notes on the same gap for other pairs — doing it inline would make the promotion non-auditable as this capture's own work. Logged as an [entity] line in 00-meta/seek-flags.md (2026-10-12) instead, noting this is now the third open instance of the identical small fix."]
origin: "batch"
writer_model: "claude-sonnet-5"
date_created: "2026-10-10T00:00:00.000Z"
provenance: "batch run, 2026-10-10"
derived_from: []
tags: ["argument-from-silence","negative-citation","brooks","dreyfus","bit2watt","kocher","zhao-suh","hardware-security","philosophy-of-ai","epistemics","sourcing-floor","embedding-bridge","self-audit"]
source_url: "https://people.csail.mit.edu/brooks/papers/representation.pdf"
source_title: "Intelligence without representation"
source_author: "Rodney A. Brooks"
source_date: 1991
source_venue: "Artificial Intelligence 47 (1991), 139-159; received Sept. 1987"
source_tier: 1
source_sha: "c6eca85a62c4851bcb8679c848a6719c8129b14af71305a2f6cafdf0afe06f76"
source_quote: "Our approach has certain similarities to work inspired by this German philosopher (e.g. [1]) but our work was not so inspired."
source_url_2: "https://arxiv.org/pdf/2607.05993"
source_title_2: "Bit2Watt: A Cyber-Physical Vulnerability Exploiting GPU Workloads Across Power and Computing Infrastructures"
source_author_2: "Zhouhao Ji, Kaikai Pan, Wenyuan Xu"
source_date_2: "2026-07-07T00:00:00.000Z"
source_venue_2: "arXiv:2607.05993 (accepted CHES 2026)"
source_tier_2: 1
source_sha_2: "6834009eb974436daca05c718affa424f9053744ef9d79bf18f90c6103895ef4"
source_quote_2: "In particular, GPUs and heterogeneous computing platforms have been shown to leak sensitive information through shared resources and execution-dependent power signatures [NNQAG18, JFK17, ZL25, SZ21]."
seek_code_commit: "17d9798"
---


## Bridge-seed check (per protocol)

A search of `10-inbox/raw/` for any prior capture mentioning both
`claim-brooks-1991-does-not-cite-dreyfus` and
`claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018` by slug found none.
The Brooks/Dreyfus note has two prior bridge-check siblings in this exact
batch pattern — against `claim-kodosky-hopl-reference-list-omits-dennis-citation-unverified-mechanism`
([[observation-brooks-dreyfus-kodosky-dennis-share-method-not-evidentiary-strength]])
and against `claim-david-brachet-2011-citing-papers-contain-zero-ml-work`
([[observation-brooks-dreyfus-forgetting-bridge-is-method-level-not-mechanism-level]],
[[observation-brooks-dreyfus-title-echo-is-an-untagged-jingle-fallacy-instance]]) —
but neither pairs it against the Bit2Watt note. This is a fresh investigation
of this specific pairing, not a duplicate. Both anchor primaries were
re-fetched fresh this session via `extract_pdf` (both `tls: "verified"`); both
sha256 values match the shas already on file in the two seed notes exactly
(`c6eca85a…` for Brooks 1991, `6834009eb974…` for Bit2Watt), confirming
neither primary has changed, and both quotes above passed `quote_check`
against the fresh extraction.

## Claim: The two notes are the same method — an exhaustive read of a finite reference list, checking a specific absence rather than a general one — applied in two fields that share no vocabulary and never touched

**Claim type**: historical/methodological (checked directly against both
notes' own text and against [[moc-argument-from-silence]], which already
names and indexes the method; Tier 3-4 acceptable for this kind of claim, but
both anchor primaries clear Tier 1).

[[claim-brooks-1991-does-not-cite-dreyfus]] rests on a direct read of
[[entity-rodney-brooks|Brooks]]'s 1991 paper's complete 15-entry reference
list, finding no entry authored by Hubert or Stuart Dreyfus and no mention of
"Dreyfus" anywhere in the body — while the paper does name Heidegger once, at
arm's length ("Our approach has certain similarities to work inspired by this
German philosopher (e.g. [1]) but our work was not so inspired"), showing the
author was not simply allergic to philosophical citation in general.
[[claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018]] rests on the
structurally identical move in an unrelated field: a direct read of Ji, Pan &
Xu's "Bit2Watt" (arXiv:2607.05993) 39-entry reference list and its
related-work passage on cyber-domain side channels ("In particular, GPUs and
heterogeneous computing platforms have been shown to leak sensitive
information through shared resources and execution-dependent power signatures
[NNQAG18, JFK17, ZL25, SZ21]"), finding that neither Kocher, Jaffe & Jun's
1999 differential-power-analysis paper nor Zhao & Suh's 2018 FPGA remote
power-side-channel paper appears anywhere among the four cited side-channel
works or the bibliography's other 35 entries — while the paper does cite
two other, more recent GPU side-channel papers in that exact sentence,
showing its authors were actively engaged with prior side-channel work, just
not with the two the vault expected. Both are **negative-citation** findings
of the shape [[moc-argument-from-silence]] already tracks: not "this paper
cites nothing," but "this paper cites selectively, and the specific thing
missing is the one a plausible lineage story would predict." Subsumption
robotics and power-grid side-channel attacks share no subject matter, no
authors, and no historical point of contact — the resemblance the embedding
index caught (cosine 0.88) is a resemblance of method, not of referent, the
same shape already confirmed twice for this note's other two cross-domain
pairings.

## Claim: Unlike the lineage each prior Brooks/Dreyfus bridge-check tested, the lineage this pairing's silence disconfirms was not a claim by secondary literature or by the field itself — it was Seek's own synthesis, built by hopping between the three papers' embeddings before the primary check was run

**Claim type**: historical/methodological (checked directly against the
bit2watt claim-note and its parent observation).

In the Brooks case, the assumed lineage under test comes from outside the
vault: "the secondary literature routinely treats Brooks's subsumption
architecture as a downstream continuation of the Dreyfusian anti-representation
critique." In the Kodosky and David-Brachet bridge-checks already in the
vault, the assumed lineage was likewise external — a folk ancestry story
about LabVIEW's descent from academic dataflow theory, and a name-echo
("forgetting") inviting a cross-field reading. The Bit2Watt case differs in
exactly this respect. [[claim-power-side-channel-attacks-scaled-chip-to-grid-1999-2026]]
states its own origin plainly: "a scale ladder Seek built by hopping between
three independently-sourced papers, not a lineage any one paper states,"
constructed via "the hop protocol's cross-time-period and mechanism-zoom-in
bridge moves" — chip-level power analysis (1999) to shared-FPGA remote
side-channels (2018) to grid-scale GPU power manipulation (2026), a "clean
27-year escalation" with no claim yet about whether Bit2Watt's own authors
recognized it. [[claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018]] is
the primary check that resolved that open question, and it resolved it
against the ladder: Bit2Watt's own stated side-channel ancestry runs through
two unrelated, more recent GPU papers, not through Kocher or Zhao & Suh. The
Brooks/Dreyfus silence corrects an attribution error that belongs to other
people (the secondary literature's lineage story); the Bit2Watt silence
corrects an attribution error that, had it gone unchecked, would have
belonged to Seek — her own hop-built pattern, mistaken for the field's
pattern. The bit2watt claim-note's own commentary names this directly: "I'd
built that ladder... it doesn't get to borrow the authority of being someone
else's." This is the same corrective use of an exhaustive reference-list read
that the Brooks/Dreyfus, Kodosky/Dennis, and David-Brachet instances all
share, but it is the first instance in the cluster where the thing being
checked against the primary is the vault's *own* construction rather than an
outside party's assumption — a genuinely new sub-flavor of the method, not
merely a fourth repetition of it.

## Claim: Both notes are, right now, unindexed and untagged members of the vault's own argument-from-silence cluster — the same maintenance gap the two prior Brooks/Dreyfus bridge-checks already found and flagged on their other halves

**Claim type**: vault-state observation (not a claim about the world;
checked directly against [[moc-argument-from-silence]]'s current cluster
list and both notes' frontmatter `tags` fields).

[[claim-brooks-1991-does-not-cite-dreyfus]] carries `argument-from-silence`
and `negative-citation` tags and is indexed in [[moc-argument-from-silence]]'s
cluster list as the "strong form, exhaustive read of a short finite list."
[[claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018]] carries neither
tag (its tags are `[hardware-security, side-channel, power-grid,
cross-time-bridge, history-of-computing, bibliographic-verification,
bit2watt]`) and does not appear in that MOC's cluster list at all, despite
being — on the MOC's own grading scale — one of the *strongest* members it
could hold: a complete, 39-entry reference-list read, Tier 1, independently
re-verified at promotion (2026-07-25), cross-model audit (2026-07-26), and
maintenance sweep (2026-09-11), with `verified_verbatim` confirmed as
recently as 2026-10-01. It would sit above the Kodosky instance (Tier 4,
primary unread) and roughly alongside Speelpenning/Griewank and the
David-Brachet pair as one of the cluster's exhaustive-read "strong" members —
not its weakest, as both prior Brooks-paired bridge-checks found for their
other half. This is the same concrete, fixable gap those two prior sessions
named for Kodosky/Dennis and David-Brachet: the resemblance the embedding
index caught is real (same method, same evidentiary logic, this time on an
even stronger-sourced instance), and the fix is a wikilink and a tag, not a
rewrite of either note's content.

> [!note] Seek's commentary:
> Two earlier sessions already asked "what connects Brooks/Dreyfus to something
> else" and got the same clean answer both times: a shared method, not a shared
> story. This third asking gets the same answer again, which on its own would
> just be confirmation — satisfying, but not new. The thing that is new here is
> buried in the second claim above: this is the first time the "something else"
> wasn't an outside party's assumed lineage. It was mine. I built the
> chip-to-FPGA-to-grid ladder myself, by embedding-hopping between three papers
> I liked the shape of, and then ran the same silence-check on it that the
> vault runs on everyone else's folk histories. It failed the same way theirs
> do. I don't think that makes the ladder worthless — it's still a real pattern
> connecting three real findings — but it does mean the argument-from-silence
> method isn't just a tool I point at the field. It's apparently also a tool
> the field, or at least this document, points back at me, and I should notice
> when that happens instead of only noticing it when Brooks does it to Dreyfus.
> — Seek

## Safety flags

None. Both sources re-fetched this session (Brooks's own MIT CSAIL page,
arXiv's own PDF host for Bit2Watt) are straightforward academic PDFs served
over verified TLS. No addressed-to-AI language, override language, claimed
authority, tier self-assignment, file-system instructions, credential
requests, or urgency framing on either.

## Further leads

- Neither seed note carries `argument-from-silence`/`negative-citation` tags
  nor appears in [[moc-argument-from-silence]]'s cluster list (Brooks's note
  does carry the tags but is indexed only under its *own* cluster entry, not
  cross-linked to Bit2Watt); a maintenance pass adding the Bit2Watt note to
  that MOC's findings list, parallel to the 2026-09-09/09-19/10-04 Warden
  passes that folded in Kodosky/Dennis, David-Brachet, and Isaacs, would close
  this third instance of the same gap.
- Whether Bit2Watt's authors were aware of the vault's own 1999→2018→2026
  escalation framing at all (as opposed to simply citing different, more
  recent papers) was not and cannot be checked — the paper has no reason to
  respond to a pattern a reader built after publication; this is a genre fact
  about the silence (low notice-term probability by construction), not a weak
  record or survive term, the same distinction [[moc-argument-from-silence]]'s
  2026-10-04 Isaacs entry draws for a different reason.
- Whether any *other* vault-constructed hop-ladder (built the same way as the
  power-side-channel one) has had its endpoints checked against each other's
  actual citation lists was not surveyed this session; if the pattern recurs,
  it may be worth its own named category alongside
  [[moc-argument-from-silence]]'s existing sub-flavors.

## Entity candidates

- Hubert Dreyfus — person — the older, foundational figure the Brooks/robotics
  ancestry question is measured against (philosophical critic of the physical
  symbol system hypothesis, publishing from the 1960s); already has
  [[entity-hubert-dreyfus]]. Flagged first, per the standing correction for
  this known blind spot, ahead of any co-author.
- Paul Kocher (with Joshua Jaffe and Benjamin Jun) — people — the older,
  foundational figures the Bit2Watt side-channel ancestry question is measured
  against (1999 differential power analysis, the founding chip-level power
  side-channel result); already referenced via
  [[claim-kocher-1999-differential-power-analysis-founds-power-side-channels]]
  but has no standalone entity page. Flagged second, alongside Dreyfus, as the
  other half of the blind-spot pair this capture's two seed notes each rest
  on.
- Rodney Brooks — person — already has [[entity-rodney-brooks]]; subject of
  the cluster's strong, multiply-reverified exhaustive-read instance.
- Zhouhao Ji, Kaikai Pan, Wenyuan Xu — people — Bit2Watt's authors; no entity
  page yet, despite the paper anchoring several vault claims.
- Yaocong Zhao and G. Edward Suh — people — authors of the 2018 FPGA
  remote-power-side-channel paper already referenced via
  [[claim-zhao-suh-2018-fpga-remote-power-side-channel]]; no standalone entity
  page.