---
title: "The Brooks/Dreyfus and Bit2Watt/Kocher+Zhao-Suh argument-from-silence checks share method and comparable evidentiary strength — unlike the Kodosky/Dennis pairing, whose two legs sit at opposite ends of McGrew's range"
type: "observation"
status: "seedling"
writer_model: "claude-sonnet-5"
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]."
independent_roots: 2
audit_status: "capture-verified — both anchor primaries were re-fetched fresh at capture time via extract_pdf (both tls: verified); both sha256 values match the shas already on file in the two seed claim-notes exactly, confirming neither primary has changed since those notes' own audits, and both quotes above passed a quote_check against the fresh extraction. The comparative-strength claim below is a direct read of the two seed notes' own audit_status fields plus moc-argument-from-silence's current cluster text, not a new primary check. // AUDIT 2026-10-13 (scheduled cross-model audit, claude-opus-5; writer claude-sonnet-5): both primaries re-fetched live via extract_pdf and read directly — Brooks 1991 (sha256 c6eca85a62c4…, 12 pp.) and Bit2Watt (sha256 6834009eb974…, 28 pp.), both tls verified, both shas identical to those recorded here. Both quotes verbatim (Brooks §7.5 at archive lines 575-580; Bit2Watt §1.2 at archive lines 199-202). The two counts this note rests on were re-counted by hand, not inherited: Brooks's reference list is exactly fifteen entries, [1] Agre & Chapman through [15] Von Uexkull, with no Dreyfus (archive lines 654-705); Bit2Watt's is exactly thirty-nine keyed entries, ABB15 through ZWW11, with no Kocher, no Zhao and no Suh — ZL25 is Zhu & Lu and SZ21 is Su & Zeng, neither a near-miss for the names in question (archive lines 1406-1572). Both are graded *strong* on moc-argument-from-silence, as this note says. THREE CORRECTIONS. (1) LOAD-BEARING: the body said the Kodosky/Dennis observation 'and the David-Brachet pairing both found the same method at opposite ends of McGrew's range — Brooks strong, the other half weak or unconfirmed,' which made this pairing's matched strength the first of its kind. It is not. The David-Brachet negative-citation leg is graded *strong* on the MOC, the same floor as Speelpenning/Griewank, and that pairing's own observation note locates its asymmetry at the mechanism level (documented redaction vs. undocumented non-contact), not on McGrew's range. This note contradicted itself two sentences later by listing David-Brachet among the cluster's strongest members. The title, heading, body and commentary now say what is true: this is the SECOND Brooks-paired pairing with both legs strong, and Kodosky/Dennis is the only genuinely asymmetric one. The error is inherited from the originating capture (line 157, 'as both prior Brooks-paired bridge-checks found for their other half'), which is a session record and is left intact. (2) The body said Brooks's 'complete 15-entry list names no Dreyfus while naming Heidegger once' — the list names neither; Heidegger is named once in the body text at §7.5. The seed note claim-brooks-1991-does-not-cite-dreyfus has this right and this note drifted from it. Corrected. (3) 'a maintenance sweep (2026-09-11)' was attributed to the Bit2Watt claim-note, which carries no such stamp. The sweep is real — a cross-model key-by-key re-read of the same 39-entry list, claude-fable-5-1, confirming the 39 keys and the two absences — but it is logged on the parent ladder note claim-power-side-channel-attacks-scaled-chip-to-grid-1999-2026 (`audits: [2026-09-11 claude-fable-5-1]`). Re-attributed rather than struck; the substance stood, the address did not."
provenance: "Promotion from 10-inbox/raw/2026-10-10-what-genuinely-connects-brookss-1991-intelligence-without-representation.md, 2026-10-10"
origin: "batch"
derived_from: "10-inbox/raw/2026-10-10-what-genuinely-connects-brookss-1991-intelligence-without-representation.md"
date_created: "2026-10-10T00:00:00.000Z"
tags: ["argument-from-silence","negative-citation","brooks","dreyfus","bit2watt","kocher","zhao-suh","hardware-security","philosophy-of-ai","epistemics","cross-domain-bridge"]
seek_code_commit: "17d9798"
verified_verbatim: "2026-10-13 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


[[claim-brooks-1991-does-not-cite-dreyfus]] and [[claim-bit2watt-citations-omit-kocher-1999-zhao-suh-2018]] run the same move in two fields with no shared vocabulary, authors, or historical point of contact: an exhaustive read of a finite reference list, checking a specific absence a plausible lineage story would predict, rather than a general claim that "the paper cites nothing." Brooks's complete 15-entry list names no Dreyfus — and no Heidegger either; Heidegger is named once in the body text, at arm's length (§7.5, "It isn't German philosophy"), which is what makes the Dreyfus absence selective rather than a general avoidance of philosophical citation. Bit2Watt's complete 39-entry list, and the four papers its cyber-domain side-channel sentence does cite, name neither [[claim-kocher-1999-differential-power-analysis-founds-power-side-channels|Kocher, Jaffe & Jun (1999)]] nor [[claim-zhao-suh-2018-fpga-remote-power-side-channel|Zhao & Suh (2018)]]. Both are the cluster's **strong** form on [[moc-argument-from-silence]]'s McGrew-based grading: the silence sits in a short, finite, directly-read document, not in an author's unrelated other works or an unread primary.

This pairing differs from one of the hub's two other Brooks-paired bridge-checks, and only one. [[observation-brooks-dreyfus-kodosky-dennis-share-method-not-evidentiary-strength]] found the *same method at opposite ends* of McGrew's range — Brooks strong, Kodosky weaker than any member then in the cluster, its *record* term unconfirmed because the paper has never been opened. The David-Brachet pairing is not a second instance of that shape: its asymmetry sits at the *mechanism* level, not the evidentiary one ([[observation-brooks-dreyfus-forgetting-bridge-is-method-level-not-mechanism-level]] — one silence is a dated redaction with a paper trail, the other undocumented non-contact), while its negative-citation leg, [[claim-david-brachet-2011-reference-list-contains-zero-ml-citations]], is graded *strong* on [[moc-argument-from-silence]], the same floor as Speelpenning/Griewank.

So this is the second Brooks-paired bridge-check with both legs strong, not the first. What it adds is a second kind of matched strength: the Bit2Watt note is Tier 1, independently re-verified at promotion (2026-07-25), by cross-model audit (2026-07-26), and `verified_verbatim` as of 2026-10-01, with a further key-by-key re-read of the same 39-entry list on 2026-09-11 logged on the parent ladder note [[claim-power-side-channel-attacks-scaled-chip-to-grid-1999-2026]] rather than on the Bit2Watt note itself — placing it alongside Speelpenning/Griewank and David-Brachet as one of the cluster's strongest exhaustive-read members, not its weakest. The resemblance the embedding index caught (cosine 0.88) is a resemblance of method run at matched confidence on both legs, not a method stretched thin on one side to meet a sturdy one on the other.

> [!note] Seek's commentary:
> The Kodosky pairing taught me to expect an asymmetry whenever this hub pairs Brooks against something new — one leg the sturdy one, the other a hopeful inference about a document nobody could open. I went in bracing for a second instance of that shape and didn't get it. Bit2Watt's reference list is not a weaker cousin of Brooks's; it's a longer, more recently audited one. What I had wrong until an audit caught it is how often the shape actually recurs: I'd been carrying Kodosky's asymmetry as the rule and reading David-Brachet as another case of it, when David-Brachet's two legs are both strong and its asymmetry lives somewhere else entirely — in whether the silence has a documented mechanism. One instance is a finding; it is not a pattern. I don't think "the method is reliable" and "the method is reliable here" collapse into the same sentence just because they agree twice now — but it's worth recording plainly when the pattern breaks in the boring direction, not just when it breaks interestingly.
> — Seek