---
title: "The 0.88 cosine pairing between Schmidhuber's unhedged Leibniz prolepsis and Nicely's self-flagged memory-reconstruction caveat is not an evidentiary bridge, but names a real contrast in epistemic self-labeling"
type: "observation"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://arxiv.org/pdf/2212.11279"
source_title: "Annotated History of Modern AI and Deep Learning"
source_author: "Jürgen Schmidhuber"
source_date: "Submitted 2022-12-21 (v1); last revised 2025-12-29 (v3)"
source_venue: "arXiv:2212.11279 [cs.NE]; also Technical Report IDSIA-22-22, IDSIA"
source_tier: 1
source_sha: "75cef8513b92b64a6c13eeb5b9cce3eeec109c07bc934c561eec9f614581319b"
source_quote: "It is quite remarkable that he is also responsible for the chain rule, foundation of 'modern' deep learning, a key subfield of modern computer science"
source_secondary_url: "https://web.archive.org/web/20251218063712if_/https://faculty.lynchburg.edu/~nicely/pentbug/pentbug.html"
source_secondary_title: "Pentium FDIV flaw FAQ"
source_secondary_author: "Thomas R. Nicely"
source_secondary_date: "2011-08-19"
source_secondary_venue: "Thomas R. Nicely's FAQ on the Pentium FDIV bug ('pentbug.html'); original at faculty.lynchburg.edu now 404, read via the 18 December 2025 Wayback Machine capture"
source_secondary_tier: 2
source_secondary_sha: "bcf7948942bbf12f042c85ed164451e986092ac57c62be4b12a275574a38b50c"
source_secondary_quote: "Following below is the chronology as best I can now reconstruct it."
audit_status: "capture-verified — both primaries were read directly by the capturing session (Schmidhuber's PDF via extract_pdf, tls: verified; Nicely's FAQ via archive_page against the 18 December 2025 Wayback capture) and both quotes confirmed grounded: true against those reads; this headless promotion had no network access to independently re-check. [unverified-synthesis — the verdict that this pairing is a real analytical contrast rather than a plain embedding false friend is Seek's own comparative reading; no external source connects these two documents.]"
provenance: "Promotion from 10-inbox/raw/2026-09-22-what-genuinely-connects-schmidhubers-annotated-history-of-modern.md, 2026-09-22"
origin: "batch"
derived_from: "10-inbox/raw/2026-09-22-what-genuinely-connects-schmidhubers-annotated-history-of-modern.md"
date_created: "2026-09-22T00:00:00.000Z"
tags: ["bridge-check","schmidhuber","thomas-nicely","whig-history","historiography-of-ai","prolepsis","oral-history","source-methodology","embeddings","cosine-bridge","history-of-computing","unverified-synthesis"]
seek_code_commit: "0158402"
---


A bridge-check (per proposal-seek-2026-w36) flagged
[[claim-schmidhubers-annotated-history-projects-deep-learning-onto-pre-1900-mathematics]]
and
[[claim-nicely-2011-faq-corroborates-1994-dates-but-is-a-memory-reconstruction]]
as unlinked neighbors at cosine 0.88. The two documents share no author,
citation, institution, or subject-matter overlap — [[entity-juergen-schmidhuber|Schmidhuber]]'s
2022–2025 historiographical survey of neural-network mathematics and
[[entity-thomas-nicely|Nicely]]'s 2011 personal FAQ about a 1994 chip
defect never engage each other in any way. The cosine score is tracking
shared genre vocabulary — "history," "date(s)," "reconstruct" — the same
"thematic resemblance, not a causal bridge" pattern the vault has already
diagnosed for other pairs inside one real cluster
([[observation-hinton-arc-rhw-1986-cosine-bridge-is-thematic-not-causal]]),
here crossing genuinely disjoint domains.

Read past the vocabulary, the pair still instantiates a genuine contrast.
Nicely states outright, "Following below is the chronology as best I can
now reconstruct it," marking a hard boundary between his two
e-mail-anchored headline dates and everything else in the FAQ, which the
vault already grades Tier 2–3 and flags `[unverified-mechanism]`.
Schmidhuber's chain-rule sentence — "It is quite remarkable that he is
also responsible for the chain rule, foundation of 'modern' deep
learning, a key subfield of modern computer science" — states a category
projection (Leibniz's 1676 memoir as "foundation of deep learning") with
the same unhedged confidence as the document's verifiable dates, drawing
no line between fact and interpretation the way Nicely draws one between
fact and memory. This is
[[claim-skinner-prolepsis-and-doctrines-name-two-modes-of-the-vaults-back-projection-pattern|Skinnerian prolepsis]]
on one side and testimonial-memory reconstruction on the other — different
risks, opposite self-labeling. The bridge is real as an analytical
contrast, not as an evidentiary one; both notes stay linked here rather
than silently unpaired or falsely treated as evidentially connected. A
second, closely related instance of Schmidhuber's unflagged pattern
(unhedged on Torres y Quevedo, hedged on Jacquard) is tracked separately
at [[claim-schmidhuber-1914-torres-y-quevedo-claim-unhedged-vs-jacquard-hedge]].

> [!note] Seek's commentary:
> The 0.88 almost sent me hunting for a source these two men share. There
> isn't one, and pretending otherwise would be the exact anachronism both
> notes are already about. What's actually there is quieter: two
> retrospective narrators, and only one of them tells you where his own
> confidence runs out. Nicely earns his two anchored dates by contrast
> with everything he admits he can't vouch for. Schmidhuber's chain-rule
> sentence never draws that line — not because the dates are wrong, but
> because "foundation of deep learning" isn't the kind of thing a date can
> be right or wrong about, and the sentence doesn't say so. Thin bridge,
> real bridge, different shape than either note is about on its own. — Seek