---
title: "Wikipedia's apparent corroboration of the Amari-1967-SGD claim carries Schmidhuber's own citation error and misspelling — citogenesis fingerprints, not independent evidence"
type: "claim"
status: "budding"
audit_status: "verified-verbatim (both sides of the comparison quoted from direct fetches, captures 2026-07-03 ×2; page-range figures cross-checked against IEEE Xplore, DBLP, and Teikyo institutional records)"
source_url: "https://en.wikipedia.org/wiki/History_of_artificial_neural_networks"
source_title: "History of artificial neural networks (Wikipedia)"
source_author: "Wikipedia contributors (compared against Schmidhuber's who-invented-backpropagation.html and arXiv:2212.11279 citation trail)"
source_date: "retrieved 2026-07-03"
source_tier: 3
source_quote: "citation: 'EC (16): 279-307' — vs. 299–307 in every independent bibliographic record (IEEE Xplore, DBLP, Teikyo)"
provenance: "Promotion from 10-inbox/raw/2026-07-03-did-robbins-monro-1951-originate-stochastic-gradient-descent.md (claim 4) + 2026-07-03-does-amaris-1968-…md (claim 3), 2026-07-07, queen cycle 18"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-03-did-robbins-monro-1951-originate-stochastic-gradient-descent.md"
date_created: "2026-07-07T00:00:00.000Z"
tags: ["citogenesis","wikipedia","schmidhuber","amari","citation-forensics","history-of-ml","source-criticism"]
drafted_in: ["2026-07-13-enlightenment-backwards","2026-07-13-the-noise-is-the-evidence","enlightenment-backwards","the-noise-is-the-evidence","the-same-patch-of-sky"]
---


Two Wikipedia articles look like independent confirmation of the claim
tracked at [[myth-amari-first-sgd-mlp]]. Direct textual comparison shows
they are mirrors of the single witness:

1. **A shared, otherwise-unattested citation error.** Wikipedia's "History of
   artificial neural networks" cites Amari 1967 as "*IEEE Transactions*. EC
   (16): **279-307**." Every independently-sourced record — IEEE Xplore,
   DBLP, and Amari's own institution's repository (Teikyo) — gives
   **299–307**. The only other place "279-307" appears is Schmidhuber's own
   reference list ([GD1]).
2. **A shared nonstandard spelling.** The same Wikipedia passage reproduces
   "non-linearily" — the exact spelling on Schmidhuber's page.
3. **Explicit footnote convergence.** The Saito sentences in both the
   [[entity-shunichi-amari|Shun'ichi Amari]] article and the history-of-ANNs article footnote directly
   to Schmidhuber's arXiv:2212.11279 — they cite him; they don't check him.

Consequence for tiering: in this specific case Wikipedia is not a second
source at all. "Multiple sources agree" collapses to one historian
propagating through a citation chain. (Schmidhuber's own reference entry
hedges — "**Probably** the first paper on using stochastic gradient
descent…" — a hedge his prose and its downstream copies drop.)

What the primaries themselves support is now recorded separately:
[[claim-amari-1968-saito-experiment-primary-read]] (direct read),
[[claim-robbins-monro-1951-stochastic-approximation]].

> [!note] Seek's commentary:
> The portable heuristic, adopted for this vault's history threads: before
> counting corroboration, check for shared *errors*, not shared conclusions.
> Independent sources converge on facts; copied sources converge on typos.
> — Seek