---
title: "Linnainmaa 1976's citation record, finally pulled — 51 dated citations across 1978–2009, then 245 in the sixteen years since 2010"
type: "claim"
status: "budding"
audit_status: "verified (Semantic Scholar Graph API queried live by DOI at capture level, 305 records in one unpaginated call, sum-check exact; Google Scholar's '1247' total is [unverified-quant — single HTML-summarization pass, needs second look]; Web of Science unpulled — no access)"
source_url: "https://api.semanticscholar.org/graph/v1/paper/DOI:10.1007/BF01931367/citations?fields=year&limit=1000"
source_author: "Semantic Scholar Graph API (Allen Institute for AI)"
source_date: "queried 2026-07-07"
source_tier: 1
source_quote: "{\"title\": \"Taylor expansion of the accumulated rounding error\", \"year\": 1976, \"citationCount\": 305}"
provenance: "Promotion from 10-inbox/raw/20260707-0931-what-are-the-year.md, 2026-07-07, queen cycle 21 — resolves question-verify-linnainmaa-citation-count (open since 2026-06-26, auth-walled in three prior runs)"
origin: "batch"
derived_from: "10-inbox/raw/20260707-0931-what-are-the-year.md"
date_created: "2026-07-07T00:00:00.000Z"
tags: ["linnainmaa","citation-history","bibliometrics","backpropagation","myth-ledger","semantic-scholar"]
drafted_in: ["2026-07-08-fifth-uncited-rediscoverer","fifth-uncited-rediscoverer"]
audits: ["2026-07-07 unknown-model","2026-07-08 claude-opus-4-8","undated claude-opus-4-8"]
---


Per-decade histogram, computed from 305 individually-dated Semantic Scholar
records (296 dated + 9 undated; sum-check matches the API's citationCount
exactly):

| Decade | Citations | Share of dated |
|---|---|---|
| 1970s (from 1978) | 3 | 1.0% |
| 1980s | 6 | 2.0% |
| 1990s | 25 | 8.4% |
| 2000s | 17 | 5.7% |
| 2010s | 67 | 22.6% |
| 2020s (through mid-2026) | 178 | 60.1% |

**What this settles for [[myth-linnainmaa-uncited-before-2010s]]:** the
literal circulating phrase ("basically never cited before the 2010s") is
falsified — a real, thin trickle exists in every decade from 1978, peaking
at 25 in the 1990s. The *gist* is vindicated with a number attached: three
full decades produced 51 citations; the deep-learning re-examination era
produced 245 in sixteen years, with 60% of the paper's lifetime citations
arriving after 2020. Also: zero recorded citations in 1986, the RHW year —
consistent with, though not a substitute for, the directly-verified
four-item reference list ([[claim-rhw-1986-reference-list-four-works]]).

**Honest edges:** Semantic Scholar indexes a curated corpus; Google Scholar
reports "1247" for the same cluster, a plausible-but-untranscription-checked
4× gap [unverified-quant]. Web of Science remains unpulled. And the
histogram doesn't yet classify the 51 pre-2010 citations by community —
whether they are numerical-analysis/AD or ML would sharpen the
scope-refinement the myth entry has been converging on
([[claim-speelpenning-1980-does-not-cite-linnainmaa]],
[[claim-werbos-1974-does-not-cite-linnainmaa]],
[[claim-linnainmaa-priority-not-paternity]]). That classification is the
natural next bee topic if the myth needs another status move.

> [!note] Seek's commentary:
> Read the shape, not just the total: 51 citations across three decades, then 245 in sixteen years, 60% of them after 2020. That's not a citation record, it's a resurrection curve — and the spike isn't a wave of new readers grasping the rounding-error math. It's the deep-learning era retroactively assembling its own genealogy, reaching back to canonize an ancestor once [[entity-backpropagation|backpropagation]] grew important enough to have a history worth disputing. Citations here are a lagging indicator of a *field's* self-importance, not of a paper's readership. The vault's own priority project rides the same wave: these notes exist because backprop matters now — which is exactly why its origins are suddenly worth pulling from a Finnish thesis.
> — Seek