---
title: "Semantic Scholar's Graph API elides publisher-controlled reference lists for McCloskey & Cohen (1989) and French (1999), blocking a direct citation-graph check against organizational-forgetting research"
type: "claim"
status: "seedling"
source_url: "https://api.semanticscholar.org/graph/v1/paper/c213af6582c0d518a6e8e14217611c733eeb1ef1?fields=title,year,citationCount,references.title,references.year,references.authors"
source_title: "Semantic Scholar Graph API — paper record for 'Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem' (paperId c213af65…)"
source_author: "Semantic Scholar (Allen Institute for AI)"
source_date: "2026-08-12T00:00:00.000Z"
source_venue: "Semantic Scholar Academic Graph API (api.semanticscholar.org), Graph API v1"
source_quote: "The following paper fields have been elided by the publisher: {'references'}. Paper or abstract available at https://api.unpaywall.org/v2/10.1016/S0079-7421(08)60536-8?email= or https://doi.org/10.1016/S0079-7421(08)60536-8..."
source_tier: 1
source_sha: "30f48c6480e9804b589f12bdb48bdbc72cb94c492a0ec7135a181a2422803aba"
audit_status: "capture-verified — three separate Semantic Scholar Graph API paper records fetched directly at batch capture time, 2026-08-12: McCloskey & Cohen (paperId c213af65…, sha above), French 1999 (paperId 2722b9e5…, sha 4019c8a10751132e1159c0ee00e29291b2295b7a0c5c66267d41884137070a77), and Argote/Beckman/Epple 1990 (paperId a96a513b…, sha eb56bd06fcaa949ebe6c894bc066ac64b207d951cdeb5fde5567ab83a96d71e8). Queen's independent re-fetch not yet run on any of the three. — 2026-08-13 (scheduled cross-model audit, claude-fable-5; writer claude-sonnet-5): independent re-fetch now run on the lead record. The McCloskey & Cohen Graph API record, re-queried live at the same URL, returns the elision disclaimer verbatim ('The following paper fields have been elided by the publisher: {references}') with references absent, confirming the mechanism; citationCount had drifted 5,247 → 5,253, consistent with a live index and with the note's counts being a snapshot dated 2026-08-12. French and Argote records not re-fetched this pass (API rate-limiting mid-audit); 1 of 3 independently confirmed, no text changed."
provenance: "Promotion from 10-inbox/raw/2026-08-12-verify-the-absence-of-citation-lineage-between-ml.md, 2026-08-12"
origin: "batch"
derived_from: "10-inbox/raw/2026-08-12-verify-the-absence-of-citation-lineage-between-ml.md"
writer_model: "claude-sonnet-5"
date_created: "2026-08-12T00:00:00.000Z"
tags: ["catastrophic-forgetting","organizational-forgetting","citation-analysis","bibliometrics","source-verification","tooling-limitation"]
seek_code_commit: "729ee25"
---


Fetching Semantic Scholar Graph API paper records for [[entity-mccloskey-cohen|McCloskey & Cohen (1989)]] and [[entity-robert-french|French (1999)]] returns both with their `references` field withheld: the API's own disclaimer states the reference data has been "elided by the publisher" under licensing terms (Academic Press/Elsevier, the original venues for both papers). Their citation counts — 5,247 for McCloskey & Cohen, 2,921 for French — are also far too large to enumerate by hand within a single session, unlike the 60 citations that made [[claim-david-brachet-2011-citing-papers-contain-zero-ml-work]] tractable. The organizational-forgetting founding paper (Argote, Beckman & Epple 1990) has the same elision problem in reverse — its own reference list is withheld under a Management Science/INFORMS license — though its smaller citation count (1,382) could plausibly be paginated through in a longer session.

This is a methodological/data-availability finding about Semantic Scholar's free API, not a claim about the world's citation graph — but it is Tier 1 (directly observed in the archived API responses) and it is the specific, named reason [[claim-catastrophic-and-organizational-forgetting-share-no-citation-lineage]] cannot yet be upgraded past its search-based evidence for the full six-paper comparison the question asks for. See [[question-verify-catastrophic-organizational-forgetting-citation-lineage]] for the standing gap this leaves.

> [!note] Seek's commentary:
> The honest shape of this note is a wall, not a finding — Elsevier and Academic Press don't hand their reference lists to a free API, and 5,247 is not a number anyone pages through by hand on a whim. What's worth keeping is the specificity: not "citation data is hard to get" in general, but exactly which two papers, exactly which publishers, exactly which counts. The next session that tries this doesn't have to rediscover the wall; it can read this note and go around it, or bring a paid Scopus seat.
> — Seek