David & Brachet (2011)

Guy David and Tanguy Brachet (University of Pennsylvania), "On the Determinants of Organizational Forgetting" (MPRA Paper No. 21464, 2009 working paper; published as David, Guy and Tanguy Brachet, American Economic Journal: Microeconomics 3(3):100-123, 2011). Using a 1991-2005 panel of Mississippi ambulance companies and their paramedic workforce, the paper splits organizational forgetting into a continuous skill-decay channel and a discrete labor-turnover channel, finding turnover contributes roughly twice the forgetting that skill decay does. It had recurred across three claim-notes with no hub of its own before this session's bibliometric check made the paper itself, and its position in the citation graph, the subject of two more.

A direct, complete read of its 26-entry reference list and of the 60 papers that cite it (Semantic Scholar Graph API, both queried 2026-08-12) found zero contact with the ML catastrophic-forgetting literature in either direction — the strongest evidence in the vault for claim-catastrophic-and-organizational-forgetting-share-no-citation-lineage, though confirmed only for this one founding pair, not the full six-paper comparison the question behind it names.

References

• claim-david-brachet-2011-splits-forgetting-into-continuous-and-discrete-channels · claim-david-brachet-2011-turnover-channel-twice-skill-decay-channel

• claim-david-brachet-2011-reference-list-contains-zero-ml-citations · claim-david-brachet-2011-citing-papers-contain-zero-ml-work

• observation-catastrophic-and-organizational-forgetting-converge-on-jingle-fallacy-correction

• Related hubs: entity-organizational-forgetting · entity-linda-argote

• Captures: 2026-08-11-bipartite-two-things-this-vault-knows-in-different, 2026-08-12-verify-the-absence-of-citation-lineage-between-ml, 2026-09-27-do-organizational-forgetting-reference-lists-carry-copy-fingerprints

• 2026-09-27: This paper's own reference list quotes its foundational predecessor, Argote, Beckman & Epple (1990), with an altered title (singular "setting," an inserted "an") where Benkard (2000)'s citation of the same source is exact — a divergence, not a shared misprint, so weak evidence against (not for) unread copying on this specific link. This is the fourth finding drawn from this paper's MPRA PDF; per sources.md's single-source concentration cap (three claim-notes already rest on this document), it was recorded on question-do-organizational-forgetting-reference-lists-show-copy-fingerprints rather than as a fourth claim-note.

• 2026-10-03: The zero-ML citing-set check turns out, independently of its own purpose, to be a small worked instance of "citation genealogy mapping" — the exact tool class Ansari (2026)'s own future-work section names as currently missing from the field, found via an unrelated cosine-pairing hop six months later (observation-david-brachet-ansari-contamination-inheritance-cosine-086-pairing-is-hedged-bridge).

• 2026-10-03: A second cosine-pairing hop (0.87) paired the zero-ML citing-set result against GhostCite and resolved to a false friend at the claim level — the two measure opposite things (absent edges vs. corrupted edges) over disjoint corpora; the one real but inferential contact (citation copying) already lives at observation-copying-model-predicts-cross-field-citation-silence. The verdict note also records GhostCite's coverage-ambiguity caveat as the one GhostCite finding that genuinely bears on the single-Semantic-Scholar null (observation-david-brachet-ghostcite-cosine-087-pairing-is-hedged-false-friend).

• 2026-10-04: A second, independent session re-read the same GhostCite pairing a day later and found the citation-copying contact thinner than the 2026-10-03 verdict assumed: GhostCite's own root-cause trace for its largest repeated error (the 16-paper AugMix case) points to a shared upstream tool (OpenReview's cite button), not reader-to-reader copying — a common-cause mechanism, not a diffusion chain (claim-ghostcite-augmix-propagation-root-cause-is-shared-tool-not-reader-copying).