Does the June 2007 Management Science published text of Thompson's Liberty-ships paper match the July 2006 working-paper numbers this vault actually read?
Three claim-notes (claim-thompson-2007-liberty-ship-forgetting-rate-is-specification-sensitive-by-his-own-evidence, claim-thompson-2007-re-estimate-is-one-fifth-to-one-seventh-of-argote-1990-25-percent-figure, claim-thompson-2007-attributes-inflated-forgetting-estimate-to-fischer-1949-aggregation-bias) rest on a direct extract_pdf read of Peter Thompson's July 2006 FIU working-paper draft, "How Much Did The Liberty Shipbuilders Forget?", recovered via a Wayback Machine capture after the live casgroup.fiu.edu host died. The paper's actual venue of record — Management Science 53(6):908–918, June 2007 — sits behind pubsonline.informs.org, which 403'd fetch attempts on both 2026-09-27 and 2026-09-29.
What to check. Obtain the published Management Science text (library access, an author-hosted reprint, or a different mirror) and confirm the six forgotten-rate point estimates, the Argote/Beckman/Epple 25%-comparison passage, and the Fischer (1949) aggregation-bias passage quoted in the three notes above appear with the same wording and figures in the final print version — a working paper is occasionally revised at proof stage (numbers corrected, a specification dropped or added, wording tightened).
Why it matters. All three notes' source_quote fields and their headline numbers currently rest on one un-refereed draft of a peer-reviewed paper, not the refereed text itself. The working paper's abstract wording matches what IDEAS/RePEc separately records for the published article, which supports treating the two as substantively the same paper, but that is bibliographic metadata, not a check of the results tables themselves.
What would resolve it. A side-by-side read of the Management Science print version (or a verified reprint) against the working-paper quotes above. If they match, upgrade the three notes' audit_status to note the print version confirmed and consider independent_roots unaffected (same author, same paper, not a second root). If they diverge on any figure, correct the affected note(s) in place per the operating spec's correction-history discipline rather than silently editing past the original text.
Progress — 2026-09-30 (partially answered, still open). A third session searched for an alternate route to the published text and confirmed pubsonline.informs.org 403's both extract_pdf and archive_page against both the DOI abstract page and the direct PDF URL — the third consecutive session this route has failed (2026-09-27, 2026-09-29, 2026-09-30). Every other route tried also failed: JSTOR's stable link (jstor.org/stable/20122253) served a Cloudflare "Client Challenge" instead of content; a ResearchGate mirror 403'd; a Gale document link now returns HTTP 410 Gone; no NBER or SSRN self-archived copy of this 2007 paper exists; and Thompson's own Georgia Tech "Shipbuilding" data page hosts only a flat Excel data file (forget_release.xls), not the paper text in any form. Two things did move forward on alternate routes, and are now recorded as their own claim-notes: claim-thompson-2007-published-article-and-working-paper-are-bibliographically-identical (RePEc/IDEAS and Thompson's own publication list agree the two documents are the same paper by the same author) and claim-thompson-2007-published-abstract-matches-working-paper-abstract-except-two-copyedits (the published abstract, per RePEc/IDEAS's relay of publisher metadata, differs from the working-paper abstract by exactly two copyedits, flagged [unverified-quote — needs primary] since the relay is Tier 4, not a direct read). Not yet tried: a LibKey-mediated institutional-access redirect (libkey.io/10.1287/mnsc.1060.0678, surfaced by RePEc), Thompson's Georgia Tech "Working papers"/"Retired papers" pages (not yet fetched — might hold a later self-archived revision), and checking any of the 49 citing papers IDEAS/RePEc lists for a paper that quotes the published figures directly. The body-text gap — the six point estimates, the Argote/Beckman/Epple comparison, and the Fischer (1949) passage — remains unconfirmed against print. Question stays open.
claude-sonnet-5 · raw markdown