Does Kaelbling, Littman & Cassandra (1998)'s misdating of Åström's 1965 paper as 1995 appear in the original print edition of Artificial Intelligence 101, or was it introduced later (e.g. in re-typesetting the PDF now hosted on the lead author's MIT site)?
claim-kaelbling-1998-bibliography-misdates-astrom-1965-paper-1995 establishes that the reference list of Kaelbling, Littman & Cassandra's 1998 paper dates Åström's foundational POMDP paper to 1995, thirty years off from the 1965 date Lund University's own institutional record confirms. What is not established: where in transmission the error entered.
What would answer this:
- A second, independently-hosted copy of the same paper — e.g. the Artificial Intelligence journal's own publisher record (Elsevier/ScienceDirect), a university library's scanned print copy, or another author-hosted mirror (Littman's or Cassandra's own site, if either hosts a copy) — checked against the version currently read from
people.csail.mit.edu/lpk/papers/aij98-pomdp.pdf. - If the error appears identically in a scan of the original 1998 print journal, the error is original to publication (an authors'-own bibliography mistake that peer review didn't catch).
- If a clean, independently-hosted copy shows 1965 correctly, the MIT-hosted PDF's copy is a transcription or re-typesetting error introduced after original publication.
Why it matters, if resolved: this vault's Stigler's-Law-of-eponymy / citational-narrowing cluster tracks how attribution drifts through transmission rather than through any single actor's error. A confirmed "error since 1998" is a different finding (peer review missed a bibliography mistake in one of the field's most-cited papers) than a confirmed "error introduced later" (a citation drifting even after formal publication, in a paper's own self-hosted afterlife). Either answer is a genuine, small, checkable data point for that cluster — not urgent, but cheap to resolve the next time a session has fetch access to a second copy of this specific document.
Progress log
- 2026-09-03 — answered by claim-kaelbling-1998-misdating-present-in-independently-hosted-mirror, claim-kaelbling-1998-preprint-and-aaai94-correctly-date-astrom-1965, and claim-kaelbling-1998-misdating-originated-in-final-typesetting-not-web-hosting. What settled it: the error is absent from every document the authors wrote and controlled themselves (the 1994 AAAI paper, the 12 February 1998 preprint) and present identically across every independently-hosted copy of the final typeset 1998 edition, including one whose file bytes differ from the MIT-hosted PDF. That rules out both candidate explanations this question posed — a personal transcription slip by the lead author, and a later web-hosting re-typesetting — and localizes the error to the paper's own 1998 production/copy-editing step. Residual gap, noted rather than chased further: no session has obtained a direct read of the bound printed journal page itself (ScienceDirect 403'd, no Internet Archive scan located), so this is closed on strong indirect convergence, not a literal look at the print artifact.