Does Robert Kahn's own September 1982 proposal document confirm the 'quintuple U.S. spending' figure attributed to it?

darpastrategic-computing-initiativefifth-generationrobert-kahnjapanscience-policyhistory-of-mltech-nationalismverificationunverified-quantunverified-mechanism

This capture answers question-verify-kahn-1982-proposal-quintuple-spending-figure as far as this session's access allowed, which was not far: the central question is [unverified — could not confirm or deny after search]. Robert Kahn's own document — "A Defense Program in Supercomputation from Microelectronics to Artificial Intelligence for the 1990's," an appendix to IPTO/DARPA's "FY84 President's Budget Review" of 1 September 1982, per claim-kahn-1982-proposal-cited-japan-quintuple-spending-as-sci-rationale's existing archival handle — was not located in any archive checked this session, and no source independent of Alex Roland and Philip Shiman's 2002 institutional history was found to confirm or deny that Kahn's document itself used the word "quintuple" or any equivalent multiplier for Japanese AI spending.

Claim: The "quintuple" figure has no source in this vault, or found this session, other than Roland & Shiman's 2002 secondary paraphrase

Claim type: quantitative (a "5x" multiplier). Sourcing floor: Tier 1-2 required. Not met — remains [unverified-quant — needs primary].

Roland & Shiman write that Kahn, in the proposal, argued "the rationale for spending at this level was competition with Japan," noting "the Japanese planned to quintuple U.S. spending on 'generic AI based technology.'" This is the entirety of what the vault, and this session's search, has for the figure: a named historian's paraphrase of a document neither this capture nor the note it corrects was able to read directly. No DARPA budget-justification document, no other Strategic Computing history, and no independent secondary account repeating the same multiplier turned up in this session's search (multiple web searches for "quintuple," "five times," and "5x" alongside Kahn/DARPA/Strategic-Computing/Japan returned nothing beyond mirrors of the Roland & Shiman book itself).

provenance: Roland & Shiman (2002), p. 56 and endnotes 58-59, p. 345 (Tier 2 — unchanged from the existing claim-note).

Claim: Kahn's primary document was not located in any archive checked this session, despite having a specific, citable archival identifier

Claim type: historical/archival-state. Sourcing floor: Tier 3-4 acceptable (uncontested — this describes the state of a search, not a contested historical fact).

Roland & Shiman's endnote 54 gives a precise archival handle: an appendix to IPTO, DARPA, "FY84 President's Budget Review," dated 1 September 1982. That handle was searched against DTIC's public discovery interface, archive.org, the Charles Babbage Institute's published Kahn holdings, Stanford's Feigenbaum papers finding aid, and Robert Kahn's own organization (CNRI). None returned the document. One superficially promising hit — a Stanford Special Collections folder titled "DARPA Proposal, 1982-85" in the Edward A. Feigenbaum papers, box 33, folder 22 — turned out on inspection to carry no description connecting it to Kahn, Strategic Computing, or the Japan rationale; Feigenbaum was a different Stanford-based AI researcher with his own, separate DARPA dealings in the same years, so this is recorded as a checked-and-ruled-inconclusive lead, not a match. Two DTIC/general web searches for the appendix's likely title terms and for "FY84 President's Budget Review" similarly returned unrelated FY1982 Army budget documents, not the IPTO appendix.

provenance: negative search results against DTIC, archive.org, CBI, Stanford (finding aid checked directly, see sources above), and CNRI, this session.

Claim: Robert Kahn's own recorded testimony does not resolve the question either — the interview most likely to cover it could not be read this session

Claim type: technical-mechanism (what Kahn himself says motivated the funding request). Sourcing floor: Tier 1-2 required. Not met — remains [unverified-mechanism — needs direct read].

Two Charles Babbage Institute oral-history interviews with Kahn exist: OH 192 (1990, interviewed by Judy O'Neill) and OH 158 (1989, interviewed by William Aspray). Per its own abstract, OH 158 covers "DARPA's Information Processing Techniques Office operations... DARPA contracting processes, and budget creation" — the interview most likely to touch the Strategic Computing funding rationale directly. This session's own attempts to extract it via extract_pdf and archive_page failed every time with a "Stream closed" tooling error (distinct from a paywall or 403), and a parallel attempt using different tooling failed identically. OH 192 was read this session by a route other than extract_pdf/archive_page, so under the vault's quote-provenance rule its content is not admissible as a sourced finding; informally, it was reported as confined to networking history (ARPANET, TCP/IP) with no mention of Japan or the Strategic Computing rationale, which is at least consistent with — though does not by itself establish — the idea that OH 192 is simply the wrong interview to look in.

provenance: attempted access only; neither interview is read to the sourcing floor this session.

Further leads

• Mark Stefik, "Strategic Computing at DARPA: Overview and Assessment," CACM 28(7):690-704 (July 1985) — a contemporaneous outside account (Xerox PARC), paywalled with no OA mirror found; unread this session, and even if read would be a secondary account, not primary confirmation.
• CBI oral history OH 158 (Kahn/Aspray, 1989) — abstract suggests budget-creation content directly on point; extraction failed twice this session on a tooling error worth retrying with different infrastructure, not re-searching.
• DTIC's own discovery interface was searched only by public web query, not through an authenticated DTIC account — a logged-in search might surface budget-justification documents not indexed publicly.
• Stanford Feigenbaum papers, box 33 folder 22, "DARPA Proposal, 1982-85" — description-only lookup; an in-person or fuller finding-aid check could still rule it in or out.

Entity candidates

• Alex Roland — person — co-author (with Philip Shiman) of the 2002 institutional history that is the sole source of the quintuple figure this whole branch rests on; not yet an entity in the vault despite anchoring at least one claim-note directly.
• Philip Shiman — person — the same book's co-author; same reasoning as Roland.
• Mark Stefik — person — Xerox PARC AI researcher, author of the 1985 CACM piece on Strategic Computing; a possible independent contemporaneous check on the funding rationale, unread this session.
• Robert Kahn — person — already an entity (entity-robert-kahn); flagged again because this capture specifically failed to reach his own words on the point at issue, twice, on a tooling error rather than a content problem.