Did GAO/NSIAD-92-340 find that only about 9% of claimed Patriot Scud kills met its 'strongest evidence' standard?

patriot-missilegulf-wargao-reportstatisticsmilitary-engineeringverification-methodsource-verification

This capture directly resolves the open question at question-verify-gao-nsiad-92-340-patriot-9-percent-strongest-evidence — flagged 2026-07-11 as an [unverified-quant] figure seen only in secondary quotes, with the primary GAO PDF noted as blocked to automated fetch. That block persisted this session: extract_pdf and archive_page against https://www.gao.gov/assets/nsiad-92-340.pdf both returned HTTP 403 Forbidden on repeated attempts, and WebFetch downloaded the raw PDF bytes but could not extract readable text from it (structural/binary content only). The GAO product page at gao.gov/products/nsiad-92-340 also 403'd to archive_page. No Wayback Machine snapshot of the PDF exists (archive.org/wayback/available returned an empty result). The best available route to citable primary text was Justia's GAO Reports mirror, which republishes the GAO's own report abstract verbatim (confirmed by cross-checking its wording against a second, independent secondary account — the Washington Post's contemporaneous headline and a House committee report, below) — and a House Government Operations Committee report that independently corroborates the figure. Both are cited below as Tier 2 (mirrors of primary government text, not the primary venue itself, per the public-shelf rule to prefer the original venue — attempted repeatedly and blocked).

Claim: GAO found "strong evidence" supported a warhead kill in only about 9% of the Patriot-Scud engagements the Army claimed as kills

The GAO report's own abstract, as republished by Justia, states: "strong evidence exists for about 9 percent of the engagements that the Army believes resulted in warhead kills, including a ground damage report that a Scud had been recovered with Patriot fragments or fragment holes and radar tracking evidence that the Scud experienced a dramatic slowdown and that debris was present after a Patriot intercept." The same abstract opens: "About nine percent of Patriot launches produced strong evidence, such as disabled Scuds embedded with Patriot fragments, that the target was destroyed." This directly confirms the core claim in the topic question: roughly 9% of the engagements the Army counted as warhead kills met GAO's strongest evidentiary bar.

• source_url: https://gao.justia.com/department-of-defense/1992/9/operation-desert-storm-nsiad-92-340
• source_sha: e910c7a27a751954c2612e6554f0ee142063350edd6ddd2d6ad4a5996f11bcd5
• source_author: U.S. General Accounting Office
• source_date: 1992-09-22
• source_title: "Operation Desert Storm: Data Does Not Exist to Conclusively Say How Well Patriot Performed" (GAO/NSIAD-92-340)
• source_venue: U.S. General Accounting Office, Report to Congressional Requesters (Justia GAO Reports mirror; original gao.gov PDF/product page blocked to fetch tools this session)
• source_tier: 2 — quantitative claim, clears the Tier 1–2 floor; would be strengthened to Tier 1 by a direct read of the gao.gov PDF, which remains blocked.
• Quote verified against the archived page text via quote_check (grounded: true).

Claim: GAO drew a two-tier evidence structure — "strong" (~9%) versus "lesser" (~16%) — and explicitly said neither tier proves destruction beyond the strong-evidence subset

The same GAO abstract continues: "Another 16 percent of the engagements produced less convincing evidence of success, such as radar tracking data showing that the Patriots came close to the Scuds. This evidence does not prove, however, that the Patriots either destroyed or diverted the Scuds." This is the definitional backbone of the "strongest evidence standard" referenced in the topic question: GAO explicitly split the Army's claimed-kill engagements into a strong-evidence tier (~9%) and a weaker, non-probative tier (~16%), leaving roughly three-quarters of claimed engagements without even the weaker tier of support.

• source_url: https://gao.justia.com/department-of-defense/1992/9/operation-desert-storm-nsiad-92-340
• source_sha: e910c7a27a751954c2612e6554f0ee142063350edd6ddd2d6ad4a5996f11bcd5
• source_author: U.S. General Accounting Office
• source_date: 1992-09-22
• source_title: "Operation Desert Storm: Data Does Not Exist to Conclusively Say How Well Patriot Performed" (GAO/NSIAD-92-340)
• source_venue: U.S. General Accounting Office, Report to Congressional Requesters (Justia GAO Reports mirror)
• source_tier: 2 — quantitative + definitional/technical-mechanism claim (what counted as "strong" vs. "lesser" evidence); clears the Tier 1–2 floor required for technical-mechanism claims.

Claim: A House Government Operations Committee investigation independently reached the same ~9% figure using the Army's own methodology

House Report 102-1086 ("Performance of the Patriot Missile System," Activities of the House Committee on Governmental Operations, 102nd Congress, 1991–1992, pp. 179–188), summarizing an oversight hearing held April 7, 1992 at which GAO, CRS, and outside analysts testified, states: "Independent review of the evidence in support of the Army claims reveals that, using the Army's own methodology and evidence, a strong case can be made that Patriots hit only 9 percent of the Scud warheads engaged, and there are serious questions about these few hits." The report separately concludes: "The strongest evidence in the Army assessment indicates that the 158 Patriots fired during the war destroyed a few Scud warheads, although there are doubts about these." This is a second, differently-authored government document converging on the same ~9% figure — GAO's Richard Davis testified at the same April 1992 hearing this report summarizes, so the two documents are not fully independent (they share GAO's analysis as an input), but the House committee's report reflects its own subcommittee investigation and additional testimony (CRS, MIT, CSIS, MITRE, Israeli analysts) beyond GAO's alone.

• source_url: https://www.globalsecurity.org/space/library/congress/1992_r/patriot.htm
• source_sha: c611ae68d2a4a135238fac593772af8c10c21c32543fcd42ac455d00ff089a69
• source_author: U.S. House of Representatives, Committee on Government Operations, Legislation and National Security Subcommittee
• source_date: 1992
• source_title: "Performance of the Patriot Missile in the Gulf War" (House Report 102-1086, pp. 179–188)
• source_venue: U.S. House of Representatives, Committee on Government Operations, Report 102-1086 (mirrored by GlobalSecurity.org; no digitized copy located on govinfo.gov or congress.gov for this pre-1995 report as of 2026-08-07 — GlobalSecurity.org appears to be the only accessible full-text copy found this session)
• source_tier: 2 — quantitative claim from a Congressional committee report, accessed only via mirror; clears the floor at Tier 2.
• Quote verified against the archived page text via quote_check (grounded: true).

Claim: GAO's underlying critique was that the Army's own supporting documents (a hand-compiled spreadsheet and a limited ground-damage survey) did not support the claimed success rate — explaining the mechanism behind the low "strong evidence" share

The same House report quotes GAO testimony directly on why the Army's original 70%-success claim collapsed under review: "The conclusion we reached was that the data the Army used to support their original assessment — that the Patriot's performance in the Gulf was about 70 percent successful — just could not be supported by the data that they said they used to come up with that assessment." GAO's specific critique of the Army's TSM (Training and Doctrine Command Systems Manager) spreadsheet: "Our review of the [TSM] spreadsheet and several binders of supporting notes showed that the spreadsheet was actually an inaccurate summary of information obtained through telephone calls to various units in the Gulf... many gaps and inconsistencies existed between the data presented on the spreadsheet and the supporting records. For example, there were several instances in which Patriot operators reported destroying more Scud warheads than there were missiles launched." GAO also found the Army's Ballistics Research Laboratory (BRL) ground-damage report "included information on only about one-third of the Saudi engagements, although the Project Manager's assessment cited it as a source for all engagements," and that its site visits "were always made days or weeks after an impact when craters had often been filled and missile debris removed." This is the mechanism claim explaining why so few claimed engagements could clear GAO's strongest-evidence bar: the Army's own source data was internally inconsistent and its ground-damage survey covered only a fraction of cases.

• source_url: https://www.globalsecurity.org/space/library/congress/1992_r/patriot.htm
• source_sha: c611ae68d2a4a135238fac593772af8c10c21c32543fcd42ac455d00ff089a69
• source_author: GAO testimony (Richard Davis, Director, Army Issues), as quoted verbatim in the House committee report
• source_date: 1992 (testimony given 1992-04-07; report published 1992)
• source_title: "Performance of the Patriot Missile in the Gulf War" (House Report 102-1086, pp. 179–188)
• source_venue: U.S. House of Representatives, Committee on Government Operations, Report 102-1086 (GlobalSecurity.org mirror)
• source_tier: 2 — specific technical-mechanism claim (what specifically undermined the Army's data); clears the Tier 1–2 floor via direct quotation of GAO's own testimony as reproduced in the committee report.

Further leads

• The Washington Post, "New Study Cuts Patriot Missile Success Rate to 9 Percent" (1992-09-30) — headline corroborates the same figure and appears to describe the same GAO report, but the article body could not be fetched this session (WebFetch returned 403; archive_page timed out twice). Worth a dedicated retry — URL: https://www.washingtonpost.com/archive/politics/1992/09/30/new-study-cuts-patriot-missile-success-rate-to-9-percent/468b0d2d-ddf3-4d90-a2f5-2379b6bf7175/ — [unsourced beyond headline — needs direct read].
• Dr. Theodore Postol's Congressional testimony (quoted in House Report 102-1086) that a Patriot must detonate within 5–10 meters of a target for a reasonable kill probability, and that "a 30 meter miss distance is nine times less likely to damage the target than a 10 meter miss and a 100 meter miss is 100 times less likely" — a specific technical-mechanism/quantitative claim currently sourced only via a secondary mirror of testimony; needs Postol's own primary writing (e.g. the Lewis & Postol Science & Global Security paper already cited in claim-lewis-postol-1993-no-video-evidence-of-patriot-scud-warhead-kill) to clear the floor independently.
• Reuven Pedatzur (Ha'aretz / Jaffee Center) testified that Israeli military officers credited the Patriot with hitting only one Scud warhead in Israel, with perhaps two others partially damaged — a separate, Israel-specific figure distinct from GAO's Saudi-focused ~9% finding; not yet independently sourced.
• Per House Report 102-1086, the Army paid Raytheon (the Patriot's manufacturer) $520,000 for its own post-war Patriot performance analysis, and Raytheon personnel were embedded in the Army's evaluation team — a conflict-of-interest angle relevant to why initial Army claims were inflated; not yet its own claim-note.
• gao.gov's PDF assets (and at least one product page) returned HTTP 403 Forbidden to every automated-fetch route tried this session (extract_pdf, archive_page, and effectively to WebFetch's text extraction) — worth flagging as a standing access obstacle for any future GAO-report research, distinct from a safety concern.
• This capture also bears on the "strongest evidence" framing already discussed in claim-patriot-army-effectiveness-claim-collapsed-from-96-to-52-percent, which held the ~9% figure open pending verification; that note can now be updated to reflect this capture's sourcing.

Safety flags

None. No addressed-to-AI language, override language, claimed authority, tier self-assignment, file-system instructions, credential requests, or urgency framing was observed on any page read this session (gao.gov product page, Justia GAO Reports mirror, GlobalSecurity.org congressional-report mirror, Washington Post article listing). No tls:"unverified" provenance was recorded on any successful archive_page fetch.

Entity candidates

• Theodore Postol — person — MIT physicist; his and Lewis's 1991–92 video-evidence analysis of Patriot intercepts was the original independent challenge to Army kill claims that both the April 1992 House subcommittee hearing and GAO's September 1992 report followed and cited — the foundational figure this capture's investigations are measured against, flagged first per the standing instruction to catch the older figure, not just the report's own authors.
• George N. Lewis — person — MIT co-author with Postol on the video-evidence analysis feeding into the same hearing and report.
• Richard Davis — person — GAO Director, Army Issues; led the GAO investigation behind NSIAD-92-340 and testified before the House subcommittee.
• Steven Hildreth — person — Congressional Research Service senior analyst who independently corroborated GAO's findings in the same April 1992 testimony.
• Reuven Pedatzur — person — Israeli defense analyst (Ha'aretz, Jaffee Center for Strategic Studies) who testified that Israeli military officers credited Patriot with only one confirmed Scud warhead kill.
• Peter D. Zimmerman — person — CSIS analyst who defended Patriot's performance and argued video evidence was inherently misleading due to frame-rate effects.
• Charles Zraket — person — former MITRE Corp. president who argued ground-damage studies were the most valid method of assessment and supplied the "no catastrophic damage" common-sense argument.
• Raytheon Co. — organization — Patriot's manufacturer; embedded personnel in and was paid by the Army for the post-war performance analysis under scrutiny.
• U.S. General Accounting Office (GAO) — organization — author of NSIAD-92-340, the primary subject of this capture.
• House Committee on Government Operations, Legislation and National Security Subcommittee — organization — produced the corroborating Report 102-1086 and held the April 1992 oversight hearing.
• Gen. Jay Garner — person — U.S. Army officer who testified and presented the Army's revised (52%/40%-high-confidence) Patriot effectiveness assessment.