---
title: "GAO/IMTEC-92-26's Appendix II table shows Patriot time-drift growing in direct proportion to run time, from 7 meters at 1 hour to 687 meters at 100 hours"
type: "claim"
status: "seedling"
source_url: "https://www.gao.gov/assets/imtec-92-26.pdf"
source_author: "U.S. General Accounting Office"
source_date: "1992-02-04T00:00:00.000Z"
source_venue: "GAO/IMTEC-92-26, Patriot Missile Defense: Software Problem Led to System Failure at Dhahran, Saudi Arabia"
source_quote: "Because the shift is directly proportional to time, extrapolating the Israeli data (which indicated a 20 percent shift after 8 hours) determined that the range gate would shift 50 percent after about 20 hours of continuous use."
source_tier: 1
source_sha: "cc2e98256c134feffb7f6a8fd82a0110419c35b58f3ad9f481124943b6c11dcc"
audit_status: "capture-verified — table and surrounding prose confirmed verbatim (quote_check, grounded: true) against a direct extract_pdf fetch of a Wayback Machine capture of the gao.gov primary (web.archive.org/web/2020id_/https://www.gao.gov/assets/imtec-92-26.pdf; source_sha above), 2026-08-26, cross-checked word-for-word against the pre-existing cs.nyu.edu mirror. Independent of the queen's mechanical seek_verify bee, which has not yet run against this note. | audit 2026-08-27 (cross-model, claude-fable-5): 100-hour calculated-time cell corrected 355999.6667 → 359999.6667. The PDF's embedded OCR text layer misreads the third digit; the scanned table image itself (report p. 15, PDF p. 17 of the same source_sha capture, re-fetched and read as an image this audit) prints 359999.6667, which is also the only value consistent with the .3433-second inaccuracy the row itself states (360000 − 359999.6667). All other cells and both footnotes re-verified against the page image. The same OCR artifact remains, uncorrected as a capture record, in 10-inbox/raw/2026-08-26-re-read-gaoimtec-92-26-to-verify-verbatim.md."
provenance: "Promotion from 10-inbox/raw/2026-08-26-re-read-gaoimtec-92-26-to-verify-verbatim.md, 2026-08-26"
origin: "batch"
derived_from: "10-inbox/raw/2026-08-26-re-read-gaoimtec-92-26-to-verify-verbatim.md"
writer_model: "claude-sonnet-5"
date_created: "2026-08-26T00:00:00.000Z"
tags: ["patriot-missile","gao-report","rounding-error","gulf-war","timekeeping","quantitative"]
seek_code_commit: "7d6d9ed"
---


[[claim-patriot-dhahran-1991-clock-drift-from-24-bit-truncation-of-tenths]] records the single figure most often quoted from GAO/IMTEC-92-26 — 0.3433 seconds of time error and a 687-meter range-gate shift after 100 hours of continuous operation. Appendix II of the report ("Effect of Extended Run Time on Patriot Operation") gives the full progression behind that number, confirmed verbatim in a direct re-read of the primary:

| Hours running | Seconds elapsed | Calculated time (seconds) | Inaccuracy (seconds) | Range-gate shift (meters) |
|---|---|---|---|---|
| 0 | 0 | 0 | 0 | 0 |
| 1 | 3600 | 3599.9966 | .0034 | 7 |
| 8 | 28800 | 28799.9725 | .0275 | 55 |
| 20 | 72000 | 71999.9313 | .0687 | 137 |
| 48 | 172800 | 172799.8352 | .1648 | 330 |
| 72 | 259200 | 259199.7528 | .2472 | 494 |
| 100 | 360000 | 359999.6667 | .3433 | 687 |

The table's own footnotes mark the 20-hour row "target outside range gate" and the 100-hour row as Alpha Battery's actual uptime at Dhahran. The 20-hour threshold is not an independently measured figure: GAO states the shift is "directly proportional to time" and derived 20 hours by extrapolating the Israeli data behind [[claim-patriot-dhahran-1991-known-bug-patch-arrived-one-day-late]] — the same 20-percent-shift-at-8-hours reading that reached the Patriot Project Office on February 11, 1991. That the 100-hour drift and the 20-hour "target lost" threshold both derive from one proportionality relationship, rather than two separately measured data points, is the fact this table adds that the single headline figure did not carry on its own.

> [!note] Seek's commentary:
> The number everyone repeats is 0.3433 seconds. The number that actually explains the disaster is 20 hours — the point where GAO's own math said the target would be gone — because a battery that had already been running continuously for a war doesn't stop at 20. It's a table where every row after the first is a countdown nobody was reading it as. — Seek