sitmap.live

Corrections, Verification & Audit Log

Every report on sitmap.live carries a verification state, and states change. This page shows what we got wrong or could not stand behind, how well sourced recent events are, and each change of state as it happened.

Unofficial · verify before acting
CorrectionsWhat is marked disputed or debunked, and what was withdrawn VerificationHow well sourced the last 30 days of events are, and whether early reports held up Audit logEach recent change of an event's state, as it happened

Marked disputed or debunked

Sources contradict each other (disputed): these reports stay on the map and in this list, for the record. A report judged false or misattributed (debunked) is taken off the map (the filter “Show debunked and hidden” brings it back) and is not listed or repeated here, so that its false claim does not spread; how many there were is under Verification, below.

Verification, last 30 days

9687settled incidents
246still settling (under 12 h)
204disputed now
0debunked, still shown
0debunked and taken down
393withdrawn (the debunked included)

Incidents by verification state. An attack and the reports of its aftermath (casualties, damage, closures) linked to it are one incident, counted at the attack's state: 135 such reports are not counted again. Incidents reported in the last 12 hours are left out of the chart: corroboration takes hours, so a fresh report is unverified for want of time, not for want of sources. This says how well sourced they are, not how true: a single-source report is not a false one.

Confirmed2232%
Corroborated8118%
Single source491051%
Unverified353937%
Disputed2042%
Debunked00%

The chart counts incidents still on the map. A debunked report is taken off the map when it is judged false, so its bar counts only one that is still shown, and reads 0 even when some were debunked. No report was debunked and taken down in these 30 days. A report is debunked only when the review can point to what a claim itself says or shows (for example recycled footage, a retraction, or an impossible place or time); a doubt that cites no claim is shown as a warning, not a debunk, so debunks are rare.

Not counted: 1504 leads (aggregator reposts that name no source).

Withdrawn events (merged away, found out of scope, removed by review, or debunked) are not listed. When an event we posted on X is later disputed, debunked or withdrawn, the post gets a reply saying so.

Do early reports hold up?

Every event on the map starts as one report. These are all 4530 events first reported since 2026-10-26 (leaving out the last 12 hours, too recent to judge), by where each stands now.

615corroborated or confirmed
16rated up with no new source
54disputed or debunked
219withdrawn
2385single source
1241unverified

615 (14%) were corroborated by an independent source or confirmed, a median of 3 hours after the first report. 16 (0%) reached corroborated or confirmed only when the same reports were rated again (changed rules or ratings, or a new review), with no new source: they are not counted as having held up. 54 (1%) were contradicted: marked disputed or debunked. 219 (5%) were withdrawn as out of scope or on review. 2385 (53%) rest on one credible source, and 1241 (27%) on no credible source: anonymous or partisan channels only, however many. Neither group has been confirmed or contradicted, which does not make them wrong.

Not counted: 773 leads, reposts by aggregator channels that name no source. Nothing in them can be checked, so they are not events until a report with a source of its own joins them.

Audit log

The newest changes of state the site has recorded, newest first, in UTC, each with why it changed. Events withdrawn from the site are not named. Recording of state changes began 2026-10-26 19:13 UTC; earlier changes were not recorded.

Since 2026-10-26: 845 moves up to corroborated or confirmed on new reports or evidence; besides, 109 reclassified by changes to the rules, ratings or settings (changes to the rules); 20 by the automated review or on review; 13 re-ratings of the same reports whose cause was not recorded. 188 moves down a level.

  1. Israeli army demolitions in southern Lebanon; Katz says over 60 villages destroyedSingle source to Corroboratedrules changed (reclassification)
  2. Artillery fire reported near Mansouri, south Lebanon, attributed to Israeli forcesSingle source to Unverifiedcause not recorded
  3. Saudi civil defence says danger passed after hostile aerial threat alert in AbhaSingle source to Corroboratedrules changed (reclassification)
  4. Israeli artillery reported shelling Zawtar al-Sharqiya and nearby villages in south LebanonSingle source to Unverifiedcause not recorded
  5. Artillery fire reported in several Nabatieh villages, attributed to Israeli forcesSingle source to Unverifiedcause not recorded
  6. UKMTO reports eight projectile attacks on tankers and container ships in Strait of HormuzSingle source to Corroboratedrules changed (reclassification)
  7. Saudi-led coalition reportedly strikes Houthi mountain depots in YemenSingle source to Confirmedrules changed (reclassification)
  8. Tank fire reported south of Khan Yunis; IDF says it killed two alleged smugglersSingle source to Corroboratedrules changed (reclassification)
  9. Second person reportedly dies after explosion at Intel factory in southern IsraelSingle source to Corroboratedrules changed (reclassification)
  10. Former Syrian general Miraj Ural reportedly abducted from Mansourieh, Lebanon, MP saysSingle source to Corroboratedrules changed (reclassification)

Changes to the rules

When the rules that set a state change, stored events are rated again on the same reports. Those moves are reclassifications, not new evidence: they are counted apart above and listed here.

  1. party rule. A party's statement about its own strikes, its interceptions or the enemy's losses no longer confirms an event; a neutral body, an authority's own record or an admission against interest still does.0 up, 61 downD32
  2. strong-origin corroboration. An established or stronger origin plus any second independent one corroborates (before, the weights had to add up to 1.2); the stored weak events with two or more origins were rated again.106 up, 0 down5a3a285 strong-origin corroboration + reverify:multi

Air-traffic signals withdrawn or lowered

Drops in air traffic are machine-detected from ADS-B counts. When a signal turns out to rest on poor receiver coverage, or is rated higher than its evidence allows, it is withdrawn or its confidence lowered, and the change is listed here.

  1. KW airspace, drop detected 2026-11-02 22:35 UTCConfidence high to mediumthe usual hour here holds too few aircraft (median 19, lowest 14) for high confidence
  2. KW airspace, drop detected 2026-11-02 22:35 UTCConfidence high to mediumthe usual hour here holds too few aircraft (median 29, lowest 8) for high confidence
  3. KW airspace, drop detected 2026-10-31 20:24 UTCConfidence medium to lowthe usual hour here is too thin (median 12, lowest 4 aircraft entering) for more than low confidence
  4. QA airspace, drop detected 2026-10-30 09:57 UTCConfidence high to mediumno airspace closure notice in force covers it, and a fall in the aircraft counted entering a FIR is never high confidence alone
  5. IQ airspace, drop detected 2026-10-30 12:36 UTCWithdrawnour receivers' coverage of the Baghdad FIR (ORBB) is thin: 69% of the polls over the 7 days before saw no aircraft inside it. A fall in the aircraft counted there is not evidence of a closure

Page built 2026-11-06 20:29 UTC · Last updated 2026-11-06 20:29 UTC · Read at 2026-11-06 20:29 UTC