Trap · reasoning move
◆ The correction nobody sees
Sounds like: “It was all over the news.”
A dramatic claim, reported fast and everywhere, becomes what people remember; the correction, days later and a fraction as loud, doesn't undo it. 'Everyone reported it' is taken as proof that it happened.
Signs
a dramatic claim within hours of an event; one original source behind many outlets; 'according to' dropped as the story travels; no correction visible where you first saw it.
Questions to ask
Who is the original source, and could they know yet? How many outlets confirmed it independently, rather than citing each other? Was it corrected, and how prominently compared with the original?
What to do
wait a day before sharing a dramatic early claim; when you've shared one that was corrected, share the correction as loudly.
Example
the al-Ahli '500 killed by an Israeli strike' headlines.
The mirror
the al-Rantisi 'list of terrorists' that was a calendar.
How it works.
In a war, the first claim about an event usually comes from one side's authorities, and outlets under time pressure pass it on, sometimes with 'according to' and sometimes without. Other outlets then cite the first ones, so a single source looks like many (see 'Citation loop'). When the claim is later corrected, the correction reaches a fraction of the audience: a large study of Twitter found that false news spread 'significantly farther, faster, deeper, and more broadly than the truth', because false stories were more novel and it was people, not bots, who spread them more — and the correction is by definition less new and less shocking than the claim. What remains in memory is the first version, often attached to an image (see 'The image does the arguing').
dataSoroush Vosoughi, Deb Roy and Sinan Aral, 'The spread of true and false news online', Science 359 (2018) 1146–1151 (2018) finding ✓
Abstract read via the Europe PMC REST API: ~126,000 stories, ~3 million people, Twitter 2006–2017; 'Falsehood diffused significantly farther, faster, deeper, and more broadly than the truth in all categories'; false news 'more novel'; bots spread true and false news at the same rate — humans spread falsehood more.
It happens in both directions, and the record is checkable.
Al-Ahli hospital, 17 October 2023: headlines worldwide reported an Israeli strike killing about 500 people, on the word of Gaza's Hamas-run authorities; the New York Times later said its coverage 'relied too heavily on claims by Hamas', and Human Rights Watch found an apparent rocket of a type used by Palestinian armed groups and called a large air-dropped bomb of the kind Israel used 'highly unlikely'. BBC News, November 2023: a presenter said Israeli forces entering al-Shifa hospital were 'targeting' medical teams and Arabic speakers, reading a Reuters report that said the force 'included' them; the BBC apologised on air and on its corrections page, saying the report 'misquoted a Reuters report' and 'fell below our usual editorial standards' (Variety). Haaretz, June 2025: the English version of a report on the aid-distribution sites said soldiers were ordered to shoot 'at' unarmed Gazans, where, as the military analyst Andrew Fox pointed out, the Hebrew had 'towards' crowds 'to distance them'; the report was cited by NPR, CNN and Reuters (The Algemeiner, 30 June 2025, a critic of the report); whether Haaretz changed the English text, and its reply, are not yet checked ⚠. From the other side: the '40 beheaded babies' reports of October 2023, which combined two separate details from an Israeli TV reporter's broadcasts into a claim she never made; the US President was 'referring to unconfirmed reports', and an Israeli official said the government had not confirmed beheadings (FactCheck.org); and the Israeli army spokesman's video from al-Rantisi hospital in November 2023 presenting a 'list of terrorists' guarding hostages that was an Arabic calendar of the days of the week, for which the army says it issued a 'prompt correction' (NBC News); whether the original video stayed up is not established by the sources read here ⚠. Open-source analysis the next day corrected both first versions. Bellingcat located the impact in the hospital's parking lot, a point that 'does not appear to be consistent with the 500, 1000 or 2000-pound bombs' Israel uses (as a PAX adviser noted). It also found that one of the images the IDF spokesman held up while saying 'no craters can be identified' 'shows what appears to be a crater'. The dead lay in a grassy area where people had likely been 'resting or sleeping' (Bellingcat, 18 October 2023).
pressNew York Times, Editors' Note on al-Ahli hospital coverage (23 October 2023), as reported by The Times of Israel (2023) incident ✓
Checked (via The Times of Israel's reproduction, 23 Oct 2023): coverage 'relied too heavily on claims by Hamas, and did not make clear that those claims could not immediately be verified'.
advocacy / viewHuman Rights Watch, 'Gaza: Findings on October 17 al-Ahli Hospital Explosion' (26 November 2023) (2023) finding ✓
Checked: 'resulted from an apparent rocket-propelled munition, such as those commonly used by Palestinian armed groups'; 'a large air-dropped bomb, such as those Israel has used extensively in Gaza, highly unlikely'; 'further investigation is needed'.
dataSoroush Vosoughi, Deb Roy and Sinan Aral, 'The spread of true and false news online', Science 359 (2018) 1146–1151 (2018) finding ✓
Abstract read via the Europe PMC REST API: ~126,000 stories, ~3 million people, Twitter 2006–2017; 'Falsehood diffused significantly farther, faster, deeper, and more broadly than the truth in all categories'; false news 'more novel'; bots spread true and false news at the same rate — humans spread falsehood more.
pressFrance 24, 'IDF found a calendar in Arabic, not a Hamas names list at hospital' (16 November 2023) (2023) finding ✓
Read (2026-09-28), Truth or Fake (16 November 2023): 'A video posted directly by Israel's Defence Forces claimed that it had found … a “list of terrorist names” in Arabic, showing each agents' rota guarding Israeli hostages under the Al-Rantisi Children's Hospital in Gaza. However, a translation of the document shows that it contains no names but instead a calendar of the days of the week.'
pressBellingcat, 'Identifying Possible Crater from Gaza Hospital Blast' (18 October 2023) (2023) finding ✓
Read (2026-09-28). Impact crater in the al-Ahli parking lot; per Marc Garlasco (PAX) 'the impact point does not appear to be consistent with the 500, 1000 or 2000-pound bombs used in Joint Direct Attack Munitions'; a grassy area with cots and blankets 'likely occupied by people who were resting or sleeping'; IDF spokesman Hagari said 'no craters can be identified', but one image he held up 'shows what appears to be a crater'; the next day IDF spokesman Conricus acknowledged the small crater.
pressNBC News, 'Information missteps have led to questions about Israel’s credibility' (November 2023) (2023) pattern ✓
Read (2026-09-28). Hagari at al-Rantisi: 'Every terrorist has his own shift'; 'Arabic speakers pointed out that the rest of the paper merely showed days of the week'; the IDF told NBC it had issued a 'prompt correction' to Hagari's calendar comment and that 'suggestions that the IDF is manipulating the media are incorrect'. Also: an Israeli spokesman's 'makeup' video, from a Lebanese film, 'had not been deleted as of Friday'.
pressThe Algemeiner, 'Haaretz Claim That IDF Was Ordered to Fire on Unarmed Gazans Refuted by Translation Discrepancies, Contradictions, and Eyewitness Accounts' (30 June 2025) (2025) ✓
Read 2026-10-02. Critics of Haaretz's 27 June 2025 report, interviewed by The Algemeiner. Says the report ‘was cited by outlets such as NPR, CNN, and Reuters’. British military analyst Andrew Fox pointed to ‘the shift in the English version of the story from soldiers firing “towards” civilians, as stated in the Hebrew original, to “at” them’; the Hebrew subheader had ‘towards’ crowds ‘to distance them’; Hebrew headline ‘Soldiers testify: IDF deliberately shoots towards Gazans near aid collection points’ vs English ‘IDF soldiers ordered to shoot deliberately at unarmed Gazans waiting for humanitarian aid’. Richard Kemp: ‘shooting towards’ is ‘quite reasonable as a means to exercise crowd control in a war zone’; Kemp also reports Gazans at GHF sites fearing Hamas. The headline's ‘Refuted’ is the paper's own verdict; Haaretz's reply is not in the article.
pressVariety, 'BBC Apologizes Over Israel-Hamas War Misreporting After Anchor Falsely Claimed IDF ‘Targeted’ Medical Staff' (November 2023) (2023) incident ✓
Read 2026-10-02. A BBC anchor twice said the Israeli military was ‘targeting people including medical teams as well as Arab speakers’ at al-Shifa, misreading a Reuters report in which an IDF spokesperson said ‘Our medical teams and Arabic speaking soldiers are on the ground to ensure that these supplies reach those in need.’ BBC corrections page, quoted: ‘This was incorrect and misquoted a Reuters report. We should have said IDF forces included medical teams and Arabic speakers for this operation. We apologise for this error, which fell below our usual editorial standards.’ The correct version was broadcast minutes later, with an on-air apology.
pressFactCheck.org, 'What We Know About Three Widespread Israel-Hamas War Claims' (October 2023) (2023) finding ✓
Read: the '40 babies beheaded' claim 'appears to be a combination of those two separate details' from the reporter's broadcasts; an unnamed Israeli official told CNN (12 Oct) the government had not confirmed beheadings; the US President was 'referring to unconfirmed reports'.
What's omitted when the correction is used as a weapon.
A correction is evidence that the outlet's process works, not only that it failed; outlets that never correct are not more accurate. And one corrected story doesn't discredit every other report from the same source: many Gaza casualty reports and many Israeli army statements have held up. The useful question isn't 'did they ever get it wrong?' (everyone did) but 'which claims got corrected, in which direction, and how fast?' On this war, the documented pattern of the biggest corrections in Western media ran mostly one way: claims from Gaza's authorities passed on before they were checked (see 'BBC' and 'The New York Times' among the organisation pages). There is also a private version: the post that quietly disappears. When Bellingcat traced 2021 footage from Syria that had been posted as 'Israeli bombing of Gaza' in October 2023, it found that 'Al-Watan has since deleted their tweet. At the time of writing, Adraee's tweet is still available'. Al-Watan is a pro-government Syrian paper; Adraee is the Israeli army's Arabic-language spokesman. A deletion without a correction tells no one what was wrong, and a post left up tells them nothing at all. Neither reaches the people who shared it.
pressBellingcat, 'Hamas Attacks, Israel Bombs Gaza and Misinformation Surges Online' (11 October 2023) (2023) pattern ✓
Read (2026-09-28). 'misinformation about incidents attributed to both sides'; the 9 Oct 'Saint Porphyrius church' video false (church: 'untouched') while geolocation 'found evidence that strikes occurred in densely populated areas in the vicinity of several schools'; a lost girl 'baselessly claimed' to be an Israeli hostage; 2021 Ariha (Syria) footage posted as Gaza by Al-Watan (deleted) and by IDF Arabic spokesman Avichay Adraee (still up at time of writing); old rocket/paraglider footage misattributed. Re-read 2026-10-13: 'Al-Watan has since deleted their tweet. At the time of writing, Adraee's tweet is still available.'
Reading the numbers: counting corrections by number, size and speed.
'Which claims got corrected, in which direction, and how fast?' can be answered for the cases on this page, and the answer depends on what is counted. By number: two corrected claims went against Israel, the al-Ahli '500 killed by an Israeli strike' headlines and the BBC presenter's misreading of a Reuters report on al-Shifa; two went in Israel's favour, the army's al-Rantisi 'list of terrorists' that Arabic speakers showed was a calendar, and the '40 babies beheaded' story that the Israeli government told CNN it had not confirmed. A fifth, critics' reading of a Haaretz translation, is disputed rather than corrected. So, on this page, the count of failed claims is two each way. Who corrected matters too: the two Israeli claims were walked back by Israeli officials, while the corrections that the outlets themselves admitted for their own coverage, at the BBC and The New York Times, concern mostly claims from Gaza's authorities repeated before they were checked, as their organisation pages record (the Times stood by its contested October 7 sexual-violence investigation). That is what the note above means by the biggest corrections running mostly one way: it describes outlets' own admitted errors, not which side's claims fail more often. By size: the al-Ahli claim attached a death toll to a single strike and ran as worldwide headlines; it is the largest of these by the claim it made, though this site has no measure of each story's reach. By speed: the army issued what it called a 'prompt correction' on the al-Rantisi calendar; The New York Times's editors' note on al-Ahli came six days after the strike (17 to 23 October 2023). Why the first version wins: false news spread 'farther, faster, deeper, and more broadly than the truth' on Twitter, in a study of about 126,000 stories, and it was people, not bots, who spread it more. The worst case for each side. For 'the media only correct against Palestinians': this page's own count is two each way, including two Israeli claims that fell apart. For 'the media are fair to Israel': the largest single false claim here, by the toll it named, ran against Israel on the word of Gaza's authorities. Holds on the evidence here: false claims from both sides were spread and corrected; the corrections reach fewer people than the claims. Open, needs reach data: whether corrections in one direction are systematically bigger. Not supported: that a correction discredits a source as a whole, or that only one side's claims fail.
What the labels (Holds, Open, Arguable …) mean
pressNew York Times, Editors' Note on al-Ahli hospital coverage (23 October 2023), as reported by The Times of Israel (2023) incident ✓
Checked (via The Times of Israel's reproduction, 23 Oct 2023): coverage 'relied too heavily on claims by Hamas, and did not make clear that those claims could not immediately be verified'.
advocacy / viewHuman Rights Watch, 'Gaza: Findings on October 17 al-Ahli Hospital Explosion' (26 November 2023) (2023) finding ✓
Checked: 'resulted from an apparent rocket-propelled munition, such as those commonly used by Palestinian armed groups'; 'a large air-dropped bomb, such as those Israel has used extensively in Gaza, highly unlikely'; 'further investigation is needed'.
pressVariety, 'BBC Apologizes Over Israel-Hamas War Misreporting After Anchor Falsely Claimed IDF ‘Targeted’ Medical Staff' (November 2023) (2023) incident ✓
Read 2026-10-02. A BBC anchor twice said the Israeli military was ‘targeting people including medical teams as well as Arab speakers’ at al-Shifa, misreading a Reuters report in which an IDF spokesperson said ‘Our medical teams and Arabic speaking soldiers are on the ground to ensure that these supplies reach those in need.’ BBC corrections page, quoted: ‘This was incorrect and misquoted a Reuters report. We should have said IDF forces included medical teams and Arabic speakers for this operation. We apologise for this error, which fell below our usual editorial standards.’ The correct version was broadcast minutes later, with an on-air apology.
pressFrance 24, 'IDF found a calendar in Arabic, not a Hamas names list at hospital' (16 November 2023) (2023) finding ✓
Read (2026-09-28), Truth or Fake (16 November 2023): 'A video posted directly by Israel's Defence Forces claimed that it had found … a “list of terrorist names” in Arabic, showing each agents' rota guarding Israeli hostages under the Al-Rantisi Children's Hospital in Gaza. However, a translation of the document shows that it contains no names but instead a calendar of the days of the week.'
pressNBC News, 'Information missteps have led to questions about Israel’s credibility' (November 2023) (2023) pattern ✓
Read (2026-09-28). Hagari at al-Rantisi: 'Every terrorist has his own shift'; 'Arabic speakers pointed out that the rest of the paper merely showed days of the week'; the IDF told NBC it had issued a 'prompt correction' to Hagari's calendar comment and that 'suggestions that the IDF is manipulating the media are incorrect'. Also: an Israeli spokesman's 'makeup' video, from a Lebanese film, 'had not been deleted as of Friday'.
pressFactCheck.org, 'What We Know About Three Widespread Israel-Hamas War Claims' (October 2023) (2023) finding ✓
Read: the '40 babies beheaded' claim 'appears to be a combination of those two separate details' from the reporter's broadcasts; an unnamed Israeli official told CNN (12 Oct) the government had not confirmed beheadings; the US President was 'referring to unconfirmed reports'.
pressThe Algemeiner, 'Haaretz Claim That IDF Was Ordered to Fire on Unarmed Gazans Refuted by Translation Discrepancies, Contradictions, and Eyewitness Accounts' (30 June 2025) (2025) ✓
Read 2026-10-02. Critics of Haaretz's 27 June 2025 report, interviewed by The Algemeiner. Says the report ‘was cited by outlets such as NPR, CNN, and Reuters’. British military analyst Andrew Fox pointed to ‘the shift in the English version of the story from soldiers firing “towards” civilians, as stated in the Hebrew original, to “at” them’; the Hebrew subheader had ‘towards’ crowds ‘to distance them’; Hebrew headline ‘Soldiers testify: IDF deliberately shoots towards Gazans near aid collection points’ vs English ‘IDF soldiers ordered to shoot deliberately at unarmed Gazans waiting for humanitarian aid’. Richard Kemp: ‘shooting towards’ is ‘quite reasonable as a means to exercise crowd control in a war zone’; Kemp also reports Gazans at GHF sites fearing Hamas. The headline's ‘Refuted’ is the paper's own verdict; Haaretz's reply is not in the article.
dataSoroush Vosoughi, Deb Roy and Sinan Aral, 'The spread of true and false news online', Science 359 (2018) 1146–1151 (2018) finding ✓
Abstract read via the Europe PMC REST API: ~126,000 stories, ~3 million people, Twitter 2006–2017; 'Falsehood diffused significantly farther, faster, deeper, and more broadly than the truth in all categories'; false news 'more novel'; bots spread true and false news at the same rate — humans spread falsehood more.
Why it works on us
- Most breaking reports turn out broadly right, so trusting the first version usually works.
- When many outlets carry the same claim, it feels independently confirmed.
- Corrections are rarely seen: they appear later, lower down, and don't trend.