Trap Β· reasoning move
β Proxy presented as outcome
Sounds like: βIt's been confirmed. / The numbers prove it.β
A warning, projection, model or indirect indicator is presented as if it were a measured outcome.
Signs
'up to', 'could', 'projected', 'at risk' in the source becoming 'are', 'have', 'confirmed' in the headline; no denominator; no evidence level.
Questions to ask
Is this measured, estimated, modelled or projected? What would the outcome look like if it were true β and did it happen? What does the source's own fine print say?
Example
'famine confirmed' when mortality was inferred, and the malnutrition evidence came by the 15% arm-circumference route that the IPC's own manual reserves for its lower-evidence category, 'Famine Likely' (see 'Israel is starving Gaza β famine has been confirmed.').
The mirror
'the famine was fake' because it was averted β ignoring the real hunger and deaths. Example C: an aerial photo of a destroyed neighbourhood offered as proof of genocide, where the photo shows destruction, not who was inside or why. Example D (mirror): 'it was all Hamas infrastructure' offered as proof that nobody was harmed.
What it is.
A projection ('could reach'), a proxy (arm-circumference instead of weight, a multiplier instead of a count) or a model is a legitimate tool β until a headline turns it into 'confirmed'. Then data-illiterate readers take a scenario as a fact, and anyone questioning it is told they are denying the suffering. The honest test is later: did the outcome occur? The 2024 Lancet letter's '186,000' (a 4:1 multiplier the authors called 'conservative', per France 24), and the 2025 famine classification whose mortality element was inferred rather than measured, are examples.
pressFrance 24, ''More than 186,000 dead' in Gaza: How credible are the estimates published on The Lancet?' (11 July 2024) (2024) finding β
Read 1 Oct 2026. Quotes the letter (published 5 July 2024, correspondence section): 'it is not implausible to estimate that up to 186,000 or even more deaths could be attributable to the current conflict in Gaza'; a 'conservative estimate' of four indirect deaths per direct death applied to the Health Ministry's 37,396 deaths of 19 June; the ratio drawn from a 2008 Geneva Declaration Secretariat report ('between three and 15 times as many people die indirectly'); 'the letter's signatories did not say why they chose four'; 'Despite it not being either a report or scientific study β¦ has made global headlines'. The letter it quotes: Khatib, McKee & Yusuf, 'Counting the dead in Gaza: difficult but essential', The Lancet 404 (2024), doi:10.1016/S0140-6736(24)01169-3 (PubMed 38976995; not open in Europe PMC; thelancet.com 403 β not yet read).
Destruction as a proxy: 'look at the ruins, so many must have died, so it must be genocide'.
Visible destruction is often used as a double proxy: ruins stand in for the death toll, and the death toll stands in for intent. Each step needs its own evidence. First, how deadly destruction is depends on who was inside and how buildings came down. The UN satellite centre counted about 81% of all structures in Gaza affected by October 2025, 123,464 of them destroyed; not all of it came with deaths: the army also demolished buildings in areas it had cleared (more than 2,500 after the October 2025 ceasefire alone, which Israeli officials tie to tunnels and booby-trapped buildings and a Gazan official quoted by the New York Times said had 'no security justification'), which destroys homes without killing their residents; what share of the total such demolitions make up is not measured here β . Comparisons show how loose the link is. In Mosul, the same UN satellite centre counted 19,888 affected structures by August 2017, about 4,773 of them destroyed and around 7,620 of the affected ones in the Old City (a preliminary analysis), and AP's investigation counted 'between 9,000 and 11,000 people' killed; in Raqqa, more than 5,000 bodies of 'civilians, prisoners, soldiers, children and fighters on both sides' had been recovered from mass graves by September 2019 (VOA), so the toll there is not a single civilian number. Destruction and death tolls track each other loosely, depending on population, evacuation and whether people could leave. In Gaza civilians were, in the UN's words in October 2023, 'trapped inside the Gaza Strip' and 'with nowhere safe to go' (UN News), and Jordan's king called taking in refugees from Gaza, for Jordan and Egypt, 'a red line' (as reported by FDD), which raises exposure. Second, a death toll, however large, is evidence about conduct and proportionality; intent is a separate question (see 'Israel is committing genocide in Gaza.'). The mirror is the same shortcut reversed: 'the buildings were empty' or 'it was all Hamas infrastructure' waves away the real deaths and the destruction of homes, hospitals, schools and water systems, which matter whatever label is used; whether that destruction was lawful is a legal question: the Rome Statute makes 'extensive destruction and appropriation of property, not justified by military necessity and carried out unlawfully and wantonly' a war crime, and likewise destroying enemy property unless 'imperatively demanded by the necessities of war' (art. 8); whether any destruction in Gaza meets that test is for investigators and courts, and no ruling on it is sourced here β .
dataUNOSAT, Gaza Strip Comprehensive Damage Assessment (11 October 2025) (2025) aggregate β
Checked 2026-09-29 against the UNOSAT map PDF itself (the unosat.org page is JavaScript-only and HDX served a challenge; read via a copy of the official PDF hosted at https://goodshepherdcollective.org/data/unosat/damage/gaza-damage-2025-10-11.pdf, an advocacy group's mirror of the UN product). Imagery 11 October 2025, published 31 October 2025: 'approximately 81% of all structures in the Gaza Strip are damaged'; 123,464 destroyed and 17,116 severely damaged; 198,273 damaged buildings in all; an 18% increase in destroyed structures since the 8 July 2025 assessment. Counts damage to structures, not deaths or who caused them.
dataAP investigation on the Mosul death toll, republished by Al Jazeera ('More than 9,000 killed in battle for Mosul', 20 December 2017); and Airwars' Mosul page (2017) aggregate β
Re-read 1 Oct 2026: 'Between 9,000 and 11,000 people were killed in the nine-month battle to recapture the Iraqi city of Mosul β¦ an Associated Press (AP) investigation has found'; 'around a third β¦ died in bombardments by the US-led coalition or Iraqi forces. Another third were killed in ISIL fighters' final frenzy of violence'; the coalition 'acknowledges responsibility for only 326'. Airwars' Mosul page (read 2026-09-29): the coalition 'used over 29,000 bombs, missiles and rockets'. Destruction figures on the page come from UNOSAT (separate card, 2026-09-30); the earlier unsupported Raqqa figure was withdrawn (2026-09-29/2026-09-29).
pressYnet, ''Israel is still demolishing Gaza' despite ceasefire, New York Times reports' (on the NYT's satellite analysis, two months into the October 2025 ceasefire) (2025) incident β
Read 1 Oct 2026 (the Times of Israel liveblog with the same report is walled). 'Citing satellite imagery from Planet Labs, the liberal U.S. daily reported that Israel has demolished more than 2,500 buildings across the Gaza Strip since the ceasefire began'; 'Israeli officials assert that these demolitions are related to the ongoing effort to dismantle Hamas' extensive tunnel network and to detonate booby-trapped buildings'; a Gazan official quoted by the NYT: 'There is no security justification for what it's doing.'
pressVOA, 'Syrian authorities find 5,000 bodies in Raqqa mass graves' (4 September 2019) (2019) aggregate β
Read 2026-09-29 (US government-funded broadcaster). 'Since the beginning of 2018, more than 5,000 bodies have been uncovered in and around Raqqa, all believed to be dumped in mass graves by Islamic State militants'; 'They have found bodies of civilians, prisoners, soldiers, children and fighters on both sides.'
dataUNITARβUNOSAT, 'Damage assessment of Old City, Mosul, Ninawa Governorate, Iraq β Imagery analysis: 4 August 2017' (published 27 November 2017), ReliefWeb (2017) aggregate β
Read 2026-09-30. 'UNITAR - UNOSAT identified a total of 19,888 affected structures within the city. Approximately 4,773 of these were destroyed, 8,233 severely damaged and 6,882 moderately damaged β¦ Around 7,620 of the total affected structures are found within the Old City'; 'This is a preliminary analysis and has not yet been validated in the field.' Counts, not a share of all buildings.
pressUN News, 'Gaza: Nowhere to go, as humanitarian crisis reaches "dangerous new low"' (October 2023) incident β
Read 2026-10-02. After Israel's order for the population to leave the north ('Some 1.1 million people'): Dujarric, 'essential to provide humanitarian access for civilians trapped inside the Gaza Strip'; Griffiths: 'one of the most densely populated places on the planet', people 'with nowhere safe to go'; WHO: 'impossible to evacuate vulnerable hospital patients from the north'.
advocacy / viewFoundation for Defense of Democracies, Flash Brief 'Egypt, Jordan, and Other Arab Governments Reject Gazan Refugees' (17 October 2023) (2023) incident β
Read (2026-09-28). King Abdullah (17 Oct 2023): no refugees from Gaza in Jordan or Egypt, 'That is a red line'; Turkey and Qatar 'refusing to take refugees even as both countries express support for the Palestinians'. Jonathan Schanzer: Egypt 'does not want to alter facts on the ground that might weaken Palestinian claims on the territory'. Richard Goldberg: 'Why is it that Israel cares more about saving Palestinian lives than the entire Arab world?'
law / courtRome Statute of the International Criminal Court, art. 8(2)(b)(xxiii) (ICC text) (1998) β
Read 30 Sept 2026: '(xxiii) Utilizing the presence of a civilian or other protected person to render certain points, areas or military forces immune from military operations'. Art. 8 also lists, read 2026-10-02 on the UN Office of Legal Affairs text (legal.un.org): (2)(a)(iv) 'Extensive destruction and appropriation of property, not justified by military necessity and carried out unlawfully and wantonly'; (2)(b)(xiii) 'Destroying or seizing the enemy's property unless such destruction or seizure be imperatively demanded by the necessities of war'.
Reading the numbers: a projection checked against the measurement that came later.
The card's test, 'did the outcome occur?', can be run on the '186,000'. The projection: a 2024 letter in The Lancet said 'up to 186,000 or even more deaths could be attributable' to the war, by applying 'four indirect deaths per direct death' to the Health Ministry's 37,396 deaths of 19 June; 37,396 Γ 5 = 186,980 (this page's arithmetic). The ratio came from a 2008 report's range of 'between three and 15', and the signatories 'did not say why they chose four'. The measurement: a household survey published in The Lancet Global Health in 2026 estimated 75,200 violent deaths (63,600β86,800) up to early January 2025, and 8,540 excess non-violent deaths. Measured that way, indirect deaths were about 0.11 per violent death, roughly one for every nine, not four for every one; the total, 83,740, is less than half the projection, and even the survey's upper bound plus the excess deaths, 95,340, is about half (this page's arithmetic). The windows differ: the letter spoke of deaths that 'could be attributable', which may include later years, so the survey does not rule out a higher eventual toll; it shows the 4:1 ratio did not hold for the first fifteen months. The worst case for each side. For '186,000 confirmed': take the survey's upper bound and all its excess deaths; about 95,000 by January 2025 (86,800 + 8,540, this page's arithmetic), with the indirect ratio a fraction of the one assumed. For 'the Gaza figures are inflated': take the survey at its lowest; its central estimate puts the Health Ministry's violent-death count 34.7% below it, an undercount, not an inflation. Holds on the evidence here: the 186,000 was a projection built on an assumed ratio, and the first measurement found far fewer indirect deaths; the ministry's count was below the survey's. Could happen over a longer window: a higher eventual toll. Not supported: that 186,000 deaths were confirmed, or that the measured toll shows the figures were inflated.
What the labels (Holds, Open, Arguable β¦) mean
pressFrance 24, ''More than 186,000 dead' in Gaza: How credible are the estimates published on The Lancet?' (11 July 2024) (2024) finding β
Read 1 Oct 2026. Quotes the letter (published 5 July 2024, correspondence section): 'it is not implausible to estimate that up to 186,000 or even more deaths could be attributable to the current conflict in Gaza'; a 'conservative estimate' of four indirect deaths per direct death applied to the Health Ministry's 37,396 deaths of 19 June; the ratio drawn from a 2008 Geneva Declaration Secretariat report ('between three and 15 times as many people die indirectly'); 'the letter's signatories did not say why they chose four'; 'Despite it not being either a report or scientific study β¦ has made global headlines'. The letter it quotes: Khatib, McKee & Yusuf, 'Counting the dead in Gaza: difficult but essential', The Lancet 404 (2024), doi:10.1016/S0140-6736(24)01169-3 (PubMed 38976995; not open in Europe PMC; thelancet.com 403 β not yet read).
data'Violent and non-violent death tolls for the Gaza conflict: new primary evidence from a population-representative field survey', The Lancet Global Health (2026) (2026) aggregate β
Spagat, Pedersen, Shikaki, Robbins, Bendavid, Hegre, Guha-Sapir, The Lancet Global Health 14 (April 2026). Read in full: 2,000 households, 30 Dec 2024 β 5 Jan 2025; 75,200 (63,600β86,800) violent deaths; 56.2% women, children, over-64s; 12,200 missing; 16,300 non-violent deaths, 8,540 excess; MoH 34.7% below the central estimate. Abstract re-read via OpenAlex 2026-10-12: 'The MoH figure for this period (49 090 violent deaths) was 34Β·7% below our central estimate'; 'violent deaths have substantially exceeded official figures whereas the demographic composition of casualties aligns with MoH reporting. Non-violent excess deaths, although substantial, are lower than some projections have suggested.' From the abstract: women, children and over-64s were '56Β·2% (95% CI 50Β·4-61Β·9) of violent deaths, totalling 42 200 deaths (95% CI 33 100-51 300)'. The age-and-sex figures used on counting-the-dead (32,900 men and 16,600 women aged 18β64 killed; 8,470 of the 12,200 missing being men of that age) come from the earlier full reading; the full text is walled today (2026-10-12) and they are not re-checked.
A live case from the other side: 'the bots went silent' (October 2026).
Between 8 and 11 October 2026 Ukrainian drones hit three data centres of Yandex, Russia's largest search engine. Charts then spread showing 'bot traffic' from Yandex's network collapsing and posts saying 'Free Palestine' and 'genocide' falling on X, read as proof that the pro-Palestinian outcry online was Russian bots. The claim suits the pro-Israel side, so it gets the same test as the others on this page: did the outcome occur? Three checks. (1) What the measure counts: in Cloudflare's terms 'a bot is any automated web request coming from a software program, not a fake social media account', and most of Yandex's comes from its search crawler (Newsweek, quoting Hadi Ghanbari). When the search engine went down, its crawler went quiet. (2) A partial day: the trend charts were captured about 18:00 UTC, 'so the last point covers only three-quarters of a day'; 18 of 24 hours is 75%, so a fall of about a quarter is what a normal day would show (this page's arithmetic; the point is America 2.0's, and i24NEWS reported that 'researchers said some of the graphs covered an unfinished day'). (3) The baseline: both terms had peaked on 7 October, the anniversary of the 2023 attack, so a fall afterwards was expected. A trend tool also 'cannot tell who posted them or whether they were automated'. The worst case for each side. For 'it was all Russian bots': grant every chart; none measures social-media accounts, and the one that counts posts covered a partial day after an anniversary peak. For 'Russia plays no part in this': the same network has been documented amplifying both sides of this conflict (see 'The argument that isn't meant to win'); this episode just doesn't show it. Holds on the evidence here: the charts behind 'the bots went silent' do not measure what they were taken to show. Open, needs posting data by account: whether Russian-run accounts on X went quiet during the outage. Not supported: that the drop in 'Free Palestine' posts was caused by the strikes.
pressi24NEWS, 'Ukraine hits Yandex data centers; Zelensky claims fewer bots, hate speech online, but no evidence of a link' (October 2026) (2026) incident β
Read 2026-10-13. 'Users pointed to a drop in posts containing phrases like βFree Palestineβ and βGenocideβ on X after the strikes, but researchers said some of the graphs covered an unfinished day'; 'Ukrainian drones hit three data centers belonging to Russian tech giant Yandex between October 8 and 11.' An Israeli news channel.
pressNewsweek, 'Could Ukraine strikes on Russian data center impact bots? What we know' (October 2026) (2026) incident β
Read 2026-10-13. Cloudflare 'reported that bot traffic from the Yandex network temporarily dropped to less than 1 percent of the maximum number of HTTP requests recorded in the past two weeks'. Hadi Ghanbari, described as 'a Russian cyber warfare expert', told Newsweek 'it is difficult to say if disinformation campaigns run through social media bots or fake accounts were disrupted. In Cloudflareβs terms, a bot is any automated web request coming from a software program, not a fake social media account'; 'Yandex runs Russiaβs biggest search engine, so most of its automated traffic comes from its crawler'.
referenceAmerica 2.0, 'The Russian Bot Collapse Report: did Russia's bots really go quiet?' (tracking page, updated 13 October 2026) (2026) finding β
Read 2026-10-13. Who runs the newsletter is not checked here. Rating: 'LOW No evidence that Russian social-media bots went quiet'. On the X trends charts (an X post of 11 October 'seen more than 600,000 times'): 'The charts were captured about 18:00 UTC on October 11, so the last point covers only three-quarters of a day; the drops of about a quarter match that. Both terms peaked on October 7, the anniversary of Hamas's 2023 attack; a fall after an anniversary spike is expected, strike or no strike.' 'The tool counts posts on a topic; it cannot tell who posted them or whether they were automated.' On Cloudflare: 'Accurate numbers, wrong label'; Cloudflare counts requests scored 'likely automated' from Yandex's own network, 'Normally 99.5% of that network's requests are automated'.
Why it works on us
- Early warnings save lives; waiting for final counts can mean waiting for the dead.
- Experts use models and proxies precisely when direct data is unavailable.
- The body issuing the figure is authoritative.
Where this trap shows up
Your side does it too β look at both columns.