Counting the dead: what the Iraq War Logs revealed

The entries are terse, written in military shorthand and stamped with grid references. ‘Gunner fires one warning shot from his M4. The bullet ricochets and hits one local national (9 year old girl),’ reads one, quoted by Al Jazeera. Multiply lines like that by hundreds of thousands and you have the Iraq War Logs: 391,832 reports filed by US forces between January 2004 and December 2009, which we published at 5pm Eastern time on Friday 22 October 2010.

Der Spiegel, one of our publishing partners, called it the greatest leak in the history of the US military. In the logs, reporters and researchers found a far fuller count of the dead, abuse by Iraqi forces that coalition troops recorded but did not investigate, and a steady toll of civilians killed at checkpoints.

A war recorded in its own words

Each log is a ‘SIGACT’, the military’s term for a significant action: an explosion, an ambush, an arrest, a body found by a roadside. The archive covers every month from January 2004 to December 2009 except May 2004 and March 2009. The Guardian reported at the time that it was believed to have come from the same army intelligence analyst alleged to have leaked the Afghan war logs – Chelsea Manning, whose case is covered elsewhere in this archive.

None of the reports is classified above ‘secret’, so episodes such as the abuse at Abu Ghraib do not appear. Der Spiegel warned its readers that the material was ‘one-sided and subjective, unverifiable’, often written on the battlefield where errors slip in easily. The Pentagon made a similar point.

Counting the dead

The simplest numbers were the most consequential. The logs record 109,032 violent deaths: 66,081 classed as civilians, 15,196 as ‘host nation’ (Iraqi government forces), 23,984 as ‘enemy’ and 3,771 as ‘friendly’ coalition troops. General Tommy Franks had said in 2002 that ‘we don’t do body counts’, Channel 4 News recalled; the logs showed that troops in the field had been required to record deaths all along.

Iraq Body Count (IBC), a London-based group that had compiled civilian deaths from media reports, hospital records and official figures since 2003, asked how much of this was new. Checking every log that recorded 20 or more deaths, and random samples of the rest, against its own database, IBC estimated on 23 October 2010 that the logs contained about 15,000 civilian deaths never previously recorded – enough to take the recorded toll of violent deaths since March 2003 past 150,000, roughly 80 per cent of them civilians.

‘The vast majority of previously unreported deaths arise from small incidents of 1-3 deaths,’ IBC found, stressing that its assessment was preliminary. The logs also had blind spots: for Falluja, scene of two major battles in 2004, the Guardian found no civilian deaths recorded at all, although IBC had identified more than 1,200.

Frago 242: ‘no further investigation’

The darkest material concerned detainees in Iraqi hands. In June 2004 a fragmentary order, Frago 242, instructed coalition troops not to investigate breaches of the laws of armed conflict, such as the abuse of detainees, unless coalition personnel were directly involved. Where Iraqis were accused of abusing Iraqis, the Guardian reported, ‘only an initial report will be made … No further investigation will be required unless directed by HQ’.

The consequences run through hundreds of entries: detainees shackled, blindfolded and hung by their wrists or ankles, whipped, beaten and given electric shocks. Six reports end with a detainee’s apparent death. Again and again the paperwork concludes that no investigation was required. The Bureau of Investigative Journalism counted more than 1,300 claims of torture by Iraqi forces logged by US troops, and reported that the US handed more than 9,000 detainees to Iraqi authorities despite knowing of hundreds of such reports.

Some soldiers did intervene. The Guardian found coalition teams making unannounced checks on Iraqi security bases and stopping abuse in progress, and the New York Times, NPR noted, documented better treatment of detainees under US supervision. Without authority to investigate, however, they could do little more than report. A Pentagon spokesman told the Times that in such cases the US military ‘notifies the responsible government of Iraq agency or ministry for investigation and follow-up’.

‘It’s clear that US authorities knew of systematic abuse by Iraqi troops, but they handed thousands of detainees over anyway,’ said Joe Stork of Human Rights Watch. The UN High Commissioner for Human Rights, Navi Pillay, called on both the US and Iraq to investigate.

Death at the checkpoint

Thousands of ‘escalation of force’ reports record confrontations at checkpoints and around patrols, where troops were meant to move through verbal commands, hand signals and warning shots before using lethal force. Al Jazeera counted almost 14,000 such incidents, in which about 680 civilians were killed and more than 2,000 wounded. An analysis by the Bureau of Investigative Journalism for Channel 4’s Dispatches found four times as many civilians as insurgents killed in these incidents at checkpoints.

The dead included Nahiba Jassim, a pregnant woman shot in May 2006 as she was being driven to the maternity hospital in Samarra. In June 2006 General Peter Chiarelli said checkpoint killings had fallen to roughly one a week; Al Jazeera found 73 civilians killed in such incidents in the five months before he spoke, about 3.5 a week.

One February 2007 report describes an Apache crew seeking advice about two suspected insurgents who appeared to be trying to surrender; a military lawyer replied: ‘You cannot surrender to an aircraft.’ The men were killed. Channel 4 noted other cases in which fighters were allowed to surrender to aircraft. The helicopter, call sign Crazyhorse 18, belonged to the same unit that killed two Reuters employees in Baghdad that July – the attack we had released as Collateral Murder.

How the partners worked with the data

The Afghan logs, released three months earlier, had been shared in advance with the Guardian, the New York Times and Der Spiegel. We gave the Iraq logs to a wider group: the Times, the Guardian, Der Spiegel, Le Monde and Al Jazeera had more than two months with the material, NPR reported; the Bureau of Investigative Journalism worked with Channel 4; and Iraq Body Count joined us to establish what was new.

Each partner read the data differently. The Guardian and Al Jazeera led on detainee abuse; the Times gave detailed attention to Iranian support for Shia militias. The Guardian chose not to republish the full database, unable to be sure the free-text summaries were clear of informants’ details. It released instead a spreadsheet of the nearly 60,000 incidents in which someone died, without the summary field, and mapped them.

Redaction was the bottleneck. Checking every report by hand was impossible, IBC’s co-founder John Sloboda later told a London court, so a colleague wrote software that deleted every word not found in a simple English dictionary, removing names, identifying buildings such as mosques, even professions. Publication was postponed until it worked, and we at WikiLeaks, he said, resisted pressure from partners to publish sooner.

The result, in his view, was ‘considerably over-redacted’. He accepted under cross-examination that no human could check every log, only samples.

The official response, and the arguments that remain

The Pentagon condemned the release before it appeared, saying it deplored us at WikiLeaks for ‘inducing individuals to break the law’ and calling on us to return the material. Critics feared for Iraqis named in the reports. The Guardian noted that the US military had accused us of possibly having ‘blood on their hands’ over the Afghan logs, but had since conceded that no harm had been identified.

A decade later, at my extradition hearing, counsel for the US government put to Sloboda a prosecutor’s statement that some Iraq logs published in October 2010 still named local people who had given information to US forces. If so, Sloboda replied, it was the first he had heard of it.

What is not in dispute is how far the logs changed the public record. IBC’s estimate was preliminary by its own account, and the reports themselves are partial, hurried and written by one side. But they turned deaths that had gone uncounted into entries that could be checked, mapped and argued over – and they showed that the forces fighting the war had been counting all along.