The Bureau is now publishing a leaked official document that records details of over 300 drone strikes, including their locations and an assessment of how many people died in each incident.
The document is the fullest official record of drone strikes in Pakistan to have yet been published. It provides rare insight into what the government understands about the campaign.
It also provides details about exactly when and where strikes took place, often including the names of homeowners. These details can be valuable to researchers attempting to verify eyewitness reports – and are often not reported elsewhere. But interestingly, the document stops recording civilian casualties after 2008, even omitting details of well-documented civilian deaths and those that have been acknowledged by the government.
Last July the Bureau published part of the document for the first time. This documented strikes, which hit the northwest tribal areas of Pakistan between 2006 and late 2009, and revealed that the Pakistani government was aware of hundreds of civilian casualties, even in strikes where it had officially denied civilians had died.
The reports are based on information filed to the FATA Secretariat each evening by local Political Agents – senior officials in the field. These agents gather the information from networks of informants in Pakistan’s Federally Administered Tribal Areas (FATA), the area bordering Afghanistan.
Now the Bureau has obtained an updated version of the document, which lists attacks up to late September 2013.
The document contains estimates of how many people have been killed in each strike, as well as whether the dead are ‘local’ or ‘non-local’ – a broad category that includes those from elsewhere in Pakistan, as well as foreigners.
When the Bureau released the first part of the report last summer, anonymous US officials attacked the document, claiming that the report was ‘far from authoritative’ as it was based on ‘erroneous media reporting’ and ‘indirect input from a loose network of Pakistani government and tribal contacts’. But the US has consistently refused to release information on what it believes has been the result of its drone strikes.
The overall casualties recorded by the document are broadly similar to those compiled by the Bureau, which uses sources including media reports, sworn affidavits and field investigations. The Bureau estimates that at least 2,371 people died in the time covered by the document (excluding 2007, which is missing from the record), while it records 2,217 deaths in total.
The document does not represent the Pakistani government’s full view of drone strikes. Alongside the Political Agents and their daily reports to the FATA Secretariat, the country’s intelligence agencies and military are each believed to collect details of attacks in separate reports. And during a recent trip to Pakistan the Bureau obtained a list of individuals killed in a single strike from a local politician.
The Pakistani government has made a series of statements on drone casualties: in March last year, officials at the Foreign Affairs ministry told UN expert Ben Emmerson, who was carrying out an investigation into drones, that at least 400 civilians – and possibly 600 – were among 2,200 drone casualties. In October, the Ministry of Defence issued a statement that contradicted this, asserting that drones had killed 67 civilians since 2008. It later retracted the statement, with unnamed senior defence officials telling The News International that the figures were ‘wrong and fabricated’.
The document obtained by the Bureau is unusual because it gives a strike-by-strike account, allowing for comparison between the government’s view of individual incidents and that of other sources.
Although the document records civilian casualties in the early years, from 2009 these almost disappear. Even well-documented cases of civilian deaths are omitted. These include at least two incidents where the tribal administration is known to have admitted to the families that it knew civilians had died.
Among the civilian deaths that go unmentioned is one of the most high-profile attacks of the past 18 months – an October 2012 attack that killed Mamana Bibi, an elderly woman, as she was in a field. Her grandchildren were nearby, and several were injured by debris.
‘If a case as well-documented as Mamana Bibi’s isn’t recorded as a civilian death, that raises questions about whether any state records of these strikes can be seen as reliable, beyond the most basic information,’ said Mustafa Qadri, a researcher for Amnesty International, who investigated the strike for a major report published last autumn. ‘It also raises questions of complicity on the part of the Pakistan state – has there been a decision to stop recording civilians deaths?’
Up to the end of 2008, the document reports where attacks have killed civilians. In this period the document lists 37 drone strikes, as well as four attacks carried out by NATO and Afghan forces – and it notes civilian deaths in 15 of the drone attacks. The document records 353 deaths in this time, of whom at least 138 are specifically described as civilians.
The document records a further 294 incidents between January 1 2009 and September 2013, when the version obtained by the Bureau ends. Only seven of these specifically mention civilian victims. Just two use the word ‘civilian’ – the others typically refer to women and children as being among the dead. A further entry states that a child was injured.
The Bureau’s data records a similar number of incidents over the same time period, but shows 53 incidents where at least one civilian death is reported by multiple credible sources – and many more where civilian deaths are possible. In total, the document records around 200 civilian deaths, including those where ambiguous language such as ‘local tribesmen’ is used – compared to a minimum of over 400 recorded by the Bureau.
A former senior FATA Secretariat official, speaking on condition of anonymity, explained that rather than attempting to establish which of the dead were believed to be civilians, agents instead categorised the dead as ‘local’ or ‘non-local’.
‘It is very difficult to report it whether this man was really a militant or a non-militant. So they found an easy way of saying it: local and non-local,’ he said.
- Chris Woods
A second local source agreed: ‘As a matter of policy, deaths in drone strikes were classified as locals and non-locals, because [the term] civilians was found to be too vague and contradictory.’ This helped to ‘avoid controversy’, he added.
The ‘non-local’ category strongly suggests that an individual is an alleged militant, the former official added. ‘Local means that they belong to that agency [tribal administered district] and you could say in general terms that they are innocent… But it is quite possible that some of them might be terrorists.’
The change in recording follows an escalation in the number of strikes in the final months of Bush’s presidency, which gathered pace under Obama. With the increased frequency of the strikes, gathering information may have become more challenging for Political Agents. Some non-combatant deaths may be missing, too, because reports are filed soon after they occur and are not later updated: several entries contain no casualty estimates at all and simply note: ‘Details are awaited.’
The former FATA official suggested that the document may have stopped regularly recording civilian casualties because of something as prosaic as a change of the personnel charged with compiling it. But other observers suggested that the cause could be less mundane.
The last drone strike in the document to use the word ‘civilian’ in describing the dead is the first of Obama’s presidency, on January 23 2009 (a strike six months later says, more ambiguously, ‘A civilian pickup was targeted’).
Amnesty’s Qadri said: ‘You cannot rule out a deliberate attempt not to include information on possible civilians or non-combatants being killed. It seems a huge coincidence that there’s this change in reporting just as Obama enters power. But whatever the explanation and despite the lingering uncertainty, we know these figures are not presenting the full picture of the US drone program.’
Chris Woods, who started the Bureau’s investigation into drone strikes and who is now writing a book on armed drones, said: ‘One of my sources, a former Pakistani minister, has indicated that local officials may have come under pressure to play down drone civilian deaths following the election of Barack Obama. It’s certainly of concern that almost all mention of non-combatant casualties simply disappears from this document after 2009, despite significant evidence to the contrary.’
‘It is feared that all the killed were local tribesmen’
A handful of entries include ambiguous language hinting at non-combatant casualties. On August 14 2010, the document records an evening strike, noting: ‘The dead included 07 Mehsuds, 05 locals and 01 unknown’. Mehsud is the name of a prominent local tribe. A field investigation by Associated Press later found that seven civilians – including a child – were among 14 to die in an attack on a house during Ramadan prayers.
And when a drone attacked a meeting of tribal elders on March 17 2011 – an attack that was condemned by the Pakistani military and civilian government – the report says ‘it is feared that all the killed were local tribesmen’.
Bureau field investigations have repeatedly encountered civilian deaths in strikes where local media have used ambiguous phrases such as ‘villagers’, ‘people’ and ‘local tribesmen’.
One entry in the file hints at problematic definitions of who is considered a ‘militant’. For a strike on April 12 2010, it records 14 deaths and three injuries, noting: ‘The killed militants also include a 12 years [sic] old child.’
‘Whatever is happening, if this document is anything to go by, it’s clear the Pakistan government’s investigations are not adequate,’ said Amnesty’s Qadri. ‘First, this table does not appear to be telling us the whole truth about casualties.
‘Secondly, what steps have Pakistan authorities taken to assist civilians caught up in these strikes like access to medical services or provide them with remedies such as access to justice or compensation? … It doesn’t seem to be the case that this record keeping is carried out so that the Pakistan state can better assist people caught up in these strikes.’
The document also barely mentions other details such as which organisation the dead are believed to have belonged to, or the names of those killed. Even when very senior militants are killed, they are almost never identified by name.
As the Bureau has found with its Naming the Dead investigation, the vast majority of those killed in drone strikes remain unidentified – only around one in five has so far been identified by name. Documents obtained by news agency McClatchy and NBC showing the CIA’s records of its drone strikes indicated that in most strikes these do not record the names of the dead either. These documents have not been published. And as the Pakistan document shows, even to the local government it is often a mystery who is dying in the CIA’s drone strikes.