Watching the detectives

Hospital bombing, Kunduz, October 2015 MSF

I wrote about medical neutrality earlier this year (see here).  As I noted then, Physicians for Human Rights stipulates that medical neutrality requires:

The protection of medical personnel, patients, facilities, and transport from attack or interference;
Unhindered access to medical care and treatment;
The humane treatment of all civilians; and
Nondiscriminatory treatment of the sick and injured.

In the wake of the US air strike on a hospital operated by Médecins Sans Frontières  (MSF) in Kunduz on 3 October, that first requirement assumes even greater significance: the obligation is not merely to exempt medical personnel, patients and infrastructure from military and paramilitary violence but to protect them from attack.

MSF provides details and updates on the strike here.  As I write, far and away the most substantial commentary on what happened – given what we know so far – is Kate Clark‘s detailed analysis at the Afghan Analysts Network here (though Matt Lee‘s angry comparison with an Israeli military attack on a hospital in Gaza is worth reading too).  As Kate notes,

Expressing distrust in the US military, NATO or Afghan government to uncover the truth, [MSF] said it wants an investigation by the International Humanitarian Fact-Finding Commission (IHFFC), a body set up by the Additional Protocols of the Geneva Conventions and, says MSF, is the only permanent body set up specifically to investigate violations of international humanitarian law. It has never been used before and, as neither Afghanistan or the United States have formally recognized the Commission, any investigation would have to be voluntary.

logo_ihffcThe IHFFC issued this statement today:

The International Humanitarian Fact-Finding Commission (IHFFC) has been contacted by Médecins Sans Frontières (MSF, Doctors Without Borders) in relation to the events in Kunduz, Afghanistan, on 3 October 2015.

The IHFFC stands ready to undertake an investigation but can only do so based on the consent of the concerned State or States. The IHFFC has taken appropriate steps and is in contact with MSF. It cannot give any further information at this stage.

Alex Jeffrey has commented briefly on the geopolitics of any investigation by the IHFFC, but there has been little or no commentary on how the US military investigates civilian casualty incidents – and this merits discussion because the Obama administration has insisted that the inquiry already under way by the Pentagon will be ‘transparent’, ‘thorough’ and ‘objective’.  And whatever may or may not transpire with respect to the IHFFC, it’s exceptionally unlikely that the US military investigation will be stopped.

I’ve worked through five investigations of so-called ‘CIVCAS’ in Afghanistan that have been released through Freedom of Information Act requests.  Each branch of the US military is required to maintain its own digital FOIA Reading Room, so that any documentation supplied in response to these requests is released into the public domain.  I should say that you need to be adept at using the search function, and to have a very good idea of what you are looking for before you start (though the Pentagon has been remarkably helpful in responding to my inquiries and questions).

It’s also fair to say that the release of investigation reports is uneven.  In the immediate aftermath of an earlier, devastating air strike on two tankers hijacked by the Taliban near Kunduz, called in by the German Bundeswehr but carried out by two US aircraft (see my extended discussion here), the United States repeatedly promised to release the investigation report: but it never did, even to the German Bundestag’s committee of inquiry, and despite repeated requests it remains classified.

There is also considerable variation in the transparency and quality of the reports that have been released: some are so heavily redacted that it is extremely (and no doubt intentionally) difficult to construct a reasonably comprehensive narrative, while others are the product of inquiries that seem to have been, at best, perfunctory.

AR 15-6 CIVCAS Uruzgan February 2010

The report into the airstrike in Uruzgan that I have been using for my analysis of the US air strike in Uruzgan in February 2010 – see ‘Angry Eyes (1)‘ and ‘Angry Eyes (2)‘: more to come – is neither.  It has been redacted, presumably for reasons of national, operational or personal security, but its 2,000 pages provide enough detail to reconstruct most of what happened.  And the investigation team was remarkably thorough: by turns forensic, sympathetic, exasperated and eventually blisteringly angry at what they found.  Whether this provides an indication of what we can expect from the present inquiry I don’t know, but it does provide a benchmark of sorts for what we (and, crucially, MSF) ought to expect.  (There are also ongoing investigations by NATO and by the Afghan authorities, but no details have been released about them either).

The strike took place on 21 February 2010, and the very next day General Stanley McChrystal (Commander US Forces – Afghanistan and ISAF, Afghanistan) appointed Major-General Timothy McHale to conduct what the US Army calls ‘an informal investigation’ into the incident that ‘allegedly resulted in the deaths of 12-15 local Afghan nationals and caused injured to others’; McHale was assisted by a team of senior officers, including subject matter experts and legal advisers:

GREGORY Angry Eyes 2015 IMAGES.139

There are two points to note here.

First, this was an investigation conducted by the US Army because the airstrike had been called in by US Special Forces and had been carried out by two US Army helicopter crews.  But the strike was orchestrated in large measure by a US Air Force Predator crew from Creech Air Force Base in Nevada; in addition to questioning the soldiers and helicopter crews involved, McHale’s team also questioned the Predator flight crew together with the screeners and video analysts at Air Force Special Operations Command at Hurlburt Field in Florida.  McHale’s report triggered a second ‘Commander-Directed Investigation’ by US Air Force Brigadier-General Robert Otto into the actions and assessments of the Predator crew; that report was submitted on 30 June 2010.  As I write, it’s not known who is leading the US investigation into the bombing of the hospital in Kunduz.  Since (on the fourth telling) the strike appears to have been called in by US Special Forces (at the request of Afghan forces) and carried out by a US Air Force AC-130 gunship this will presumably be a joint investigation.

Second, the term ‘informal investigation’ is a technical one; certainly, on McHale’s watch the conduct of the inquiry was remarkably rigorous.  US Army Regulation 15-6 sets out how the Army is to conduct an investigation:

‘The primary function of any investigation or board of officers is to ascertain facts and to report them to the appointing authority. It is the duty of the investigating officer or board to ascertain and consider the evidence on all sides of each issue, thoroughly and impartially, and to make findings and recommendations that are warranted by the facts and that comply with the instructions of the appointing authority.’

Here is the distinction between informal and formal investigations (I’ve taken this summary from a US Army Legal Guide here; the full version, specifying the conduct of an informal investigation, is here and here):

Informal investigations may be used to investigate any matter, to include individual conduct. The fact that an individual may have an interest in the matter under investigation or that the information may reflect adversely on that individual does not require that the proceedings constitute a hearing for that individual. Even if the purpose of the investigation is to inquire into the conduct or performance of a particular individual, formal procedures are not mandatory unless required by other regulations or by higher authority. Informal investigations provide great flexibility. Generally, only one investigating officer is appointed (though multiple officers could be appointed); there is no formal hearing that is open to the public; statements are taken at informal sessions; and there is no named respondent with a right to counsel (unless required by Art 31(b), UCMJ); right to cross-examine witnesses; etc….

“Generally, formal boards are used to provide a hearing for a named respondent. The board offers extensive due process rights to respondents (notice and time to prepare, right to be present at all open sessions, representation by counsel, ability to challenge members for cause, to present evidence and object to evidence, to cross examine witnesses, and to make argument). Formal boards include a president, voting members, and a recorder who presents evidence on behalf of the government. A Judge Advocate (JA) is normally appointed as recorder but is not a voting member. If a recorder is not appointed, the junior member of the board acts as recorder and is a voting member. Additionally, a non-voting legal advisor may be appointed to the board. Formal AR 15-6 investigations are not normally used unless required by regulation.’

In setting all this out, I should add two riders.  In treating MG McHale’s investigation in such detail, I don’t mean to imply that I fully concur with its analysis.  This is a judgement call, of course: the redactions make it difficult to press on several key issues, all of which relate to who knew what when and where (more to come on this).  And neither do I mean to suggest that any US military investigation into what happened in Kunduz, however probing, would be adequate. As MSF’s Chris Stokes has said, ‘relying only on an internal investigation by a party to the conflict would be wholly insufficient.’  But if the report is conducted with the same careful attention to detail – and if it is released with minimal redactions – it would provide a necessary resource for all those involved in and affected by this truly appalling incident.

More to come – I hope.

UPDATE (1):  The US investigation is headed by Brigadier-General Richard Kim.  Nancy Youssef reports that his arrival in Kunduz was delayed ‘because of instability in the northern Afghan city.’ As with the Uruzgan air strike in 2010, the video recording from the AC-130 gunship that carried out the attack, together with audio recordings of conversations between the air crew and ground troops, will be of great importance.  According to Youssef, these show that ‘rules of engagement—the guidelines for the use of force—were misapplied.’  (In the Uruzgan case, the radio conversations between the air crew(s) and the Joint Terminal Attack Controller on the ground were released in redacted form in response to a FOIA request; apart from a single image of the strike, however, the video remains classified.)

I’ve previously noted the debate surrounding the Pentagon’s new Law of War manual which was issued in June 2015; since the US has admitted that the strike on the hospital was carried out within the US chain of command, section 7.17 on ‘Civilian hospitals and their personnel’ is particularly relevant (see also the Guardian report here):

During international armed conflict, civilian hospitals organized to give care to the wounded and sick, the infirm, and maternity cases, may in no circumstances be the object of attack, but shall at all times be respected and protected by the parties to the conflict.

7.17.1 Loss of Protection for Civilian Hospitals Used to Commit Acts Harmful to the Enemy. The protection to which civilian hospitals are entitled shall not cease unless they are used to commit, outside their humanitarian duties, acts harmful to the enemy. Acts Harmful to the Enemy. Civilian hospitals must avoid any interference, direct or indirect, in military operations, such as the use of a hospital as a shelter for able-bodied combatants or fugitives, as an arms or ammunition store, as a military observation post, or as a center for liaison with combat forces. However, the fact that sick or wounded members of the armed forces are nursed in these hospitals, or the presence of small arms and ammunition taken from such combatants and not yet handed to the proper service, shall not be considered acts harmful to the enemy. Due Warning Before Cessation of Protection. In addition, protection for civilian hospitals may cease only after due warning has been given, naming, in all appropriate cases, a reasonable time limit, and after such warning has remained unheeded.


The obligation to refrain from use of force against a civilian medical facility acting in violation of its mission and protected status without due warning does not prohibit the exercise of the right of self-defense. There may be cases in which, in the exercise of the right of self- defense, a warning is not “due” or a reasonable time limit is not appropriate. For example, forces receiving heavy fire from a hospital may exercise their right of self-defense and return fire. Such use of force in self-defense against medical units or facilities must be proportionate. For example, a single enemy rifleman firing from a hospital window would warrant a response against the rifleman only, rather than the destruction of the hospital.

MSF has consistently denied that anyone was firing from the hospital; it has also insisted that it received no advance warning of the attack – on the contrary, MSF ensured that all US and Afghan forces had the co-ordinates of the hospital, and made frantic phone calls to try to stop the bombing once it started.

UPDATE (2):  A team from the Washington Post has produced a remarkably detailed report, ‘based on multiple interviews in Afghanistan and the United States with U.S. and Afghan military officials, Doctors Without Borders personnel and local Kunduz residents’; it includes maps and a graphic showing exactly what an AC-130 is capable of.


As you can see, the illustration makes much of the aircraft’s concentrated firepower, unleashed as it circles counter-clockwise around the target in a five-mile orbit, but the AC-130 also has an extensive sensor suite on board (see ‘Angry Eyes (1)‘: an AC-130 was involved in the early stages of the Uruzgan incident).  The reporters do note that the aircraft is equipped with ‘low-light and thermal sensors that give it a “God’s eye [view]” of the battlefield in almost all weather conditions’ – but, as I’ve tried to show in my posts on Uruzgan (and as we know from other sources!), there’s no such thing as a God’s eye view.  Even so, the aircrew can surely have been in no doubt that they were bombing a hospital.

Predator View

I’m in Zurich – many thanks to Benedikt Korf for the invitation and the wonderful hospitality – for a seminar with Benedikt’s doctoral students on the long form version of ‘The Natures of War’ (they have what I hope is the penultimate draft; I’ll post the final version once I’ve reworked it after the seminar) and then a public lecture: ‘Angry eyes: the militarization of vision and modern war’.

Part of my argument in the lecture is about the narrowness of ‘Predator View’, so let me explain what I mean.  My starting point is the illusion that the use of the intelligence, surveillance and reconnaissance capabilities of remote platforms like the MQ-1 Predator or the MQ-9 Reaper produces a transparent battlespace: in effect, a version of Donna Haraway‘s ‘God-trick’, the ability to see everything from nowhere in particular.  The most succinct version of this is the treatment of laser targeting as ‘the eye of God’:

Eye of God.001

In fact, ‘Predator View’ is remarkably narrow. First, and most directly, a common complaint is that the view provided by its Full Motion Video is literally too narrow – ‘like looking through a soda straw’.  It’s for this reason that pilots and sensor operators also use FalconView, a Windows-based mapping tool that is supposed to provide them with situational awareness.  I’ve taken the image below from a presentation by Lt Col Paul Hastert (USAF) on ‘Spiral development in wartime’ – the screenshot is the FalconView image, the small square the field of view from a Predator – but the most detailed account of FalconView is Jon Lindsay‘s ‘”War upon the map”‘: User innovation in American military software’, Technology and culture 51 (3) (2010) 619-51.


There have been several later releases of FalconView, and a version is now even open source, so it’s probably not surprising that the US military also uses Google Earth.  In fact, in two of the air strikes I consider in detail it was, in part, juggling these different views that contributed to wrong targeting decisions.  (There are projects that provide a wider field of view – the nine-camera suite of Gorgon Stare is in operation and its second increment is already in use in Afghanistan – and others that integrate imagery from multiple sources, like Raytheon’s Zeus, which is intended to supplant FalconView as the ‘primary mission execution tool‘).

But in the air strikes that I consider those involved in these networked operations are also busy communicating via radio and – crucially – via the typed messages displayed in mIRC‘s online chat windows.  The image below is taken from the US military’s multi-service manual on Tactical Chat (see also the discussion at Public Intelligence here).

mIRC 3 chatrooms

In the case of the ‘friendly fire’ incident in the Sangin Valley in 2011, the pilot and sensor operator were unaware of crucial entries made by image analysts in mIRC and of private (or ‘whisper’) chats to the Mission Intelligence Co-ordiator.  That’s not entirely surprising.  Once permission has been given to engage a target, ‘KILL CHAIN” is typed into the room to minimise extraneous chatter that might otherwise visually distract and so ‘degrade’ the strike.  According to the Mission Intelligence Controller for the Sangin incident:

‘When we got a standby for a 9-line [strike briefing] we put [KILLCHAIN] in the mIRC room to shut down chatter so we can be heads down supporting the pilot and the sensor … as they are getting ready for a possible Hellfire shot our whatever action… That’s our job, to be their third set of eyes…’

That third set of eyes is vital, though of little use if its responses are shut down.  In this case the pilot was paying no attention to mIRC because he was fully occupied:

‘I did not see the chat in the main mission room after seeing muzzle flashes. From that point on, my focus was clear concise communications with the [Joint Terminal Attack Controller] and ensuing [Close Air Support] comms, 9 line etc. At this point I was also focused on keeping the aircraft in position to maintain FMV on the individuals. I was also getting the aircraft into the [Weapons Engagement Zone]. Winds were 30 knots out of the west so I had to make sure I kept the aircraft position and direction in cross check.’

All of this suggests that it is unduly narrow to focus the analytical gaze on the Full Motion Video feeds.

But there’s more.  Militarised vision is not limited to Predators and their successors, and it’s important to consider the clarity and resolution of the imagery captured and transmitted from other, conventional aircraft like the AC-130 (first image below), the B-1 bomber, F-15E strike aircraft or attack helicopters (second image below)  – again, this was important in two of the strikes I consider in detail.

AC-10 Uruzgan.001

MMS from attack helicopter - Uruzgan.001

This leads me to two other senses in which the critical emphasis on ‘Predator View’ is too narrow.  It’s a mistake to fasten on one incident – even what has become the iconic ‘signature strike’ coordinated by a Predator on three civilian trucks in Uruzgan – because that misses the continuities, repetitions and transformations in Close Air Support (which is where most of the mistakes are made).

Better, I think, to widen the analytical gaze, which is why I examine the Kunduz air strike on 4 September 2009 (in which no remote platform was involved), the Uruzgan air strike of 21 February 2010 (which was co-ordinated by a Predator but carried out by two attack helicopters), and the Sangin Valley air strike of 6 April 2011 (which was executed by a Predator).  The Granai strike is included because it trigged McChrystal’s Tactical Directive of 6 July 2009 to minimise civilian casualties: ‘‘We must avoid the trap of winning tactical victories – but suffering strategic defeats – by causing civilian casualties or excessive damage and thus alienating the people.’  In all three cases I’ve worked through thousands of pages of official investigations – one by the German Bundestag since the ISAF report remains classified (Kunduz), and the others by the US Army and US Air Force (Uruzgan) and the US Marine Corps (Sangin Valley).

4 US air strikes in Afghanistan.001

And finally, it is also necessary to consider the multiple viewing positions involved in networked military violence.  There is a de-centralised, distributed and dispersed geography of militarised vision whose fields of view expand, contract and even close at different locations engaged in the administration of military violence.  And in all three incidents it turns out that vital mistakes in the interpretation of imagery were made at operations centres in theatre on the ground.

This has prompted me to radically revise the argument I originally set out in ‘From a view to a kill’ (DOWNLOADS tab).  There my focus was on the geography shown in the first map below – like every other commentator I’ve read, incidentally – but I’ve now widened the angle to take in the geography shown in the second map and this substantially changes the story.

Dispersed vision Uruzgan (1).001

Dispersed vision Uruzgan (2).001

It was what happened at the operations centres at Special Operations Task Force 12 in Kandahar and Combined Joint Special Operations Task Force in Bagram that turned out to be every bit as important as the Predator crew’s obvious desire to ‘go kinetic’.

I make similar arguments about Kunduz, where the F-15E pilots were clearly reluctant to strike and the ground commander and JTAC at the Forward Operating Base overrode their objections.

Sangin is even more instructive, because the official investigation of this ‘friendly fire’ incident that resulted in the deaths of two Marines largely exonerated both the Predator crew and the ground commander and JTAC at the operations centre – but a detailed, spirited counter-memorandum from the US Marines commandant refuted the findings line by line, castigated the attempt to pin the blame for what happened on the young, inexperienced lieutenant leading the patrol, and excoriated the ground commander and JTAC back at the patrol base.  More on this soon.

The (long!) essay on these  strikes is the last chapter I need to complete for The everywhere war. There is a strong sense, so I shall argue, in which it was the view ‘on the ground’ that mattered much more than any ‘light of God’ from above.  And as for transparency….