Eyes in the Skies

Posted on April 9, 2016 by Derek Gregory

I’m just back from a wonderful time at UC Davis, where I was speaking at a symposium called “Eyes in the skies: drones and the politics of distance warfare.” It was a creative program, packed with insights from Caren Kaplan and Andrea Miller, Priya Satia and Joe Delappe.

On my way back to Vancouver on Wednesday I received an invitation from Britain’s Guardian (in fact, the Sunday version, the Observer) to write something around that very question using the Gavin Hood film “Eyes in the sky” as a peg.

It’s just been published and you can find it here.

At Davis I’d been giving what I think will be my final presentation of “Angry Eyes” (see here and here), so I was still preoccupied with remote platforms and close air support – not the contradiction it sounds – rather than targeted killing (which is the focus of the film). The published version has, inevitably, been edited, so I’m pasting the full-length version below and added some links that might help. There are still lots of short-cuts and elisions, necessarily so for anything of this length, so I hope readers will forgive the inevitable simplifications.

***

Gavin Hood’s Eye in the Sky is a thrillingly intelligent exploration of the political and ethical questions surrounding drone warfare. It’s been carefully researched and is on the cutting-edge of what is currently possible. But there’s a longer history and a wider geography that casts those issues in a different light. As soon as the Wright Brothers demonstrated the possibility of human flight, others were busy imagining flying machines with nobody on board. In 1910 Raymond Phillips captivated crowds in the London Hippodrome with a remotely controlled airship that floated out over the stalls and, when he pressed a switch, released hundreds of paper birds on the heads of the audience below. When he built the real thing, he promised, the birds would be replaced with bombs. Sitting safely in London he could attack Paris, Berlin – or Manchester (a possibility that understandably prompted questions about navigation).

There has always been something hideously theatrical about bombing, from the Hendon air displays in the 1920s featuring attacks on ‘native villages’ to the Shock and Awe visited on the inhabitants of Baghdad in 2003. The spectacle now includes the marionette movements of Predators and Reapers whose electronic strings are pulled from thousands of miles away. And it was precisely the remoteness of the control that thrilled the crowds in the Hippodrome. But what mattered even more was surely the prospect Phillips made so real: bombing cities and attacking civilians far from any battlefield.

Remoteness’ is in any case an elastic measure. Human beings have been killing other human beings at ever greater distances since the invention of the dart, the spear and the slingshot. Pope Urban II declared the crossbow illegal and Pope Innocent II upheld the ban in 1139 because it transformed the terms of encounter between Christian armies (using it against non-Christians was evidently a different matter). The invention of firearms wrought another transformation in the range of military violence, radicalized by the development of artillery, and airpower another. And yet today, in a world selectively but none the less sensibly shrunken by the very communications technologies that have made the deployment of armed drones possible, the use of these remote platforms seems to turn distance back into a moral absolute.

But if it is wrong to kill someone from 7,500 miles away (the distance from Creech Air Force Base in Nevada to Afghanistan), over what distance is it permissible to kill somebody? For some, the difference is that drone crews are safe in the continental United States – their lives are not on the line – and this has become a constant refrain in the drone debates. In fact, the US Air Force has been concerned about the safety of its aircrews ever since its high losses during the Second World War. After Hiroshima and Nagasaki the Air Force experimented with using remotely controlled B-17 and B-47 aircraft to drop nuclear bombs without exposing aircrews to danger from the blast, and today it lauds its Predators and Reapers for their ability to ‘project power without vulnerability’.

It’s a complicated boast, because these remote platforms are slow, sluggish and easy to shoot down – they won’t be seen over Russian or Chinese skies any time soon. They can only be used in uncontested air space – against people who can’t fight back – and this echoes Britain’s colonial tactic of ‘air policing’ its subject peoples in the Middle East, East Africa and along the North-West Frontier (which, not altogether coincidentally, are the epicentres of todays’ remote operations). There are almost 200 people involved in every combat air patrol – Nick Cullather once described these remote platforms as the most labour intensive weapons system since the Zeppelin – and most of them are indeed out of harm’s way. It’s only a minor qualification to say that Predators and Reapers have a short range, so that they have to be launched by crews close to their targets before being handed off to their home-based operators. This is still remote-control war, mediated by satellite links and fiber-optic cables, but in Afghanistan the launch and recovery and the maintenance crews are exposed to real danger. Even so, Grégoire Chamayou insists that for most of those involved this is hunting not warfare, animated by pursuit not combat [see here, here and here].

Yet it’s important not to use this aperçu to lionize conventional bombing. There is an important sense in which virtually all aerial violence has become remarkably remote. It’s not just that bombing has come to be seen as a dismal alternative to ‘boots on the ground’; advanced militaries pick their fights, avoid symmetrical warfare and prefer enemies whose ability to retaliate is limited, compromised or degraded. When he was Secretary of Defense Robert Gates acknowledged that the US had not lost a pilot in air combat for forty years. ‘The days of jousting with the enemy in the sky, of flirting daily with death in the clouds, are all but over,’ writes the far-from-pacifist Mark Bowden, ‘and have been for some time.’ The US Air Force goes to war ‘virtually unopposed’. In short, the distance between the pilot in the box at Creech and the pilot hurtling through the skies over Afghanistan is less than you might think. ‘Those pilots might as well be in Nevada’, says Tom Engelhardt, ‘since there is no enemy that can touch them.’

This suggests that we need to situate armed drones within the larger matrix of aerial violence. Bombing in the major wars of the twentieth century was always dangerous to those who carried it out, but those who dropped bombs over Hamburg or Cologne in the Second World War or over the rainforests of South Vietnam in the 1960s and 70s were, in a crucial sense, also remote from their targets. Memoirs from Bomber Command crews confirm that the target cities appeared as lights sparkling on black velvet, ‘like a Brocks firework display.’ ‘The good thing about being in an aeroplane at war is that you never touch the enemy’, recalled one veteran. ‘You never see the whites of their eyes. You drop a four-thousand-pound cookie and kill a thousand people but you never see a single one of them.’ He explained: ‘It’s the distance and the blindness that enables you to do these things.’ The crews of B-52 bombers on Arc Light missions dropped their loads on elongated target boxes that were little more than abstract geometries. ‘Sitting in their air-conditioned compartments more than five miles above the jungle’, the New York Times reported in 1972, the crews ‘knew virtually nothing about their targets, and showed no curiosity.’ One of them explained that ‘we’re so far away’ that ‘it’s a highly impersonal war for us.’

Distance no longer confers blindness on those who operate today’s drones. They have a much closer, more detailed view of the people they kill. The US Air Force describes their job as putting ‘warheads on foreheads’, and they are required to remain on station to carry out a battle damage assessment that is often an inventory of body parts. Most drone crews will tell you that they do not feel thousands of miles away from the action: just eighteen inches, the distance from eye to screen.

Their primary function is to provide intelligence, surveillance and reconnaissance. This was exactly how the Wright brothers thought military aircraft would be used – in July 1917 Orville insisted that ‘bomb-dropping’ would be at best a minor role and almost certainly useless, though he was speaking before the major air offensives in the final year of the war and could have had no inkling of what was to come in the Second World War. The Predator and its precursors were designed to identify targets for conventional strike aircraft over the Balkans in the 1990s, and thirty years later it is still those ‘eyes in the sky’ that make the difference. Although drones have been armed since 2001, until late 2012 they were directly responsible for only 5-10 per cent of all air strikes in Afghanistan. But they were involved in orchestrating many more. Flying a Predator or a Reaper ‘is more like being a manager’, one pilot explained to Daniel Rothenberg: ‘You’re managing multiple assets and you’re involved with the other platforms using the information coming off of your aircraft.’ In principle it’s not so different from using aircraft to range targets for artillery on the Western Front, but the process has been radicalized by the drone’s real-time full-motion video feeds that enable highly mobile ‘targets of opportunity’ to be identified and tracked. In the absence of ground intelligence, this becomes crucial: until drones were relocated in sufficient numbers from Afghanistan and elsewhere to enable purported IS-targets in Syria to be identified, most US aircraft were returning to base without releasing their weapons.

Armed drones are used to carry out targeted killings, both inside and outside areas of ‘active hostilities’, and to provide close air support to ground troops. Targeted killing has spurred an intense critical debate, and rightly so – this is the focus of Eye in the sky too – but close air support has not been subject to the same scrutiny. In both cases, video feeds are central, but it is a mistake to think that this reduces war to a video game – a jibe that in any case fails to appreciate that today’s video games are often profoundly immersive. In fact, that may be part of the problem. Several studies have shown that civilian casualties are most likely when air strikes are carried out to support troops in contact with an enemy, and even more likely when they are carried out from remote platforms. I suspect that drone crews may compensate for their physical rather than emotional distance by ‘leaning forward’ to do everything they can to protect the troops on the ground. This in turn predisposes them to interpret every action in the vicinity of a ground force as hostile – and civilians as combatants – not least because these are silent movies: the only sound, apart from the clacking of computer keys as they talk in secure chat rooms with those watching the video feeds, comes from radio communications with their own forces.

In contrast to those shown in Eye in the Sky, those feeds are often blurry, fuzzy, indistinct, broken, compressed -– and, above all, ambiguous. How can you be sure that is an insurgent burying an IED and not a farmer digging a ditch? The situation is more fraught because the image stream is watched by so many other eyes on the ground, who all have their own ideas about what is being shown and what to do about it. Combining sensor and shooter in the same (remote) platform may have ‘compressed the kill-chain’, as the Air Force puts it, and this is vital in an era of ‘just-in-time’, liquid war where everything happens so fast. Yet in another sense the kill-chain has been spectacularly extended: senior officers, ground force commanders, military lawyers, video analysts all have access to the feeds. There’s a wonderful passage in Brian Castner‘s All the ways we kill and die that captures the dilemma perfectly. ‘A human in the loop?’, Castner’s drone pilot complains.

‘Try two or three or a hundred humans in the loop. Gene was the eye of the needle, and the whole war and a thousand rich generals must pass through him… If they wanted to fly the fucking plane, they could come out and do it themselves.’

This is the networked warfare, scattered over multiple locations around the world, shown in Eye in the Sky. But the network often goes down and gets overloaded – it’s not a smooth and seamlessly functioning machine – and it is shot through with ambiguity, uncertainty and indecision. And often those eyes in the sky multiply rather than disperse the fog of war.

Angry Eyes (2)

Posted on October 7, 2015 by Derek Gregory

This is the second installment of my analysis of an air strike orchestrated by a Predator in Uruzgan province, Afghanistan on 21 February 2010; the first installment is here.

(4) Command and control?

What was happening in and around Khod was being followed not only by flight crews and image analysts in the continental United States but also by several Special Forces command posts or Operations Centers in Afghanistan. In ascending order these were:

(1) the base from which ODA 3124 had set out at Firebase Tinsley (formerly known as Cobra);

(2) Special Operations Task Force-12 (SOTF-12), based at Kandahar;

(3) Combined Joint Special Operations Task Force – Afghanistan (CJSOTF-A) based at Bagram.

Once the ODA 3124 left the wire, command and support passed to SOTF-12; the OD-B at Tinsley had limited resources and limited (and as it happens intermittent) communications access and could only monitor what was happening.

That was normal, but in fact both higher commands did more or less the same: and the investigating team was clearly appalled. At SOTF-12 all senior (field grade) officers were asleep during the period of ‘highest density of risk and threatening kinetic activity’ (although they had established ‘wake-up criteria’ for emergency situations). The Night Battle Captain had been in post for just three weeks and had been given little training in his role; he received a stream of SALT reports from the Ground Force Commander of ODA 3124 (which detailed Size of enemy force, Activity of enemy force, Location and Time of observation) but simply monitored the developing situation – what one investigating officer characterised as ‘a pretty passive kind of watching’.

The same was true at CJSOTF-A (the staff there monitored 15-25 missions a day, but this was the only active operation that had declared a potential Troops in Contact).

When the more experienced Day Battle Captain entered the Joint Operations Center at Kandahar and was briefed by the Night Battle Captain he was sufficiently concerned to send a runner to ask the Judge Advocate, a military lawyer, to come to the JOC. He believed the occupants of the vehicles were hostile but was not convinced that they posed an immediate threat to troops on the ground: ‘I wanted to hear someone who was extremely smart with the tactical directive and use of CAS [Close Air Support] in a situation I hadn’t seen before’.

This was a smart call for many reasons; the commander of US Special Forces, Brigadier General Edward Reeder, told the inquiry: ‘Honestly I don’t take a shit without one [a JAG], especially in this business’. Significantly, the Safety Observer at Creech testified that there was no ‘operational law attorney’ available onsite for aircrews conducting remote operations; conversely, JAGs were on the operations floor of CENTCOM’s Combined Air and Space Operations Centre at Ul Udeid Air Base and, as this case shows, they were available at operations centers established by subordinate commands in-theatre.

The JAG at Kandahar was not routinely called in for ‘Troops in Contact’ but on this occasion he was told ‘my Legal Opinion [was] needed at the OPCENT and that it wasn’t imminent but they wanted me to rush over there right away…’

Meanwhile up at Bagram Colonel Gus Benton, the commanding officer of CJSOTF-A, was being briefed by his second-in-command who understood that the Ground Force Commander’s intention was to allow the three vehicles to move closer to his position at Khod. He thought that made sound tactical sense.

‘I said that … is what we did, we let them come to us so we can get eyes on them. During my time I never let my guys engage with CAS if they couldn’t see it. I said that is great and COL [Benton] said “that is not fucking great” and left the room.’

At 0820, ten minutes after the JAG entered the JOC at Kandahar, while he was watching the Predator feed, the phone rang: it was Benton. He demanded Lt Colonel Brian Petit, the SOTF-12 commander, be woken up and brought to the phone:

He spectacularly mis-read the situation (not least because he mis-read the Predator feed). It was true that the vehicles were in open country, and not near any compounds or villages; but Benton consistently claimed that the vehicles were ‘travelling towards our objective’ whereas – as MG McHale’s investigating team pointed out to him – they were in fact moving away from Khod.

There had also been some, inconclusive discussion of a possible ‘High Value Target’ when the vehicles were first tracked, but the presence of a pre-approved target on the Joint Prioritised Effects List (Benton’s ‘JPEL moving along this road’) had never been confirmed and the Ground Force Commander had effectively discarded it (‘above my authority’, he said).

Certainly, the JAG at Kandahar read the situation differently:

When Benton rang off, the JAG went over to the Day Battle Captain and Lt Col Petit and recommended an Aerial Vehicle Interdiction (AVI) team be called in for a show of force to stop the vehicles without engaging the occupants in offensive action.

They agreed; in fact another Task Force also watching the Predator feed called to make the same suggestion, and the Fires Officer set about arranging to use their Apache helicopters to conduct an AVI:

The Fires Officer had been responsible for setting up the Restricted Operating Zone for aircraft supporting the ODA – de-conflicting the airspace and establishing what aircraft would be available – but its management was de-centralised:

‘I establish the ROZ, give the initial layout of what assets are going on, and then I pass that to the JTAC [Joint Terminal Attack Controller with the Ground Force Commander at Khod]. I pass the frequencies to the assets and the JTAC controls them from there.’

At 0630, long before all this frantic activity at Kandahar, the two OH-58s had arrived at a short hold location beyond the ‘range of enemy visual and audio detection’, and at 0730 they had left to refuel at Tarin Kowt. The Day Battle Captain and the Fires Officer both thought they were still off station. In fact, the helicopters had returned to hold at Tinsley/Cobra at 0810 and flat pitched to conserve fuel (which means they landed and left the rotor blades spinning but with no lift); thirty minutes later the JTAC called them forward and the Predator began to talk them on to the target.

The Day Battle Captain had another reason for thinking he and his colleagues in the JOC had more time. He maintained that the helicopters had been brought in not to engage the three vehicles but to provide air support if and when the ‘convoy’ reached Khod and the precautionary ‘AirTic’ turned into a real TIC or Troops in Contact:

‘… the CAS brought on station for his [the Ground Force Commander’s] use was not for the vehicles but for what we thought was going to be a large TIC on the objective. The weapons team that was pushed forward to his location was not for the vehicles, it was for the possibility of a large TIC on the objective based on the ICOM chatter that we had.’

That chimes with Benton’s second-in-command at Bagram, who also thought the Ground Force Commander was waiting for the ‘convoy’ to reach Khod, but neither witness explained the basis for their belief. It was presumably a string of transmissions from the JTAC to the Predator crew: at 0538 he told them the Ground Force Commander wanted to ‘keep tracking them and bring them in as close as we can until we have CCA up’ (referring to the Close Combat Attack helicopters, the OH-58s); shortly before 0630 he confirmed that the Ground Force Commander’s intent was to ‘permit the enemy to close, and we’ll engage them closer when they’re all consolidated’; and at 0818 he was still talking about allowing the vehicles to ‘close distance.’

Yet this does not account for the evident urgency with which the Day Battle Captain and the JAG were concerned to establish ‘hostile intent’ and ‘immediate threat’. When the vehicles were first spotted they were 5 km from Khod, and when they were attacked they were 12 km away across broken and difficult terrain: so what was the rush if the Ground Force Commander was continuing to exercise what the Army calls ‘tactical patience’ and wait for the vehicles to reach him and his force?

In fact, the messages from the Ground Force Commander had been mixed; throughout the night the JTAC had also repeatedly made it clear that the ODA commander’s intent was ‘to destroy the vehicles and the personnel’. The Ground Force Commander insisted that ‘sometime between 0820 and 0830’ he sent a SALT report to SOTF-12 to say that he was going to engage the target. Unfortunately there is no way to confirm this, because SOTF’s text records of the verbal SALT reports stopped at 0630 for reasons that were never disclosed (or perhaps never pursued), but it would explain why the JTAC’s log apparently showed the JAG contacting him at 0829 to confirm there were no women and children on the target. It would also account for testimony by one of the screeners, who realised that the helicopters were cleared to engage at 0835, ten minutes before the strike, when the NCO responsible for monitoring the Predator feed at SOTF-12 ‘dropped’ into the ‘ISR’ (I presume the relevant chat room window), and in response:

‘The MC [Mission Intelligence Co-ordinator at Creech] passed that the OH58 were cleared to engage the vehicles. We were all caught off guard… It seemed strange because we had called out that these vehicles were going west. I don’t know how they determined these vehicles to be hostile… I brought up a whisper [private chat] with the MC, I said are you sure, what are the time frames when they will be coming in, and the MC responded saying we don’t know their ETA and at that moment the first vehicle blew up…’

Should those watching the events unfold have been taken aback when the vehicles were attacked? According to the pilot of the Predator, he and his crew were surprised at the rapid escalation of events:

‘The strike ultimately came a little quicker than we expected…. we believed we were going to continue to follow, continue to pass up feeds… When he decided to engage with the helos when they did, it happened very quickly from our standpoint. I don’t have a lot of info or situational awareness of why the JTAC decided to use them when they did. When they actually came up … the JTAC switched me on frequencies. So we weren’t talking on the frequency I was talking to him on a different frequency to coordinate with the helos.

But their surprise was as nothing compared to the reaction of most observers when the first vehicle exploded. The officer in charge of the screeners and imagery analysts who had been scrutinising the Predator feed at Air Force Special Operations Command at Hurlburt Field in Florida couldn’t believe it:

The Day Battle Captain testified:

‘I did not feel that the ground force commander would use any kind of close air support whatsoever to engage those vehicles… Based on the information that I had and looking at the vehicles move away it did not appear that they were moving towards the ground forces…

… as we were watching the Predator feed the first vehicles exploded. And everyone in the OPSCEN was immediately shocked… The amount of time from when that course of action approved by the SOTF commander to when we actually saw the strike occur there was no time, there was not adequate time to inform the ground commander that that was the course of action decided by the CJSOTF commander… I have phones ringing left and right, talking to people, trying to explain things, you know we look up on the screen and it happened…’

The Fires Officer:

‘I don’t think at any time anyone communicated to the GFC [Ground Force Commander] not to strike these vehicles because it is not something that we normally do. We feel that if he is in contact with the Predator and the OH-58s that we sent out to screen which we were not aware of and he is on the ground he generally has a pretty good picture of what is going on. He might be more privy to some conversation that he had with the OH-58 than what we know about. We normally give the GFC pretty big leeway on how they operate and the same with the JTAC because he has control of the assets and I am not going to try to take his assets away.’

In short, the investigation concluded that the Ground Force Commander never knew that an Aerial Vehicle Interdiction was being arranged, and neither of his higher commands were aware that he had cleared the helicopters to attack the three vehicles.

But, as I will show next, what lay behind these failures of communication was a de-centralised, distributed and dispersed geography of militarised vision whose fields of view expanded, contracted and even closed at different locations engaged in the administration of military violence. Far from being a concerted performance of Donna Haraway‘s ‘God-trick’ – the ability to see everything from nowhere – this version of networked war was one in which nobody had a clear and full view of what was happening.

Part of this can be attributed to technical issues – the different fields of view available on different platforms, the low resolution of infra-red imagery (which Andrew Cockburn claims registers a visual acuity of 20/200, ‘the legal definition of blindness in the United States’), transmission interruptions, and the compression of full-colour imagery to accommodate bandwidth pressure. So for example:

But it is also a matter of different interpretive fields. Peter Asaro cautions:

‘The fact that the members of this team all have access to high-resolution imagery of the same situation does not mean that they all ‘‘see’’ the same thing. The visual content and interpretation of the visual scene is the product of analysis and negotiation among the team, as well as the context given by the situational awareness, which is itself constructed.’

The point is a sharp one: different visualities jostle and collide, and in the transactions between the observers the possibility of any synoptic ‘God-trick’ disappears. But it needs to be sharpened, because different people have differential access to the distributed stream of visual feeds, mIRC and radio communications. Here the disposition of bodies combines with the techno-cultural capacity to make sense of what was happening to fracture any ‘common operating picture’. As one officer at Kandahar put it:

‘We didn’t have eyes on, minus ISR platform, that we can all see, who watches what? All the discrepancies between who watches what. What I see may be different from what someone else might interpret on the ISR… ISR is not reliable; it is simply a video platform.’

He was talking specifically about the multiple lines of communication (and hence bases for interpretation) within his Operations Center: now multiply that across sites scattered across Afghanistan and the continental United States and it becomes clear that the contemporary ‘fog of war’ may be as much the result of too much information as too little.

To be continued.

Angry Eyes (1)

Posted on October 1, 2015 by Derek Gregory

I promised to post the notes for my presentation of ‘Angry Eyes: the God-Trick and the geography of militarised vision‘, and this is the first instalment (illuminated by some of the slides from the presentation). This isn’t the final, long-form version – and I would welcome comments and suggestions on these notes – but I hope it will provide something of a guide to where I’m coming from and where I’m going.

In many ways, this is a companion to ‘Dirty Dancing: drones and death in the borderlands’ (I’ll post the full text version of that shortly; until then see here, here and here), but that essay examines aerial violence in the Federally Administered Tribal Areas of Pakistan, tracing the long history of air strikes in the region, from Britain’s colonial ‘air policing’ of its North West Frontier through the repeated incursions by Afghan and Soviet aircraft during the Soviet occupation of Afghanistan (which are missing from most critical accounts) to today’s drone strikes directed by the CIA and air raids conducted by the Pakistan Air Force. ‘Angry Eyes’ focuses instead on a series of US air strikes inside Afghanistan.

(1) Eyes in the Sky

The history of aerial reconnaissance reveals an enduring intimacy between air operations and ground operations. Balloons and aircraft were essential adjuncts to army (and especially artillery) operations; before the First World War most commentator insisted that the primary use of military aircraft would be to act as spotters for artillery, enabling the guns to range on distant targets, and that bombing would never assume a major offensive role. As I’ve noted elsewhere, Orville Wright was among the sceptics: ‘I have never considered bomb-dropping as the most important function of the airplane,’ he told the New York Times in July 1917, ‘and I have no reason to change this opinion now that we have entered the war.’ For him – though he did not altogether discount the importance of striking particular targets, like the Krupp works at Essen – the key role of the aeroplane was reconnaissance (‘scouting’) for ground forces, including artillery: ‘About all that has been accomplished by either side from bomb dropping has been to kill a few non-combatants, and that will have no bearing on the result of the war.’ That was, of course, a short-sighted view – even in the First World War aircraft carried out strikes against targets on and far beyond the battlefield – but the sharper point is that the importance of aerial reconnaissance depended on a version of what today would be called networked war (albeit a desperately imperfect one) (see my ‘Gabriel’s Map [DOWNLOADS tab]; for the pre-war history of bombing, see here; for the bombing of Paris in the First World War see here).

Over the next 50 years the technologies of vision changed dramatically: from direct to indirect observation, from delayed to real-time reporting, and from still to full motion imagery.

And the ligatures between seeing (or sensing) and shooting steadily contracted until these functions were combined in a single platform – notably (but not exclusively) the Predator and the Reaper. Even then, wiring aerial operations to ground manoeuvres often (even usually) remains central:

Remarks like these speak directly to Donna Haraway’s cautionary critique of the ‘God-Trick’: the claim to see everything from nowhere, or at least from a privileged ‘vanishing point’. This has been made explicit by Lauren Wilcox in Bodies of Violence:

… the satellite systems and the drone’s video cameras mean that the bomber’s eye view is the God’s eye view of objectivity… this myth is put into practice in the apparatus of precision bombing, in which the view from above becomes the absolute truth, the view from nowhere.

And – Haraway’s point, which has been sharpened by Wilcox – is that this view from nowhere is, in some substantial sense, a view from no-body (and even of no-body). Here is Owen Sheers in his novel I saw a man:

“A U.S. drone strike.” That was all the press release said. No mention of Creech, screeners, Intel coordinator, an operator, a pilot. It was as if the Predator had been genuinely unmanned. As if there had been no hand behind its flight, no eye behind its cameras.

The appeal to the divine is thus more than a rhetorical device. One Predator pilot admitted that ‘Sometimes I felt like God hurling thunderbolts from afar.’ As Wilcox notes, then,

‘Precision bombing reproduces the illusion of a disembodied subject with not only a privileged view of the world, but the power to destroy all that it sees…. The posthuman bodies of precision bombers, relying on God’s eye, or panoptical, views are produced as masterful, yet benign, subjects, using superior technology to spare civilians from riskier forms of aerial bombardment.’

And yet there have been seemingly endless civilian casualties…

(2) Killing and casualties in Afghanistan

Throughout the US-led occupation of Afghanistan, air strikes have been the overwhelming cause of civilian deaths caused by coalition forces:

As Jason Lyall‘s marvellous work shows (below), air strikes have been concentrated in the south. I should note that the title of his map refers explicitly to ISAF air operations – I’m not sure if this includes those conducted under the aegis of Operation Enduring Freedom, a separate US-UK-Agfghan operation, although a primary source of his data is USAF Central Command’s Airpower Statistics. It makes a difference, for reasons I’ll explain later; the strike that is my primary focus took place in the south (in Uruzgan) but was in support of a Special Forces operation conducted under OEF.

In any event, most of those strikes have been carried out from conventional platforms – strike aircraft or attack helicopters – not drones (though notice how the data on weapons released from Predators and Reapers was rapidly removed from the regular Airpower Statistics issued by US Central Command):

This relates to a specific period, and one might expect drone strikes to become even more important as the numbers of US ground troops in Afghanistan fall. Even so:

in many, perhaps most of those cases drones have provided vital intelligence, surveillance and reconnaissance (ISR) capabilities: in effect, they may well have orchestrated the attacks even if they did not execute them;
according to the Bureau of Investigative Journalism [‘Tracking drone strikes in Afghanistan‘], ‘Afghanistan is the most drone bombed country in the world… Research by the Bureau… has found more than 1,000 drone attacks hit the country from the start of 2008 to the end of October 2012. In the same period, the Bureau has recorded 482 US drone strikes in Pakistan, Yemen, Somalia and Libya’; and
where drones have also carried out the attacks, Larry Lewis’s analysis of classified SIGACT data shows that ‘unmanned platforms [are] ten times more likely to cause civilian casualties than manned platforms’ (see also here)

There have been two main forms of air strike in Afghanistan. First, the US military carries out so-called ‘targeted killing’ there as well as elsewhere in the world; it has its own Joint Prioritized Effects List of people deemed to be legitimate military targets (see here and here), and the supposed capacity of its drones and their crews to put ‘warheads on foreheads’ means that they are often involved in these remote executions.

Even so, these operations have certainly caused the deaths of innocent civilians (see, notably, Kate Clark‘s forensic report on the Takhar attack in September 2010: more here).

Second, the US Air Force also provides close air support to ground troops – and civilian casualties are even more likely to result from these situations, known as ‘Troops in Contact’.

As Marc Garlasco noted when he was working for Human Rights Watch:

“When they have the time to plan things out and use all the collateral damage mitigation techniques and all the tools in their toolbox, they’ve gotten to the point where it is very rare for civilians to be harmed or killed in these attacks. When they have to do it on the fly and they are not able to use all these techniques, then civilians die.”

That said, it is simply wrong to claim that the US military is indifferent to civilian casualties. There have been several major studies of civilian casualties (see also here and here).

In addition, the juridification of later modern war means that military lawyers (JAGs) are closely involved in operational decisions (though the laws of war provide at best a limited shield for civilians and certainly do not outlaw their deaths); Rules of Engagement and Tactical Directives are issued and modified; and investigations into ‘civilian casualty incidents’ (CIVCAS) are established at the commander’s discretion. Of most relevance to my own argument is General Stanley McChrystal‘s Tactical Directive issued on 6 July 2009.

This was not window-dressing. Here is Michael Hastings in his by now infamous profile of McChrystal in Rolling Stone (8 July 2010):

McChrystal has issued some of the strictest directives to avoid civilian casualties that the U.S. military has ever encountered in a war zone. It’s “insurgent math,” as he calls it – for every innocent person you kill, you create 10 new enemies. He has ordered convoys to curtail their reckless driving, put restrictions on the use of air power and severely limited night raids. He regularly apologizes to Hamid Karzai when civilians are killed, and berates commanders responsible for civilian deaths. “For a while,” says one U.S. official, “the most dangerous place to be in Afghanistan was in front of McChrystal after a ‘civ cas’ incident.” The ISAF command has even discussed ways to make not killing into something you can win an award for: There’s talk of creating a new medal for “courageous restraint,” a buzzword that’s unlikely to gain much traction in the gung-ho culture of the U.S. military.

Indeed, McChrystal’s actions were fiercely criticised: see Charles Dunlap here (more here).

(3) Predator View

And so I turn to one of the most extensively documented CIVCAS incidents in Afghanistan: an attack on three vehicles near the village of Shahidi Hassas in Uruzgan province in February 2010, which killed at least 15-16 civilians and injured another 12. This has become the ‘signature strike’ for most critical commentaries on drone operations:

In the early morning of 21 February 2010 a US Special Forces team of 12 soldiers (these are always described as Operational Detachment Alpha: in this case ODA 3124) supported by 30 Afghan National Police officers and 30 Afghan National Army troops flew in on three Chinook helicopters to two locations near the village of Khod. This is an arid, mountainous region but Khod lies in a river valley where an extensive irrigation system has been constructed to create a ‘green zone’:

On a scale from 0 to 2, this was a ‘level 1 CONOPS’, which means that it was judged to pose a ‘medium risk’ to the troops with ‘some potential for political repercussions’. These are usually daylight cordon and search operations with air support. In this case the mission was to search the compounds in and around the village for a suspected IED factory and to disrupt ‘insurgent infrastructure’.

The Taliban clearly knew they were coming. While the troops waited for dawn the scanners on their MBITR radios picked up chatter urging the mujaheddin to gather for an attack, and they passed the frequency to an AC-130 gunship which was providing air support; through their night vision goggles the troops could see figures ducking into the cover provided by the irrigation ditches; and communications intercepted by other support aircraft, including a manned electronic signals intelligence platform referred to only by its call sign ‘Arrow 30’, confirmed a strong Taliban presence.

There were reports of vehicles moving towards the village from the south, and then headlights were detected five kilometres to the north. The AC-130 moved north to investigate. It had an extensive sensor suite on board but its resolution was insufficient for the crew to detect whether the occupants of the vehicles were armed [PID or ‘positive identification’ of a legitimate military target], and so they co-ordinated their surveillance with a Predator that had taken off from Kandahar Air Field and was controlled by a crew (call-sign KIRK 97) at Creech Air Force Base in Nevada. In addition to its Multi-Spectral Targeting System, the Predator was equipped with an ‘Air Handler’ that intercepted and geo-located wireless communications; this raw signals intelligence was handled by an ‘exploitation cell’ (almost certainly operated by a National Security Agency unit at Kandahar) who entered their findings into one of the chat-rooms monitored by the Predator and other operations centres that were involved in the mission.

When the AC-130 started to run low on fuel, the Predator took over ISR for the duration of the mission. The JTAC could not see the trucks from his position on the ground, and neither did he have access to the full-motion video feed from the Predator – the ODA was not equipped with a ruggedised laptop or ROVER [Remote Operational Video Enhanced Receiver] that should have been standard equipment (‘There’s one per base, and if it goes down you’re out of luck’) – and so he had to rely entirely on radio communications with the flight crew.

Throughout the night and into the morning the crew of the Predator interpreted more or less everything they saw on their screens as indicative of hostile intent: the trucks were a ‘convoy’ (at one stage they were referred to as ‘technical trucks’); the occupants were ‘Military Aged Males’ (’12-13 years old with a weapon is just as dangerous’); when they stopped to pray at dawn this was seen as a Taliban signifier (‘I mean, seriously, that’s what they do’); and when the trucks swung west, away from the direct route to Khod, this was interpreted as ‘tactical manoeuvring’ or ‘flanking’.

Eventually the ground force commander with ODA 3124 became convinced of hostile intent, and anticipated an imminent ‘Troops in Contact’. This in turn prompted the declaration of a precautionary ‘AirTIC’ to bring strike aircraft on station since the Predator only had one Hellfire missile onboard. The ground force commander was annoyed when fighter aircraft arrived (call-sign DUDE 01) – ‘I have fast movers over my station, my desire is to have rotary-wing aircraft’ – because he believed the engine noise would warn the target. In fact, the JTAC who had access to intercepts of Taliban radio communication confirmed that ‘as soon as he showed up everyone started talking about stopping movement’; coincidentally, as it happened, the vehicles immediately swung west, heading away from Khod.

Two US Army Kiowa combat helicopters (OH-58s, call-sign BAM-BAM) were now briefed for the attack.

Their situation map (below) confirmed this as a landscape of ever-present threat, and this imaginative geography was instrumental in the reading of the situation and the activation of the strike:

Meanwhile, the Predator’s sensor operator was juggling the image stream, switching from infrared to ‘Day TV’ and trying to sharpen the focus:

The helicopters had their own sensor system – a Mast Mounted Sight (MMS) – but its resolution was low (see below); they were also reluctant to come in low in case this warned the target:

In any case, there were severe limitations to what the pilots could see:

So, for all these reasons, they were reliant on what the Predator crew was telling them (they too had no access to the FMV feed from the Predator). They lined up for the shot, and the Predator crew keenly anticipated being able to ‘play clean up’. ‘As long as you keep somebody that we can shoot in the field of view,’ the Predator pilot told his Sensor Operator, ‘I’m happy.’

Throughout the mission the Predator crew had been communicating not only by radio but also through mIRC (internet relay chat); multiple windows were open during every mission, but KILLCHAIN was typed into the primary chat room to close down all ‘extraneous’ communications during the final run so that the crew could concentrate on executing the strike.

The ground commander through the JTAC cleared the helicopters to engage: ‘Type Three’ on the slide above refers to a control situation in which the JTAC can see neither the target nor the strike aircraft and wishes to authorise multiple attacks within a single engagement. According to the US Air Force’s protocols for terminal control:

Type 3 control does not require the JTAC to visually acquire the aircraft or the target; however, all targeting data must be coordinated through the supported commander’s battle staff (JP 3-09.3). During Type 3 control, JTACs provide attacking aircraft targeting restrictions (e.g., time, geographic boundaries, final attack heading, specific target set, etc.) and then grant a “blanket” weapons release clearance to meet the prescribed restrictions. The JTAC will monitor radio transmissions and other available digital information to maintain control of the engagement.

Hellfire missiles from the helicopters ripped into the trucks, and when the smoke cleared those watching – from the helicopters and on screens at multiple locations in Afghanistan and the continental United States – began to suspect that women and children were clearly in the field of view. A team from ODA 3124 was helicoptered in to co-ordinate the evacuation of casualties and to conduct a ‘sensitive site exploration’.

It turned out that the occupants of the vehicles were all Hazaras who were vehemently anti-Taliban (4,000 Hazara had been massacred by the Taliban at Mazar-i-Sharif in August 1998); they were going to Kandahar for a variety of reasons – shopkeepers going for supplies, a mechanic going to buy spare parts, students returning to school, patients seeking medical treatment, others simply looking for work – and they were travelling together (‘in convoy’) for safety through what they knew was Taliban territory. When civilian casualties were eventually confirmed – which is a story in itself – General McChrystal set up an Informal Investigation.

Most commentators (including me in “From a view to a kill”: DOWNLOADS tab) have endorsed the central conclusion reached by the Army investigation:

But this was not the conclusion reached by the USAF Commander’s Directed Investigation into the actions of the Predator crew (which McChrystal ordered when he received McHale’s report). Major-General Robert Otto conceded that ‘the Predator crew’s faulty communications clouded the picture on adolescents and allowed them to be transformed into military-aged males’, but he insisted that their actions were otherwise entirely professional:

‘Upon INFIL and throughout the operation, extensive Intercepted Communications (ICOM) chatter correlated with FMV and observed ground movement appeared to indicate a group of over thirty individuals were an insurgent convoy…. Kirk 97 did not display an inappropriate bias to go kinetic beyond the desire to “support the ground commander”. The crew was alert and ready to execute a kinetic operation but there was no resemblance to a “Top Gun” mentality.’

The reference was to a statement made to McHale’s team by a captain at Creech:

Be that as it may, the last clause– the desire ‘to help out and be part of this’ – is, I think, substantial. As I argued in “From a view to a kill”, most Predator and Reaper crews insist that they are not thousands of miles from the battlefield but just eighteen inches away: the distance from eye to screen. There is something profoundly immersive about the combination of full-motion video and live radio communication; perhaps the crews who operate these remote missions over-compensate for the physical distance to the troops on the ground by immersing themselves in a virtual distance that pre-disposes them to interpret so much that appears on their screens as hostile and threatening.

So far, so familiar. But two qualifications impose themselves.

First, virtually all the published accounts that I have read – and the one that I have published – draw on a detailed report by David S. Cloud, ‘Anatomy of an Afghan war tragedy‘, that appeared in the Los Angeles Times on 10 April 2011, which was based on a transcript of radio communications between the Predator crew, the helicopter pilots and a Joint Terminal Attack Controller (JTAC) who was relaying information to and from the ground force commander. But according to Andrew Cockburn, McHale’s original investigation compiled a hand-drawn timeline of events that ran for 66 feet around the four walls of a hangar he had commandeered for his office; his investigation ran to over 2,000 pages of evidence and transcripts. It’s a complicated, composite document: a record of transactions – of conversations, negotiations and interrogations inflected by the chain of command – made at different times, in different places and under different circumstances. Redactions make inference necessarily incomplete, and there are inevitably inconsistencies in the accounts offered by different witnesses. So I need to be cautious about producing a too coherent narrative – this is not the tightly integrated ‘network warfare’ described by Steve Niva in his excellent account of Joint Special Operations Command (‘Disappearing violence: JSOC and the Pentagon’s new cartography of networked warfare’, Security dialogue 44 (2013) 185-202).

Still, when you work through those materials a radically different picture of the administration of military violence emerge. In his important essay on ‘The necropolitics of drones’ (International Political Sociology 9 [2015] 113-127) Jamie Allinson uses McHale’s executive summary to demarcate the kill-chain involved in the incident:

The US military “kill chain” involved in the Uruzgan incident comprised ground troops, referred to in the text as “Operational Detachment Alpha” (ODA), the Predator Drone operators based at Creech Air Force Base in Nevada, the “screeners” processing information from the Predator video feeds at Hurlburt Field Base in Florida, and helicopter gunships known as ‘OH-58D’ in the text. The helicopters fired the actual missiles: but this was on the basis of decision made by drone operators based on their interpretation of what the screeners said.

But – as I’ll show in the second instalment – the kill-chain was far more extensive and included two Special Forces operations centers at Kandahar and Bagram that were responsible for overseeing and supporting the mission of ODA-3124.

The significance of this becomes clearer – my second qualification – if the air strike in Uruzgan is analysed not in isolation but in relation to other air strikes that also produced unintended casualties. As a matter of fact, official military investigations are required to be independent; they are not allowed to refer to previous incidents and, indeed, JAGs who advise on targeting do not routinely invoke what we might think of as a sort of ‘case law’ either. But if this air strike is read in relation to two others – an attack by two F-15E strike aircraft on a tanker hijacked by the Taliban near Kunduz on 4 September 2009, and an attack carried out by a Predator in the Sangin Valley on 6 April 2011 – then revealing parallels come into view. All three were supposed to involve ‘Troops in Contact’; the visual feeds that framed each incident – and through which the targets were constituted as targets – were highly ambiguous and even misleading; and the role of the ground force commander and the operations centers that were supposed to provide support turns out to have been critical in all three cases.

To be continued.