The emergence of lethal surveillance: Watching and killing in the history of drone technology

Introduction

There is a growing body of scholarship on drone technology across critical geography and security studies that is attentive to the materialities of contemporary war and that seeks to place the drone within a wider set of practices and assemblages and histories. The drone here is analyzed as a complex object emerging out of tangled histories of colonial control and state violence (Shaw and Akhter, 2012, 2014; Satia, 2014), assassination (Grayson, 2012), air power strategy (Kaplan, 2006), and public debate and secrecy (Walters, 2014). Particular attention is given to the aerial view from above and the space of air more broadly in shaping contemporary drone wars (Adey et al., 2011, 2013). As Tyler Wall and Torin Monahan remind us, ‘Drones are a combination of the new and the old: a new aerial surveillance and killing system with capabilities previously not offered by conventional air power, coupled with an older cosmic view of air mastery through technological speed, verticality, and vision’ (Wall and Monahan, 2011: 241). In his recent writings on drones and modern warfare, Derek Gregory has also advocated taking into account a historical genealogy of the drone. As he shows, situating the drone within a longer history of bombing and visual surveillance helps us to better understand the geographies of war that the drone shapes and is embedded in today (Gregory, 2013). Across his writings on drones, this genealogy allows Gregory to see the new visibilities produced by the contemporary Predator as effects of a privileged and one-sided view that often renders invisible the violence and history of colonialism that it is shaped by and continues to produce (Gregory, 2011b). It further allows Gregory to place the drone within a contemporary shift in warfare more generally, arguing that the drone is one of the technologies of what he calls ‘everywhere war’ – war that is characterized by ‘event-ful’ violence, that can occur anywhere, and that is more spatially blurred, or ‘slippery’, than the spaces of traditional warfare (Gregory, 2011a).

In this article, I add to these studies of contemporary drone wars by examining several significant periods in the development of drone technology from the beginning of the 20th century to 11 September 2001. Turning to this history reveals two trends in Western warfare as they relate to the drone: (1) the increasing importance of intelligence, surveillance, and reconnaissance (ISR) and (2) the development of dynamic targeting. These trends converge today in a practice I call lethal surveillance, where ISR capabilities are linked directly to targeted killing in an attempt to close the temporal and spatial gap between the two. In other words, lethal surveillance is a practice in which mechanisms of surveillance and knowledge production and decisions on life and death have become one and the same. What is distinctive about today’s drone is therefore not its ability to minimize risk by removing the human from war or to target at a distance – these are important characteristics of the drone today as well as historically, but they do not explain why the drone now has become such a central technology of war. Rather, its significance lies in how it mobilizes the twin histories of the practices of ISR (and the use of information in war more generally) and lethal targeting. Through this examination it becomes clear not only that identifying lethal surveillance helps us to better understand the complexities of contemporary drone wars but, further, that lethal surveillance reflects a tendency across modern Western warfare and violence and should point us toward a much longer history of scientific development, war, and liberal governance when analyzing the current deployment and use of drones.

Viewing the drone through the lens of the emergence of lethal surveillance makes two major contributions to our conceptual and political understandings of contemporary war and security. The first is that we do not view the drone in isolation, but rather see it as intimately connected to other military and security practices, including domestic state surveillance, military mapping, and special operations targeting programs, to list a few. Related to this, lethal surveillance is a practice of modern warfare and security that is perhaps most clearly seen in the contemporary drone strike but is, importantly, not limited to the drone. Turning to the history of drone development helps us to see the range of practices and technologies lethal surveillance encompasses. Doing so causes us to ask new and different questions about drone warfare, and the security landscape more generally, as I explore in the last section of the article. Second, on a more political register, lethal surveillance helps us begin to understand why killer drones are being used increasingly today. As mentioned earlier, more mainstream narratives that see the drone as a result of more sanitized warfare or risk reduction are complicated by the tendency toward lethal surveillance across the history of Western modern war. Effective critiques of drone warfare, therefore, must account for this history of violence.

In the next three sections, I focus on significant periods in the history of US drone development: the early years during World Wars I and II, the Cold War, and the 1990s. I selected these periods because they highlight key developments in drone technology across the 20th century. This is by no means meant to be a complete or linear history of drone technology; rather, the engagement with programs within these periods highlights the aspects of lethal surveillance that today come together in contemporary drone warfare. The research for the first period came primarily from the British National Archives, where I accessed records on pilotless aircraft from various military aviation departments, aircraft production files, the War Office, and the offices of the Armed Forces, among others. Primary research for the second and third periods came from a variety of sources, including the National Security Archives maintained by George Washington University, congressional committee and US General Accounting Office (GAO) reports, and military and policy reports from various US government agencies and think-tanks. At the end of the article, I discuss how these three periods help us to see the emergence of lethal surveillance today and the significance of lethal surveillance for our study of contemporary drone technology.

Early years: Pilotless bombing

Perhaps surprisingly – because the idea and development of unmanned aircraft today is so pointed toward the future – the history of the drone traces alongside, and even predates, the history of manned aircraft. ¹ If we take into account the use of balloons and even the history of automata more generally, the history of drone technology could be traced back even further (New York Times, 1863). Here I focus on the development of drones around World Wars I and II and after, and across these time periods we see through this lens of drone technology some notable key changes in how modern Western war is waged. More importantly, though the focus shifts at different times between surveillance and targeting, we can see a tendency toward lethal surveillance reveal itself across these periods.

Early research and development on what we can identify as crude precursors to today’s Predator technology dates to the beginning of the 20th century and is shaped by the events, challenges, and technologies used in World Wars I and II. A focus here on two major programs in the United Kingdom ² serves to demonstrate the desired goals at the time for developing drone technology for use in war as well as the challenges faced in achieving these goals. ³ These programs were set within a broader context of a radically changing landscape of modern warfare, both in terms of the range and scope of technologies of killing deployed and the spaces in which war was fought, notably with a significant extension of the battlefield into the air. ⁴ In comparison to the battlespaces of today’s Western wars, World War II is best characterized as exemplary of ‘traditional’ modern war, with relatively fixed territories and aims against fixed and defined targets. Despite this, it is important to remember that at this time Western powers were trying to understand and develop the concepts and practices of strategic and tactical bombing (Van Creveld, 1989: 191; McFarland, 1995: 3–4). Therefore, to a large extent the early drone programs reflect less an understanding of the potential for unmanned technologies in war as distinct from manned ones, than the working out of the possibilities of aerial bombing in war more generally.

The Larynx and the Ram, two major early UK projects, produced different conceptions of how unmanned aircraft might be used as a weapon of war. The Larynx was initially designed to be mechanically controlled and to fly a predetermined path or ‘home’ in on a target. At the end of its flight, which ranged from 200–500 miles, it was programmed to either drop bombs or dive into the target itself, effectively becoming its own bomb. The Ram, by comparison, was smaller and designed to be wirelessly controlled from an operator in a nearby manned aircraft. It got its name from its ability to ‘ram’ into enemy air formations, in this way acting as a decoy to draw enemy fire or for bombing enemy formations while piloted aircraft remained at a safe distance.

The development of both technologies was characterized over this period by a series of starts and stops due to a variety of factors, but namely because of budgeting priorities – funding was often interrupted because of the war effort – and technical setbacks primarily related to achieving an adequate degree of accuracy from the weapons. The Larynx and the Ram were seen as desirable for three main reasons.

First, they extended the range and kind of attack, whether against a target on the ground or another in the air. In part, the aim of extending the point of attack beyond the human pilot reflected a goal of limiting risk to the pilot’s life, but of greater concern was figuring out how to fly at night or in poor weather conditions, in which the human pilot was at a disadvantage. The work on the Larynx and the Ram also spurred new ideas about how the military might attack from the air, and the drone fostered the concept of the long-range attack. ⁵ It also opened up the possibility of continuous (and inexpensive) bombing, to effect the ‘local collapse of morale’. ⁶

Second, and related, the development of these drone programs was promoted because of the perceived ‘moral effect’ it would have on the enemy. There are two aspects to this perception. First, an unmanned long-range bombing attack was seen to be so terrifying and unstoppable – in both its continuity and its surprise – that it would not only cause crippling physical damage, but would also have a significant impact on the resolve of the civilian population and the enemy’s military. ⁷ Second, the Air Ministry may have come to this conclusion because of its own fears. ⁸ Reports of German pilotless aircraft and reports of attacks by the German V-1 were a clear motivator for continuing work on the Larynx in particular. ⁹

Third, the Air Ministry saw the potential of pilotless aircraft to reduce costs and risks to personnel as an advantage over manned aircraft. Unlike today’s discussion of limiting risks in the sense of saving soldiers’ lives, the Air Ministry primarily framed the reduction of risk to personnel in the ability to increase the reach of the aircraft by flying in areas previously seen as too dangerous. Also of concern was the potential of the drone to provide a more cost-effective means of bombing due to fewer personnel on staff and potentially cheaper building materials. Promises of mass-producing the Larynx, for example, put the costs of the aircraft significantly lower than its manned counterparts, although this was never realized in practice. ¹⁰ The inability to radically drive down cost was another reason the drone programs were put on hold. ¹¹

While never making it to the mass-production phase, both the Larynx and the Ram underwent significant tests. In early tests of the Larynx one observer noted that ‘some of the aeroplanes went out of control and performed evolutions unknown to man’. ¹² Tests in 1925 and then again in 1928 in Iraq were of mixed success, contributing to the intermittent starting and stopping of the research program. The largest challenge to developers was accuracy due to the reliance of the Larynx’s mechanical control on unpredictable meteorological data. ¹³ Frustrated with this lack of reliability, one official noted in 1925 that the Larynx was essentially ‘a very long-ranged but very inaccurate and expensive gun’. ¹⁴ In an effort to increase the accuracy, a radio-controlled version was proposed, but controlling it wirelessly at such a great distance outside of visual range proved difficult. Similar technical difficulties plagued the Ram. Not only was controlling the Ram from the control aircraft difficult mid-flight – as one operator observed, ‘one loses all sense of distance between the target and the Ram’ ¹⁵ – but the wireless technology at the time was not sufficient for accurate control. ¹⁶ Because of these difficulties of achieving sufficient control and accuracy, as well as limited funding due to the war effort, both projects were shuttered by the end of World War II in favor of greater focus on long-range missile development. It can be argued, however, that these early experiments laid much of the groundwork for the missile programs. ¹⁷

Cold War: Unmanned reconnaissance

Missiles and nuclear warheads framed the security landscape in the years after World War II. While research on drone technology halted briefly after the war, it reemerged in the United States during this second period with the drone conceptualized in a radically new way. The drone now served as a tool for ISR, and was one of the many technologies developed by the US government to gather an array of intelligence data through electronic means, including data from photographic imagery, radar, and other electronic signals. This period set the groundwork for what has now developed into electronic or information warfare (Poteat, 1998).

Intelligence, surveillance, and reconnaissance have always been significant aspects of warfare (Keegan, 2003), but there is a notable shift after World War II where information takes on new strategic and tactical roles (Bousquet, 2009). The Cold War marks the beginning of electronic and information warfare with the growth of signals intelligence (SIGINT) often in favor over human intelligence (HUMINT). As a result of this shift, emphasis was placed on gathering large amounts of data in a variety of forms, and it is at this time that systems analysis was pioneered by analysts, for example at the RAND Corporation, in order to make sense of it. ¹⁸ In addition to gathering intelligence, the United States also sought to intercept, destroy, or fool enemy radar and other electronic communications, and drones were used to some extent for these purposes as well (Ehrhard, 2010: 25).

The biggest motivation for investing again in drone development programs is tied to the success and limitations of the U-2 spy plane. This context of manned aerial reconnaissance is crucial to understanding how the drone was conceptualized and why it was ultimately not that popular and did not get developed further during this time. ¹⁹ Gathering information from the other side of the Iron Curtain was one of the biggest security challenges throughout the Cold War, and the means of gathering this intelligence shifted significantly from human to technological. The kinds of data sought changed as well, with emphasis placed on radar, electronic signals, data transmissions, and the like. The development of the U-2, which carried on board a variety of sensors, was led by the CIA with support from the Air Force in response to the need to penetrate Soviet territory while avoiding surface-to-air missiles. Not only did this require a technological breakthrough – the ability to fly at high altitude and later to fly with stealth – but it also marked a notable policy shift in using aircraft for intelligence gathering. As a CIA history noted, ‘Peacetime reconnaissance flights over the territory of a potential enemy power thus became national policy. Moreover, to reduce the danger of conflict, the President entrusted this mission not to the armed forces, but to a civilian agency – the CIA’ (Pedlow and Welzenbach, 1992: 322). Furthermore, with the development of the U-2 and other reconnaissance capabilities, we see the increasing desire for faster means of gathering information. As President Eisenhower remarked after being shown photographs from an early 1956 U-2 flight which made multiple passes in a short period of time over Egypt in response to a joint English and French attack there, ‘Ten-minute reconnaissance, now that’s a goal to shoot for!’ (Pedlow and Welzenbach, 1992: 119). ²⁰

While U-2 missions were largely successful during the Cold War in obtaining information, there was often reluctance to use the aircraft because of the potential political risk of the plane coming down in Soviet territory. Some government contractors, including Ryan Aeronautical Company and Lockheed (Skunk Works), began developing reconnaissance drone prototypes in the late 1950s in part in an effort to minimize this risk. After Francis Gary Powers was shot down over the Soviet Union in 1960, Ryan obtained a contract from the Air Force to develop its Firebee target drone into a reconnaissance plane that would eventually be nicknamed the Lightning Bug (Wagner, 1982: 15). As U-2 flights were halted after the Powers incident, a Ryan official noted: ‘We believe that for some missions the cockpit must be taken out of the airplane, put on the ground, and let the pilot fly the vehicle from there’ (Wagner, 1982: 13). Similarly, in 1962 Lockheed developed the Tagboard, which was designed to be released from another aircraft (Rich and Janos, 1994: 262). Through the Douglas Aircraft Company, the CIA also directly sponsored the development of Aquiline, a long-range reconnaissance drone, and Axillary, a short-range drone, although both projects were tabled by the early 1970s (Pedlow and Welzenbach, 1992: 339).

Both the Tagboard and the Lightning Bug flew missions over China and across Southeast Asia in the late 1960s with varying degrees of success. They generally flew pre-programmed missions and took photographs. They were also used to identify and map enemy missile sites, either with radar and other monitoring equipment or by acting as a decoy to draw surface-to-air missile fire (Schuster, 2013). This work carried on into the Vietnam War, where it is estimated that drones were flown in approximately 3,000 mission flights (Comptroller General of the United States, 1981: 1). There it seems that the drone missions expanded somewhat and the drones were used for taking photographs of targets for airstrikes, recording post-bombing damage, uncovering and identifying new targets, and dispersing propaganda material and electronic listening devices (Comptroller General of the United States, 1981: 2). In an assessment of the use of drones in Vietnam made at the beginning of the 1980s, the Government Accounting Office noted that the ‘most advantageous’ military use for drones was in hostile environments or for boring or tiring missions. Despite some promising success, ‘due to the shortage of RPV [remote piloted vehicle] equipment, many of these applications were severely handicapped’ (Comptroller General of the United States, 1981: 2).

This observation by the GAO points to a common theme and an intriguing question of drone research and use throughout the Cold War period: despite the usefulness and stated need for the technology, it never really got off the ground, primarily due to lack of program investment, interest, and funding. There are three main reasons for this. The first is the ‘pro-pilot bias’ within the Air Force; drones were seen as ‘too drab and unexciting’ and as limiting pilot career advancement (Comptroller General of the United States, 1981: 20). In fact, it was the Army that continued a consistent drone research and development program after the Vietnam War. The second reason was the lack of knowledge across the government about the various deployments of drones because the programs were so classified and compartmentalized. This also limited the programs’ public visibility. Third, in part because of this limited visibility, the drone competed with the more popular satellite and stealth technology programs being developed at the time. Despite being promoted as less expensive and less risky politically when compared to the U-2, the small group of researchers and officials who promoted drones generally could not raise the profile of their programs. ²¹ As the editor of Armed Forces Journal International noted in 1981, ‘RPVs may have met their enemy. Could it be us?’ (Wagner, 1982: iv).

While the reconnaissance drone programs were overshadowed and ultimately shuttered during the Cold War, they clearly contributed to a larger effort of extending the geographical range and technological capability of ISR. As an early history of the drone written at the time concluded:

And where from here? The applications of a plane that thinks and responds as though it were a pilot are unlimited. Already the Firebee has proven it has the eyes of a man through television. It can be remotely controlled by a ‘pilot’ from a presentation of television pictures secured by a camera in the vehicle and relayed in real time. Digital programmers give it a capability of thinking and responding at precise intervals to complicated commands. With a photographic memory, it is capable of returning intelligence in an environment too hostile for man to survive. This is what RPV is all about. (Wagner, 1982: 91)

This comment clearly anticipates the next 20 or so years of drone development.

Kosovo air war: The Predator and dynamic targeting

The two most successful drone research and development programs pursued by the US government in the 1990s – the Predator and the Global Hawk – sought to increase ISR capabilities of the drone as well as its data transmission capability. ²² US military engagements during this decade emphasized high-tech air power and surgical strike operations, particularly in ones like the NATO airstrikes over Kosovo, which were now labeled as operations of ‘humanitarian war’. With a focus on ‘effective and swift’ airstrikes, the United States could engage ‘on behalf’ of others while limiting risk to itself. ²³ This logic of limiting risk through a focus on airstrikes clearly carried over to the second Iraq war after 9/11 with the first phase concept of ‘shock and awe’, although it should be noted that despite these operational goals, the wars in both Kosovo and Iraq lasted much longer than predicted by the military. Also during the 1990s, the role of air power and intelligence gathering began to shift to become more engaged in the process of dynamic targeting, where new targets are identified as the operations continue (called ‘flex targeting’ in the Kosovo air war) (Cordesman, 2003: 116). ²⁴ Because of all of these changes, the necessity of the transmission of real time information became increasingly clear.

In the 1980s some limited drone research continued within the military, mostly by the Army. As mentioned previously, that the drone programs stayed alive in the Army is most likely an indication of the pilot bias within the Air Force, although the commitment to developing drone programs increased across the Department of Defense (DoD) slowly over the decade (US General Accounting Office, 1984: 18). The Army’s program focused on target designation and ISR, seeking to provide the ‘eyes of man’ on the battlefield to remote viewers. As a GAO report notes, ‘the system is to extend the eyes of the brigade and division elements during combat to as far as the range of their artillery weapons’ (Comptroller General of the United States, 1981: 12).

In the mid-1990s the Defense Advanced Research Projects Agency (DARPA) established a high-altitude unmanned aerial vehicle (UAV) program, which led to the development of two of the primary larger drone systems used today: the Predator and Global Hawk. The DARPA program came out of an effort by Congress at the end of the 1980s to consolidate the DoD’s UAV development programs. One of the requirements for DoD UAV development was ‘Long Endurance Reconnaissance, Surveillance, and Target Acquisition (RSTA) Capability’ (Drezner et al., 1999: 5–6). As a RAND Corporation report explains:

The intent was to provide warfighting commanders in chief (CINCs) with the capability to conduct wide-area, near-real-time RSTA, command and control, SIGINT, electronic warfare, and special-operations missions during peacetime and all levels of war. The CINCs would be able to exercise this capability against defended and denied areas over extended periods of time. (Drezner et al., 1999: 6)

Drones would allow for persistent or loitering surveillance capabilities compared to the episodic surveillance provided by satellites and manned aircraft (Ehrhard, 2010). Other requirements included short-range drones and medium-altitude endurance drones. The Predator, considered a medium-altitude drone, was bid on and developed by General Atomics in 1995, envisioned primarily for electronic and photographic intelligence gathering. Teledyne Ryan (previously Ryan Aeronautical Company) began work on the Global Hawk (larger and higher-altitude than the Predator) in 1995 as well.

These programs were put to the test in the late 1990s, particularly in the US and NATO airstrikes over Kosovo. NATO used a variety of drones for intelligence gathering over the Balkans from 1995–99. These included the Predator, the Pioneer (developed by the Navy), and a German drone (Dixon, 2000: 4). As the conflict continued, however, and in particular when Operation Allied Force intensified the bombing of Kosovo in 1999, drones began to play a larger role in real-time target identification and assessment in ongoing operations (Dixon, 2000: 5). Operation Allied Force was in part significant because it ‘saw an unprecedented use of unmanned aerial vehicles’ (US DoD quoted in Cordesman, 2003: 299).

Incorporating drones into the targeting process – the ‘hunter-killer’ strategy – proved beneficial for a number of reasons. At an operational level, drones shortened the amount of time it took to go through the targeting process. In relaying real-time data, drones could direct a bomber to a target, watch the target (often undetected), and assess the aftermath of the bombing (Dixon, 2000: 7). This helped facilitate the development of more dynamic targeting. As one Air Force General described:

We must fully develop the technology and tactics to rapidly strike targets. To do this, we need equipment that will provide real-time imagery and target location directly to our fighter and bomber crews. This will allow us to reduce the barriers between the ‘sensor’ and the ‘shooter’ in the targeting cycle – what we call ‘attacking the seams’. (Cordesman, 2003: 117)

Drones were also used after the end of the airstrikes to monitor peacekeeping operations and treaty enforcement. And more generally, the use of drones in the targeting support capacity also proved beneficial in maintaining public support for NATO intervention. The drones could carry out important reconnaissance work without risking pilots’ lives, especially since NATO had set a limit for how low manned planes could fly in an attempt to avoid casualties. Billed as a ‘humanitarian war’, NATO’s intervention focused predominantly on airstrikes, and the drone allowed it to both decrease risk as well as save money. ²⁵

The use of drones by NATO in Kosovo presented some challenges, however. While making the targeting process more efficient, there were still places where the time between target identification, assessment, and attack could be further shortened. At least two aircraft, manned or unmanned, had to confirm the target. The target’s location had to then be relayed to the bomber aircraft or to a missile control center. Each extra step took more time and increased the risk of communication error in the transmission of data. The more dynamic the targeting process became during the bombing campaign, the greater the need for a speedier and more accurate process became as well. Arming the Predator in 2001 was seen as one way of further contracting the kill-chain by putting more of its functions into one platform.

Second, while the drone itself could be navigated using GPS satellite links, when a drone had identified a target there was a difficulty in communicating its precise location – in translating the imagery that the drone was producing and feeding back into geographical coordinates. This was partially solved by the development of the PowerScene computer program, which mapped the drone video feed into a terrain visualization program. From this the coordinates could usually be determined and sent to the bombing aircraft (Dixon, 2000: 16). In general, the military’s handling of geographic data, and especially the interoperability of this data between systems, needed improvement. As Cordesman concluded, ‘More generally, Kosovo reinforced the need for a comprehensive US effort to create a detailed topographic map of the entire earth for military purposes’ (2003: 304). In a Borgesian development, after Kosovo the DoD considered creating a ‘Global Information Grid’ which ‘would cover the entire world, and link all of the service and intelligence systems together and which is tied to both operators and support elements … It is unclear how quickly such a system can be created, but it conceptually would change much of the US approach to warfare’ (Cordesman, 2003: 299).

Finally, the production of real-time video feed from the drones created some new and unexpected problems, which one assessment of Operation Allied Forced termed ‘UAV micromanagement’ (Dixon, 2000: 19). Video feeds could be sent to and monitored from practically anywhere – for example, as far away as Washington, DC – creating a situation where more senior commanders and officials could be involved in day-to-day operations. As one report noted:

According to CAOC [Combined Air Operations Center] staffers, the General (SACEUR) would on occasions telephone the CAOC demanding that UAVs break off from their tasking and go and look at things of interest to him. [General] Clark was in daily telephone contact with Kosovo Liberation Army chief, Hashim Thaci, and immediately after these conversations would dispatch a UAV to look at what often turned out to be spurious targets. (Dixon, 2000: 12)

Despite these challenges, with NATO’s use of drones in Kosovo it became clear that drone reconnaissance functions could be linked directly to real or near-real time targeting. In fact, these functions were seen to make the targeting process faster, more efficient, and more effective in a more dynamic targeting environment. Linking reconnaissance to increased lethality, Cordesman writes:

Kosovo again demonstrated the need for theater-level expeditionary capability to rapidly deploy the intelligence, reconnaissance, targeting, and battle damage assessment assets needed to get maximum benefit from both air power and long-range land artillery systems. The combination of JSTARS, the ABCCC, U-2, unmanned aerial vehicles (UAVs), and satellite reconnaissance coverage – plus target analysis – proved critical in giving NATO strike-attack sorties more lethality. (Cordesman, 2003: 296)

Furthermore, the focus on air power, and the use of drones in particular, allowed NATO countries to limit the political risk of sending troops into combat – setting a future precedent of a desire for limited deployment.

The emergence of lethal surveillance and contemporary drone warfare

The drone programs in the 1990s, particularly the Predator and the Global Hawk, saw the most consistent development and success compared to any other US drone program prior to then. However, the 9/11 attacks and the subsequent invasion of Afghanistan brought unprecedented interest in the drone. A slew of money came from Congress to arm the Predator and to produce more unmanned aircraft, which were quickly deployed to Afghanistan and then Iraq. The Global Hawk drone was also sent to Afghanistan, with mixed success in the field, even though it had not yet graduated from the development phase. The popularity of drones with Congress grew to such an extent that, in 2007, Congress required the DoD to pursue unmanned aircraft acquisition programs over manned ones, unless the need for a manned aircraft could be justified (Gertler, 2012). In military operations in Iraq and Afghanistan drones have played a key role in both counter-terrorism and counterinsurgency operations, ²⁶ providing ‘eyes in the sky’ with the ability to monitor the movement of individuals for long periods of time and also to provide precision targeting capability. Furthermore, there is a push to make drones more automated, stemming from the increasing use of ‘pattern of life’ analysis and signature strikes by the Obama administration.

Perhaps more significantly, the use of drones has also expanded widely beyond the battlefields of Iraq and Afghanistan, and this aspect of contemporary drone wars is ever evolving. Armed drones have been used to extend and prosecute the global war on terror to Pakistan, Yemen, and Somalia by both the military and the CIA. One of the most high profile cases of targeted killings in the last few years might be that of Anwar al-Awlaki on 20 September 2011, significant in part because he was an American citizen. ²⁷ The United States also deployed drones to assist with regime change in Libya, and the use of drones was a part of the Obama administration’s case for arguing that the US military engagement in Libya was not actually a war. ²⁸ Further, unarmed drones are now widely used by US Customs to patrol the US–Mexican border and by state and local police departments. ²⁹

While this is only a cursory sketch of recent developments, what is clear is that the contemporary landscape of the use of drones is complex and evolving. However, bringing the aforementioned history to bear on the present, we can see that since 9/11 the processes of surveillance and targeting have started to collapse into each other, forming a single process of lethal surveillance. From these three historical periods we see first that, over time, intelligence and surveillance takes on a more prominent role. In this capacity drones initially play more of a sideline or supporting role, for example providing a nonhuman alternative to other surveillance and reconnaissance technologies during the Cold War. Drones did not offer significantly different capabilities in terms of intelligence gathering, but they could mitigate the potential political fallout of a downed manned spy plane. The drones built by Ryan and Lockheed during this time were therefore one of the many technological solutions for mass information gathering behind the Iron Curtain.

In the 1990s, however, drones like the Predator were no longer a ‘nonhuman’ alternative, but instead provided a new ‘more than human’ (or ‘beyond human’) capability. In the Kosovo air war, for example, targets could be better observed through drones compared to manned planes. Drones became important tools on a complex and dynamic battlefield for shortening the time between identification and execution of targets and for assessing bomb damage, due to their ability to stay above the area observed. With drones like the Global Hawk and the Predator, this trend has only intensified with increased emphasis on data-processing capacities and automation. These drones are essentially information-processing machines (especially taking into account the larger assemblage of operator bases and command centers that the drone is embedded in), and this makes the drone significant for being one of a number of technologies that are attempting to gather and process data with increasing speed and greater analytical capabilities. The push to automate drones is a part of this development, and is important in the context of lethal surveillance less for the ethical dilemmas posed by robotic killing (although these are surely not to be ignored) than because of what this shift to automation means for information in war. What automation signals in part is the speeding up of war and its related information overload, such that we as humans can no longer participate in war in the same way because it moves too fast and the analysis is too complex. Because of this, proponents of automation argue that machines (because they are better data possessors) will inevitably be necessary as well as better at identifying targets and making decisions to kill.

Second, this trend of developing greater and more complex information gathering and processing capabilities is tied to changes in bombing and targeting more generally in Western military engagements, and over these same periods we see dramatic changes in targeting, from a focus primarily on targeting fixed sites or areas that are determined ahead of time to an ever more dynamic targeting process. During World War II and much of the Cold War, the targets that were selected for real or potential bombing campaigns (such as nuclear deterrence plans) were primarily fixed in place, for example infrastructure targets like bridges, roads, or military sites, or even towns or cities. The earliest drones, like the Larynx, were designed to target these kinds of fixed locations. And drones used during the Cold War assisted in locating and identifying them, along with a host of other technologies. In the 1990s, however, targeting practices began to change. Targets during this time were still largely fixed sites, but target lists were updated more regularly, creating more dynamic lists. Thus the space of the battlefield – and, more specifically, where bombs were dropped – was viewed as increasingly dynamic. This was fine-tuned even more in the post-9/11 campaigns in Afghanistan and Iraq (e.g. the ‘shock-and-awe’ air campaign at the start of the war). The war on terror, however, brings a significant shift in this development, where not only is targeting accelerating today, but there is also a qualitative change in the target from the site to the individual (Gregory, 2014: 14).

With the emergence of lethal surveillance after 9/11, we have therefore a merging of two key processes of modern warfare in intelligence gathering/information processing and targeting. How these have each changed and developed over the 20th century impacts how they are deployed today as well as how, in fusing together, they produce new practices and landscapes of war. In other words, the military’s capability now to identify and kill a person through the same action, while technically new, is the culmination of a longer modern history of warfare inflected with developing histories of scientific knowledge production and practices of killing and control.

This history, and the emergence of lethal surveillance, should cause us therefore to also ask different questions and widen our view when trying to make sense of today’s drone wars. For example, given the growing importance of information and ISR to warfare and the longer modern history of scientific and technological development these information gathering practices are embedded in, what kinds of logics of mastery and control form the foundation of the expanding use of drones? Or, with the collapsing together of techniques of knowledge production and targeting as drone strikes become more individuated and rely on pattern of life analysis, how might we understand changes in the way liberal war is now waged as biopolitical practices are being increasingly deployed in warfare? Or finally, in what ways does the practice of lethal surveillance exceed the drone, pointing to connected practices such as the NSA surveillance and data collection programs, the rise of the Joint Special Operations Command (Niva, 2013), or US border security and domestic police practices? These questions, which are only a few that present themselves from this history, require a greater critical engagement not only with today’s deployment of drones, but also with how techniques of power and knowledge production are merging with practices of killing and control in new and evolving ways.

Viewing the drone through the lens of lethal surveillance supports recent scholarship on the material histories of contemporary practices of war, but it also challenges or complicates them and our understandings of war and security today. For example, in his analysis of Western air power, Mark Neocleous argues that through the development of air power, and especially its early application to colonial control, we can see that war and police must be thought of as interrelated. War, in other words, has always been police power, where police is understood, following Foucault, as governing bodies and ordering populations, and is more generally tied to the maintenance of social order (Neocleous, 2013: 580). Drones, he argues, make this police basis of war clear, where drones allow for the ‘dream of a permanent police presence across the territory’ (Neocleous, 2013: 590). Neocleous’s argument points to the importance of knowledge production for the development of air power, which we see clearly in the emergence of lethal surveillance. Yet lethal surveillance points less to a tendency of police power across the 20th century than to the importance of war and violence for practices of knowledge production, scientific development, control, and targeting, and to the relation among these practices. Perhaps today’s drone reveals police power as always already war power (Foucault, 2003: 15) and helps us to further articulate both sides of the intimate relationship between war and police power that Neocleous (2014) develops.

In a similar register, the emergence of lethal surveillance through drone warfare shows that, while not always technically possible, the linking of surveillance and killing – the hunter-killer strategy – has long been an aim of Western military strategy. ³⁰ Further, the US military itself has often seen this as a more-than-human technical challenge. What this results in is not bureaucratic death-dealing with increasingly automated and unaccountable decision-making devoid of human thought (Shaw and Akhter, 2014), but instead the culmination – or clear expression – of the violence of Western thought itself.

Finally, and connected to this point about seeing clearly the connections between killing and knowing through the contemporary drone, the frame of lethal surveillance and the history it emerges from provides a way, following Caroline Holmqvist, into the human experience of war. As she writes:

The question becomes one of how to integrate accounts of the real/material – of the actual injury sustained by actual people in contexts of war – with accounts of how we come to see what we see, know what we know and think what we think about war: accounts of the epistemologies of war. (Holmqvist, 2013: 548)

These registers are collapsed in the practice of lethal surveillance, making it especially difficult to see the effects of war on human bodies, on political structures, on human thought. As Holmqvist argues, it is necessary to think war in human terms, and – especially with the growing emphasis of automation in drone development – these terms include the longer modern practices of war and violence, scientific development, and Western governance that have shaped contemporary drone wars. An effective critique of drone warfare will need to understand and untangle these interconnected histories and practices that shape lethal surveillance.

Conclusion

The purpose of this article and of exploring earlier drone programs has been in part to show the importance and usefulness of turning to the history of drone development for making sense of contemporary drone warfare. This research adds to a growing body of scholarship in critical human geography and critical security studies that focuses on the materialities and complex histories of security and military practices. Here I have argued that taking this history into account reframes the questions we ask about drone warfare and that this frame should be centered on the emergence of lethal surveillance, where we see two aspects of modern war – the development of ISR and dynamic targeting – become increasingly intertwined with each other.

The emergence of lethal surveillance and the rise of drone warfare signal more than a changing mode or practice of warfare. Lethal surveillance and the drone strike reflect a tendency present within a longer modern history of Western warfare and violence that, as we have seen in ISR and targeting across these periods, lies at the intersection of scientific development and knowledge production and practices of killing and control. This not only allows us to connect the drone to other practices of war and security, both historically and in the present, but also requires us to critique the drone strike not as an anomaly in the practice of Western war but as a practice that falls squarely within the history and development of Western warfare and violence.