Unmanned: Drones, Data, and the Illusion of Perfect Warfare

by William M. Arkin

  Simulators for high-speed pilot training have been around for a long time, and flattened terrain is common in simulating urban environments in both image generators and video games. Single-player shooting games abound, and there are even sophisticated immersion games used in large-scale simulations in controlled environments. But it is only since the late 2000s that the density and complexity of data needed to mimic a real world in 3-D have been available for common use—no special computers and no proprietary software are needed for a soldier to go to Khairabad. The village and its surrounding mountainous terrain are optimized for conducting ground combat simulations: every potential sniper nest, every observation point, every escape route, every suspicious line of fire that can reach the road has been identified. All of the wonders of a new generation of map products can be developed for both operations and analysis. The randomness of normal village life can be switched on, and entering the village, one can have an active roadside market, foot and vehicular traffic, villagers with pack animals, open doorways, translucent windows and blind stairways, character models and vehicles selected from Afghanistan-unique content libraries.

  From another library, users can then select the insurgent stronghold scenario, the IED detection and defeat mission or the sniper attack, or they can practice in real time with air spotters and fighter aircraft flying overhead. Insurgents burying an explosive fitted with a cell phone detonator create simulated dust that is potentially visible from the air; patched roads are created to act as potential telltale signs; the buried demons are given distinct digital signatures to appear warmer to thermal sensors. They even are detectable on a molecular level through more sophisticated eyes in the sky. The simulation comes with the ability to visualize the same terrain in the form of magnified-view scopes and laser designators on board various aircraft; simulated data from simulated aircraft sensors speaking to simulated laptops through simulated networks. Drones are there, too, fully integrated into every scenario and yet also unexceptional in every way. As the second decade of US war in Afghanistan began, drones were hardly remarkable to the soldier in the field anymore. Small unmanned ground vehicles had also become ubiquitous, and MetaVR could boast of tunnels modeled with geometry and textures to allow simulated investigation by tiny robots, remotely controlled and monitored SUGVs, as the military calls them.3

  Attack the network. Understand the enemy: with all of the superior technology and with all of the force of the Machine applied all the way to the edge, the upper hand doesn’t go to the strong and good-looking—to some modern-day Gilgamesh and Enkidu armed to the teeth and itching to do battle. It goes to those who prepare. Preparation of the battlefield is the term du jour.

  As is customary in the world of the military, preparation exists in accordance with new doctrine, and it is standardized and reduced to processes. Learn and follow the formal steps involved in Preparation of the Battlefield—Define the Battlespace Environment, Describe the Battlefield Effects, Evaluate the Threat, Determine Threat Courses of Action. It is a world of the same software, the same maps, the same symbology and terminology, the same in every unit, transposable anywhere, a training range that is not a shooting gallery yet is just as important, and even more so as troops and the real machines that we associate with war go home and the Data Machine remains behind. Preparation of the battlefield became so important that in 2004, the most elite special operations forces instituted rotations of the data as well as the soldiers. Units would collect onto portable hard drives the imagery and video and databases for their area of Iraq or Afghanistan, wiring a reachback loop so that the stateside unit could monitor, watch, study, and even sit in on video teleconferences of their replacements to keep up to date and focused.4

  Absent a specific area of operations with real-world intelligence, PMESII is the universal base layer—political, military, economic, social information, infrastructure; data slotted into the proper fields of databases to deconstruct the operating environment into features, networks, and nodes turned ripe for further designation as “of interest” and finally as potential targets. On top of it all is layered the threat stream, the take from the satellites and aircraft and drones and aerostats and ground systems, from the human watchers and the remote collectors who reside in the digital domain. In a practical sense, this means reports and photographs and intercepts and alchemic discoveries all drifting through a stationary corner, a sort of pollen that adheres and a sediment that collects, ever so subtly revealing some Holmesian clue. Befitting an army that operates with the pretension that every soldier is a sensor and that throws around terms like “strategic corporal” to refer to the importance of interpersonal relations; and exemplifying an institution that can’t help but systematize observation and relations into yet another acronym and process, the final layer is labeled the human terrain system.

  When General Petraeus became Iraq commander, he pushed human terrain hard. Within weeks of the successful invasion of the country, the army admitted to itself that Iraq was an enigma to almost all American soldiers, and though there were a few Iraqi experts, they were almost exclusively assigned to the search for Saddam’s WMDs.5 With the preponderance of human activity in analysis and support of the Data Machine, just being in Iraq also didn’t necessarily mean learning anything about the country. As one Predator pilot stationed at Balad Air Base wrote: “Miniature golf was about as close as a majority of the troops on base would ever get to the battlefield. Few would ever interact with the Iraqis; some would never even see an Iraqi.” When Iraqi insurgents fired mortars onto the base, they nicknamed the base Mortaritaville. The odds against an individual getting hit by a mortar round were about the same as being struck by lightning somewhere inside the “Tornado Alley” of Oklahoma and Texas.6 That also meant that those who did interact with the Iraqi population outside the fences were that much more likely to be injured or killed.

  Hearts and minds, information operations, human factors, soft power, human terrain: it doesn’t really matter what you call it. While the hunt was on and the insurgency was growing, and while counter-IED became the overwhelming mission, Petraeus’s ivory tower goal to pacify the Iraqi population grew in competition with an overwhelming juggernaut. They tried: native speakers enlisted and contracted for their language skills were anointed cultural advisors. Anthropologists and social scientists were given jobs on the battlefield and sent out Margaret Mead–style to understand a people who had remained wholly foreign. Female engagement teams were formed to penetrate the fairer side; coalition and then even Afghan partners were brought into the secret spaces. The Pentagon created a cadre of AFPAK “hands,” six-week wonders tutored in Dari and Pashto to man a burgeoning community of intelligence fusion centers. And back in Florida, CENTCOM activated an Afghanistan Pakistan Intelligence Center of Excellence. “Attack the network” also turned to human terrain. “Once coalition forces separate the enemy from the people, they bring in indigenous police forces to hold the security gains and then build trust and confidence as well as conduct reconstruction,” wrote Brigadier General Anthony Tata, deputy director of the IED fight. “There is no greater ambassador for the American people than a war-fighter on the ground interfacing with the local population; and that works.”7

  It is such baseless pretension, this “working” at some village level because of a shared cup of tea, while at the same time the very power of the Machine and the attractiveness of the data are that the soldiers don’t have to learn any actual human language. An extraordinary range of data appeared to turn the population into data: human terrain data and key leader engagements (KLEs) that collect data about tribes, family connections, language inflections. The connections and the sources of violence are all pooled into a single sociocultural knowledge base and then fed into the Combined Information Data Network Exchange (CIDNE, pronounced “Sidney”), a Middle East–wide events tracking system of the United States and NATO that seeks to record everything from commanders’ local expenditures for school repairs to IED attacks. Thus
, at every command brief, at every shift change, hourly, daily, weekly, and monthly, or on demand whenever, the significant activities (the SIGACTS) are gathered—the latest operational and enemy activity melded with human observations—information that is constantly collected and entered into databases, where it is passed along to higher and subordinate headquarters.8

  Everyday war is a challenge and “a roll of the dice,” says one military intelligence officer of his typical experiences operating in Iraq. Getting from base to base and from camp to engagement areas, whether for resupply or combat, depended upon the arithmetic of the roll. In Iraq and Afghanistan, like a sophisticated game of Battleship or Minesweeper, intelligence at this level—at the village level—became almost exclusively focused on finding and neutralizing IEDs. “There are only so many avenues of approach into and out of the platoon and company patrol areas,” says the officer. The trick was determining what routes and areas were dangerous; varying one’s own routes so that patterns were less obvious, and then, most importantly, predicting which engagement area the enemy was going to use next so that forces could either attack it or avoid it.9

  Though the systems available are an overwhelming word list of acronyms, there are basically two parts to the convention of getting through today and increasing the odds for tomorrow. In Everybad, wherever it is, pattern analysis has become the ubiquitous mission preparation: think all-news AM radio that provides traffic alerts “on the 8s” transformed into an automobile-mounted GPS that magically ingests slowdowns and accidents and then places it all in a life-and-death, no-mistakes instant navigator that not only provides warnings but also can assess the possibilities and dangers of alternate routes. Whether unit analysts input their SIGACTs into an older flattened Time/Event Plot Wheel or use the density plot software contained in the Distributed Common Ground System, the dual goal is not just to report bad traffic but to create a record of date, time, and location patterns to track enemy networks so as to know who is responsible for the jam and where they are at any given moment. And then the pseudoscience of this new type of warring is to predict where they will be next.

  Any unit that ventures out into the unknown can create a mathematical grid: each kilometer of road is given a unique identifier within fixed named areas of interest, many of the latter as small as a two-block area; as many as 300 in a company area of operations may support about 180 infantrymen, masterminded by a handful of data gatherers. Every SIGACT is associated with a specific geographic point and instantly logged to show which grid is quiet, active, more active, and even more active; up to 10,000 SIGACTs at a time can be tracked at the unit level. With engagements plotted by location, and by hour and day of the week, and with percentages established by commanders for thresholds—say 15 percent to designate a high-risk area—windows of engagement with the highest probabilities of success are established: a location where indirect fire such as a mortar is known to have come from, a line of fire that snipers have consistently employed, IEDs that have been found and detonated. When it all comes together, when the data is collected and entered properly, when the servers serve and all the knowns are collected, the probability of success is increased: one NCO describes his experiences with convoy operations, remarking that the data in his area of operations showed that only about 5 percent of the roadside bombings occurred between the hours of 2400 and 0400, not only determining the optimum time to conduct movements but also providing immeasurable psychological reassurance to his soldiers.10

  The SIGACTs system, which can precisely recount how many and precisely where suicide attacks or roadside bombings have occurred, is also overwhelmingly just a tactical reporting system. It is not intelligence per se, and though it is precise, it is little more than day-to-day reporting of data. Its corrosive influence at higher levels was seen in 2007 when Secretary Gates found “confusion over how the war was actually going,” with wide divergences between Washington intelligence analysts and commanders on the scene, differences that some tried to repair by working together and sharpening the questions asked. By June 2008, Gates let loose in a videoconference: “‘I don’t have a feel for how the fight is going!’ he said. ‘I don’t think the president has a clear idea either.…’”11

  Contributing to the problem of assessment are the bifurcated objectives and missions. Military units on the ground only care about the SIGACTs, about force protection, about their immediate needs. Targeting and a bigger counterterrorism war, truly taken out of the hands of the conventional units, demand different data, and different decisions relating to Zarqawi-type high-value targets. Add to this that in intelligence terms, there has always been a tension between collection of information and the preservation of a capability to collect more. When something is found, when something is heard and geolocated, is it a clue to follow and understand, or is it a target to kill? In immediate self-defense, the answer is always “kill,” but with so much data flowing, and the objective of the data itself not just to kill and not be killed but also to build, the pendulum seems to have swung almost completely over to the kill side.12

  The initial battle against IEDs sought to reduce the number of soldier casualties and improve various direct methods of detecting the devices and jamming their detonation signals. And those efforts were largely successful, the math suggesting that the number of casualties declined even as the number of attacks escalated.13 But the advantage still stayed with the stealthy attacker, and the area of interest wasn’t limited to any set of finite Everybads. That is when going after the IED network itself—financiers, the makings of explosives, engineers, drivers—gained greater traction, even if in Iraq that meant Iran and in Afghanistan, that meant Pakistan. Then came Yemen, Saudi Arabia, Somalia, Syria, Lebanon, Uganda, North Korea, China, Georgia, Uzbekistan, Turkmenistan, Russia, Everybad. It is no wonder that later on, every new country and every new insurgent or terrorist group present gigantic and constant political blind spots to the United States. Even in Iraq, knowledge would seemingly disappear in less than three years after the withdrawal of US forces, the activity given wholly over to a certain kind of intelligence that means data gatherers.

  Wherever Everybad is, it is also about as far from the Pentagon as one can get, from the “analysis” done on high, from the intelligence centers and the war rooms with their eye-popping video walls, and from the conference rooms where decisions of life and death are so easily made (and not made). Here the war is wagered in the billions, the metrics of success as much a mix of data from the field, polling numbers, and atmospherics absorbed from the news media as they are based upon a captain’s feel, or his commander’s feel, or his commander’s, and on and on. At the decision-making apogee, the words “national” and “strategic” aren’t meant to be hierarchical or superior, but here, the pretensions hold, the big picture is located. Are we losing the war? That depends. Based upon the objectives laid down in the president’s statement of x-day, based upon the political realities, based upon the resources applied, based upon comparison to yesterday or last year or ten years ago, based upon the absence of a 9/12, based upon consideration of all the limitations, the compromises, the fears?

  Or based upon the reports from Everybads near and far, where all the high percentages can be made to appear so much bolder than all the lows. Stoplight charts—green, yellow, red—and other familiar templates turn complex data into an answer, an actual PowerPoint briefing that is meant to summarize all that is surveyed and the fortunes of battle at some moment in time. None of it is meant to deceive or mask when red is actually red, when things really are bad. But far away from Everybad, it is intrinsically a color-coded existence that analyzes and briefs never to get to red by combining all of the greens and the yellows with all of the other hues and their big-picture rationalizations to bring data and the politically desirable end color into alignment. Think a television commercial, necessary and instantly recognizable but also subliminal; sometimes just vacant space and the requisite stare, sometimes creative and heroic and noticed, s
ometimes an actual bathroom break. At the national level, there are simulations and there are choices made regarding which route to follow, but mostly the national SIGACTs are constructed to always be cautiously green or optimistically yellow, to always leave open the political options.

  War has always had this divide, the view from the foxhole so different from that at the top. That is how even in the face of disastrous firefights and the perpetual Catch-22s of anytime and anywhere, a war can be going well to those higher up. And that is how, even when things are relatively quiet in the hooch, when thirty days without an accident tick by on the orderly room bulletin board and the PowerPoint briefing is solid green, things can also be for shit. When Afghanistan and Iraq descended into frontless battles, after the Taliban and Saddam were deposed, after al Qaeda was scattered, the battlefield literally became the domain of some infinite arithmetic. Terrorism remained the bigger threat, but the war became achingly local, and self-defense became seeking to get the enemy’s unmanned robots before they maimed and killed.

  An army marches on its stomach, and since the one thing that can’t be stomached is too much hardship or friendly injury and death, protecting the force (known as force protection) became an industry in itself. Just as the IED and special operations worlds (and the CIA and the DEA) developed their own collection platforms, an entire other world of intelligence collectors emerged just to keep an eye on the bases and the soldiers—unmanned ground sensors, robots, closed-circuit television towers, tethered balloons. And all of this requires the unique parts and service personnel and engineers and the helium, another layer upon another layer, each with their own gizmos and hierarchies and economies behind them. More materiel to move from base to base, to network and serve, more unmanned helpers to relieve man of the risks, and yet to increase the risks intrinsically as Everybad feeds every new war. And all producing yet more data.