The axe at my side, in which my arm trusted,
the dirk at my belt, the shield at my face,
my festive garment, my girdle of delight…
TABLET VIII, EPIC OF GILGAMESH
I told my publisher that if I bought my own drone, I could tell the story, understand the allure, and perhaps even convey the complexity of the Machine.
My first stop was eBay, especially after I read that a Philippine man had pled guilty to trying to auction parts from a military reconnaissance drone on the website.1 A search for “drone” yielded over 5,000 active listings as of mid-2013, including everything from women’s clothing to action figures. The top listing, though, was a Parrot 2.0 App-Controlled Quadricopter, $293.95 or best offer—which I dismissed as a toy—followed by a $679 DJI Phantom. And then there was an Oktokopter DJI Wookong for $6,400. The cost was ten times as much as the Phantom but came with the caveat: “Please note that this equipment is not a toy.”
Before your own eyes glaze over, let me say that I wasn’t even sure if the Wookong had a camera, and I was hearing the voice of some fast-talking car salesman in my head: four on the floor, carburetor this, muffler that, souped up; the salesman in my head was lovingly stroking this baby. I’ve never really cared much about cars and don’t have a pilot’s license, and I’ll admit that it was all pretty much Greek to me. So I plunked down my $300 and ordered the Parrot to start.
What a fabulous toy!
Without its Styrofoam bumpers—aka “the indoor hull”—protecting the four propellers, it looks like a big bug. Though it’s not bigger than a laptop—a really small laptop—it has a forward-looking high-definition video camera and a non-high-definition video camera underneath. The attraction for a neophyte like me is that you can download the app and operate it with your iPhone or iPad. Charge the battery and it’s ready to go.
In my living room, admiring my new Parrot, I joined the AR.Drone Academy online to track my flights and to find other Parrot users. I read the (surprisingly short) operator’s manual (clearly, nobody reads anymore). I watched a variety of arty and instructional videos online that made it look really easy. And I decided on an inaugural indoor flight with the protective bumpers on. It looks pretty fragile, I thought. I’ll just hover around the room, check out the controls, and see how the camera works.
My heart pumping, I hooked up the battery, tested the propellers as instructed, plugged in a USB thumb drive to store the recorded video, connected my iPhone via the drone’s own wireless receiver, set the software for indoor flight, and put everything on the lowest possible speed and altitude settings. My Parrot took off and hovered; think tiny helicopter, the four propellers whirling in unison to provide lift and control direction. With thumbs on the two dots that simulate joysticks on the iPhone touch screen, I made the slightest movement and the Parrot darted away, almost twinkling at me like some Pixar character, animated and alive, as I later surmised, because it clearly had a mind of its own. Less than ten seconds into my first flight, I unceremoniously crashed into the window.
The second attempt was no better—I hit the chimney of the wood stove.
By now I was laughing and so was my wife, who was recording my endeavor. But it was very frustrating, and I told myself I’d get the feel for it.
The first flight outdoors didn’t go much better. I was worried about the wind—this baby only weighs fourteen ounces, less than a small box of pasta. I set the maximum altitude at three meters just to be safe. The barn, two trees, the house, the driveway: before the battery ran down, I’d crashed into them all. I just couldn’t get the hang of the controls and couldn’t for the life of me figure out how to get the drone to come back once it darted away. Specks of dirt in the gears that turn the propellers were my first maintenance job, but other than that, my Parrot actually proved pretty hardy. And not only that, the video produced was beautiful and unintentionally hilarious.
The Parrot came to my brother-in-law’s wedding in Maine the next week, where I hoped to find a big enough flat space far away from any water to practice. It’s very pesky, I told my stepson, Galen, age eighteen, who badgered me to break it out so we could fly. I’m not going to let you fly it until you watch the instructional videos, I told him. Two minutes later, he was back and ready.
It wasn’t like I was teaching him to shoot a gun, but as owner and commander of my own embryonic squadron, I became ridiculously officious, carefully showing him how to connect it to the phone, how to do the self-test, instructing him to land it on my command, fretting about the wind and using words like “aloft” like I knew what I was talking about. It was just breezy. Then I let him take the controls. The wind from the ocean definitely buffeted and pushed the Parrot around, but Galen almost effortlessly got the hang of it right away, flying it up to the top of the house we were staying in, crossing the road between two electrical wires—“Watch out!” I yelled—darting the Parrot back to the front lawn, flying it this way and that, and then landing it. He loved it, I loved it; I was humbled. When I debriefed Galen after this and other missions, he said the trick for him was calibrating himself to the rhythm of the drone, what actual pilots call “feeling” the plane, though unlike the pilot of a manned airplane, a drone pilot doesn’t actually feel the inertia and acceleration caused by a gust of wind.2
They call Galen’s generation “digital natives.” His instinctive aptitude truly says something about our society, and about the expansive world of the Data Machine that drones represent. Whatever happens in the wiring of a brain that allows a young child to so easily pick up a second language, the digital natives have acquired a new way of absorbing and interacting with our wholly digital world.
Still, it drives me to distraction when I watch how Galen or any of his contemporaries operate; they have multiple things going on at once, on the laptop, on the iPod, on the phone, on the TV. They’ll have several chat sessions open on a variety of appliances, texting on the phone as well; they’ll be watching a show and watching a ball game; they’ll have a YouTube video running; they’re listening to music; and sometimes they’ll have a couple of homework assignments going, sometimes involving real books but more often than not digital pablum and hands-on projects. The relentless demand to command all this data is only compounded by the speed at which music and videos are transmitted globally, not just to the home Wi-Fi network but also through the satellites and cables and fiberoptic networks needed to move it all. Galen and his generation didn’t conform to the machinery of our day; the Machine conforms to their expectations.
And that wedding in Maine? It was truly beautiful. After a week of dreary rain and fog, the sun shone brightly at the top of Cadillac Mountain at the moment of the ceremony, and everything went off without a hitch. Everyone, even my father-in-law, age eighty-four, had their own digital camera and was snapping away, the big old cameras with the bulging lenses dusted off, powered up, and brought out for the occasion as well, but more often than not it was just the ubiquitous smartphones, especially among the digital natives. There we were, recording everything, an NSA-like acquisition, each wedding participant—myself included—more comprehensive and efficient collectors than the old-fashioned single authoritarian wedding photographer of old, gigabytes upon gigabytes so easily collected, a massive and dispersed digital record. Those little LED screens also served, though, as a kind of obstacle, removing the participants by one step from the event. And then I saw our own family intelligence agency choking on data as everyone compiled their mass of photos and signals, the designated nerd spending hours on end putting them all on a DVD, the task of sorting through them all left to the bride and groom back at the family CIA.
As wedding turned to reception and after-party, the picture taking hardly slowed. This is my Parrot’s way as well. In flight, one doesn’t turn the video camera on and off—the data collection is automatic and constantly streaming like a data waterfall. Video continues, only constrained by onboard power, which constrains the flight. Movement of the video from camera to thumb drive to iPhone to computer (or the Web) is slow and intrusive to the operation of the drone itself. Storage being so cheap these days, I was able to use a 128-gigabyte USB thumb drive, its capacity exceeding the amount of video that could be shot by the drone in a single mission, given the thirty-minute life of the battery. And with my mighty thumb drive, I’d never run out of storage space. So my squadron’s second requisition, after my ginormous data center, was going to have to be a second battery, and maybe a third, for sustained operations.
More drone flights followed: my friend Peter came to visit for the weekend, and to my surprise, he was fascinated with the drone. A decade older than me and a journalist and writer of the old school, which means two-finger typing and a deep love of musty archives—the real stuff—he’d never seemed to me to be much of a technology guy. So we went down to the school, where the size of the playing fields was sure to keep us out of trouble. I was getting better at flying, practice and help from Galen paying off. But on flight segment three, the Parrot sailed away and just kept going. It took us a good half hour to find it in the trees, ominously close to Barnard Brook.
My brother, an ultralight enthusiast who built and flies his own airplane in Virginia, asked if my new drone carried a weapon, certainly an apropos quip given how one type seems to rule over all others. I guess I could use it kamikaze style, I responded, but the weapon would have to be awfully light. He suggested attaching something called a servo and dropping a paintball, but I was thinking more of the aptly named and supersecret Pyros or some other variety of the forearm-sized weapons increasingly becoming the side arms of the uniformed digital natives.
Another training session with Galen, as he finally taught me how to just nudge the controls and get the Parrot to come back and we flew the drone higher and higher, over the roof of the house, peering now into all of South Pomfret village: the general store, the library, homes, and beyond. Winds at 120 feet required our smart little craft to constantly compensate to right itself, and the video, though beautiful at that altitude, was so jerky that watching it was enough to cause vertigo. But I was thinking that pretty soon I’d be able to bring the drone along to people’s homes and parties, weather permitting, second battery now part of the unit supply, and do little surveys and demonstrations, and peer into neighbors’ backyards, conduct some real espionage.
A short conversation with an attorney introduced me to tort law and the detail that Vermont might have some special rules. “If you look into a neighbor’s yard it probably is not a violation of their privacy,” he said, “but if you get on a ladder to do so…”
Now, in addition to needing an insufferable multitasking digital native assistant, and a budget, and someone to handle the accumulating mass of pictures and video, I also needed a squadron lawyer.
I’m afraid of my drone. There: I admit it. I still want to get that singular beautiful picture from a unique vantage point to serve as an illustration for this book, but I’m worried about trying to master something that isn’t about anything except mastering something. And I’m worried about all that video: Why am I even keeping it all? It’s a great illustration, my Parrot, of the fact that I don’t want to spy on my neighbors, and though I own a new toy, I’m not really much of a hobbyist or a flier. I won’t be getting my wings, but the drone has shown how the military and intelligence communities have impetuously moved forward with their drone acquisitions: You gotta fly it. You gotta communicate with it. You gotta keep it in good repair. You gotta have power and lots of it. You gotta point it at something and that’s gotta have a reason. You gotta look at the video and you gotta store it and then be able to find it later. You can always use a better camera, which means more storage space and more issues with where and what you can record. And more range is useful, too, which means longer-range communications, which means a better controller, and maybe even a staff. Pretty soon, you’ve got a squadron; you’ve got dozens, hundreds, thousands.
Is that it? The acquisition of toys, toys turned into tools into weapons of war? Whatever you call it, we have created a killing Machine. We need to examine whether it is effective, to ask first of all two fundamental questions: whether it is even the right strategy to pursue to defend against terrorism, and whether it secures our future. And we also need to understand the Data Machine itself, whether it is spending our national treasure wisely or doing its given job well. But what if there are legality and economy and even oversight, no matter how effective or ineffective they might be? What if the making of the killing Machine in itself, the creation of this other black box–packed world of Gilgamesh’s, changes the real world, changes us? That’s my discovery with my own little Parrot and my deep descent into the world of the unmanned. We do have thousands of drones, and the Machine already rules. Somehow, though, it seems exactly what we want, what we crave in this life separated from the reality of war.
I went to look at the digital natives to seek an answer, and the sight was not pretty. One brainiac with a physics doctorate arrived in Kandahar in 2013 to serve as a kind of science officer with Combined Joint Task Force Paladin, the multinational counter-IED authority then still busily at work.
The Georgia Tech graduate’s job title reads “operations research analyst”: part staff officer, part intelligence analyst, part scientist, a new kind of government issue sent into combat, but a fighter who would be unrecognizable to an Eisenhower or even a Schwarzkopf. Put through his paces at basic in a coddled and lecture-filled predeployment training building in Maryland, the modern-day Poindexter is not an actual military officer or soldier or spy, nor is he one of those dreaded contractors; he’s a bona fide army civilian employee, the kind that used to populate the arsenals in the rear areas doing the industrial weapons work but that now are essential unlaborers of the Data Machine. Like his volunteer brethren who serve in uniform, and yet totally different, he is there on the battlefield serving his own one-year combat tour because actual soldiers, no matter how good they are, need something beyond even the perfected Niagara of the user-friendly Machine to make sense of the big data of war.
Poindexter’s focus is significant activities (SIGACTs) and statistical trends. He works alongside scores of others in the local intelligence fusion center. In each of the regions, there are counterpart centers and task forces with analysts and subject matter experts (SMEs) numbering in the hundreds who toil away, an activity duplicated endlessly in each of the districts and base hubs and units. They are backed up by thousands at higher headquarters, with more than half of the boots on the ground provided by men and women of statuses other than soldier. They are backed up by tens of thousands more working 24/7 shifts in the United States. More thousands populate similar fusion centers slightly different than counter-IED. There they wrestle with the reporting tools and databases of local and third-country national vetting and hiring, counterintelligence and insider threats (“green on blue”), counternarcotics, counter–threat finance, counter–human trafficking, countercorruption, warrant-based targeting, information security, physical security, and base and force protection. If a Poindexter supports the world of special operations, he does the same cubicle work that prepares and supports Village Stability Operations (VSO), Foreign Internal Defense (FID), Counterinsurgency (COIN), psychological operations, and irregular warfare. If he directly supports killers in the black world to find and fix the high-value targets and individuals, he resides in a similar world of the x-men. The higher any unlaborer goes, the more distinguished the arsenal of databases and networks—MagicDesk, Fire Truck, SKOPE, JIANT, SOIS, PDAS—the software and networks of the new gods.
Even as troops left Afghanistan, collection never faltered. Just in Kandahar, at just one enormous military and intelligence hub, more than a dozen different types of planes and drones flew at the end of 2013. Their black boxes are all the latest models. Hovering above all is the multi-INT Persistent Ground Surveillance Systems (PGSS), a combined balloon and unmanned ground sensor grid keeping watch on direct threats to every hub. Technicians fiddle with the latest black boxes in this world as well: the austere location force protection kits and the shooter detection systems. Also flowing in are the signals from fixed intercept sites and vehicle-mounted stations, even the take from the close-access technical reconnaissance of the x-men. Out on patrol, the actual warriors conduct sensitive site exploitations, collecting documents, hard drives and disks, cell phones and anything else that might provide a clue; they conduct meticulous forensics of what can only be called crime scenes; and they deal with the people, carrying out key leader engagements, interrogations, screenings, and background checks; and, most important these days to populate the databases for tomorrow, they collect biometric data.
This is a different kind of army: it doesn’t advance per se, setting up camp and pitching pup tents as it halts for the night. The Poindexters and their subordinate plain old analysts are more like Dilbert technicians filling air-conditioned offices—planning, scheduling, monitoring, scanning, collating, translating, geolocating, data-pulling, processing, formatting, chatting, and briefing. Day in and day out, they prepare pattern of life studies, predictive analysis, historical threat analysis, District Narrative Assessments (DNA), intelligence preparation of the battlefield, Intelligence Summaries (INTSUMs) and graphic Intelligence Summaries (GRINTSUMs). They are literally the masters of software, valued for their digital acumen, almost all of them commercial products, technicians and unlaborers present and accounted for in attending to the Data Machine. There are dozens if not hundreds of additional black box software packages and databases of all of the three-letter agencies that connect the unlaborers. Their world is a self-contained and self-generating society within itself, not the real world.
In my experience of trying to tease out some intersect between the secret life of the information warriors and war, I discovered that video games and television series like Game of Thrones unite digital natives and their elder brethren-in-arms in some common and enduring language. And since I know nothing of the most current games or fads, I’ve tried to talk about Star Trek and have come up with an interesting conversation. Poindexter and his ilk can completely get into debating Spock versus Data, the next-generation science officer: who’s smarter, who’s a better character, who is more likable or who they would like to spend eternity with on a deserted planet. Amongst people of a certain generation—mine—Spock wins hands down, not just because he is a first love but also because he was an introduction to the future; charismatic and fearless, with that mystical otherworldly quality characterized by all the Vulcan mumbo jumbo and the mind meld. To the digital natives of the Poindexter generation, Data is the man. He is not just Spock 2.0 but a walking thesaurus intrinsically more lovable than the half-man with no emotions, a striver who wants external input and wants to collaborate, a techie who can learn at an astronomical rate, funny precisely because he is unfunny, an overtalking and eager oid that crosses the line from a that to a who.
There is no right answer in this debate, which is itself highly symbolic: an equal number of fanboy scholars seem annoyed that Data can babble on uncontrollably as are annoyed about Spock’s monotone and the fact that his every line ends with the always anticlimactic “illogical.” Data is endearing precisely because of his obvious almost-human flaws; it is in fact the nonmachine who seems more robotic.
The prime difference, of course, is that Data is a robot, an android with a positronic brain, a Hollywood tip of the hat to the great Isaac Asimov, who first conceived of such a brain in 1950s science fiction. Spock, on the other hand, is an alien from another galaxy, and though he has a human mother and a Vulcan father, in the end, he is not even on our historical horizon, while here is the truth: Data is completely logical.
Data the machine, on the other hand, can be more likened to the NSA or the Data Machine itself—he can literally listen to 100 separate songs at once and discern all of them. It is an incredible achievement and a necessary demonstration. Yet as a character, Data sought—with the processing of enough information—to be more like a human and even to tackle the human soul, a struggle not unlike the human struggle going back six thousand years to the time when the story of Gilgamesh was first conceived. Spock is in many ways another in a long line of Gilgameshes, that fantastic character that is only part human. In his detached way, he has to be likened to the modern-day national security wonk, the hyperexpert pursuing über-objectives with superhuman focus, the man or woman who will do almost anything to save lives, the unflappable officer you want on the bridge, unless it is a bridge to nowhere.
While science and society debate the potential of artificial intelligence and the limitations of actual software and models in becoming self-aware or producing humanlike reasoning, even the skeptics admit that a fusion agent like Data might be able to crunch the numbers with great accuracy to determine where a human target or an improvised explosive device might be hidden, even if, as one scientist friend says, it might also have nothing to say about what to do about it next. The popular nightmare and a staple of science fiction and movies going back to 2001: A Space Odyssey is somehow that the machines go beyond their programming and threaten humans and humankind. Thus the egghead advisors in today’s military formations don’t even pretend to be working for the purpose of unlinking warfare from its human dimensions; it is merely that the Machine has already become too complex for ordinary humans to master.
Autonomy is coming precisely to solve this problem. The vigilant Data Machine already supports mission tasks that must be accomplished on a scale beyond human capability. Translation software, artificial intelligence, and electronic means of processing raw data—signals and imagery—are already managing the glut.3 The military isn’t shy about calling for “appropriate levels of autonomy”4 in future development, with the increased use of autonomy and autonomous behaviors not just improving performance but also providing cost savings.5 The increasing volume of data, combined with an ever-changing operational environment in the limitless number of possible Everybads, the air force says, demands innovative approaches in order to translate raw data into intelligence and deliver it rapidly.6 It can all sound like another Gnat-750-to-Reaper continuum or a WARHORSE-to-ACES HY reduction in weight and improvement in performance, except that lurking in the dark corners are the mad scientists who do foresee “Complete Kill Chain Weapons,” that is, machines that would make every decision from detection to the kill.
Though concerns about “killer robots” are valid, and there are already weapons that people should be concerned about that essentially kill automatically on the battlefield, the reality is that automated killing is inimical to military culture and the concepts of just warfare that most militaries rely on. Again, the military is quite aware of the shift and the danger, stressing the difference between performance and “execution,” the military’s actual word. “Autonomous mission performance may demand the ability to integrate sensing, perceiving, analyzing, communicating, planning, decision-making, and executing to achieve mission goals versus system functions,” the Pentagon says.7 Systems, the military says, are “only as good as the software writer and developer because the control algorithms are created and tested by teams of humans.” And that is where we will be as long as machines can’t adapt to changes as well as predict what will happen next, as the human brain does. Killer robots are probably not the only place to look. Much more likely is human-robot teaming at the small-unit level: unmanned ground systems that guard bases or conduct logistics or even conduct missions such as remotely conducted, “nonlethal” crowd control; “dismounted offensive operations; and armed reconnaissance and assault operations,”8 with everything under human control at some command post until that line is crossed or some undefined accretion of capability is achieved.9
Will the Machine someday decide who lives and dies? In some ways, the technologists can’t contain themselves in seeing exponential growth and development in the future. Hence we hear not just of Complete Kill Chain Weapons but also of a future of “swarming” autonomous drones, actually operating “out of the loop,” meaning without human intervention.10 I simply wonder how different that really is from what we have today.