Desperate measures for desperate times, Swaringen guessed. Couldn’t have another terror attack on US soil. Nobody would get reelected to office.
Several more incidents, or situations in the parlance, came and went. Clark Barter was there for some of them, but was gone for others. David Swaringen had the reins in Barter’s absence. Swaringen was still not completely certain of his own ability and judgment, but he felt more confident. When people looked to him for guidance and direction, he felt slightly less like an impostor, slightly more capable of making good decisions.
It was a hard thing, to order the death of another person. But it would have been even harder to stomach another terrorist attack. At least, that’s the way Swaringen assuaged the brief pangs of guilt he still occasionally felt as he watched people’s innards splatter. Thermal imagery left little to the imagination. It was graphic, and the bowel-shaking reality could not be hidden behind jargon and euphemism. The bland technical terms — intercept, incapacitate, neutralize — did no justice to the splash of blood and guts. The feedback was immediate, visceral, unmistakable. David Swaringen was having people killed.
He didn’t feel great about it, but neither did he feel all broken up, either. In his estimation, the people who were involved in the kinds of things that Clark Barter had described, peeling the skin off of other people’s faces, slicing off their genitalia, beheading them while chanting religious slogans, those kinds of people deserved to have their innards aerated, to have their bodies laid open and left for the wild dogs and insects.
But a question remained in Swaringen’s mind. It was an important question, but one that had so far been entirely taken for granted.
Who decided which names went on the list?
And how did they know they were putting the right names on the list?
The identification criteria in use on the operations center floor were rigid and rigorous, and Swaringen had little doubt about the accuracy and precision of the process of matching faces in the world to names in the database. But Swaringen could not vouch for the process used to identify people to be put on the list of “bad humans” in the first place.
Sure, some people just needed to die. But how did they know who fell in that category? Were there fringe cases? Line-straddlers? Could-go-either-way examples? If so, who made the life-or-death determination?
Swaringen gathered his gumption and asked Barter. The old man’s reply was characteristically unhelpful. “That’s protected under a different caveat.” Which was security-speak for “You’re not cleared for that information.”
Which struck Swaringen as odd. He had the authority to kill people, but not the authority to understand why they were dying.
As the hours and days passed, and the body count climbed, and the number of people in detention rose, the question grew in prominence in Swaringen’s mind: who’s the mastermind behind the list, and what process do they follow?
How do we know we’re killing the right people?
Swaringen decided to find out for himself.
Sometimes the best way to ask a sensitive question was not to ask it at all. Sometimes it was better to ask an oblique question instead, one that was related but not uncomfortably so. Like an administrative rather than substantive question, or a technical one, in lieu of a pricklier moral or philosophical one.
Sometimes it was better to aim for practical rather than political.
Swaringen took this tack in his effort to determine the origin of the list, the set of “bad human” names that lived in the biometric identification database. If one of the zillions of surveillance cameras on planet Earth found your face and matched it against the biometric data points in the system, the computer notified an operator on the command center floor. The operator notified his supervisor, who made phone calls, or radio calls, or sent commands via text or chat. And then you were investigated, prosecuted, and maybe even executed.
So the list was an important thing. Swaringen cringed to ponder the consequences if there was a flaw in the process.
And it wasn’t an idle fear. America had done it before. At one time, US forces paid more than a year’s salary to dirt-poor villagers in Iraq and Afghanistan in exchange for turning in their neighbors. Obviously, the quality of intelligence gleaned from such arrangements was even worse than abysmal. Guantánamo swelled as villagers ratted each other out to the Americans for princely sums. Perhaps the effort did ensnare a legitimate terrorist or two, but in the main, the policy provided a terrific way for villagers to rid themselves of economic, social, and romantic rivals. Mountains of cash had been paid to place thousands of innocent people in detention, with no opportunity for a day in court.
Swaringen hoped it wasn’t such a Gestapo-like arrangement that had produced the names on the list of people the drones were searching for. The whole thing had an Orwellian flavor to it. Worse, even.
Swaringen couldn’t bear to think about what might’ve been going on behind the scenes. And he couldn’t just let it rest.
He walked down the hall from the operations center after his shift, pausing at locked doors along the hallway to read the office placards. It didn’t take long to arrive at his desired destination: the IT department. To most people, computer networks worked on the PFM principle. Pure Freaking Magic. But Swaringen knew it wasn’t magic. It was human effort. The roomful of computers and video monitors in the operations center all required care and feeding. They required network connections, and network connections required servers, and servers required IT personnel to keep them running smoothly.
Swaringen knocked on the door.
“I’m still pretty new here,” he said to the IT technician who answered, “and I’m hoping to understand things a bit better. I’d like to learn more about our process, about how things work.”
The technician was eager to oblige. He walked Swaringen through a tour of the IT spaces, showing him the rack of servers that transferred data to and from the field, filtered it, prioritized it, and displayed it on the monitors in the command center.
Swaringen was surprised to learn that for every hour of video displayed on one of the monitors in the operations center, there were fifty more hours of undisplayed video. That tidbit dramatically changed his estimate of the number of drones airborne at any given time. The number had to be staggering. Well into the thousands. He supposed that he had read something about drone production programs in the newspaper at some point, but he hadn’t paid much attention at the time. Now, however, Swaringen understood that the production scale had to rival the industrial mobilization during World War II.
“What happens to the unused video?” Swaringen asked.
“It isn’t unused,” the technician said. “Just because human eyes don’t see everything, doesn’t mean we’re not looking at everything closely. We have over a hundred algorithms at work on every single second of video, looking for specific things. The high priority feeds are sent into the command center.”
Swaringen pursed his lips, thinking. “How do you know the algorithms are looking for the right things?”
The tech smiled. “You never know, really. It’s always a work in progress. We’re constantly tweaking parameters, making adjustments, analyzing results.”
“That’s a high-stakes game,” Swaringen said.
“Damn right it is.” There was pride in the tech’s voice.
“So really,” Swaringen said, “the algorithms are running the show.”
The IT tech held his palms up. “Whoa. I’m not allowed to say anything like that. Stuff like that could get me fired. But I can say that nobody on the command center floor ever sees a snippet of video that the algorithms don’t screen for relevance and intelligence value.”
“How does the machine know?"
The tech smiled. “Machines don’t know. Just like humans don’t know. Computers draw conclusions based on algorithms and heuristics, exactly like our brains do. The algorithms grade their own performance based on parameters we set, and they’re cons
tantly improving, training themselves for better performance.”
“Artificial intelligence?”
The technician gave Swaringen a crooked smile. “No intelligence is artificial,” he said. “We prefer to call it nonbiological intelligence.”
Swaringen nodded. A troubled look crossed his face. “The algorithms aren’t determining which names go on the wanted list, are they?”
The technician shook his head. “No, not yet. But obviously, with seven billion people on the planet, the algorithms are doing some pre-screening. Computers prioritize the people to watch. But analysts do the watching.”
Swaringen frowned, pensive.
“At least, that’s what I’m told,” the tech said, palms raised again. “Really, the list just shows up.”
“It just shows up?”
The technician nodded. “That’s right. Once or twice a month, they hand-deliver a thumb drive to the ops center floor. It contains the names and associated biometric markers in a spreadsheet.”
“What kind of biometric data?”
“Everything you can think of. The distance between a person’s eyes. Between the corner of his mouth and his nostrils. The length and width of his chin. The length and width of his brow. The angle of his ears, the length of his neck. Fingerprint data, obviously. In some cases, unique patterns in the irises. It’s beyond comprehensive, and it’s the main reason the systems are so accurate.”
Swaringen pondered. “These people aren’t volunteering this information, I gather.”
The tech laughed. “Sure they are. By walking around in public. NSA is tapped into every ATM in the country, and probably half the cash machines in the world. Every time anyone withdraws cash, their face is recorded. The biometrics are measured, categorized, and stored in the database. Same goes for all those traffic cameras you see.”
“How in the hell did we get permission to do that?”
“I don’t think we asked for permission,” the technician said. “I think we figured out that we could do it, so we did.”
Swaringen grimaced, uncomfortable. Not because he was a Luddite. He knew that computers made better medical decisions than doctors did. Computer-controlled robots performed better surgery than human surgeons did. He knew that computers made better flying decisions than pilots did. He knew that computerized robots made better assembly workers than humans. So there was no reason to think that computers couldn’t be used to reliably identify behavior patterns.
The problem was making sure they were identifying the right behavior patterns. Humans were very complex. Behavior was not a monolithic thing. There were degrees and shades of gray, degrees and shades of intent. It was hard enough for humans, even with the brain’s massively parallel processing, to accurately ascertain intent and affiliation. Swaringen hoped the computers were better at it than humans were, but he had his doubts.
But there were humans in the loop determining the list, according to the technician. Swaringen wanted to know more about how they did it. “Tell me who delivers the list every week.”
The technician shrugged. “Just a courier. I should really learn her name, but I haven’t.”
“A female courier?”
“That’s right. Same lady every time.”
“Do you think she’ll talk to me?”
“Couldn’t hurt to ask.”
Swaringen left the IT center with an uneasy feeling in the pit of his stomach. If anything, the technician’s description of the way things worked at the NSA’s command center generated more questions than answers. Sure, computers were smart. But you had to program them. You had to tell them what to look for. Which meant that you had to know what to look for. And you had to figure out how to turn that into machine language, to make sure nothing was lost in translation. The whole thing was fraught.
It was relatively easy to screen for intent in an email. You just looked for a particular set of words, in any language that you cared to exploit. And it was no longer a secret that NSA had been reading the world’s email for a long time.
But even email analysis wasn’t straightforward. People communicated with nuance, innuendo, metaphor, euphemism. Especially people who were planning illegal activities, who knew they were being watched. Those people were especially careful not to broadcast their intent. They used codes, indirect phrases, misdirection. So deciding the right words to look for wasn’t nearly as straightforward as it seemed.
And Swaringen had a hunch that the problem of determining intent and affiliations was all the more fraught when information was sparser. As in video surveillance, for instance. Maybe you could figure out a person’s circle of friends and acquaintances. Maybe some of those affiliations were so obviously “bad” that they were unambiguous. But Swaringen figured such circumstances were rare. Criminals and terrorists were careful. If not by nature, then by necessity. If you were a terrorist, the whole Western world was interested in hunting you down. Vast sums of treasure and talent had been invested to make you dead, as soon as possible.
Which led Swaringen back to the seminal question: who decided who lived and died, and how did they decide?
The IT technician had given him an important clue. The courier was female. Obviously, the courier had to be cleared for the information she was carrying, which meant she was on the Penumbra access list. Swaringen had only ever seen two or three females on the Penumbra ops center floor. It wouldn’t take him too long to figure out which one was the courier.
He would approach her in much the same way he approached the IT technician. He would tell her, truthfully, that he wanted to understand more about the process, that he wanted to learn more about the intelligence gathering system.
Ultimately, Swaringen wanted to talk to the person who put the names on the list.
26
Mark Severn had been yanked from the field. He had been ‘made,’ hunted down, even, in Budapest. Hence the fake death, in order to avoid a very real death. That meant that Mark Severn’s duties had to fall to someone else.
Sam drew the short straw.
She pointed out the obvious: she had been made as well. They’d followed her all the way from DC to Hungary.
“Not the same,” Davenport said. “You mopped up most of the muscle they sent after you.”
“Most. But not all.”
“They don’t know you like they know Severn,” Farrar said.
“Bullshit. They know me on sight.”
“Some of them do, yes,” Davenport said. “But as I mentioned, you thinned the herd considerably. I really think the risk is drastically reduced, especially since you’re back on our home turf.”
Sam frowned. “I’d really rather take that vacation,” she said.
Davenport smiled. “Weren’t you the one telling me just a couple of days ago that we couldn’t possibly drop this case, that all sorts of messed-up things were happening that we needed to run to ground?"
“Yes,” Sam said. “And I was right, as usual. I just don’t want to be the one chasing it down. I’m tired, and I need a break.”
“Unfortunately, Sam,” Farrar said, “we’re a bit thin at the moment.”
“Bullshit,” Sam said. “You have a floor full of monkeys sitting at computers instead of pounding the pavement.”
Davenport and Farrar shared a look. “Not all assets are created equal,” Farrar said.
“And you and Dan already have a number of hours invested in this,” Davenport added. “You’re making good headway. You made a few important discoveries. You have momentum. If you can hold out just a bit longer, get to the end of this thing, you’ll be able to take that vacation.”
“And then some,” Farrar added. “I can authorize up to a week off with pay. It won’t count against your annual leave.”
“A bribe?” Sam asked.
“No, Sam. An expression of my genuine thanks for your effort and sacrifice.”
Sam ultimately capitulated. Truth be told, she wanted to be the one to break the case. But she didn’t want to w
ant it. Because of Brock.
She thought about how to break the news to him. She didn’t expect the conversation to go well.
Boston was definitely on her list, but not at the top. Because rarely in the counterespionage business was it easy to figure out who ordered a killing based on who had done the killing. Like nearly everything else in the modern world, assassinations were usually outsourced.
For that reason, and operating under the assumption that Janice Everman of the US Department of Justice was, in fact, a murder victim, Sam decided it would be important to figure out what she was working on immediately before her death. It sounded fairly obvious, but those kinds of simple questions usually led you in the right direction. Life was messy and chaotic and organic and nonlinear, but it was rarely overly complicated, when it came right down to it.
So she moseyed over to the Department of Justice, paused at the front desk, showed her badge, and asked the desk attendant whether he knew who had replaced Janice Everman. Sam figured Everman’s replacement was as good a place as any to start.
“Do you mean in the Security and Public Policy division?”
“Is that where Janice Everman worked?”
“Janice Everman was the security and public policy department,” the attendant said. “She was a force of nature, and we all miss her.”
“Who replaced her?”
“Mr. France,” the attendant said, making a face as if he’d just bitten into a lemon. Evidently, Mr. France wasn’t well-liked, Sam surmised. The attendant gave her Jonathan France’s office number, pointed the way to the elevator, and wished her luck. Which was strange, Sam thought.
She rode the elevator, pondering what might have happened to Janice Everman. It wasn’t difficult to imagine a litany of enemies that anyone working in the Department of Justice might accumulate. Especially someone who could be characterized as a ‘force of nature.’
The Essential Sam Jameson / Peter Kittredge Box Set: SEVEN bestsellers from international sensation Lars Emmerich Page 72