Burn-In

by P. W. Singer

  “Agent Keegan, I wondered when you might call,” Shaw said in a disembodied voice.

  “That doesn’t look like Washington.”

  “I’m at home. With all that is going on, I thought you might need something calming to look at.”

  She wondered if that too was the truth; he could be anywhere, just pushing out a different feed.

  “Thank you. As you’ve probably seen, we’re trying a different way of reconnecting to the natural state of things.” She opened her Watchlet’s lens and extended her arm, so it showed the kitchen sink, the candles, and then the darkness outside. Keegan wanted him to see up close what had happened to the city, reminding him that he’d never know what this was like in real life.

  “I see that,” said Shaw. “You will be glad to hear that the president has assured me that power will be restored within the next two days. Our own estimates are at least ninety-six hours, but it is still a storm that can be weathered.”

  “I don’t think the real storm is over,” said Keegan. “The threat against the Washington area, perhaps other cities, remains urgent.”

  “Yes, I have been briefed on that. In fact, the president’s weighing martial law. Ostensibly the added military resources in the streets are to aid in recovery, but they would really be to stop Jackson Todd.”

  “It won’t work,” Keegan said.

  “And why not?” said Shaw.

  “They don’t have TAMS.”

  “I was told the TAMS program was shut down,” he said.

  “Yes and no,” she replied. “I have it back.”

  “That is an interesting development,” he said, no visible evidence of surprise, but something Keegan guessed he hadn’t been briefed on. “And you are letting me know because?”

  “I have the tool, but not the data. The Bureau can’t provide the kind of information we need.” She left it vague, not specifying the real reason she didn’t have access.

  “Legally speaking, that is correct. So what you are asking for is the ‘open kimono’ package, full cloud search access?”9

  “Yes. We need full access,” said Keegan, not wanting to repeat the creepy phrase businesses used for open networks. “You said I had your support. This is the time.”

  “Three point five million dollars per minute,” he said matter of factly.

  “What?”

  “Three point five million dollars per minute. That’s what we charge client companies for access to that kind of data.”

  “I’m pretty sure you already know I can’t afford that.”

  “Yes, I know that, and so much else, which is exactly what you are asking for here. We have more insight into every American than their own government does. Sensors in their vizglasses. Sensors in their heartburn medicine. Implants in their house. Implants near their kidneys. Chips in their cats. Chips in their chips. All that computing on what is becoming an endless edge of an ever-growing network then reporting back into the cloud anything and everything.10 And then we are able to combine and mine that nearly infinite information to gain insights beyond people’s wildest dreams. Sometimes even the most disturbing ones of their dreams, revelations of their psychology, the true dimensions of their personality, to the extent that the algorithm now knows more about them than even they do.11 And, unlike what we charge the companies to get the fruits of it, people give us all that for free. Not just what they are doing and thinking, but how to change what they do and think.12 They give us control of their lives, without reservation, in exchange for us giving them free access to the services and goods that we will charge them dearly for.”

  “There are literally thousands of lives on the line right now,” said Keegan, not needing another lesson from Shaw on just how influential he was. “I think it’s children’s lives,” she added, knowing that wouldn’t sway him, but still needing to say it.

  “Yes, Agent Keegan, this is the time, isn’t it?” The view wavered and panned to show green fields, the sunset moving to the side. That kind of acreage in the world’s most expensive zip code served as another reminder of how power now came from the control of information.

  “People see who they want when they think of me and the work that I do, but they rarely understand its true purpose. My actions are felt profoundly, even if they are not widely known.”

  The screen pivoted to show Shaw’s face, a smile emerging on it. He really had been there all the while. “This will be one of those times. I will give you what you need.”

  The condescension made Keegan bristle, which made her keep pushing. “Thank you. And one more thing—we need a vehicle. Something to get around a city gone to hell.”

  American Legion Memorial Bridge

  Washington, DC

  They slalomed between the darkened cars abandoned on the bridge, their own lack of headlights making the graveyard of technology feel even more disconcerting.

  The Range Rover Encounter steered itself through the shadows using a fly’s-eyes mix of LIDAR, thermal cameras, and other sensors. It had driven itself to Keegan’s condo not twenty minutes after her call with Shaw; she had no idea from where.

  But it was exactly what they needed—military grade, albeit wrapped in luxury. She’d seen the ads. It was supposedly designed for rugged journeys exploring the world’s last few unconnected zones, like a last-chance glacier safari in upper Greenland.13 The reality was that it was mostly used by diplomats in war zones and dilettantes in suburbia. The sleek SUV melded an exterior of up-armored protection with interior features like the built-in champagne chiller mounted between the hand-stitched red leather seats.14 More pertinent to tonight’s drive, the Encounter had a diesel engine and drive-by wire control system hardened against pulse waves and a massive antenna meshed into the roof itself, providing a Navy warship’s equivalent of bandwidth to pull down from a dedicated satellite connection. Typically used by rich kids to stream VR games in the back seat without suffering the indignity of a microsecond pause in download; tonight, it would be used to mine Shaw’s world-spanning cloud database, searching through zetabytes of information every second.

  “Alright, start searching through the cloud for anything you can that is relevant to our investigation,” Keegan said out loud, partly as a command and partly to get used to talking to the new TAMS in the same way. “We know Todd’s going for something that has to do with the tenth plague, Death of the Firstborn. That’s the big one, the one that finally convinced the pharaoh to let Moses’s people go. Between that and the symbolism, it likely has something to do with kids.”

  “That aligns with current model projections,” TAMS said.

  Keegan wondered if she should think of it as TAMS or TAMS-2. And then she just set it aside. It was a TAMS with the same data and same memories, just up to the point that she’d turned it off.

  “We also know Todd still has access to Preston’s automation software,” she said. “Check what matches up, cross-referencing large populations of minors and current implementation of systems based on Preston’s code.”

  “OK,” said TAMS. “Would you like me to include past implementations?”

  “Yes, good call,” Keegan said, noting how it had packaged the suggestion that she had given the wrong command as a polite request. Was this another one of the user-interface updates from the other FBI field offices’ TAMS that she’d just not noticed before? Or was it a feature only of wherever Modi had gotten this one from? “Just because somebody installed Preston’s software once and never updated it, doesn’t mean they aren’t still using it.”

  The machine plugged itself into the vehicle’s entertainment port for even faster access to Shaw’s cloud network. Keegan snorted a laugh at the sight of its metallic arms on the luxury vehicle’s passenger-side pullout table made of rare rosewood. “And don’t scratch the trim; I’m not sure I can afford the deductible.”

  “OK,” said TAMS, keeping its arms still, meaning it had already run those calculations.

  The air warmed slightly as TAMS ran its searches.
The machine occasionally emitted a faint hum as it opened up each new layer of data in Shaw’s cloud.

  Keegan could do little but wait, so she pulled the car to a stop at the halfway point of the bridge. It was as logical a place as any, as they literally didn’t know which side of the river Todd was on. Looking out into the abyss-like darkness of the waters below made it feel like the potential places were just as infinite as the data.

  “There are 152 children between the ages of eight and seventeen at a Red Cross disaster relief shelter established at the National Zoo,” said TAMS, starting to report matches. “The National Zoo operations center utilizes Professor Preston’s open-source automation software.”

  Keegan shook her head. “Not it,” she said. “Tigers aren’t the problem. It’s technology that has to be the killer.”

  TAMS identified another connection point. “There are 167 children of indeterminate age from Baltimore, recently loaded onto three Star Choice LLC school buses. They are traveling to an aid facility set up at the US Naval Academy in Annapolis, which uses the software in its network as well.”

  “Something Defense Department related could be significant,” she mused, pulling off the pair of tortoiseshell and titanium vizglasses fastened into the charging clips on the ceiling. Shaw had certainly provided the whole package, all she needed to work, in a bubble of luxury. It wasn’t a sign of added care; rather, the network of support built around the superrich couldn’t even contemplate anything else.

  The lens projected the data stream pulled from Shaw’s KloudSky network. Keegan’s vision filled with everything from social media posts by the kids on the bus to satellite tracking of the vehicles’ movement history, derived from rider activity metadata over the last several years. Was this what it felt like to be Shaw, an infinite amount of other peoples’ data to be mined to your own personal purpose?

  “They’re already past Bowie,” Keegan said. “And each bus has a human driver. Doesn’t seem to fit the profile of Todd’s attacks so far.”

  But just what was that profile? Not just the means but the ends? Everything Todd had done so far had pulled on the thread of human powerlessness, the sense that death could be random, but only when ascribed to a machine’s actions, not a person’s. The machines—really, the code running it—would have to be the ones taking lives.

  TAMS moved on to a new cluster of data. “There are ninety-three newborn infants at the National Neonatal Intensive Care Unit of Children’s Hospital, which relies on—”

  Her sharp intake of breath and rapid rise in pulse rate were enough to prompt the machine to dive further. Without being asked, a satellite image of the hospital popped into her lens view.

  “That’s a hell of a lot of preemies for one hospital,” she said. “Why so many at that one NICU?”

  The vizglasses then changed to a map of Washington, DC, marked by a series of overlapping circles, showing the blast effects of the power outages. A series of dots then emerged, the locations of the area’s hospitals, most of them inside the circles. “Multiple area hospitals were taken offline in the recent power outages,” TAMS summarized. “Emergency services have moved their patients to parallel units at any nearby operative hospitals, along with their equipment.”

  “How does it overlap with the threat profile?”

  “Patient care in the NICU is coordinated by a system that automatically registers patient identity and manages their condition and treatment. Among its capabilities are oxygen-level monitoring and drug delivery.15 This system’s software is also open-source algorithms originally created by Preston.”

  TAMS shifted Keegan’s visual to display the hospital’s operating system, showing the patient records as multicolored lollipop-like icons within an administrative form layout. Rather than the customary images of faces, however, the accompanying picture for each patient was just a barcode. She realized they were the identifiers worn on their ankle bands. Human babies’ faces were too mashed up for facial recognition software to register them yet, plus they were just data in some form to the network anyway.

  “However, the systems are air-gapped from the Internet and have asynchronous cloud access,” the robot explained.

  “So, Todd would have to be at the hospital in person . . . I see four in the map outside the affected areas.”

  The map rebuilt itself, with the functional hospitals highlighted, and a graph stacking up the number of patients in their NICUs.

  “Children’s National has the largest number, followed by Fairfax, Georgetown University, and Sibley,” TAMS reported.

  Dammit, they were spread all over the city. Which one? Children’s National had the most cases, making it the most damaging target to hit. But Fairfax was on the same side of the river as Todd’s house, easier for him to reach while on the run.

  As Keegan churned through the parameters, TAMS highlighted Georgetown University Hospital, overlaying it with a photo of a tiny baby linked up to a respirator, clearly taken through a window. Then, it projected a second photo of a tired, smiling mother in a wheelchair, holding a baby as she was being wheeled out through a pair of sliding glass doors.

  “KloudSky records include photos of Todd’s son and wife posted online fourteen years ago,” it said. “Metadata markers show their upload was made via a wireless network run by Georgetown University Information Services. Subsequent credit card company records confirm Todd making multiple purchases at the Georgetown University Hospital gift shop.”

  Keegan knew that hospital. She knew that NICU. “Haley was born there,” she said quietly, trying not to personalize it but unable to avoid it. The memory of that NICU was as much a part of her as her time in the Sandbox, and ten times as scary. The image forever seared into her brain was of Haley’s glassy eyes and button nose, all that peeked out from the light blue plastic thermal wrap and cotton that cocooned her inside the machines that kept her alive.

  “Yes,” TAMS said, matter of fact in its knowledge of all the moments in her life.

  “How many?” she whispered, her mind coming to grips with it.

  “There are presently seventy-seven infants in the Georgetown University Hospital NICU.”

  Across her field of vision, the lollipop-like icons reappeared, then their demographics. Thirty-two boys. Forty-five girls. She shifted in her seat at the realization of how many lives not yet lived depended on this decision. “Georgetown Hospital seems the most likely then?” she asked.

  The machine responded immediately. “Yes. It is a probabilistic exploration, but that is what the model indicates.”

  “Whatever you want to call it, it’s little better than a hunch . . . But that’s what good agents sometimes have to do.”

  Even if they were wrong, the machine might as well learn from the experience.

  Georgetown Neighborhood

  Washington, DC

  As the Encounter drove itself through the closely packed row houses of Georgetown, Keegan let her eyes wander, taking in the two sides of the tragedy, evident just from the outside of people’s homes. In the glimpse of an elderly man in a sleeping bag, curled up on his porch, probably too fearful to be in the same room with machines that couldn’t be trusted. Through the next-door house’s living room window, a mother reading a book to her kids by candlelight, seizing the opportunity to reconnect while all other connections had gone down. “There’ll be a lot of memorable family moments made tonight . . . as long they don’t accidentally burn down the city,” Keegan thought to herself.

  As they drove farther up to Georgetown Heights, into the wealthier part of the neighborhood, the row homes turned into mansions. Keegan caught sight of the running lights of a large National Guard octocopter loitering overhead. TAMS marked it as providing network connectivity, an emergency deployed hot spot. Not surprising, Keegan thought, that the politically connected were the first ones to be reconnected.

  TAMS broke the silence. “I have now obtained an answer to your query on the date of April 28.”

  “What?” K
eegan asked. TAMS must be choking down all the data in Shaw’s cloud, churning through all the vestigial requests, in this case from literally days back. “What was my question?”

  “During the operation to arrest Gregory Heath, you inquired about a real estate transaction.”

  “Huh?”

  Noah Reddy’s voice then echoed out of TAMS’s speaker. “This really is beautiful land, though.” Keegan’s stomach clinched as she relived her last moments with her friend.

  “It is. Doesn’t seem fair,” she heard her own voice say. “TAMS, just how did these fuckers afford a place like this?”

  TAMS turned to face her and then said in its distinctive monotone, “With my increased access to data, I can now answer that question.”

  “You’re scouring real estate records?” Keegan could not conceal her annoyance at the waste of TAMS’s processing power. That was the problem of putting parallel computing into a cloud that big, all the queries running off in different directions.

  “Yes, I have also achieved access to additional financial and legal data through the broader private cloud accounts provided by Mr. Shaw.”

  “Not the time, TAMS. We have a more important investigation.”

  “This information is relevant to our investigation,” TAMS replied.

  “OK. What’d you find out that’s so important then?”

  “There were four Cayman Islands–registered shell companies involved in the real estate transaction: PS2 LLC; Brilliance LLC; SCIKITE LLC; KronoGraf LLC,” TAMS said, as it projected a series of records and lines linking legal documents. “The purchase followed a sequential transaction, which originated with a Bitcoin mining collective, Militas Mining.” A line then connected to highlighted text from the Todd investigation file. “An offshore bank account at Can-Fin Global was accessed by a server in the geographic location of Todd’s neighborhood. This account received four deposits from Militas Mining, using them in the purchase of nanocomposite paper sheets. These are of the type utilized in the design of the improvised electronic pulse weapon used by Todd.”