Chapter 18
DIA Headquarters
Joint Base Anacostia-Bolling
Washington, D.C.
Kyle Ryan was in the zone. His fingers flew like an over-caffeinated pianist, alternating between two computer keyboards. Triple monitors on the desk in front of him flickered obediently. He interacted with the machines with fleeting ease, a familiarity that bordered on the ethereal.
Kyle was oblivious to the room’s coolness, a long-sleeve chill that kept the equipment running smoothly.
The gorgeous day outside had gone unnoticed; he’d arrived well before sunrise and, as a matter of security, the workspace had no windows.
On the far wall a software engineer with a sense of humor had hung a banner with an AI-generated image: poised on a rock in Rodin’s The Thinker pose was the robot from the movie The Terminator.
The computer stations all around were littered with take-out food containers and empty energy drink cans.
It was cluttered and frenetic. Fast-paced and impulsive.
And to Kyle it was glorious.
He and the boys were on a roll.
Kyle had joined the DIA only recently, shunning a promotion to O-4 in the Navy for a GS-13 civilian job.
He’d hit the ground running, and not in a figurative sense.
His first field assignment, an operation in Central Africa, had turned deadly.
He had lost teammates and become entangled in an attempted coup.
It was a harrowing experience, but a trial by fire that he’d survived.
Soon after returning stateside, his life had taken yet another unexpected vector.
He was handpicked to join an elite group of DIA programmers on a top secret AI initiative.
Initially, Kyle had thought his superiors simply wanted to keep him out of the field.
But as he learned more about the project, as he recognized its potential, his reservations vanished.
His memories from Angola were still raw and jagged, but they bothered him least when he was here, immersed in this new challenge.
He and a select group of DIA computer experts were putting the final touches on a new and exciting technology.
Historically, the DIA had operated under a largely strategic mandate.
It provided intelligence to the various military services with an eye toward the long term.
Yet the agency also maintained a secondary mission to support current operations.
In essence, the DIA served as a funnel for raw material from the intelligence gatherers.
The NRO and NSA were experts when it came to vacuuming up vast amounts of imagery, SIGINT, and ELINT.
The sheer volume of information, however, had surpassed the ability of analysts to decipher it.
This meant that valuable intel was wasted due to its perishable nature.
The DIA was ordered to come up with a fix, and a number of ideas were floated.
The most promising was Cyber Cell 6.
The question of whether there were actually five other cyber cells at the DIA had become a running joke among the team.
The very existence of CC6 was highly classified, its compartmentalization strict.
The unit’s workspace was a SCIF, regularly swept for eavesdropping, and everyone’s comm devices were checked at the outer security station.
Few others in the headquarters building even knew what they did.
The concept was illustrative of the advances in artificial intelligence.
AI was the new buzzword throughout the intelligence community.
It meant a great many things to a great many people, yet the basic idea was clear.
Massive computing power, properly harnessed and directed, could have impacts on national security that rivaled, and perhaps even surpassed, the advent of nuclear weapons.
Deep in the basement of DIA headquarters, cables and relay networks had been installed to connect to a massive new data center in rural Virginia. The project had been given the moniker MAADN, a necessary condensation of the formal project title: Metadata Analysis and Distribution Network.
It was a mouthful of a name for an undertaking that was, in concept, quite straightforward: to distill massive amounts of raw data into usable, time-sensitive intelligence.
In essence, MAADN was “the ultimate intelligence analyst”—this the opinion of President Jack Ryan when he had approved the classified funding request. In early testing, MAADN had been modified to add capabilities not originally envisioned.
Aside from crunching data from other agencies, it proved surprisingly adept at linking into networks in real time, both open-source and hacked, that could be leveraged to support operational missions.
There were nine technicians in Cyber Cell 6, a mix of software and systems specialists. They worked in three shifts, a necessity since MAADN itself did not sleep. It ran 24/7/365.
Kyle had arrived at five that morning expecting to continue the testing they’d been conducting for weeks—formal validation of the operational software package. Then he’d gotten a call from Bubba Pettigrew at ONI.
He had met Bubba through his sister Katie, which allowed him to put a face with the name—or more truthfully, with the accent.
Bubba had asked for immediate assistance with a high-priority mission.
While the details of CC6 remained highly classified, the DIA had recently put out feelers to selected intelligence agencies, ONI among them, that it was looking for real-world challenges to test the capabilities of a new data-crunching behemoth.
The situation presented by Bubba this morning seemed ideal.
An Air Force C-32A had gone down in Turkey with the loss of all on board, including the secretary of commerce.
Kyle was typically too engrossed in work to follow national news—and truth be told, he thought it often unreliable—but the disaster in Bodrum had hit the news cycle like a tsunami; even he had caught wind of it on his drive to work.
When Kyle ran the idea by his shift team, they were all in favor.
In the room with him were two other DIA civilians: David Craterly and Mustafa Hosny, better known as Moose.
The three had been working together for over a month.
Two other shift teams rounded out the oversight on MAADN, working alternating schedules, and everyone met once a week for coordination.
The request from ONI was squarely in MAADN’s wheelhouse.
The baseline ask was for CC6 to gather and analyze any intelligence that might relate to the crash.
An initial study of the black box data suggested that the aircraft might have fallen victim to a GPS spoofing attack.
If confirmed, it could be an ongoing threat to United States naval and air assets in the region, if not across the globe.
Yet before MAADN could do its magic, Kyle and his team had to define the search.
“I’ve got some updated NRO tracks,” Craterly announced, the map on his main screen a spaghetti-like clutter of orbital profiles.
“Can we control?” Kyle asked.
“Nope. Manipulation not authorized, but one live sweep should be passing over our site in a few minutes.”
“Bring it.”
Craterly began typing.
Given the high profile of the accident, Kyle reckoned they could probably get authority to take control of birds.
But that would take time, and Bubba had emphasized that speed was of the essence.
Soon Kyle saw the crash scene in a multispectral mesh, various bandwidths filtered and fused to create a sharp high-res image.
He saw the wreckage clearly and could make out a dozen people combing over it.
He studied each one closely, but the person he was looking for didn’t stand out.
Bubba had told him his sister was on-site, and Kyle didn’t doubt it.
Last week, she and John Conza had been on a cushy TDY to Italy.
But if there was a crisis in the neighborhood?
Katie seemed to find action like a moth found light.
“Moose, how’s that history coming?”
Dark-skinned and of Egyptian heritage, Moose had been born in Palo Alto, the lone offspring of a Stanford computer professor and a Silicon Valley circuit designer.
He had his father’s dark features, his mother’s slight build, and a brain that combined the best of them both.
He’d graduated top of his class at Caltech, both bachelor’s and master’s, then postponed his PhD in favor of some real-world experience.
And in his chosen data-crunching realm, there was nowhere more real than CC6.
Craterly wheeled his chair across the room like a Mario Kart character—in their spare time they cranked up the classic game, and at six foot two, two hundred sixty pounds, he generally played Bowser to Moose’s Luigi.
Craterly was a graduate of Rensselaer Polytechnic Institute, a leader among the world’s scientific universities.
“Tell me again exactly what we’re looking for? ” he asked.
“There’s suspicion that this jet might have suffered an electronic attack that screwed with its navigation systems. We’re searching for hardware in the area that could have initiated something like that.”
“I need parameters,” Moose insisted as he typed feverishly at his workstation.
“GPS spoofing hardware to begin. Let’s give it a ten-mile radius from ground zero.”
“Mobile?” Craterly questioned.
“Not sure. Enter it as a secondary parameter.”
“If it was mobile, it would be gone now, right?” Moose asked.
“Excellent—we’ll look for what’s different. Program comparative looks at the scene, before and after the time of the crash at two-hour intervals. Extra points for heavy-duty vehicles, something capable of hauling generators and electronic equipment.”
“SIGINT,” Craterly suggested. “If this involved GPS jamming, it might show up in regional signal captures.”
“How do you figure?” Kyle asked.
“A false signal could screw up navigation for cars, boats, mobile phones, anything that uses GPS to compute position and time. Some of those devices will send silent auto-alerts to the system administrators. We should look for reported GPS anomalies within a fifty-mile radius, say two hours before or after the crash.”
“I like it,” said Moose as he typed nonstop.
“Heat!” Craterly added. “The more power, the more effective jamming is.”
“Nice,” said Kyle. “High value to unusual infrared signatures.”
“We won’t get IR for the time of the crash,” Moose countered. “I took a glance at the airport weather. Cloud cover was thick.”
A valid point, Kyle thought. Clouds attenuated IR signatures.
“But it was clear three hours before the crash,” Craterly countered. “We back up and look for IR in an extended window. A vehicle positioning for an electronic attack would likely have been in place long before the event.”
The strategy session went on for another ten minutes, all three of them launching ideas like so many clay pigeons. Those that weren’t shot down were added to the search equation.
At the end, Kyle reviewed the parameters they’d come up with. Everyone was satisfied.
“There’s one more thing,” Kyle added. “Katie says one passenger appears to be missing from the wreckage. It’s possible they never got on the flight. We need to find out who it is.”
This strategy debate was more straightforward. Ten minutes later, Moose hit the Send button.
At something near the speed of light, their commands flashed beneath the Potomac on an SMF fiber-optic cable to a destination forty miles away.
MAADN was waiting.
In the last decade, Northern Virginia had become the dominant venue for data centers in the United States.
The region was to AI what Silicon Valley was to microchips.
The reasons were steeped in practicality: abundant power-supply and transmission infrastructure, proximity to submarine cable landing stations, and low-latency connections to insatiable government users across the D.C. metroplex.
The DIA’s newest warehouse in southern Loudoun County was, in appearance, like a hundred similar ventures.
The main exterior difference was the level of security on the perimeter—double-high fences rimmed with concertina, monitored surveillance, and a military police contingent.
Inside the building, the difference was a matter of scale.
Ten football fields of climate-controlled space was home to thousands of graphic processing units, neuromorphic processors, and ASIC chips.
They hummed in parallel, lights blinking like a galaxy of binary heartbeats.
MAADN received the instructions from its operators and immediately set to the task. For six seconds the AI supercomputer strategized and divided tasking. Within a minute, information was cascading in from six of the nation’s foremost intelligence-gathering agencies.
Without so much as a paused breath, MAADN went to work.