“What can you tell us about the damage?” Barker asked.
“Pretty near total. Coal structures, and particularly soft coal seams, aren’t very strong to begin with. As I understand it, what miners do is extract most of the coal throughout a large area, leaving pillars of unmined coal to support the roof. Then they mine the pillars, too, replacing them with big table-shaped supports. Finally, they remove those, too, as the machinery retreats. Eventually, the ceilings come down on their own, so it doesn’t take much to cause a collapse. And these weren’t small explosions. They registered on seismic recorders a long way off. Mine management says it expects all, or almost all, the rooms and tunnels collapsed. That means all the equipment and other underground infrastructure was destroyed or buried, too.
“Thanks,” Barker said, turning to the audience. “Anybody have any questions for Graham? Okay, Mary?”
“How hard do the mine owners think it will be to reopen the mine?”
“From what I’ve heard, it’s not so much a matter of getting back in operation as starting all over again. Except for some of the vertical shafts, this mine no longer exists. The other problem is that so many mines were hit there’s no telling how long it will take to replace all the destroyed and buried heavy equipment. Some of it is generic, or assembled from off-the-shelf components, like the elevators. But the rest, like the coal extraction machines, are unique to this kind of work. Coal mining hasn’t been a growth industry for a while, so if the destruction elsewhere is as complete as it is here, it’ll be years before the manufacturers still in business can replace what’s been lost.”
“So how long do they think it will take to get things running again?”
“The big question right now is whether they’ll reopen this mine at all. Demand for coal has been dropping for years, and it would take a huge investment to get this one back in operation. The company may decide to just take the insurance money and shut things down for good.”
“Thanks, Graham. Good report. Stay safe.”
* * *
“So how did your predictive model do?” Barker was debriefing Frank and Shannon in his office after the team meeting.
“Not so well,” Frank admitted. “This was kind of off the scale of anything we expected. That said, the attack still shows our approach is solid even if it fell outside the parameters we built into the model.”
“How’s that?”
“First, because the attacks occurred just a few hours after an announcement was made with a very big impact. For the first time, the amount of carbon dioxide in the atmosphere exceeded 400 parts per million for an entire month. That’s the highest level in millions of years. It’s also a big deal because that’s the point where scientists say a major ocean level rise becomes inevitable. So, it’s a degree of CO2 pollution we never wanted to reach.”
“Still, it’s just a number,” Barker said. “It’s not as if we’re safe if CO2 plateaus at 399 but screwed if it hits 400.”
“That’s another reason why it fits right into our theory that negative warming announcements trigger attacks. By the rough calculations we’ve made, the amount of coal capacity taken offline correlates well to what it would take to put us back under 400 PPM for the time being. That tells me we should put the model project into overdrive. Of course, that would take more people.”
“You’ve convinced me. Let me see if I can sell it upstairs, and I’ll let you know.”
“Sounds good. Incidentally, I doubt you want me to lead that effort.”
“Really? Why?”
“Building out the model is just a numbers exercise. I’d rather focus on figuring out who’s behind the attacks and how to stop them. Would that work for you?”
Barker paused. “Yeah, I guess. Finishing up the model won’t take any special skills. So, sure. Why don’t you stay with what you’re good at? Shannon, how about you?”
“I’d like to keep working with Frank.”
“I thought you’d say that.” He paused again. “Okay. That also makes sense. Frank still needs a partner here at the NSA. But I’ll need your help to transition the predictive model project to someone else.”
“Of course.”
Barker stood up. “Then, in that case, the two of you better beat it so I can tackle the latest important but complicating discovery you’ve dumped on my desk.”
* * *
“Nice job ducking the dreary stuff,” Shannon said on their way back to Washington.
“Actually, that wasn’t it. I’m just not a manager type.”
“So, I’m told.”
“‘So, you’re told?’ How’s that?”
“I’m sorry – I didn’t mean that the way it sounded.”
Frank frowned. He wasn’t used to anyone knowing much about him and preferred anonymity.
“How should it have sounded?”
“I guess – I don’t know – Jim Barker just mentioned you liked working alone when he asked me to be your go-to guy at the NSA.”
Well, what could he say? He’d always been ill at ease around people. Half the time when he spoke with someone he felt like he was talking over a telephone link with a time delay. He’d start talking over them, never quite able to get the timing right, and feeling increasingly awkward by the minute. And with most people he felt like he’d just woken up from a twenty-year nap, unable to conversationally synch up with the reality everyone else lived in.
“Well, that’s so,” he said. “I’m not exactly a people person.” He left it at that, and they drove in silence for a while.
“I hope you didn’t mind me asking to continue to work with you?” Shannon said at last.
“No – no. Not at all. But I was a bit surprised.”
“Why?”
He regretted the words immediately. He didn’t want to admit he assumed folks felt as awkward around him as he did around them.
“I don’t know. No reason, I guess.”
She laughed. “Fact is, I’ve been enjoying being your Girl Friday. I’m not used to working with a famous hacker. It’s been interesting.”
He reddened. “Well, the North Korean thing was a while ago.”
“But not the Caliphate attack –”
“Wait a minute – Jim told you about that, too? What else did he share?”
“I guess he also told me you discovered someone was trying to hack the last election and stopped them just in time. Is there more?”
“No! That’s all there is. Anything else I’ve done has been extremely boring.”
“Well, what we’re working on now certainly isn’t. And I can see why you like being able to follow up on anything you want to. I hate being tied down to one piece of something bigger I can’t help manage, so I’m thrilled to be along for the ride.” She turned and gave him an even warmer smile than usual.
* * *
“So, did you ask Shannon out yet?”
“Please, Marla, if you keep doing this, I won’t answer the phone when you call.”
“You don’t have a choice. You’re my dad. It’s in the contract.”
“Not so. The contract only says I have to tease you.”
“I haven’t believed that since I was five years old.”
“More like seven, actually. You know, I always meant to type up a phony contract with the name of the hospital at the top so I could prove they made a father agree in writing to tease his daughter every day. I’m sorry I never got around to it.”
“Nice try. You still haven’t answered my question. Have you asked Shannon out?”
“No! Are you happy now?”
“Of course not. Not until the answer is yes. And don’t say ‘I’ll think about it’ again. It won’t work. What’s the problem?”
“The problem is I’m too used to being alone.”
>
“People born with only one leg probably get used to that, too. But they take a new one when it’s offered.”
“So, what I’m hearing is you’re suggesting I need an artificial relationship to make up for my congenital social shortcomings.”
“Whatever floats your boat. Anyway, asking her out to dinner isn’t inviting her to move in.”
“Well, sure. But if it turns into something, then there are, well, expectations.”
“Good grief, yes. Like having someone to share and do things with. Now, wouldn’t that be just terrible?”
“Look, I’ve got to go. Talk to you later.” He hung up before she could respond. He stared at the phone for a moment and stood up. Time for a walk.
Hands shoved in his pockets, he set out for the Mall. The fact was, he’d started thinking he should ask Shannon out for dinner. But he hadn’t done anything about it. Why?
How had he become so solitary, anyway? The way he remembered it, he’d always had at least a few friends as a child. In college, there had been places near campus where he could hang out. After his freshman year, he practically lived in a dusty, dingy coffee house in the basement of a university building, becoming part of its motley bunch of regulars. There was Mary Pat. She ran the place. And Freddie, an out-of-work, overweight, diabetic ex-con who lived with his mother.
He stopped in front of the Lincoln Memorial and watched the school groups climbing the steps. Who else used to hang out at the coffee house? Right – Randy, an ex-stockbroker who flipped out on LSD back in the late 1960s and lived in a tent in the middle of a big park at the end of a subway line. There was a law student, whose name he couldn’t remember, who seemed to only leave to attend classes. He drew cartoons and sometimes played the piano there – badly. And then there were the anonymous, compulsive Go-playing engineering students. He used to roust them out at midnight by blasting Jimi Hendrix’s version of “All Along the Watchtower” on the sound system. He smiled at the memory. Had he started to change even then?
Now that he thought of it, of course, he had. He never actually did anything with any male friends. If he was in a relationship, he had all the company he needed. And when he wasn’t, he could fall back on the no-risk, no-personal-investment habit of hanging out where he knew people. At least, until he got married.
He turned away and stared down at the reflecting pool, its long sides converging toward the point that was the Washington Monument.
By the time his marriage broke up, the guard had changed at all his old haunts. The scruffy coffee house wasn’t there at all; the university had turned it into a sterile satellite cafeteria with plastic tables and chairs. And he was painfully aware of how much older he was than the new regulars at the hangouts that were left. Whatever meager social skills he had weren’t up to the challenge of working his way back in.
He turned and headed home, staring at his feet. That was twenty years ago. He’d withdrawn deeper into himself each time his increasingly infrequent female relationships ended, becoming more awkward and reticent each time. He couldn’t imagine anyone wanting to be around the cocooned outcast he was now. After all, why would they?
His phone buzzed in his pocket. The text read “Pick you up for work tomorrow?” It was from Shannon.
6
Who? Me?
Shannon frowned. “So, I’ve pulled together the dates and exact times of all the announcements to date, or as close to the exact times as I could get. What should I do next?”
Frank hesitated before responding. If his hunch was wrong, he’d look more like a doofus than a famous hacker.
“How about you graph how quickly a responsive attack occurs after an announcement. Some I don’t really care about, like the ones that happen at night at a facility that doesn’t have round-the-clock work shifts, because we assume those were deliberately delayed. What I’m curious about is the time lag between when news breaks and the response begins. Does it vary? Is it always the same? Is there some other pattern?”
“Okay. I’ll do that.” She returned to her desk and started tapping away. A few minutes later, she came back from the printer with a piece of paper in her hand and a fascinated look on her face. “Here you go,” she said.
He pushed back from the desk and looked at the graph. Bingo. There was a big spike. More than eighty percent of the daytime attacks were launched within thirty minutes of the triggering event. Just as he’d hoped.
“Well, you look happy,” Shannon said. “What is that graph telling you?”
“That a robot is triggering the attacks.”
“A robot? Seriously? You mean a real robot, like in science fiction?”
“Not if you’re thinking of a physical robot. But if you mean an artificially intelligent software program, then yes, exactly. What this graph indicates is that all these attacks were set up in advance. As soon as news of a certain type of event hits the Internet, the software calculates the CO2 impact and then launches an appropriate attack. I expect if we look deeper into the data, we’ll find the remaining twenty percent relate to safety delays and unique events requiring more sophisticated impact calculations.”
“How can you tell all that?”
“Here – look at the timing. For these ultra-sophisticated attacks to launch this quickly, someone – or something – would have to watch the news twenty-four hours a day. He’d also need to immediately match a specific attack to the greenhouse gas impact of a specific announcement. To do that, the attacker would need a dual database-driven model, just like the one we’re building, with one database filled with targets, the other with specific events, and each of them ranked by greenhouse gas output. After someone’s done all that – and, of course, planted all the malware at the targets as well – all they’d need to add would be some code to make the match and automatically trigger the attack. That last bit would be a pretty basic robotic program, but a robot nonetheless.”
“Makes sense. Does the graph tell you anything else?”
“Not directly, but – no surprise to you by now, I expect – it makes me want to review more information. Could somebody do a big data analysis to see how many announcements of negative climate change data have been made, large and small, since the attacks began?
“I’m sure they can. Then what?”
“Whoever is behind this must have installed malware at a heck of a lot of targets, because it seems like they’ve always got a right-sized and nationally-appropriate target in inventory. But I’m not sure that’s the case. I’d like to see whether any announcements weren’t followed by attacks. If there have, that might allow us to make some helpful inferences.”
“Like what?”
“Well, like whether the impact of some announcements is too big or too small to react to. If so, we can use that information to tune up our projection model. If we see trigger events the attacker usually would but hasn’t reacted to, what’s the reason? Does it not attack certain countries? Or maybe its inventory of compromised targets isn’t so big after all? Let’s ask for that information and see what we get.”
“Okay, will do. By the way – I packed enough food for two today. Want to break for lunch?”
“You didn’t have to do that.”
“I know. But I’ve seen what you bring to eat.”
“What’s wrong with a granola bar?”
“Nothing, other than the fact that it’s almost nothing. I’ll get what I brought from the refrigerator and meet you in the conference room.”
“Can I suggest a lunch topic?” Shannon said when she met him there, setting a chicken salad in front of him. “I don’t know a lot about robotics and artificial intelligence, so if that’s going to be a big deal on this project, I should buy a few books and study up on it. But in the meantime, maybe you can give me an introduction.”
“Sure. I guess the first thing to under
stand is that just like anything else with computers, it all starts with ones and zeros and gets built up from there. And again, just the same, it’s all logical and hierarchical. So, you can’t just start at some high level and write a complicated program. You have to start with ‘if this, then that’ statements and the like and work up from there. That’s tedious and time-consuming.
“Next, AI is different from other types of programs, like ones that solve mathematical problems. With those, there’s only one right answer. AI can be especially useful where there may not be just one right answer. Or maybe there is, but you need to come up with a much more complicated approach to get there. Like image or speech recognition. Those are really tricky challenges, where the data can be very ambiguous.”
“So how do you go about it?”
“Computer scientists have come up with a lot of different ways, but some techniques are fundamental to just about all of them. One of them involves choosing the most likely outcomes instead of computing them all.”
“Wait a minute – how does a program know what’s more likely?”
“Good question. There are again several approaches, and more being developed all the time. For example, an AI program can use information loaded into a database for it to draw on, or it can use information it’s added to the database itself.”
“Where does it get new information from?”
“Here’s an example. Let’s say you’ve created a computer program to figure out how to get from the start of a maze to the end. If the program tries one way and hits a dead end, it would add that information to its database so it doesn’t follow the same route again. Every time it starts over, it has fewer options to try, and eventually it tries the one that will solve the maze. And if you extend the maze, it doesn’t have to start all over again, because it already has the right route to the new starting point in its database.”
The Turing Test: a Tale of Artificial Intelligence and Malevolence (Frank Adversego Thrillers Book 4) Page 5
