In those moments I am drifting from wakefulness, I can only hope we do not get one of Barrens’s colleagues knocking on my door about noise complaints. It strikes me that the entire day today, not once did I think of Minnow the cat, and I fall asleep, probably more pleased than I ought to be.
6
Creating the data-mining swarm is not too hard. It needs search functions, secure access protocols, statistical tools, some plug-in modularity, and some machine learning to optimize performance over time. It will mostly be a fancy search engine that will trawl public forums and sites, and penetrate the secure data archives accessible with Callahan’s hacks, and other databases as bypasses are found to get to them. I discard the most robust of the evolutionary algorithm techniques and re-implement what I have gotten to work before in my class projects. I build it around a loosely coupled neural net of particles that can alter the weights of the communications between each individual subunit, and a simplified algorithm that will allow the swarm limited self-modification over how each unit evaluates search parameters and how each tags retrieved data for assembly by the collective. The function for testing the swarm’s output is fitted to Callahan’s trove of files, and can itself be tweaked with new results. Coding it takes two weekends, since most of the module functions are code blocks I’ve recycled, or from the toolkits of Barrens’s mentor.
Debugging the little monsters takes a month.
Barrens helps with the testing on an isolated server I’ve built, a computational cube that’s just a stack of tablets wired together for more power. He has a knack for putting in just the right mix of inputs and training data to crash the thing. I teach him about the program as I add in a more robust set of debugging functions—when this thing starts to get going, we’ll probably need them for future fixes.
Finally, when a week straight of his putting it through its paces fails to crash it, it’s probably stable enough to use. We’ll still need to modify it as we go along, but it’s good enough. Now I just need to tap into the rebellious youth Lyn and I pretended to have when we were kids, to make sure that the spread of my artificial infection is as untraceable as can be.
Lyn was always the better programmer. It’s why she gets paid so much to be in Information Security. She did not have the temperament and physicality to make it as a field agent, but she is one of the best neuralhacks under Research and Development, and not just Network Administration.
At seventeen, I learned my best tricks from her. It is also when we discovered some of the peculiarities of the Nth Web architecture.
The false dawn has begun to slip up at the edge of the false sky, its red-gold light cutting through the shadows in an alleyway between the Londinium Center for Fine Arts and the Museum of Ethics, revealing two girls in tan jumpsuits and boots and gloves and face masks climbing into and out of the recycling bins.
I’m sweating, and breathing hard, and it’s not from the physical effort or from the smell.
Could you look any more anxious, Hana? Calm down. We’re just students rooting through the junk for a project.
We think our thoughts directly to each other, telepathically. It is sloppier and more difficult than using the Implant to send e-messages to each other, but this way, there is no record of the messages going from Implant to the Web and to Implant. Nobody hears our thoughts but us, and no record exists but in our memories.
I’m nervous, okay? You should be nervous too. If you get written up, you’re never getting into ISec.
Sure I could. We’re just exploring the system—nothing malicious intended. Lyn gives me her brightest, most confident grin. It’s even true.
Standing knee-deep inside a blue garbage bin full of dusty coils of wiring and a variety of shattered appliances and electronics, I finally spot a cracked black slab of crystal. My student amplifier glimmers to life.
“Found one.”
Lyn hops out of the adjacent garbage bin, calls out, “Don’t shatter another one!”
“It happened once, okay? Lighten up.”
I don’t need the amp to lift such a light mass up into the air, but I need the practice of regulating the extra power available through the grid. My talent’s been going through its last surge of growth, and while it makes big jobs easier, I had to relearn how to handle delicate objects.
My control is mostly back to normal. The hours of juggling eggs with my mind at night has restored my fine mental control, but it still takes more focus than I’m used to.
So, yes, I pulverized the first tablet we found this morning. Too excited, maybe, or worried that any moment someone in a uniform with a loud voice is going to walk up to us and ask exactly what we think we’re doing. The next two tablets we found had chemical contamination in the circuitry and could not be made functional without completely deconstructing them back into raw plastech and resynthesizing the circuits.
This one is perfect. A wave of my hands, and the screen flickers. Lots of static. This part I can do better than Lyn. She’s a better programmer, but is mediocre with plastech manipulation and fabrication.
I think-tap the small app I customized into my amplifier. It draws on my talent, and under the loose guidance of my thoughts, the program slides psychic tendrils through the circuits, finding the loose connections, repairing them. It takes twenty minutes, and then I directly flare some more power and fuse the cracks on the casing.
“Almost brand-new.”
“Great!” Lyn almost snatches it from me in her haste. “Did you remember to burn out the transmitter?”
I didn’t actually.
“Of course I did.” Surreptitiously, I send a last telekinetic spike of energy into the device and melt the wireless components as I was supposed to.
She grins at me. She probably felt the surge through her fingers. “Suuuure you did. Let’s get going.”
I can’t help shaking my head. “You are way too enthusiastic about getting into the sewers.”
“Hush now. Do your thing, muscles!”
I take a deep breath. A manhole is right in the alley. Now, I need to draw some real power from the grid—lifting the two-hundred-kilo armored disk out of the way is more than I can do without an amplifier. A few blue sparks of waste energy ripple around the edges.
“Wahoo!” Without a beat of hesitation, she drops down into the darkness. Lyn has some bruiser in her; not enough that she could have a combat role, but she can dance with preternatural grace and has no fear of a five-meter drop.
Gingerly, I climb down the ladder, then lower the access door back over us with a clang. In the darkness, the utility lights on our shoulder pads and belts automatically come on, drawing on our psi to light the way.
For all her bluster about not being afraid, Lyn is the one who insists on our attempting our experiments only through an untraceable tablet wired into a maintenance port in the sewers. She is the one who taught me about the curious phenomenon of dangling IDs—code identifiers for individuals who don’t seem to exist anywhere in the system. Every person who has ever lived on the ship has a unique ident code; when one dies, the code is supposed to be locked out of the system.
But in her restless, sleepless nights of exploring the Nth Web, Lyn found active identifiers that don’t belong to anyone. We had thought they were just a myth.
The air down here is fetid and sour, but not toxic. We try not to look at the things floating in the ankle-deep water.
She plugs a jury-rigged adaptor cable into the centimeter-wide, triangular maintenance port and plugs the other, tiny needle end into our refurbished handheld terminal. “Testing and … it works.”
Something bumps my calf and I hold in a shriek.
“Lyn, I swear, if a rat climbs up my pants, I am going to murder you.” I shiver a little in the hot, humid shaft.
“I’ll treat you to lunch at that crepe place you like every day until you’re sick of it if that happens, okay? Now, what have we here?”
Both of us have our hands on the tablet. She leads now because she is
the better neuralhack, but I ghost along behind her, watch as she runs her custom-made masking algorithm, which sets up a subnet of proxies scattered around the hardware network of Analytical Nodes upon which the Nth Web runs.
Finally, she starts trying out one ident code after the other, logging them into the system through a maze of accesses that would take hours for any system monitors to notice.
“Huh. They all work. They … they all work.”
The bemused expression leaves Lyn’s face, and she’s not smiling anymore.
Perhaps she’s finally thinking through what it might mean.
“How can so many of them work?” Now, her cheeks are starting to redden. She is upset. “I figured maybe a couple of them would. But we’ve gone thirty-two for thirty-two.”
“Let me have a turn, all right?”
I imagine that Lyn is thinking furiously about why so many vulnerabilities are slipping through the cracks of the Nth Web’s security.
But I also have some experiments I want to run. Now that we don’t have to worry about tripping an unauthorized-access alert that can trace to our real identities, what I want to learn most of all is just how big the Nth Web is.
I plug in my apps, which start to use a special data signal I found while messing with a buggy Nth Web protocol assignment. It is a code for a data packet that ignores security between the nodes and accesses the hardware directly. The only thing the signal does is cause one Analytical Node to ping all adjacent nodes, which then send an echo back to the source. Limited as it is, no student should have access to something like this.
My program sends out these special pings from the nodes closest to our access point, measures the delay it takes for each reply, then accesses the next ring of nodes to send a new set of signals. In this way, it slowly constructs a map of how the nodes are connected. Then it’s my turn to think that surely something is very, very wrong. The map is the way I imagine it to be, at first. Then, as more pings go out and more nodes answer, it keeps growing, and growing. From dozens of points of light, it becomes thousands, then tens of thousands. There are so many connections to display and so many echoes returning that the limited processor in the tablet slows to a halt, freezes completely.
“What in the … How can there possibly be so many nodes on the ship? There’s … there’s thousands of times the computational power we could possibly need for every function, and ninety-nine percent of it is just dormant, with no program processes running on them.”
Lyn frowns. Down here in the dark, both of us have found something we would have been happier not knowing. All our lives, we have been told that we have to conserve resources for the long, long journey to our new home. So why would more computers than we are supposed to have power to run be built into the Noah? “If the system is that large, that could explain why so many dangling ident codes slip through the garbage-collection programs. They must get copied back onto some of those redundant systems, and maybe it takes ages to propagate the sweeps.”
I try to think of reasons why. “Perhaps all those extra nodes are needed for something. Maybe for when the Noah reaches Canaan?”
Lyn unplugs the tablet and carefully and thoroughly smashes it under a glowing, psi-enhanced kick. She lets the pieces float away.
“I just changed my mind. I think, very strongly, that we should forget this.” Her voice is grave, and the words are slow.
Yeah. All right.
It doesn’t feel like a joke anymore, the stories we tell each other late at night, of overcurious students being taken away by gray-uniformed officers only to return dazed, unable to remember quite what their class project was. And of course, there are stories of those who never return at all.
How much more is there to the ship that people are not supposed to know?
Lyn and I eventually spoke of those discoveries again years later and laughed at ourselves. As we were promoted through the ranks, we convinced ourselves that what seemed like vast and dangerous secrets were ordinary redundancies built into the system, and functions of bureaucratic red tape such as the LTI database, which we wanted to think was the source of the dangling IDs. That the ghost ident codes belonged to individuals who were temporarily taken out of the system, unsolved murder victims hidden away in the LTI databases indefinitely or until their killer dies and his revealed memories “solve” the crime, at which point the records are restored to the larger system. But I don’t think so anymore.
Back then, it had seemed like the sign of something bigger than Lyn and me, something terrifying. Something that could get Information Security appearing at our door late at night.
That cold fear creeps up my spine again as I consider what Mincemeat could be. Killer? Monster?
Lyn’s method of masking system access still works. With it, my data-miner begins to infect its way across the nodes of the Noah. I take an untraceable tablet I restored out of a junk pile, wire it into a maintenance port behind the Moskva Mathematika Center, run a masking function through proxies and a dangling-ID log-in, and upload the first piece of the swarm. It starts in the Statistical Research Center’s cluster of Analytical Nodes, where its multiple threads, even if someone notices them, could be mistaken for a statistical probe through the databases. From there, it replicates onto adjacent Analytical Nodes, into the financial system, health records, employment records, and further and further out across the ship.
After another week of waiting and wondering, the first useful data retrievals start coming in. I am eager to do something other than refine my program and comb through Callahan’s trove and stare at the image of Barrens’s weblike map of the victims and possible relationships and the possible profile of a killer that might do this. It’s more question marks than anything else. Another month goes by before we have a new lead to chase down.
The human element of detecting is not without boredom either.
I waste an hour going back and forth along the West Twelfth Park on foot, trying to find a person of interest I have dug up on my very own. It feels as if the bronze statues of the historic explorers of Earth are mocking me with their far-off gazes and optimistic poses. James Cook, in particular, regards me, brows tilted just so, his eyes simultaneously knowing and condescending. I have passed his green-tinged statue three times.
I promise myself that when the opportunity arises, I will have Cook’s statue relocated to some lightless corner of one of the museum archives.
Finding my man would be easier if I had messaged the man first and actually made an appointment. The small weed of anxiety has grown larger. I avoid leaving tracks in the system as much as I can.
I do my deep-searches anonymously on public terminals, mask my accesses and queries with false trails to other nodes throughout the system. I change my pattern of purchasing groceries, no longer always picking them up from the same vendor each time. I let Barrens do his watchdog best. Just in case.
These worries arise more easily at night, or in the dark office of Long Term Investigations. Walking through the park under the bright daytime skyscape fades the misty fears, until I think of how Mincemeat might have found and eliminated Callahan, a canny veteran gifted with both bruiser and touch talents.
“Seen Cal spar with Enforcers and give them a run for their money,” Barrens once said. “Take a real monster to do that to Cal so quick, without a fight that would tear up the whole apartment block.”
Bees buzz in the air. Lavender blossoms along the pebble-textured paths through the park. Gold and orange and silver carp swim idly through the streams connecting the ponds spread out through the park. Oak trees dapple the grass with the shadows of the leaves. Children are everywhere. Some sit and laugh at the sight of a fuzzy Welsh corgi scampering and barking, and a few particularly lucky ones pet it and giggle when it licks them. It must belong to a whole team from some department—no one individual can afford a dog, other than the department chiefs and higher-ups in the chain, the ones who own actual houses as individuals, rather than renting. If any one o
f those top-class officers were out here in the park, an ominous escort of black-armored Enforcers would be keeping an eye on everyone around.
Finally, I spot the man I am looking for. Keepers are required to bring the children in their care out for “fresh air” and greenery a minimum number of hours of the day depending on the child’s age, and this is the closest park or garden to his residence.
He is young, for a Keeper. Just twenty years old, still a bit coltish and lean, not yet come into the weight he looks as if he will put on later. His mix of looks calls to mind long-lost regions of the Mediterranean—dark hair, olive skin, sharp features. He gently rocks a baby carriage, singing something without words, slow and soothing. He has no permanent partner anymore, just a rotating series of substitutes—this, and the drop in his household leisure spending, is what drew the attention of my AI probes.
Today, he is alone, and he looks somewhat lost and unsure compared to the other Keepers, who are all older than him, watching their assigned charges, all working in pairs carefully selected for balancing personalities and psychological traits. He keeps checking his psi-tablet, as if looking through a guidebook. He stands up, checks his watch, sits again. His eyes are tired, and his ocher Keeper’s apron is rumpled and spattered with globs of myriad yellow and brown hues.
“Hello? Mr. Gorovsky, right?”
“Um. Yes. Apollo.” He shakes my hand. His fingers are soft, his grip is rather limp. It does not match his listed profile.
“My name is Hana. I’m here to ask how you’ve been doing since your partner’s Retirement.”
He looks askance at me. Not at all subtly, he is noting the emitter pattern of my Implant. “Um, you don’t look like a Behavioralist.”
The Forever Watch Page 7
