Kill Process (2 page)

I tuck the coffee under my stump, swipe my badge, hold the door with my foot, then take my coffee in hand again. I’ve done this one-handed entry ritual since I lost my arm.

I wend my way through a mess of desks to my distant corner, a few other early risers there. By common social agreement we ignore each other until the niners arrive.

I take a seat, plug my laptop in, and leave the headphones on the desk. I want my ears free to warn me of anyone’s approach. I’ve got my back and right side to a wall, a safe corner I usurped when I moved to the Portland office.

The data queries I run this morning hit our backend database, spidering through social graphs and past behavior, feeding the custom algorithms I wrote to profile our users.

When someone uses Tomo, we log everything. What status updates they saw, whose photos they clicked, what profiles they opened, everything they posted, read, uploaded, or downloaded, as well as when these things happened and where they were when they did it. Printed out, as happens when the court subpoenas someone’s profile, a user’s account history spans thousands of pages.

I used to be Tomo’s chief database architect, employee number forty-eight. After I returned to work, I couldn’t handle the Palo Alto headquarters anymore. In some ways, my new role in the Portland office suits my needs better. I’m a data analyst, writing code that interfaces with our databases and optimizes ad placement. The job gives me the time and access I need to pore through our endless stores of user data.

Want to know what an obsessive mother looks like? She visits the Tomo profiles of her children 11.6 times a day, every day.

I spot drug dealers by their endless private messages, filled with dozens of varying nicknames to disguise the drugs their clients ask for. Even as fast as street names change, it only takes a few dozen messages to ferret out the connection between old names and new ones, and watch brand-new drugs come online. I can watch them in real-time, if I want. The suppliers, of course, don’t use Tomo. They’re rightfully too paranoid.

I even watch the watchers. An FBI research analyst views thousands of profiles of people he doesn’t know each day, and he logs in and out through different accounts. We track everything, even which accounts are used on the same computer, so we can build up inferred associations, even if two accounts lack explicit connections.

Child pervs are easy. They’re the ones viewing profiles of kids under thirteen. Tomo officially doesn’t allow kids under thirteen, of course. Still, it’s easy enough to identify them by their obsessions over music, actors, TV shows, even their word choices. There’s also the curious phenomenon of duck faces, the attempt to make sexual faces that seems to peak around twelve. I got tired of looking at them, so I wrote a duck face eliminator plugin. It detects pursed lips and replaces them with the Mona Lisa’s smile. Much more peaceful.

I know who clicks on the links about wardrobe malfunctions, and who loves their mom and who doesn’t. I can spot who’s cheating on their spouse, and who has ever thought of it. We retain those drunken chat messages forever. In short, I know everything.

If the police had unfettered access to Tomo’s data and knew what they were doing, they’d spot the criminals in an hour. Crime would drop overnight, and it would only require imprisoning half the population of the United States.

Angels we’re not, with the possible exception of Thomas.

The NSA has unlimited access to our data, although nobody knows what they do with it. The raw data isn’t of much use unless you know how to find the patterns. That’s where I shine, even though my manager doesn’t understand a tenth of what I do. He still thinks I pick ads out of a database to place on webpages, although I’ve tried to explain profiling countless times.

Of course, I’m not interested in the millions of crimes and morally bereft activities I unearth. It’s incidental, the sort of stuff anyone familiar with machine learning and a few weeks to spare could do. For the most part, I don’t give a damn. We’re all guilty of something.

No, what I’m interested in is a little harder to find, a little more specific. My own personal demon.

E
MILY PLOPS
into the booth across from me with a sigh of the world-weary.

“You are not going to believe my morning. I’m in the middle of getting Freddie ready for preschool, and the damn cat pukes in my laptop bag. It’s not enough I still had to change a diaper. I’m already running late, so I have no choice but to bring the bag full of bile with me, drop Freddie off, and when I arrive at work, run to the nine o’clock meeting. I take out my laptop to present to the executive managers. It’s covered with a layer of stinking yellow puke and matted fur, but it’s too late to do anything, so I give the presentation and pretend everything is fine.”

She emits a guttural yell of despair. I smile. Emily’s life is so ordinary, it makes me want to hug her. Maybe this is the life I could have had, if everything hadn’t gone awry. I’m not sure I would choose kids, but I’d like to live my life totally out there, brutally honest, without these half-lies, calculations, and suspicions that eat away at me like toxic waste my body can’t expel.

“Did you order yet?” she says. Without waiting for a response, she reaches out an arm to block a waitress carrying food to another table. “The salad with the chicken breast on top, no skin, no cheese, light on the dressing, and coffee, lots of coffee.”

I’m not required to talk a lot with Emily, which I like. It’s different than when we were in elementary school, when I was the fearless leader.

“How’d the presentation go?”

“Awesome, of course. A little puke wasn’t going to stop me. The executive team loved it, although one guy kept trying to edge away from my laptop.” She shakes her head and grabs a pickle from the bowl on the table. “That computer was only two months old, but it’s not like I can wash the thing. I told my boss, who’s still pissed I forgot the last one in a taxi. ‘Blah, blah, six notebooks in six months, blah, blah.’ The hell with them, I make the company several million a year, they can damn well buy me a new computer any time I want.” She says this last bit waving around her pickle like a school teacher wielding a pointer.

“So how’s your job going? When are they going to make you manager?”

Of course, I don’t want to be a manager. Even when I was high up in the food chain, my role was technical, not managerial. These days, I like my quiet corner and my unlimited database access, and the last thing I want is attention. If I could be invisible, that would be best.

This doesn’t fit Emily’s picture of the world where everyone should seize the opportunities that come their way with gusto. She wants me to once again climb the ranks of the company, but those days are over for me.

“No promotion yet,” I say. I grab onto one exciting thing to tell her. “I discovered if you rotate an ad on a page a tiny bit, so the crookedness is imperceptible at first glance, in certain cities it increases the click-through rate. In Cincinnati, a three-degree clockwise twist increases the click-through rate ten percent. A one-degree left turn is more effective in Philadelphia.”

I’m right on the border of where Emily accuses me of speaking a foreign language.

“That’s exactly why you should go for a promotion,” she says, nodding. “How many ideas do you have? How many can you test in a week? One, maybe two. If you were a manager, you could farm out your ideas to an entire team and test ten or twenty concepts in a week.”

Our food drops with a thud on the table. The waitress tears off the receipt, depositing illegible scrawl onto the table.

“Jesus, Angie, how can you eat that?”

I ordered the house burger, a colossal construction five inches across, with an accompanying mound of onion rings. One-handed burger eating isn’t pretty, and I’d never order anything like it with anyone besides Emily or Thomas. I don’t work out, but thanks to good genes and the occasional hike or bike ride, I’ve managed to maintain a reasonable physique considering that I mostly sit on my butt all day and code. Besides coloring my hair every month at the salon, I’m not big on appearances.

“It’s delicious,” I say. “How can you eat tasteless salad?”

“I eat salad so I feel light and ethereal. It gives me energy so I can kick butt.”

Emily is strange sometimes.

“Database queries can’t be written on rabbit food. I need protein.”

“That explains the onion rings, then.”

Looking at the mound of greasy batter, I resolve to eat only half of them.

“Tell me what Freddie’s doing.”

She launches into a long explanation of Freddie’s gastrointestinal system which somehow segues into his new vocabulary. Although my kidless coworkers would be bored to tears, I revel in the normalcy of it all.

*     *     *

That night I microwave a burrito when I get home. On occasion, I let myself have a night off between projects, but not often. The anxiety eats at me. On average, eleven women die each day at the hand of a husband or partner. Maybe I can save one.

Gary is the top candidate for my next target. He’s a banker, barely on Tomo at all except to post the occasional picture of his new BMW or home entertainment system. Pictures come in floods only on vacations, when we receive a long sequence of Helen. Helen at dinner, Helen in the BMW, Helen in a bikini. At Gary’s forty-fifth birthday in London, Helen’s makeup is on thick. I enlarge the photo to see if there’s a bruise underneath. There’s a slight darkening, but I can’t be sure. I look at more photos, wondering if I’m really dying a little bit each time I see one, or if it only feels that way.

All database queries return results in a particular order, such as alphabetical by name or sorted by age. When someone visits Tomo, they see a sequence of posts in reverse chronological order.

My custom queries are ordered by potential threat and need for intervention: How likely is he to be abusing her? How likely is she to help herself or obtain help from someone else? There’s only one of me, and so many millions of women.

The typical domestic violence victim will live with the abuse for two to three years before they seek help, and it usually takes five or more attempts before they finally extract themselves from the relationship. That is, if they can. Seventy-five percent of battered women who are killed by their abusers are murdered when they attempt to leave the relationship.

This last statistic drives my extreme approach. No sane person wants to go around killing other people, but I explored all the other options. Try to get the woman out, and you might indirectly kill her. Use punitive measures against the abuser, such as ruining their finances or getting them fired, and they’ll take out their anger and frustration on their victim. Expose their abuses in the hopes of getting them arrested, and they might go free, in which case the repercussions fall on the partner. If I knew of anything else that guaranteed results, I’d use it, but I don’t. The outcome of my kill process is deterministic.

And whatever excuses we might make for them, whatever their own pasts, abusers make a choice to hurt the very people they profess to love: at that point, they forfeit their right to mercy. The fact that our institutions are weak and ineffectual, that they don’t protect the victim, is something we’ve allowed to persist. I won’t.

Right now, my algorithms tell me it’s very likely Gary’s abusing her, and very unlikely she’s going to self-rescue, at least not in time, not before the danger to her life grows critical.

Helen doesn’t reply to her mother’s messages, even though they’re connected on Tomo. Helen’s Tomo activity has gone passive; she’s reading without posting. She’s clicked on links her friends shared for domestic abuse hotlines. Many times. Most of them weren’t actually posted by her friends. I inserted them in her stream, visible only to Helen.

Thanks to the Tomo app permissions on her phone, I have access to her call history. Even with all those clicks on the abuse hotlines, she’s never actually placed the call, at least not from her personal phone. It’s almost certain Gary monitors her phone usage, so if she makes the call, she’d better be ready to go.

Tomo’s search engine is predictive, like all searches these days. Before Helen ever hits enter, the browser sends up the partial string that’s been entered to see what might match, so we can suggest likely terms in a dropdown list. This data isn’t normally saved as it would overwhelm our logs, however I can turn it on for a given account as I did two weeks ago for Helen and Gary.

Maybe the most damning piece of evidence is the time Helen typed into the search “how to kill yourself.” She never hit enter, never went on to see all the help the net has to offer on that particular topic. Later the same day, she looked up her mother’s profile eighteen times and her father’s profile twelve. She composed a message to her mom six times, and cancelled each time. She couldn’t send a simple “I miss you,” even though she stared at the text for twenty-three minutes.

I rest my head on the cool metal case of my laptop, and let the sharp edge bite into my forehead. I focus on the pain and try to still the trembling in my hand.

I know what it’s like when even suicide doesn’t seem like an escape. It’s possible to become so fucked in the head that you still love the person that does such terrible things to you. These assholes use warped logic and emotions as weapons even the most intelligent person can fall victim to, leaving you believing you’re responsible for the situation. All the while, they’re killing you from the inside, bit by fractured bit.

Nobody deserves to have their love repaid like that.

Helen and Gary live in Beaverton, a suburb twenty minutes outside of Portland. This vastly simplifies my ability to take care of Gary.

*     *     *

Finally the day arrives when my preparations are complete. I wake, go to work, put in a full day, then come home. I try to take a nap and fail. Later that night I pop a caffeine pill and get ready to leave.

I leave my cellphone at home, in front of the television, set to stream Star Wars episodes 4 and 5. The mobile phone runs code to act as a remote control. It will pause and resume the movies at pseudo-random periods to simulate bathroom and snack breaks, at which point it will report accelerator data consistent with those activities. It took two weeks to patch the core OS to make it do what I wanted. It’s an alibi if I need it, but more importantly, it keeps my data profile normalized, so I don’t ever pop up on anyone’s radar in the first place.