No Harm Can Come to a Good Man

by James Smythe

  ‘It’s fine,’ Amit says. He takes his number, in case he needs picking up, and he watches the cab leave. He turns and walks up the path, and then he’s walking on the glass. It’s immaculate, and the sun’s reflection bounces off it, so bright that he can barely stand to look.

  It’s hard to see the door at first, until Amit spots a handle moulded from the same glass as the rest of the building. It’s unlocked and he steps inside. It’s freezing cold in the building. There is a desk here, and then staircases rising from either side, entwined, a helix. Above the lobby, a sign hangs. ClearVista: the numbers don’t lie. Underneath it, a giant video screen showing the ClearVista advertorial. Here is the woman getting the job she desired. Here’s the couple with their new baby. Everything here is so simple, so easy to explain away. ClearVista leads to happiness and contentment, and the answer that you want. They’re selling a vision of a future that you want to will into being. They can add Homme’s video to this, Amit thinks, and barely skip a beat. Here’s a man who is under-qualified, somehow proving himself by default; and then here’s Laurence, buried in the annals of history. Hoist on his own petard.

  There is nobody at the desk, but there is a button and a screen. He presses it. The screen clicks to life, with a face. A poorly animated face, like the ones in their promotional videos, but this one a somehow beautiful but generic woman. She is an amalgam and he finds it impossible to place her ethnicity. She reminds him of so many actresses, but isn’t like any of them, not when you focus.

  ‘Please state your name,’ she says. Her voice is clipped. He recognizes it; the voice from the telephone service.

  ‘Amit Suri,’ he says.

  ‘Please hold,’ she says. She closes her eyes. She looks peaceful, Amit thinks. It’s a trick; an exaggerated visual representation that she is doing something. She opens her eyes again. ‘You’re here to see Mr Hershel,’ she says. His name is the only part of the sentence that sounds perfect, unclipped.

  ‘Yes,’ Amit says.

  ‘Please hold. I will send for him.’ Amit looks around, but there are no chairs. The face folds itself away, disappearing to a fixed point on the screen, and the screen goes blank. Amit walks around, looking behind the desk. There’s no movement, so he waits, and he looks through the glass, and up the stairs. The building goes on and on, but he can’t see any rooms, or doors, or people. There’s only the building itself.

  He waits and waits, and there’s no sign, so he presses the button again. The face comes back and asks him for his name. He says it again, and again it rolls its eyes back in false thought.

  ‘I will tell Mr Hershel of the urgency of your enquiry. Please hold.’ Amit shakes his head and walks back. In the distance he hears the footsteps of somebody approaching. They echo throughout the building, because there is no other noise. He looks up the stairs, craning his neck to see expensive black shoes, tan slacks, and a man in a blue jacket coming down the stairs. Amit knows him; he saw him in the airport, way back. After Laurence’s first incident, when they first received the report. Laurence said that he kept seeing him and Amit didn’t believe him.

  ‘He’s waiting upstairs for you,’ the man says. He seems resigned to something; that his being here isn’t his choice.

  ‘Who are you?’ Amit asks.

  ‘I told you before: I’m really nobody,’ the man says. ‘He’s waiting.’ He stands to one side of the staircase, to let Amit through. Amit walks past him, and up the glass staircase. The first floor stretches back, and finally there are rooms here, brushed-silver walls and darkened glass doors. He opens them, one by one, but there is nothing inside them. No furniture, no staff. There’s no sign that there has ever been anything else here. Up the next flight of stairs, and then the stairs end. The building stretches up, but there are no more floors. There are more rooms here, but Amit knows where he’s going. At the far end is a boxed-off room, larger than the others, and he can see a wall of computers lining the back wall. Lying on the floor in front of them is Hershel. He has his eyes open, and he stares at Amit.

  ‘I’m sorry,’ he says, before Amit can hear it. But he can see his lips form the words.

  Hershel looks like shit, Amit thinks. He’s a wreck. Protein bar wrappers are tucked in the room’s solitary trashcan, and empty tins of soda. He pushes himself to his feet as Amit opens the room’s solitary door. It’s freezing inside. Amit sees little jets of air pumping out by the servers, the banks of blinking lights and whirring hard drives kept artificially cooled. Hershel stands in front of them, unsteady on his feet.

  ‘Look,’ he says. There’s a crack in his voice. ‘This wasn’t meant to happen.’

  ‘You did this?’

  ‘No! The algorithm did.’

  ‘You made it, Hershel.’

  ‘Amit,’ Hershel says, ‘you don’t understand. I’m serious, here. This isn’t something I intended. You think I knew?’

  ‘I think you’ve conned people before. I think you’ve blackmailed and cheated, and I think you’ve found another way to do it.’

  ‘No. I didn’t ask you for money, did I? I didn’t … I didn’t blackmail you. I could have done. ClearVista could have done.’

  ‘And how much press has ClearVista had because of this? You released this shit into the world. It’s a pox, and you unleashed it!’

  ‘I didn’t release anything.’ He puts his hand on the servers. ‘They’re not doing anything now, you know. It runs the equation and then it sleeps. It answers the phone and diverts the calls, and then it sleeps. It runs this entire business, and then it sleeps. You think it’s insidious, but it’s not. It’s not an AI. It doesn’t know. It just does what we told it to do before, and it’s carried on doing it. It sees equations, and it fills them.’

  ‘You told it to show Laurence like it did.’

  ‘No! Jesus no.’

  ‘So why did it? Why did it think he couldn’t be President? Why did it make that video?’ Hershel looks away, staring through the glass of the inside wall, and the outside. The view is warped.

  ‘Okay,’ Hershel says. ‘I’ll explain.’

  ‘This all began about a decade ago. The stock markets have been using algorithms for years and years now, to predict and control their numbers. It’s a science, not an art. But sometimes things go wrong. Back at the start of the last decade, the New York stock exchange’s algorithms just went haywire. Died. Over a period of thirty minutes, the markets suffered the largest crash in their history. Nobody could explain it – and everybody panicked. It was their algorithms. The software found a glitch and it compensated. Fucked the markets entirely. Know how they fixed it? They switched it off and then switched it on again.

  ‘The algorithm that I helped build isn’t all that different to what they use. It’s mostly data mining and compositing information. It’s taking sources and trying to make them work in conjunction. I was here from the start. They hired me – you remember that – after I dropped out: because I dropped out, probably. Paid me to do this, and they built a team from the ground up. I did the stuff where it drags data from the rest of the world and makes it work for the algorithm. So, let’s say you come to ClearVista and you ask it about … fuck, I don’t know. Whether you’ll get a loan from a bank. It takes everything. It takes who you are, but it takes everything else. Loan rates, who applies for loads, what job you do, your credit rating, whether people with your credit rating get loans, whether they pay them back. All that stuff. I built the software that did that and then, when I was done, they offered me another role. Supervising the algorithm’s day-to-day behavior. Watching it. Here’s the thing; it grows. Not like it’s intelligent, nothing like that, but it learns. It’s designed to, and it can’t help it, because the Internet tells it everything it needs to know. But sometimes shit goes wrong, and they needed somebody to make sure it didn’t. So I took the job, and they gave me shares, and they paid me a wage. I could monitor the thing from home – it’s an algorithm, doesn’t need any managing or anything like that – so I did. Seemed
easier.

  ‘Then I thought about using it myself. I like gambling, dude, and the market was going to fall out of it. Look at it now! Soon as this became commercial, I knew there wasn’t going to be any gambling on anything. Nothing would be a risk any more. So when it was just me, I made a few bets on things – sports, mostly, the Superbowl and some soccer games – and I started winning. Not every time. It’s not infallible. But it picked up a lot, and I was teaching the algorithm more and more. Really getting to those bits of the Internet – of data – that it wasn’t reaching before. It used everything: the type of grass on the fucking pitch, the wind-speed, the health of the players. I couldn’t do big numbers, because I was being watched. I’m not allowed to gamble, but I am allowed to work for ClearVista. And that’s a joke in itself.

  ‘The algorithm didn’t work every time. It works based on things that it can determine, but there are margins. Human error. Bet on the Kentucky Derby? Can’t do that. Horses fall, and you can’t call it. Laurence … He’s like a horse. He’s a loose variable. I didn’t have anything to do with Laurence’s prediction. You can’t alter it or trick it. Do you know how many people work here now? Me. Just me. Everybody else is a shareholder or silent partner. No staff, because the algorithm runs the phones. Plugins do the rest. Customer service? All that is automated and the servers self-maintain. The orders come in; the money follows; the numbers do the rest. That’s how your video was generated. I saw it, and everything’s automated after that. I didn’t want it to get crazy; but it was new, you know, something worth checking out.

  ‘So, I hired somebody to watch Laurence – the guy you met downstairs. He works for ClearVista. Used to be security. Now … oh, I don’t know. I wanted to make sure that I knew how Laurence reacted. Because I had a theory. Let’s say you want to use ClearVista before you go to a bank for a loan, so you ask the algorithm if you’re likely to get one. The people who apply for the loan; what if they were in a social group likely to apply for a loan? And what if that social group was also likely to ask ClearVista for a result before they went through the paperwork? If they’re unsure enough to ask ClearVista, the algorithm will know that. Do you see what I mean? It takes itself into account. Homme, he was the first potential nominee to ask the questions that he did. He got there first, Amit, and he asked what would happen, and ClearVista did what it could. As soon as somebody else asked if they would be president, the algorithm adapted. It worked out that he would ask that question, and it knew that it existed. It knew how this would work. If Homme’s odds were 62%, Laurence was going to lose. The algorithm works in absolutes. If Laurence lost, he was going to have a breakdown, because everything said that he would: his medical results, his son’s death, and the fact that he was in therapy. Don’t ask me how, but the algorithm gets results we can’t. Private records. It knows everything. Everything. So, the video was going to show Laurence having a breakdown. That’s his end result, and the first video it booted out, right? But then you asked for a second video, after the plane; after he’d seen the first video. It knew he had watched the first, because this was a rework, and so the breakdown – the only inevitable part of this, if he lost the race, which he was now going to do, absolutely, no fucking question – was going to be worse. It drew on everything. Gun licenses. It knew that would come into play, because he was a soldier, and because he owned one. It made its own myth, Amit. It’s self-aware, but not in the way that people think of computers as being. It can’t think on its own, but it knows that it exists; so the algorithm cannot help but reflect itself now in everything that it does.

  ‘And the worst thing? It released the video itself. The algorithm sent it to the news, because they were the ones who hated Laurence; and Laurence’s breakdown was inevitable. It wrote itself into the prediction, and then all it could do was fulfil it. That’s logic; that’s the algorithm we built.

  ‘This ends with a gunshot, Amit. It can’t end any other way, now.’

  Amit rubs at his face. ‘How sure are you?’ he asks.

  ‘Positive. I mean, there’s math to back it up. I could show you the breakdowns of where the decisions were made.’

  ‘It’s not been tampered with?’

  ‘No. It’s impossible to fuck with this.’ Hershel stands up and paces. ‘I keep thinking about whether I should feel guilty for this. Because I helped make this, you know? And it’s the reason that this is happening. If the algorithm didn’t work … I don’t know. Maybe I’m wrong. But this gets worse. You didn’t show Laurence the video I sent you, did you?’

  ‘No,’ Amit says. ‘I don’t know how he would react.’

  ‘Yes you do,’ Hershel says. ‘I lied to the system. I asked for another go, and another. I’ve got videos that take this further and further.’ He turns and brings them up on the screen; so many windows, and all of them showing Laurence. Amit glances at the frozen frames, and then looks away. They’re worse than anything; in some of them, things that he would never wish to see. ‘I could be wrong about all of this, but there’s one thing that I can’t change; Laurence was always going to have a breakdown. He was never going to be president.’ He minimizes the windows. ‘I could be wrong, of course.’ He smiles slightly, because he knows that he isn’t.

  ‘I have to go,’ Amit says. He opens the door. ‘You got a car?’

  ‘Outside, but I’ll be—’

  ‘The least you can do is give me the keys to your fucking car, Hershel!’

  Hershel nods and pulls them from his pocket.

  ‘I’m sorry,’ he says. ‘Tell the Walkers that I’m sorry as well, will you?’ Amit doesn’t stop. He keeps going, through the glass building, down the glass stairs, past the man in the blue jacket and out of the glass doors, into the thick, grimy heat.

  He syncs his phone with the car and calls Jessie, and he explains to her about the algorithm. ‘There’s nothing to clear Laurence’s name,’ he says, ‘no conspiracy, nothing underhand. Laurence has a got a gun, and he’s dangerous.’

  ‘Then we go on the air with the story as it is. A misunderstanding. Laurence Walker, a man on the edge. He can come on and explain.’ She pauses. ‘People love a comeback.’

  ‘There’s no comeback from this,’ he says. He tries to call Deanna, but there’s still no answer. There’s nothing he can do. He pounds the steering wheel with his hands, leaving welts across his palms. The sun is setting, night coming. He sees the horizon as he approaches the freeway, and then there it is, in the distance; drooping as it folds behind the mountains.

  The next flight to Syracuse is full, Amit discovers, and the one after that. Next closest would be a flight to JFK, leaving just before midnight, getting him there sometime in the middle of the night. It would be faster to do that and drive than wait for the morning Syracuse planes, the assistant tells him; so he uses his air miles for a ticket on that flight, and he walks through the airport, back and forth. The shops shut, and the cleaners come out, and the only people waiting are those who he’s booked onto a flight with. They sit in the departure lounge, some of them on their computers, some reading, some sleeping; and then they’re called to the plane and they queue slowly. Amit is sitting by himself in a row and once they’re in the air he puts the arms up and lies flat across the seats; he pulls the complimentary blanket over himself and tries to sleep. The plane rumbles beneath him, and every so often a beep comes from somewhere else in the cabin and it makes him sit up, thinking that they are about to land; but then he looks out of the window and sees the blackness below, occasionally the threaded veins of lights that mean there are cities and towns and roads; and then they all drop away and he doesn’t know if it’s because there is nothing there, or just a cloud beneath him, obscuring his vision. He lies back and tries to sleep, but then he pictures the videos in his mind; and then there’s that beep in the cabin, and it keeps happening to him, over and over, scaring him that something is about to happen.

  16

  Laurence stands at the barbecue on the decking and he sprays fuel onto the c
oals. He lets it splash, because he loves the showmanship of this: the only time that he can cook and it be a display. He wears an apron that he found in the basement, left here in a box of things by the previous owners – it’s black and dusty, water damage up the side of it, and was likely used for manual labor rather than cooking, by the looks of it – but it’s all the same when covered in the splashes of the fat from the burgers. He steps back.

  ‘Nearly got me,’ he says. Deanna, Lane and Alyx wait inside, setting the table, putting salad into a bowl.

  ‘I want to go home,’ Alyx says, and she starts crying, so Deanna holds her and tells her that they will, soon enough.

  ‘Don’t lie to her,’ Lane says.

  ‘I’m not,’ Deanna replies. They all sit and wait for the food, and then they eat, but it’s overcooked, charred so much as to almost be inedible; but Laurence eats it all. His, and then Lane’s, when she pushes it to one side in favor of the leaves and chopped tomatoes. When he’s finished he talks to them all, looking at them one by one.