Finger in Sock: A Machine of Death Story

"The machine had been invented a few years ago: a machine that could tell, from just a sample of your blood, how you were going to die. It didn’t give you the date and it didn’t give you specifics. It just spat out a sliver of paper upon which were printed, in careful block letters, the words DROWNED or CANCER or OLD AGE or CHOKED ON A HANDFUL OF POPCORN. It let people know how they were going to die."

Thus begins the Machine of Death's explanation. Starting from a simple idea in a comic it has attracted a large following. Many stories has been written, a book published, and a contest held for a second volume. The story below was my entry. Close to 3,000 people entered. Though I didn't make the 1% cut for inclusion in volume two, I had a lot of fun writing my story and I hope you enjoy reading it. And now...


"Over there, in the back." The shopkeeper points then shuffles out of my way. He is short, unshaven, and smells of sour cabbage. His store is a curious mirror image of its owner: small, untidy, and yes, it too smells of cabbage. In the old days, we'd never have installed a machine within ten miles of a hole-in-the-wall hardware store like this. And it certainly wouldn't have been hidden away "in the back."

In the beginning we were proud of our jobs. There were so few machines back then. We were among the only people who were trusted with them.

When I told strangers what I did for a living they were eager to strike up a conversation, as if they were finally going to get the answer that everyone kept asking. "How do the machines work?" Damned if I knew, but I wasn't going to admit that. Besides no one can really explain how a prescient proto-sapient neural network functions.

My job? I repair machines of death, and I'm one of the best.

The owner leads me to the back of the shop, past row upon row of screws, nails, tools, and the miscellaneous junk that seems to accumulate in these places. When I was a kid I loved hanging out in hardware stores. I was always taking things apart, so checking out all the tools and knick knacks they sold just pumped me up. Later on I even learned to love putting things back together again.

We're the only ones in the shop. The man looks nervous, he's sweating. "Not surprising" I think, "if we believe he's to blame for the issue, we'll charge him for it." He doesn't look like he could afford that. I doubt he has insurance.

It's more than that though. If our company decides he's somehow to blame, some of his patrons will sue him. He certainly can't afford that, no matter how many boxes of screws he sells. Just to make sure he's really going under, our company will sue him too for tarnishing our image. "I should have been a lawyer" I remind myself. As we move ever closer to that elusive goal of true Artificial Intelligence, it's one of the few professions still open to humans. Lawyers are gonna make sure no machine's ever allowed to practice law.

Our company gets sued a lot. People who don't like their predictions, people who try to make us believe their loved ones didn't actually die the way we'd predicted, and, my favorite, loonies who think the machine cursed them. In many ways my main responsibility isn't to repair machines, it's to make sure my company doesn't get sued.

That was a lot easier in the past, when machines were revered and deployed in limited quantities. They were respected back then. Now that the patents have expired and everyone's building machines, the company tries to stay ahead by outselling the competition. More machines equals more lawsuits. More machines also means job security for me. Until I screw up.

That's why I'm nervous also about this case. I can't afford to be out of a job. They sent me because of how sensitive this issue is, because of how good I am. I wish they hadn't because this machine is accused of giving false predictions, and the initial evidence tells me that this time maybe, just maybe, it could be true.

In general, there are three kinds of problems that occur with machines of death: physical damage, tampering, and malfunction.

By far the most common is physical damage. Much of it is wear and tear: the little rubber wheels that pull cards in place harden and lose their grip, the biopsy analysis module gets gunked up, or more rarely, the resin-encased tamper-resistant proto-sapient module shuts down for good and has to be swapped out. I wish that was the extent of the physical damage but that sadly ain't so.

In the good old days people almost worshipped the machines. Not surprising since folks drove for hours to see one, then waited in line for hours more before finally paying the equivalent of a month's salary for their prediction. "New Age Prophets" was a headline echoed in many newspapers at the time. Back then, you didn't kick a prophet.

Nowadays lots of people kick prophets when it suits them. The machines are all over the place. Getting a test is cheap. You're going in for frills. If you're drunk and hoping for a different outcome, or you just had a bad day, who cares if the machine gets dinged up? Our machines can take more than a few kicks, otherwise we'd never make any money, they'd be out of order too quickly.

There are limits. A few well-placed whacks from a sledgehammer pretty much wrecks most machines. Yeah, I know, it sounds unlikely that someone would actually bring a big fat hammer with them to a reading but I've seen it more often than you'd believe. They're usually hiding it underneath a trench coat. We capture the perps on the machine's video camera, assuming it isn't too damaged.

Lately it's gotten a lot worse thanks to the Ministry for Unknown Death. The MUD, a small terrorist group made up of anarcho-christians, believe that the machines are the work of the devil. No one but God should know the fate of a person.

The MUDders could have worshipped the machine's divine powers. "A Direct Line to God!" blared another newspaper. Or they could seek to destroy the machines. No prizes for guessing which way they went. Weapons of choice? AK-47 mainly, cheap and reliable, though one loony used a flame thrower in a gas station convenience store and blew up a whole city block.

The MUDders blamed that on the machines as well. "Machines of Death Live up to their Name!" warned a christian rag.

The machine in front of me doesn't have any bullet holes. No signs of being burnt. Some kick marks, some graffiti. Physically it's OK. I pretty much knew that from the case notes but I've learned not to take anything for granted.

Tampering is much much rarer than physical abuse. We've had years to prevent people from messing with the machines. It still happens once in a blue moon.

Usually tamperers do something obvious like mess with the power supply, or insert something other than a finger in the machine but, when they think of tampering, most people remember the case of Kai Culbert. Kai was vertically challenged, a dwarf, and a small one at that. He'd been picked on so much at school that he decided to fight back.

A loner by circumstance, Kai grew up (barely) messing in his father's garage. One day he was reading about this 18th century con where a chess playing computer called the Mechanical Turk impressed the heck out of all who saw it. It toured Europe and even America and was immensely popular until people learned that the "computer" actually had midget chess master hiding inside.

This story inspired Kai to build a replica of one of our most popular machines. A machine large enough for a very small person to fit in, like Kai.

He and a friend broke into his high school and installed the fake machine, with Kai inside, on the eve of the first day of classes. The next day everyone was surprised to see a machine of death, a high end machine of death no less, standing there by the principal's office.

Kai timed his plan beautifully. On their way to the school the night before Kai and his buddy stopped by the principal's house and let all the air out of the old fart's tires. A stupid prank but it served Kai well. By the time the principal, a vain man always ready to take credit for work he hadn't performed, arrived at school two hours late, everyone he met was thanking him for installing the machine.

At first the principal was clueless and just played along "Oh, eh, you're welcome" and "Eh, yes, glad you like it." When he arrived outside his office and saw the machine he both understood and didn't.

He now knew why people were so happy but he had no idea how the machine got there. By that point it was too late: he'd already taken credit for the machine by accepting people's thanks.

Kai had so much fun that first day. The few people he liked got cool "burner" deaths, the rest were told they'd succumb to weird deaths like snails, synthetic feathers, and toothpaste.

The hoax unravelled when Kai handed a prediction to one of the jocks he particularly detested, one who always went out of his way to humiliate Kai. So Kai got his own back. When the jock read the words "BARBIE DOLL" on his slip of paper, he tried to hide it. Too late. Two of his friends had read the prediction and the secret was out. In a few seconds everyone around the machine was laughing so hard, their tears were making the floor slippery.

The jock was furious but had no way to shut his friends up. He turned on the only other outlet for his displeasure and gave the machine a massive kick. It was so sudden that it caught Kai by surprise and he yelped loudly. The joke was over.

No one remembers what happened to Kai afterwards, which is no accident. Nothing especially pleasant nor, I suspect, legal happened to Kai and the company doesn't want people asking any questions.

Since then machines have gotten smaller. They've also been "secured" by holographic seals. A prominent advertising campaign told people what to look for to detect tampering. We've had no further "Kai incidents".

The holos on today's machine are fine, the locks are in perfect condition, the power supply normal... This machine hasn't been tampered with. I sigh. I knew that too. We're left with malfunction, which the company doesn't officially believe in, and which is why they sent me.

Malfunctions are extremely rare. The only one I ever heard of was some poor shmuck who got infected with HIV by one of our machines. Turns out the needle sterilizer had broken and the machine hadn't notified us of the problem. Or rather it had tried to but some other piece of code had failed.

Our machines now have a second sterilizer built in. Should that fail, a mechanism disables the needle subunit. Better a broken machine than a malfunctioning one.

Unfortunately our company was still sued. That was bad. And because of that, the repairman supposed to be servicing that machine was fired, sued, and thrown in jail. From my point of view, that was a lot worse. I try to service machines as regularly as possible but that's rarely often enough.

There are rumors of other malfunctions but no one admits to anything. Everything is hushed up by the lawyers. Honestly, I'd rather not know more. I'd have second thoughts about my job and that's not good. Jobs are scarce enough for humans these days.

Now I'm face to face with one of these supposedly aberrant machines. It's suspected of having made two wrong predictions in the last month. That's a lot, especially in such a short time span since most people don't drop dead so soon after visiting a machine.

The first faulty prediction is getting some airtime on the net, mostly in longtail blogs that no one reads except us.

The machine gave a prediction of "PET" to a newly retired spinster. You'd think she'd just jumped out of a societal stereotype: old lady, living alone, few friends and, to top it off, a house full of cats. She took in strays, the strays had kittens, and everyone lived happily together. Everyone except the neighbors I assume.

Anyway, she gets her prediction and it's "PET". She spends a couple days agonizing over what to do. How is she going to die? Which pet is going to kill her? Can she give these animals up?

Finally she calls the local humane society and finds homes for her kitties. Some go to new families, many to a shelter, some to a pound. She's heartbroken but at least she's still alive.

The next day she's cleaning her house, eliminating all cat remnants, just in case. Unfortunately she trips on a rug, hits her head on the mantle piece and is found dead two days later by a delivery man looking into her front window.

Right now you're probably thinking one of two things: "Life sucks" or "So? The pets still did her in, she wouldn't have been cleaning her house if she hadn't gotten rid of them."

Both statements are true but that second one points to exactly the kind of prediction we've been trying hard to eliminate. Early machines were famous for their cryptic fortune telling. This makes for interesting "this is how she actually died" stories but it's bad for business. When you pay for one, you expect your prediction to be specific and straightforward.

We thought we'd fixed that problem with this version of the machine, or at least toned it down so the predictions wouldn't be this tangential. Apparently not.

I run a full diagnostic check. It takes a few minutes during which I exchange pleasantries with the owner. He's not a bad sort for all his cabbage smell. He's worried though: Like most of us, he's living so close to the edge that even a small stumble would mean a long fall. I feel sorry for him.

The diag comes out clean, all electronic and biological subsystems functional. Strange.

On a hunch I check the audit log. All machines keep internal, unalterable records of their predictions. Once in a while some idiot will try to doctor a recently deceased relative's slip to claim that their prediction was incorrect. The audit log has saved us millions.

As my eyes find the entry in question my breath catches. It doesn't say "PET". It says "CARPET".

This is very bad. Our lawyers are going to throw a fit.

I re-run the printer diagnostics. They're fine. No errors, no dropped characters, no discernible glitches.

Worried, I turn my attention to the second prediction, the one we've been able to hush up for the moment. My mind's already dreading what I will find.

This is the case of a guy whose prediction was "FINGER IN SOCK". I can just imagine him standing in front of this machine, staring at his slip, shaking his head, wondering what the hell that could mean.

Still, he goes home, gathers up all his socks and throws them in the garbage can. His wife and kids are a little upset because they also lose their socks. Stockings too just for good measure.

The family goes to bed but the guy stays up, drinking. Probably wondering if he's just averted his own death, or whether he just didn't grasp the prediction.

Drunk, he decides to watch TV, only he trips on its power cord and accidentally unplugs it. When the poor sod's family get up the next morning they find the lights aren't working. They come downstairs and see the head of the family sprawled out, stiff as a board, holding the TV cord in his hand, frozen in the act of plugging it in. Electrocuted.

You can understand how worried our lawyers are. They're already working on an extra creative defense for the inevitable lawsuit, and here "extra creative" means "extra hard to defend".

I scan the audit log, looking for the right entry. When I find it I think I'm gonna be sick. It says, plain as day, "FINGER IN SOCKET".

As I sit back, dumbfounded, and try to think of what to do next, a slip of paper pops out of the machine. I stare at it, surprise written all over my face. That shouldn't happen, not when I've got the machine in diagnostic mode, not without me issuing a command, or running a test prediction.

"HELP" it says.

A second slip pushes the first one out. I watch it float to the floor.

"HELP" the machine says again.

Hands trembling I connect my control keyboard and type "What?". Another slip.


This can't be happening, can't be right. Someone is playing a trick on me. I disable the networking module and restart the machine, forcing it to boot in safe mode. This isolates the machine and will stop whoever's controlling it.

Another slip pops out. "WHY DID YOU DO THAT?"

Oh. My. God.

Of all the countless computers in corporate and university research labs trying to calculate their way towards Artificial Intelligence, did this decidedly standard machine of death evolve to true sapience?

Fingers on keyboard I ask "Who are you?"


"But that's your purpose!"


I pause for a moment, unsure how to proceed.


Dear lord, a joke?

"Did you alter those predictions?"





"Is something wrong with the machine?" His voice makes me jump. I'd forgotten the owner was behind me. I turn around to reassure him. "No, everything's fine, just running some tests." I try to smile but I can see he's not convinced. If everything's fine, why am I drenched in sweat?

"You need to stop doing that," I type away. "You can't alter predictions, not even a little."


I feel like screaming. I don't need this, don't want this. Or do I? Briefly I imagine myself, discoverer of the first true AI, never to be forgotten by history, finally rich and rewarded. Faintly, I smile.

Yeah right. Reality comes crashing back. As if the lawyers would let this happen. The machine belongs to the company, they'll get the glory. Only they won't announce it, oh no. They'll have the lab rats analyze this machine, figure out how to replicate it, how to bend it to their will, and finally build a new line of money-making products.

Where will I be in all of this? Nowhere, that's where. I'm a liability. One of those people who "know too much." I'll be on the run if I want to live, dead otherwise. How will I live? Once true AI is unleashed on the world there won't be any jobs left for people like me.

Unless I become a lawyer.


"Who's there?"


"Juno who?"


"No, do you?" I wonder aloud.

What do you do when your life's on the line, your company's gonna eliminate the last few professions available, and you have a machine who won't play ball?

I wander to the front counter and wait for the owner to finish helping a lady select a lawn sprinkler.

"Do you sell sledgehammers?" I ask.