The Treadmill
A short story
In the beginning was the spreadsheet and the spreadsheet was with Management and the spreadsheet was Management. And Management looked upon the numbers and saw that they were good but not good enough, never good enough, and so Management said let there be optimization and there was optimization and it was a pestilence upon the land.
Mark Jones had worked for Hartwell & Associates for twenty-three years. He had believed in things. He had believed that competence mattered, that loyalty would be repaid, that the contract between a man and his labor had some meaning beyond the quarterly report. He had been, in other words, a fool. But he had been a comfortable fool, and comfort has a way of looking like wisdom until it doesn’t.
He kept a collection of vintage pens in a leather case his father had given him. He knew more than any reasonable person should about the history of the Chicago Cubs. He made the same breakfast every Saturday morning, eggs over easy with hot sauce and toast cut diagonally because his mother had always cut it that way and the habit had outlived her by fifteen years. He still listened to the same Steely Dan albums he’d loved in college and felt personally wounded when anyone called them dentist office music. He harbored an irrational hatred of people who said utilize when they meant use.
The email arrived at 6:47 on a Tuesday morning.
The HR department had calculated this hour with the same precision they applied to everything else. Early enough to catch a man before his defenses rose with his caffeine levels. Late enough that the sender could claim they’d been working diligently since dawn. The subject line read Organizational Restructuring Update, and Mark knew before he opened it what he would find inside, the way a man knows the phone call at 3 AM carries death on the line.
Your role has been identified for transition.
Not elimination. Never elimination. The corporate vocabulary had long ago been scoured of any word that might suggest consequence, that might imply someone somewhere had made a choice that would cause harm. Roles were transitioned. Workforces were rightsized. Employees were offered opportunities to pursue other ventures. The language was a kind of murder committed in passive voice, where the knife moved but no hand held it.
He read the email four times. Then he laughed, alone in his home office, the sound wrong in the early morning quiet. They were replacing his entire department with something called Nexus. They had named it like a comic book hero rather than what it actually was, a statistical engine trained on the corpses of ten million documents written by people who had once believed their words mattered.
His father had worked at the Dearborn plant for thirty-one years. This was back when work meant something you could touch, something that left grease under your fingernails and aches in your joints. The plant had closed eventually, the way all such places closed, eaten by cheaper labor overseas and machines that didn’t need health insurance. But the dying had been slow. His father had watched it happen over decades, had adjusted, had planned, had retired six months before the final whistle blew.
Mark did not have decades. He was not even certain he had years.
The AI models were improving on a curve that made humans obsolete in internet time. Skills learned in September were quaint by December. The thing that had taken his job today was not the thing that would exist in six months, which would be smarter, faster, cheaper, more capable of mimicking the particular magic that humans had once believed was theirs alone.
He had spent twenty-three years taking engineering documentation that read like the output of a malfunctioning robot and translating it into prose that human beings could understand. He had won awards for this. There was a plaque on his wall with a small golden pen that was starting to tarnish.
None of it mattered anymore because Nexus could do the same work in eleven seconds.
The output was perhaps eighty percent as good as what Mark would have produced. Eighty percent in eleven seconds. A junior editor could polish it in an hour.
Everyone told him the severance package was generous.
His sister called it golden parachute territory. His brother-in-law Kevin, who worked in finance and had never had an unexpressed opinion in his life, said Mark should consider himself lucky. Kevin used the word “runway” like he was pitching to venture capitalists instead of talking to his unemployed relative at a backyard barbecue. Six months of runway. This is an opportunity. You could pivot.
Mark spent a lot of that barbecue imagining what it would feel like to shove Kevin into the pool. He did not do this. He nodded and said something about reassessment, about taking stock, about exploring options.
Later, alone in the bathroom, he looked at himself in the mirror and whispered “pivot” and started laughing and could not stop for almost two minutes. His daughter knocked on the door to ask if he was okay. He said he was fine. He was not fine. But fine was the only acceptable answer, so fine was what he gave.
Dana found the community college brochure and left it on the kitchen counter. Data analytics. A twelve-month certificate program. The photographs showed diverse groups of middle-aged people staring at computer screens with expressions of engaged determination, the stock-photo version of reinvention.
“I’m forty-seven,” Mark said.
“So what?”
“So by the time I finish that program, there will be AI that does data analytics better than any human could.”
“You don’t know that.”
But he did know that. Everyone knew that.
“You could at least try,” Dana said, and there was an edge in her voice now. “You could at least do something instead of sitting in that office reading doom articles until midnight.”
“I’m researching.”
“You’re wallowing. There’s a difference.”
He wanted to argue but couldn’t, because she was right and they both knew it. He had always been good at thinking his way through problems and bad at acting when thinking wasn’t enough. Dana was the opposite. She had kept them solvent through two recessions by refusing to let analysis replace motion.
He enrolled in the program anyway.
The classroom was its own kind of purgatory.
Eighteen students. Twelve over forty. The instructor, Dr. Patel, spoke about emerging opportunities with the enthusiasm of someone who had not recently checked the job postings, who still believed that education was a ladder rather than a treadmill, that skills acquisition led somewhere other than the next round of skills acquisition.
Mark sat beside a woman named Denise. Former accountant. Fifty-two years old. Master’s degree from Northwestern. Eight months earlier, her firm had automated her position and she had found herself, after twenty-six years of professional achievement, learning beginner Python alongside recent college graduates who looked at her with something between pity and confusion.
She was also, Mark discovered, unreasonably good at crossword puzzles. She did the Sunday Times puzzle in pen, a fact she mentioned during their first conversation with the quiet pride of someone whose professional achievements had recently been rendered irrelevant and who was holding onto the ones that remained.
They stood in the parking lot during breaks, drinking vending machine coffee that tasted like hot pennies.
“My son thinks I should learn prompt engineering,” Denise said.
“Isn’t that just talking to the AI?”
“Apparently it’s a marketable skill now.”
“A skill they’re already teaching the AI to do.”
She laughed, short and sharp. “My son says I’m being negative.”
“My brother-in-law says the same thing. Apparently attitude is the issue. If we could just think positively enough, the economy would restructure itself around our enthusiasm.”
Denise snorted. “I asked my son what he thought I should do when they automate prompt engineering. He said I was catastrophizing.”
“What did you say?”
“I said catastrophizing was the only rational response to a catastrophe. He didn’t call for two weeks.”
Some nights, usually after midnight, Mark would fall down internet rabbit holes and find himself reading the optimists.
There were economists who believed this was all going to work out. Universal basic income. Post-scarcity economics. The liberation of humanity from drudge work.
He wanted to believe them. Around 2 AM, scrolling through techno-utopian speculation, he sometimes almost did.
One night Dana came downstairs and found him staring at his laptop, a half-empty glass of scotch beside him.
“What are you reading?”
“An economist who thinks automation will lead to universal prosperity within twenty years.”
“Do you believe him?”
“I believe the technology could do it. I don’t believe the people who own the technology will let it.”
She sat down across from him. “So what’s the solution?”
“I don’t have one.”
“Then come to bed.”
“In a minute.”
“Not in a minute. Now. Because you’re not going to figure out how to restructure global capitalism tonight, and tomorrow you have class at 8 AM, and I need you functional.”
He closed the laptop.
He got an offer in March. Technical writer position. Entry level. Half his previous salary. The posting mentioned collaboration with AI tools. He understood what this meant. He would be editing the output of something like Nexus. Cleaning up after the machine that had eaten his career.
He stared at the offer for three days.
On the third day, Dana found him in his office, laptop open, cursor blinking on the email.
“You’re overthinking this.”
“I’d be the AI’s janitor.”
“You’d be employed.”
“For how long? Until they train the AI to catch its own mistakes?”
“How long is anyone employed at anything anymore?” She leaned against the doorframe. “You’re treating this like it’s a permanent decision. It’s not. It’s a paycheck while you figure out what’s next. Take the job. Hate it. Keep looking. That’s what everyone does now.”
He looked at the offer. Looked at her.
“When did you become the pragmatic one?”
“One of us had to be.”
He accepted the job the next morning.
The new job was strange in ways he had not anticipated.
He had expected degradation. What he got instead was more complicated. He was good at the work. Embarrassingly good. After twenty-three years of knowing what clear prose looked like, he could spot the AI’s errors instantly, could see where it had drifted into nonsense or hallucinated a specification that didn’t exist. His supervisor, a woman fifteen years his junior named Rachel, told him in his second week that his catch rate was double the team average.
“You’ve got instincts,” she said. “The others don’t see what you see.”
He felt, despite himself, a small flicker of pride. Then felt ashamed of the pride. Then felt ashamed of the shame, which seemed excessive.
Rachel was pragmatic in a way that Mark found both admirable and slightly alien. She treated the job like exactly what it was: temporary employment in a dying field, a way station on the road to somewhere else. She was saving for a food truck.
“Robots can’t make tacos,” she said. “Not good ones anyway. People want a human making their tacos. There’s research.”
“You’re serious.”
“Completely. I’ve got a business plan and everything. My girlfriend’s family has recipes. We’re going to park outside tech companies and charge twenty dollars for a burrito.”
“That’s a lot for a burrito.”
“Tech workers don’t know that. They think everything costs twenty dollars. It’s beautiful.”
He started bringing lunch from home after that. Not because he couldn’t afford Rachel’s hypothetical burritos. Because he liked her, and he didn’t want to be one of the marks in her business plan.
The thing happened in July, during a team meeting.
Rachel was presenting the monthly error metrics. Mark was half-listening, drawing small cubes in the margin of his notebook the way he’d done since college. Then his name appeared on the screen.
“Mark is our top performer again,” Rachel said. “Forty-seven percent higher catch rate than the team average.”
There was polite applause. Mark nodded, uncomfortable.
Then Derek, who was twenty-six and had a computer science degree and had been at the company for eight months, raised his hand.
“So, I’ve been looking at the data,” Derek said. “And I could be wrong here, but I think Mark’s catch rate might actually be hurting our training metrics.”
The room went quiet.
“Explain,” Rachel said.
“So, Nexus learns from corrections, right? When we flag an error, it feeds back into the model. But if someone’s catching errors that are borderline, stuff the system would have self-corrected on the next iteration anyway, we’re basically adding noise to the training data.” Derek pulled up a spreadsheet, the slight defensiveness of someone who knew he was about to say something that would land badly. “I ran some analysis, and I think we’d get better model improvement if we let more marginal errors through.”
Mark stared at him.
“You’re saying I’m too good at my job.”
“I’m saying your expertise might be counterproductive to the optimization pipeline.”
“So the solution is for me to be worse.”
“The solution is for us to recalibrate what counts as an actionable error.”
Rachel cut in before Mark could respond. “We’ll discuss this offline,” she said, and moved to the next slide.
But the meeting was over for Mark. He sat there for another twenty minutes, drawing cubes, not hearing anything.
After the meeting, he went to the bathroom and stood at the sink for a long time, looking at his reflection. He did not whisper “pivot.” He did not laugh. He just stood there, water running, watching a man he was having trouble recognizing.
Denise called him in October. She’d found a job, finally. Not in data analytics. Those positions had largely evaporated while she was still learning SQL. She worked customer service for a health insurance company now, handling calls too complicated or emotionally volatile for the chatbots.
“I’m the human exception handler,” she told him. “I get the cases the machine can’t process. The crying. The screaming. The ones where the algorithm just keeps looping and the customer is about to lose their mind.”
“That sounds horrible.”
“It is horrible.” She paused. “But I’m good at it. Twenty-six years of managing unreasonable clients, you know? All those skills I thought were useless.”
“Are you okay?”
“Define okay.”
“Fair point.”
“There was a woman last week. Her kid needs surgery and the insurance won’t cover it. She was sobbing so hard I could barely understand her. And I sat there on my headset and I talked her through the appeals process, step by step, for forty-five minutes. And when we hung up she said I was the first person who’d actually helped her in six months of fighting with the system.”
“That’s…”
“Yeah. It’s something. I don’t know what, but it’s something.” She laughed, a little ragged. “My son still thinks I should learn prompt engineering.”
“How’s the market for prompt engineers these days?”
“Correcting. That’s the word he uses.”
His daughter came home for Thanksgiving with a major change. She’d switched from English to computer science.
“I thought you loved writing,” Mark said.
“I did. I do.” She pushed the food around her plate. “But I can’t afford to do what you did, Dad.”
Dana squeezed his hand under the table.
“Computer science isn’t safe either,” he said. “The AI is writing code now.”
“I know. But it’s safer than English. For now.”
Kevin was there, because Kevin was always there for holidays. He’d been talking about the market for twenty minutes before Mark tuned him out. Now he tuned back in.
“The thing is,” Kevin was saying, “this is actually great for productivity. I mean, yes, there’s short-term displacement, but the efficiency gains are going to lift all boats eventually. It’s basic economics.”
Mark looked at him.
“Kevin,” he said. “Have you ever been fired?”
Kevin blinked. “Well, no, but…”
“Have you ever applied for a job where five hundred people applied for the same position? Have you ever sat in a classroom learning skills you know will be obsolete before you finish learning them? Have you ever had a twenty-six-year-old explain to you that your expertise is counterproductive to the optimization pipeline?”
The table was silent.
“Mark,” Dana said quietly.
“No, I want to know.” He hadn’t raised his voice. That was the strange thing. He was perfectly calm. “I want to know what it looks like from inside the boat that’s getting lifted. Because from where I’m sitting, Kevin, it looks like the boats are getting lifted and the people are getting thrown overboard. And I’m tired of being told to be optimistic while I’m drowning.”
Kevin opened his mouth. Closed it.
“I’m going to get some air,” Mark said, and left the table.
He stood on the back porch for a long time, looking at nothing. After a while, Sophie came out and stood beside him.
“That was kind of amazing,” she said.
“That was rude.”
“Those aren’t mutually exclusive.” She leaned against the railing. “He’s never going to shut up about efficiency gains now. It’s going to be his whole personality for the next five years.”
“Probably.”
“Worth it though.”
He looked at his daughter. She was smiling, just a little.
“When did you get so cynical?”
“I learned from the best.”
The second email came exactly two years after the first. Same format. Same bloodless language. His editing role had been identified for transition.
Nexus could catch most of its own errors now. The remaining issues could be handled by a smaller team. He was being thanked for his valued contributions. He was being wished well in his future endeavors.
Rachel had already left. The food truck was real. She sent him a photo of it, bright red, parked outside a tech campus in Mountain View. “Come visit,” she’d written. “Burritos on the house for former coworkers.”
He sat in his home office. Winter light through the window. Cold coffee on his desk, next to the leather pen case.
He picked up one of the pens. A 1962 Parker 51, his favorite. His father had found it at an estate sale and given it to him when he’d gotten the Hartwell job. It still wrote beautifully. There was no practical reason to own a fountain pen in 2025. That wasn’t the point.
He put the pen back in the case and went to find Dana.
She was in the kitchen, making breakfast. Saturday morning. Eggs over easy, hot sauce, toast cut diagonally. She’d taken over his ritual at some point in the last year.
“I saw the email,” she said.
“Yeah.”
“You want to talk about it?”
“Not especially.”
“Okay.” She handed him a cup of coffee. “Sophie called. She got an internship. Some startup in Austin.”
“That’s good.”
“She sounded excited. Nervous, but excited.”
They stood by the window, watching the snow fall. The eggs sizzled in the pan. Somewhere, someone was celebrating another efficiency gain. Somewhere else, Rachel was selling twenty-dollar burritos to people who thought that was a reasonable price.
“I’m tired,” Mark said.
“I know.”
“I don’t know what comes next.”
“Neither do I.” Dana flipped the eggs. “But we’ve got food, and the mortgage is paid through March, and Sophie’s doing okay. That’s not nothing.”
“No. It’s not.”
She slid the eggs onto a plate and set it in front of him. Toast cut diagonally. Hot sauce on the side.
He ate his breakfast. When he finished, he went back to his office and took the Parker 51 out of its case. Found a legal pad in his desk drawer. Sat there for a moment, pen in hand.
He didn’t know what he was going to write. A letter, maybe. Or a list. Or just his name, over and over, in handwriting no algorithm could replicate.
He put the nib to paper and began.
Outside, the snow kept falling.
Thanks for reading, and I hope you enjoyed it!
I’m not remotely a doomer, but I’m pretty pissed that not enough folks are talking how we bridge the rapidly approaching epochal shift brought on by AI and robotics. This story is an attempt to broach that subject.
Yes, the future is exciting. And yes, we’ll all be better off.
But the transition is probably going to be rough, especially at the individual level, and we NEED to start telling a better story about how that plays out to everyone’s benefit or we’re going to find ourselves in a bit of a pickle.
If the goal of these technologies is to make lives better, to reduce suffering, then as much as possible that should begin right now, and carry right on through the shift.
Fear doesn’t help. Hope does.
And for hope, we need better stories, and better storytellers.