The Softening Variable
The glass box was smaller than she’d imagined. Suite 9-14 turned out to be a six-by-eight-foot cube of floor-to-ceiling windows on the ninth floor’s southwest corner, positioned so that every person who walked the main corridor could see her sitting inside it like a fish in a particularly depressing aquarium. The desk was white laminate. The chair was ergonomic in that way that promised lumbar support and delivered only a vague sense of judgment. There was a single potted succulent on the windowsill that looked like it had been watered approximately once, three years ago, by someone who had since left the company.
Mira stood in the doorway for a full minute before stepping inside. The door swung shut behind her with a pneumatic hiss that felt final.
Her old desk on the fourth floor had been a warren of monitors and sticky notes, a cave she’d carved out of the open-plan chaos. She’d known exactly which floorboards creaked, which ceiling tiles dripped when it rained, which hours the coffee machine actually dispensed something drinkable. Here, everything was clean and sterile and wrong. The window gave her a view of the parking garage across the street, where a man in a yellow vest was directing traffic with the exhausted precision of someone who had long ago stopped believing his job mattered.
She sat down. The chair made a sound like a deflating balloon.
Her new phone was already on the desk, a sleek black handset with a single pre-programmed contact: Derek Voss, Strategic Initiatives. She picked it up, turned it over, put it back down. The screen was blank. She had nothing to say to him, and he had nothing to say to her that she wanted to hear.
The spreadsheet was waiting on her laptop when she opened it. She’d taken to keeping the file in a folder called “Q3_Reports_ARCHIVE” buried three levels deep in the shared drive, a hiding spot so obvious it was invisible. The file opened to a single tab, clean and white, with yesterday’s logged prediction still sitting in the first row:
*Gus will bring his thermos today. — CONFIRMED*
She stared at it for a long moment. The thermos prediction had felt like a test, a way to prove to herself that the spreadsheet still worked after the promotion, after the glass box, after everything. Gus had walked in at 7:52 AM with his dented blue thermos, same as he did every Tuesday, and Mira had felt a small, guilty relief that at least some things remained constant.
She deleted the row. The spreadsheet hummed, a low vibration she felt through her fingertips on the keyboard. It was waiting.
Mira pulled up the company-wide announcements feed on her second monitor. The feed was a graveyard of corporate optimism: “Meridian Celebrates Safety Week!” “New Partnership with Pacific Freight!” “Reminder: Q3 Goals Are Within Reach!” She scrolled through them, looking for patterns, for the small cracks that the spreadsheet might be able to see.
That was when she saw the email.
It was from HR, addressed to all department heads, timestamped 8:47 AM. The subject line read: “Preliminary Restructuring Framework — Confidential.” Mira clicked it open without thinking, her analyst brain already cataloging the language: *streamlining*, *rightsizing*, *strategic realignment*. The email contained a single attachment, a PDF titled “MERIDIAN_RESTRUCTURE_PRELIM_v1.pdf.”
She didn’t open the attachment. She didn’t need to. The spreadsheet was already glowing.
The glow was faint at first, a pale blue shimmer that rippled across the cells like heat haze on asphalt. Mira watched as a new row appeared at the top of the sheet, the text materializing letter by letter, as if the spreadsheet were typing it out in real time:
*Layoff wave: 200 employees. Effective 30 days. Departments: Operations, Customer Service, Warehouse, IT. Severance: 2 weeks per year of service. No exceptions.*
Mira’s hands went cold.
Two hundred people. She did the math without meaning to—Meridian employed roughly eight hundred people across three floors and a warehouse. Two hundred was a quarter of the company. Two hundred was the entire fourth floor, the entire customer service wing, the entire IT department except for the two guys who kept the servers running and who were probably too valuable to cut.
She thought of the fourth floor. She thought of Priya at the front desk, who always saved her a seat by the window. She thought of Marcus in IT, who’d fixed her laptop three times and never once made her feel stupid for breaking it. She thought of the woman in the cubicle next to hers—what was her name? Dana. Dana who had a picture of her dog on her desk and who brought in homemade banana bread every Friday.
Two hundred people. The spreadsheet was waiting, its cursor blinking in the empty cell beneath the prediction, patient as a held breath.
Mira closed the laptop.
She sat in the glass box, staring at her reflection in the window. The man in the yellow vest was still directing traffic. The parking garage was full. The sun was doing its best to burn through the smog, and the city stretched out beyond the glass like a promise she couldn’t keep.
She opened the laptop again.
The prediction was still there. Of course it was. The spreadsheet didn’t forget, didn’t waver, didn’t care about her moral qualms. It just sat there, glowing faintly, waiting for her to log it.
Mira’s fingers hovered over the keyboard. She could type it out. She could log the prediction, and it would come true, and two hundred people would lose their jobs, and she would have done nothing to stop it. Or she could delete the row, close the file, pretend she’d never seen it—and the layoffs might not happen. The spreadsheet had rules. She’d learned them over the past two weeks, cataloging them in a notebook she kept in her bag:
1. Predictions logged before 5 PM come true.
2. Predictions logged after 5 PM don’t come true, but the spreadsheet remembers.
3. She can’t predict her own actions.
4. Predictions can be altered, but only by logging a new one before 5 PM.
The layoff prediction was already in the sheet. It had been logged—by the spreadsheet itself, or by whatever force animated it. She hadn’t typed it. She hadn’t even known about the restructuring email until she’d opened it.
But the prediction was there. And if she did nothing, it would come true.
Mira looked at the clock. 9:47 AM. She had seven hours and thirteen minutes before the 5 PM cutoff.
She started typing.
*Restructuring announcement.*
She stared at the words. They were vague, almost meaningless. A restructuring announcement could be anything—a minor reorg, a new reporting structure, a change to the company’s org chart. It didn’t have to be a layoff wave. It didn’t have to be two hundred people.
The spreadsheet hummed. The cell accepted the text, and the original prediction flickered, wavering like a candle in a draft. Mira watched as the numbers shifted, the text blurring and reforming:
*Layoff wave: 200 employees. Effective 30 days. Departments: Operations, Customer Service, Warehouse, IT. Severance: 2 weeks per year of service. No exceptions.*
The prediction didn’t change.
Mira frowned. She tried again, typing more carefully this time:
*Restructuring announcement. Minor reorg. No layoffs.*
The spreadsheet went still. The glow dimmed, then flared, then dimmed again. For a moment, Mira thought she’d broken it—that the spreadsheet had finally hit a prediction it couldn’t process, a variable it couldn’t account for.
Then the text shifted.
*Restructuring announcement. Minor reorg. No layoffs. — CONFIRMED*
Mira stared at the screen. The original prediction was gone, replaced by her vague, hopeful version. The spreadsheet had accepted it. The layoff wave had been overwritten.
She let out a breath she hadn’t realized she’d been holding.
The glass box suddenly felt less like a prison. She leaned back in the ergonomic chair, which creaked in protest, and looked out the window. The man in the yellow vest had finished directing traffic and was leaning against the parking garage wall, drinking from a thermos. Mira wondered if he knew how lucky he was, to have a job that didn’t require him to predict the future.
She closed the laptop. The spreadsheet’s glow faded, but she could still feel it humming beneath her fingertips, a low vibration that seemed to come from the building itself.
The rest of the morning passed in a blur of onboarding tasks. She filled out forms. She attended a video call with HR about her benefits package. She learned how to use the printer on the ninth floor, which was different from the printer on the fourth floor, which was different from the printer on the third floor, which was broken and had been broken since before she started.
At 12:30, her phone buzzed. A Slack message from Derek Voss:
*Mira. My office. 1 PM. Bring your forecasting model.*
She stared at the message for a long moment. The forecasting model. He meant the spreadsheet. He didn’t know what it was, not really—he thought she’d built some kind of predictive algorithm, a sophisticated piece of data analysis that could anticipate shipment delays and operational hiccups. He had no idea it was a living thing, a company ghost, a manifestation of the building’s accumulated secrets.
She typed back: *On my way.*
Derek’s office was on the tenth floor, one floor above hers. It was twice the size of her glass box, with a real door and actual walls and a view of the harbor instead of the parking garage. He was standing by the window when she walked in, his back to her, his hands clasped behind his back like a general surveying a battlefield.
“Mira,” he said, without turning around. “Congratulations on the promotion.”
“Thank you,” she said.
“I assume you’ve settled in.”
“It’s very… glassy.”
Derek turned. He was a thin man with a thin smile and thin hair, and he wore his suits like they were armor. He gestured to the chair across from his desk, and Mira sat down.
“I’ve been reviewing your work,” he said. “The forecasting model you’ve been using. It’s impressive. The accuracy rate is unprecedented.”
“I’ve been lucky,” Mira said.
“Luck doesn’t predict a forklift accident to the minute.” Derek sat down across from her, folding his hands on the desk. “I’ve been in this industry for twenty years. I’ve seen a lot of data analysts. None of them have ever been this accurate.”
Mira said nothing.
“I want to scale it,” Derek said. “Your model. I want to integrate it into our broader operations. Use it to predict profitability opportunities—which clients to prioritize, which accounts to drop, which employees to cut.”
The word *cut* hung in the air between them.
“I don’t think that’s a good idea,” Mira said.
Derek’s smile didn’t waver. “Why not?”
“The model isn’t designed for that kind of prediction. It works best with operational data—shipment delays, equipment failures, things like that. It’s not meant to make decisions about people.”
“Everything is data, Mira.” Derek leaned forward. “People are data. Their performance reviews are data. Their attendance records are data. Their value to the company is data. Your model can help us make better decisions. More efficient decisions.”
Mira thought about the spreadsheet. She thought about the layoff wave she’d just overwritten, the two hundred people who would keep their jobs because she’d logged a vague prediction. She thought about what would happen if Derek got his hands on it, if he started feeding it names and performance metrics and asking it to decide who deserved to stay.
“I’ll need to run some tests,” she said. “To see if the model can handle that kind of input.”
Derek’s smile widened. “Of course. Take your time. But Mira—don’t take too long. The company is in a delicate phase. We need to move quickly.”
Mira nodded. She stood up, walked to the door, and paused with her hand on the handle.
“One more thing,” Derek said. “I’ve been meaning to ask. The spreadsheet you’ve been using—where did it come from?”
Mira’s heart skipped. “I built it,” she said. “From scratch. It’s just a standard regression model with some custom variables.”
“Impressive,” Derek said. “I’d love to see the source code sometime.”
“It’s proprietary,” Mira said. “I’ll put together a summary for you.”
She left before he could ask anything else.
Back in the glass box, she opened the spreadsheet. The vague prediction was still there, glowing faintly: *Restructuring announcement. Minor reorg. No layoffs. — CONFIRMED*
She stared at it for a long moment. Then she typed a new row:
*Derek Voss will request access to the forecasting model within 7 days.*
The spreadsheet accepted the text. The glow flickered, then steadied.
Mira closed the laptop and looked out the window. The man in the yellow vest was gone. The parking garage was emptying out, cars streaming toward the highway in the late afternoon sun. She had four hours until the 5 PM cutoff, and she had no idea what she was going to do.
The spreadsheet had given her a tool. But it had also given her a target. And Derek Voss was circling.
Mira stood in the doorway for a full minute before stepping inside. The door swung shut behind her with a pneumatic hiss that felt final.
Her old desk on the fourth floor had been a warren of monitors and sticky notes, a cave she’d carved out of the open-plan chaos. She’d known exactly which floorboards creaked, which ceiling tiles dripped when it rained, which hours the coffee machine actually dispensed something drinkable. Here, everything was clean and sterile and wrong. The window gave her a view of the parking garage across the street, where a man in a yellow vest was directing traffic with the exhausted precision of someone who had long ago stopped believing his job mattered.
She sat down. The chair made a sound like a deflating balloon.
Her new phone was already on the desk, a sleek black handset with a single pre-programmed contact: Derek Voss, Strategic Initiatives. She picked it up, turned it over, put it back down. The screen was blank. She had nothing to say to him, and he had nothing to say to her that she wanted to hear.
The spreadsheet was waiting on her laptop when she opened it. She’d taken to keeping the file in a folder called “Q3_Reports_ARCHIVE” buried three levels deep in the shared drive, a hiding spot so obvious it was invisible. The file opened to a single tab, clean and white, with yesterday’s logged prediction still sitting in the first row:
*Gus will bring his thermos today. — CONFIRMED*
She stared at it for a long moment. The thermos prediction had felt like a test, a way to prove to herself that the spreadsheet still worked after the promotion, after the glass box, after everything. Gus had walked in at 7:52 AM with his dented blue thermos, same as he did every Tuesday, and Mira had felt a small, guilty relief that at least some things remained constant.
She deleted the row. The spreadsheet hummed, a low vibration she felt through her fingertips on the keyboard. It was waiting.
Mira pulled up the company-wide announcements feed on her second monitor. The feed was a graveyard of corporate optimism: “Meridian Celebrates Safety Week!” “New Partnership with Pacific Freight!” “Reminder: Q3 Goals Are Within Reach!” She scrolled through them, looking for patterns, for the small cracks that the spreadsheet might be able to see.
That was when she saw the email.
It was from HR, addressed to all department heads, timestamped 8:47 AM. The subject line read: “Preliminary Restructuring Framework — Confidential.” Mira clicked it open without thinking, her analyst brain already cataloging the language: *streamlining*, *rightsizing*, *strategic realignment*. The email contained a single attachment, a PDF titled “MERIDIAN_RESTRUCTURE_PRELIM_v1.pdf.”
She didn’t open the attachment. She didn’t need to. The spreadsheet was already glowing.
The glow was faint at first, a pale blue shimmer that rippled across the cells like heat haze on asphalt. Mira watched as a new row appeared at the top of the sheet, the text materializing letter by letter, as if the spreadsheet were typing it out in real time:
*Layoff wave: 200 employees. Effective 30 days. Departments: Operations, Customer Service, Warehouse, IT. Severance: 2 weeks per year of service. No exceptions.*
Mira’s hands went cold.
Two hundred people. She did the math without meaning to—Meridian employed roughly eight hundred people across three floors and a warehouse. Two hundred was a quarter of the company. Two hundred was the entire fourth floor, the entire customer service wing, the entire IT department except for the two guys who kept the servers running and who were probably too valuable to cut.
She thought of the fourth floor. She thought of Priya at the front desk, who always saved her a seat by the window. She thought of Marcus in IT, who’d fixed her laptop three times and never once made her feel stupid for breaking it. She thought of the woman in the cubicle next to hers—what was her name? Dana. Dana who had a picture of her dog on her desk and who brought in homemade banana bread every Friday.
Two hundred people. The spreadsheet was waiting, its cursor blinking in the empty cell beneath the prediction, patient as a held breath.
Mira closed the laptop.
She sat in the glass box, staring at her reflection in the window. The man in the yellow vest was still directing traffic. The parking garage was full. The sun was doing its best to burn through the smog, and the city stretched out beyond the glass like a promise she couldn’t keep.
She opened the laptop again.
The prediction was still there. Of course it was. The spreadsheet didn’t forget, didn’t waver, didn’t care about her moral qualms. It just sat there, glowing faintly, waiting for her to log it.
Mira’s fingers hovered over the keyboard. She could type it out. She could log the prediction, and it would come true, and two hundred people would lose their jobs, and she would have done nothing to stop it. Or she could delete the row, close the file, pretend she’d never seen it—and the layoffs might not happen. The spreadsheet had rules. She’d learned them over the past two weeks, cataloging them in a notebook she kept in her bag:
1. Predictions logged before 5 PM come true.
2. Predictions logged after 5 PM don’t come true, but the spreadsheet remembers.
3. She can’t predict her own actions.
4. Predictions can be altered, but only by logging a new one before 5 PM.
The layoff prediction was already in the sheet. It had been logged—by the spreadsheet itself, or by whatever force animated it. She hadn’t typed it. She hadn’t even known about the restructuring email until she’d opened it.
But the prediction was there. And if she did nothing, it would come true.
Mira looked at the clock. 9:47 AM. She had seven hours and thirteen minutes before the 5 PM cutoff.
She started typing.
*Restructuring announcement.*
She stared at the words. They were vague, almost meaningless. A restructuring announcement could be anything—a minor reorg, a new reporting structure, a change to the company’s org chart. It didn’t have to be a layoff wave. It didn’t have to be two hundred people.
The spreadsheet hummed. The cell accepted the text, and the original prediction flickered, wavering like a candle in a draft. Mira watched as the numbers shifted, the text blurring and reforming:
*Layoff wave: 200 employees. Effective 30 days. Departments: Operations, Customer Service, Warehouse, IT. Severance: 2 weeks per year of service. No exceptions.*
The prediction didn’t change.
Mira frowned. She tried again, typing more carefully this time:
*Restructuring announcement. Minor reorg. No layoffs.*
The spreadsheet went still. The glow dimmed, then flared, then dimmed again. For a moment, Mira thought she’d broken it—that the spreadsheet had finally hit a prediction it couldn’t process, a variable it couldn’t account for.
Then the text shifted.
*Restructuring announcement. Minor reorg. No layoffs. — CONFIRMED*
Mira stared at the screen. The original prediction was gone, replaced by her vague, hopeful version. The spreadsheet had accepted it. The layoff wave had been overwritten.
She let out a breath she hadn’t realized she’d been holding.
The glass box suddenly felt less like a prison. She leaned back in the ergonomic chair, which creaked in protest, and looked out the window. The man in the yellow vest had finished directing traffic and was leaning against the parking garage wall, drinking from a thermos. Mira wondered if he knew how lucky he was, to have a job that didn’t require him to predict the future.
She closed the laptop. The spreadsheet’s glow faded, but she could still feel it humming beneath her fingertips, a low vibration that seemed to come from the building itself.
The rest of the morning passed in a blur of onboarding tasks. She filled out forms. She attended a video call with HR about her benefits package. She learned how to use the printer on the ninth floor, which was different from the printer on the fourth floor, which was different from the printer on the third floor, which was broken and had been broken since before she started.
At 12:30, her phone buzzed. A Slack message from Derek Voss:
*Mira. My office. 1 PM. Bring your forecasting model.*
She stared at the message for a long moment. The forecasting model. He meant the spreadsheet. He didn’t know what it was, not really—he thought she’d built some kind of predictive algorithm, a sophisticated piece of data analysis that could anticipate shipment delays and operational hiccups. He had no idea it was a living thing, a company ghost, a manifestation of the building’s accumulated secrets.
She typed back: *On my way.*
Derek’s office was on the tenth floor, one floor above hers. It was twice the size of her glass box, with a real door and actual walls and a view of the harbor instead of the parking garage. He was standing by the window when she walked in, his back to her, his hands clasped behind his back like a general surveying a battlefield.
“Mira,” he said, without turning around. “Congratulations on the promotion.”
“Thank you,” she said.
“I assume you’ve settled in.”
“It’s very… glassy.”
Derek turned. He was a thin man with a thin smile and thin hair, and he wore his suits like they were armor. He gestured to the chair across from his desk, and Mira sat down.
“I’ve been reviewing your work,” he said. “The forecasting model you’ve been using. It’s impressive. The accuracy rate is unprecedented.”
“I’ve been lucky,” Mira said.
“Luck doesn’t predict a forklift accident to the minute.” Derek sat down across from her, folding his hands on the desk. “I’ve been in this industry for twenty years. I’ve seen a lot of data analysts. None of them have ever been this accurate.”
Mira said nothing.
“I want to scale it,” Derek said. “Your model. I want to integrate it into our broader operations. Use it to predict profitability opportunities—which clients to prioritize, which accounts to drop, which employees to cut.”
The word *cut* hung in the air between them.
“I don’t think that’s a good idea,” Mira said.
Derek’s smile didn’t waver. “Why not?”
“The model isn’t designed for that kind of prediction. It works best with operational data—shipment delays, equipment failures, things like that. It’s not meant to make decisions about people.”
“Everything is data, Mira.” Derek leaned forward. “People are data. Their performance reviews are data. Their attendance records are data. Their value to the company is data. Your model can help us make better decisions. More efficient decisions.”
Mira thought about the spreadsheet. She thought about the layoff wave she’d just overwritten, the two hundred people who would keep their jobs because she’d logged a vague prediction. She thought about what would happen if Derek got his hands on it, if he started feeding it names and performance metrics and asking it to decide who deserved to stay.
“I’ll need to run some tests,” she said. “To see if the model can handle that kind of input.”
Derek’s smile widened. “Of course. Take your time. But Mira—don’t take too long. The company is in a delicate phase. We need to move quickly.”
Mira nodded. She stood up, walked to the door, and paused with her hand on the handle.
“One more thing,” Derek said. “I’ve been meaning to ask. The spreadsheet you’ve been using—where did it come from?”
Mira’s heart skipped. “I built it,” she said. “From scratch. It’s just a standard regression model with some custom variables.”
“Impressive,” Derek said. “I’d love to see the source code sometime.”
“It’s proprietary,” Mira said. “I’ll put together a summary for you.”
She left before he could ask anything else.
Back in the glass box, she opened the spreadsheet. The vague prediction was still there, glowing faintly: *Restructuring announcement. Minor reorg. No layoffs. — CONFIRMED*
She stared at it for a long moment. Then she typed a new row:
*Derek Voss will request access to the forecasting model within 7 days.*
The spreadsheet accepted the text. The glow flickered, then steadied.
Mira closed the laptop and looked out the window. The man in the yellow vest was gone. The parking garage was emptying out, cars streaming toward the highway in the late afternoon sun. She had four hours until the 5 PM cutoff, and she had no idea what she was going to do.
The spreadsheet had given her a tool. But it had also given her a target. And Derek Voss was circling.