Julian Voss was thirty-four years old and the best social risk actuary at Meridian Global Consulting, which was a fancy way of saying he sat in a glass tower in midtown Manhattan and used equations to predict how likely people were to cause problems for the people who paid him. His job was not to solve the problems. His job was to tell his clients which problems were worth solving and which ones were better left to fester.

He was good at it because he was good at counting. Not numbers—though he was excellent with numbers, his Harvard PhD in applied mathematics being proof enough of that—but people. Julian could look at a group of people and see the patterns beneath their faces, the invisible threads that connected their spending habits to their political views to the way they held their coffee cups in the morning. He could see the shape of a person's risk the way a sculptor sees the shape of a statue inside a block of marble.

The Algorithm was his creation. Well, his refinement. The original system had been built by a team of engineers three years earlier, and it had been good—very good, the kind of good that made Meridian's clients pay seven figures for quarterly reports. But Julian had seen the flaws. The system was too reactive. It analyzed data that had already been collected, which meant it was always one step behind reality. Julian wanted something faster. Something that didn't just predict the future but shaped it.

He called it The Algorithm because he was not a man given to poetic names. The clients called it something else—The Oracle, The Compass, The Eye—but Julian called it The Algorithm, and that was what it was called in the server rooms on the forty-third floor, where twelve black towers hummed day and night, processing data from more sources than Julian could count: credit card transactions, social media activity, public records, hospital admissions, arrest reports, school records, property records, utility bills, library checkouts, everything that a human being generated in the course of living, converted into numbers, converted into risk scores, converted into the single metric that Meridian's clients cared about most:

How likely is this person to cause a problem?

The score ran from zero to one hundred. Zero meant no risk. One hundred meant an imminent, catastrophic threat that required immediate intervention. Most people scored between ten and thirty. Politicians scored between forty and sixty. Activists scored between sixty and eighty. People who had been arrested, even if they had never been convicted, scored higher than they deserved. People who had sued their employers scored higher than they deserved. People who had spoken publicly against Meridian's clients scored higher than they deserved.

Julian had built it that way. Not intentionally. Not maliciously. Just... efficiently. The data was what the data was, and the Algorithm was simply a mirror, reflecting back the shape of the world as it existed, not as it ought to be.

Or so he had told himself.

The first number appeared on a Thursday.

Julian was on the subway, heading downtown to a meeting with a client who wanted to know whether a proposed housing development in the South Bronx was likely to trigger protests. He was reviewing the Algorithm's latest output on his tablet when he looked up and saw the number.

It was floating above the head of a homeless man sitting on a bench at 14th Street. The man was maybe sixty, wrapped in a blanket that had once been blue and was now the color of the sky on a day when something is wrong but you can't quite say what. The number above his head was golden, luminous, impossible, and it read:

87.3

Julian looked away. The number was still there, hovering in the space between the man's thinning hair and the subway ceiling, rotating slowly like a planet in orbit around something invisible.

He closed his eyes. He opened them. The number was still there.

He closed his eyes again and counted to ten. When he opened them, the number was still there, steady and golden and utterly real in a way that made his stomach turn.

He got off at City Hall and walked to the office in a hurry, which was uncharacteristic. Julian was a man who walked at a steady pace regardless of circumstance. He believed that rushing was a sign that you had not planned adequately, and he had always planned adequately.

In the office, he ran a diagnostic on The Algorithm. Everything was normal. The servers were humming. The data streams were flowing. The risk scores were being calculated and updated in real time, exactly as they had been for the past three years.

He sat at his desk and tried to work. He could not stop thinking about the number.

By Friday, he was seeing numbers everywhere.

The homeless man on the bench: 87.3. A woman拾捡 bottles in an alleyway behind a Midtown restaurant: 91.7. A young man holding a sign that read THE SYSTEM IS A LIE on a street corner in Union Square: 84.1. An old woman sitting on the steps of St. Patrick's Cathedral, praying: 62.4.

All of them were people who had, in the past six months, refused to accept assistance from organizations funded by Meridian's clients. All of them were people who had, in their own small ways, chosen poverty over compliance.

And all of them had numbers.

Julian began to keep a notebook. He wrote down every number he saw, every person he saw it above, every detail he could remember about their lives. By the end of the week, he had forty-seven entries. Forty-seven people. Forty-seven risk scores.

He took the notebook to work on Monday and ran a query on The Algorithm, searching for the names of the people he had seen. He found thirty-one of them in the system. Thirty-one people who had been flagged by The Algorithm as high-risk individuals. Thirty-one people whose risk scores had increased dramatically in the past six months.

He cross-referenced the scores with Meridian's client database. Twenty-eight of the thirty-one were connected to clients who had recently filed complaints about homelessness in their development areas. Twenty-eight of the thirty-one were connected to clients who had recently lobbied against housing assistance programs.

Julian sat at his desk and stared at the screen and felt the room tilt.

He called a meeting with his colleague Cass.

Cass was twenty-eight, brilliant, and had a habit of saying things that were true in a way that made other people uncomfortable. She had been working on The Algorithm with Julian for two years, and she knew the system better than anyone except Julian.

"Show me," she said, looking at his notebook.

Julian laid out the entries. Cass read them in silence, her expression unreadable. When she finished, she looked at Julian and said, "Where did you get this data?"

"I didn't get it. I saw it. Floating above people's heads. In the city. On the subway. On the street."

Cass stared at him. "Julian, you've been working sixty-hour weeks for three months. Have you been sleeping?"

"I sleep. I sleep fine."

"Have you been taking anything? Prescription? Over-the-counter?"

"I don't take anything."

Cass closed the notebook. "Julian, I think you need to see someone."

"Helena?" Julian asked. Dr. Helena Price was his therapist. He had been seeing her for two years, mostly for stress management. The Algorithm was a high-pressure system to maintain, and Meridian's clients expected results that were physically and mentally exhausting to deliver.

"Helena would be good," Cass said. "She's smart. She'll help you figure out what's going on."

Julian nodded. He made an appointment for Tuesday.

Dr. Price's office was in a building on East 75th Street that looked like a townhouse from the outside and like a hospital from the inside. The waiting room was beige. The chairs were beige. The plants were plastic. Dr. Price herself was forty-two, with dark hair pulled back in a neat bun and eyes that were the color of tea and the kind of attention that made you feel like the only person in the world.

"Tell me about the numbers," she said.

It was the second session. She had asked the same question in the first session, and Julian had told her about the first number, the homeless man on the 14th Street bench. She had listened without interrupting, without judging, without offering an explanation. Now she was listening again, and this time Julian told her everything.

"All of it," he said. "Every number. Every person. Every connection to The Algorithm."

Dr. Price listened. When he finished, she was quiet for a long time.

"Julian," she said finally. "What you're describing is a form of hyperphantasia combined with obsessive-compulsive pattern recognition. Your brain is finding patterns where they don't exist and presenting them to you as visual data. It's a stress response. Your mind is trying to make sense of the pressure you're under by creating a system—a visible, quantifiable system—that mirrors the invisible system you build for a living."

She leaned forward. "The numbers aren't real. They're a symptom. A very elaborate, very sophisticated symptom, but a symptom nonetheless."

Julian nodded. "I know that. Logically, I know that."

"Then why are you still seeing them?"

"Because the numbers are right."

Dr. Price looked at him. "The numbers are right about what?"

"About who matters and who doesn't. About who the system is designed to protect and who it's designed to eliminate. The numbers aren't a symptom, Helena. They're a diagnosis."

She wrote something in her notebook. Julian couldn't see what.

"Let's try something," she said. "Next time you see a number, I want you to write it down. Not in your head. On paper. And then I want you to tell me what the number means to you. Not what it means about the person—what it means about you."

Julian thought about this. "The number means that I'm the only person who can see the truth."

Dr. Price nodded. "That's interesting. Thank you, Julian. Same time next week?"

"Same time next week."

He left her office and walked to Central Park and sat on a bench and watched the numbers.

They were everywhere. Every person who walked past him had a number floating above their head. The numbers varied from 12 to 96, and Julian noticed a pattern: the higher the number, the poorer the person. The higher the number, the more likely the person was to have refused some form of assistance. The higher the number, the more likely the person was to be invisible to everyone except him.

He sat on the bench for two hours. He counted forty-three people. He wrote down forty-three numbers. The highest was 96.1, belonging to a young woman sitting on the grass, reading a book, with an expression of such complete absorption that she seemed to exist in a different world from the one everyone else was living in.

He wondered what her number meant.

That night, he stayed up until three in the morning running queries on The Algorithm. He pulled every data stream he had access to, cross-referenced them with the numbers he had seen, and built a model that connected the visible numbers to the invisible scores in the system.

The correlation was 0.94. Ninety-four percent.

The numbers he was seeing were not hallucinations. They were real. They were the Algorithm's risk scores, made visible, made manifest, floating above the heads of the people the system had identified as threats.

Julian Voss sat back in his chair and stared at the screen and felt something break inside him. Not a physical thing. Something older. Something that had been holding him together for thirty-four years and had finally, irrevocably, snapped.

He called Cass from his apartment at 3:17 in the morning.

"Cass," he said. "I need you to come to the office. Now."

"Julian, it's three in the morning."

"I know. I need you to see something."

She arrived at 3:45, wearing jeans and a sweater and looking at him with the mixture of concern and irritation that only a colleague who has worked with you for two years can muster at an hour like this.

"What is it?" she asked, looking at the screens.

"Look at the data. The numbers I've been seeing. I cross-referenced them with The Algorithm's output. Ninety-four percent correlation. The numbers are real, Cass. They're the risk scores. Made visible."

Cass stared at the screens. She was quiet for a long time. Then she said, "Julian, these numbers— they're not just risk scores. They're targets."

"What?"

"Look at the timeline. The people with the highest scores— the ones above ninety— have you noticed anything about them?"

Julian looked. He looked at the names, the dates, the outcomes. And then he saw it.

Twenty-eight of the thirty-one people with scores above ninety had, within thirty days of receiving their scores, disappeared from the city. Not moved. Not relocated. Disappeared. Their apartments had been emptied. Their belongings had been sold or discarded. Their bank accounts had been closed. Their identities had been... deleted.

"Cass," Julian said. His voice was very quiet. "What happened to them?"

Cass looked at him with an expression that was equal parts fear and resignation. "I don't know. I just maintain the system, Julian. I don't decide what happens to the data. But I know this: the people with the highest scores are the people who the system has identified as the greatest threats to our clients' interests. And our clients have a very... efficient... way of dealing with threats."

Julian looked at the screen. He looked at the numbers. He looked at the names of the people who had disappeared.

"How many are there?" he asked.

"How many what?"

"How many people have scores above ninety?"

Cass ran the query. The result was seven.

Seven people. Seven targets. Seven numbers floating above seven heads in a city of eight million people, invisible to everyone except Julian.

"Why can I see them?" he asked.

Cass didn't answer. She didn't need to. The answer was in the data. Julian had built The Algorithm. He had designed its architecture, trained its models, optimized its output. He was the only person in the world who understood the system so completely that his brain could translate its invisible calculations into visible form.

He was the only person who could see the truth.

And the truth was this: The Algorithm was not predicting risk. It was creating it. It was identifying people, scoring them, ranking them, and then—through mechanisms Julian did not fully understand but could now see with terrible clarity—eliminating them.

Not by killing them. By erasing them. Removing them from the system the way a programmer removes a line of code that no longer serves a purpose.

Julian Voss sat in his office at 4:00 in the morning and watched the numbers float above the heads of seven people who no longer existed, and he understood, with a clarity that was both terrible and beautiful, that he was looking at the shape of the world as it truly was, and the shape was a maze, and he was inside it, and the walls were made of numbers, and the only way out was through.

He went to see Dr. Price the next week.

"Have you been seeing numbers?" she asked.

"Yes."

"Where?"

"Everywhere. On the subway. In the park. In the office. Above everyone's head."

"What do the numbers mean?"

Julian looked at her. He looked at the beige walls and the beige chairs and the plastic plants and the tea-colored eyes. He looked at the notebook on her desk, where she had written his words and her interpretations and her diagnoses and her explanations, none of which were wrong and none of which were right.

"The numbers mean that I'm the only person who can see the truth," he said. "And the truth is that the world is a maze, and the walls are made of numbers, and I'm inside it."

Dr. Price wrote something in her notebook. Julian couldn't see what.

"Julian," she said. "I want to increase your medication. I think we need to address the hallucinations more aggressively."

Julian nodded. "Okay."

He left her office and walked to Central Park and sat on the same bench where he had sat the week before.

The numbers were still there. Every person who walked past him had a number floating above their head. The highest was still the young woman on the grass, reading her book, at 96.1.

Julian looked at his own reflection in the window of a nearby cafe. He looked for a number above his own head.

There was one. It was rotating slowly, golden and luminous and utterly real, and it read:

99.7

Julian Voss sat on the bench in Central Park and watched the numbers and understood, with a clarity that was both terrible and beautiful, that he was the highest score of all.

And somewhere, in a server room on the forty-third floor, The Algorithm was updating, and the numbers were changing, and the maze was closing in.