The data looked wrong. Daniel Reeves knew it looked wrong because he had looked at it a hundred times and the numbers never changed.

Deep Space Monitoring Center, Building 4, Room 217. Third floor, east wing. The kind of office that existed only in government buildings—beige walls, flickering fluorescent light, a desk that wobbled because one leg was shorter than the rest. Daniel sat at that desk every day from nine to five, analyzing radio signals from deep space, running them through algorithms that filtered noise from signal, signal from noise, and then filing the results in a database that nobody read.

It was honest work. Boring, but honest. And Daniel liked honest.

The anomaly was in dataset 4471-B, a routine scan of the Orion sector that had come in on a Tuesday morning. The signal was weak—barely above the background noise—but it had a pattern. Not the random pattern of cosmic radiation, not the structured pattern of a pulsar. Something else. Something that looked, if you squinted at it hard enough, like compression.

Daniel had run it through every filter. He had cross-referenced it with every database. He had even pulled the raw data and looked at it himself, staring at the waveform on his monitor until his eyes burned. The result was always the same: something in the Orion sector was being compressed. Spatially. Structurally. The signal was getting denser, more concentrated, as though the space it occupied was collapsing inward.

He took the findings to Mr. Kowalski on Thursday.

Kowalski was forty-five, built like a fire hydrant, and had been Daniel's supervisor for three years. He wore suits that were always slightly too tight and carried a coffee cup that said WORLD'S OKAYEST MANAGER in letters that had faded to illegibility.

"Compression," Kowalski repeated, reading Daniel's report over the top of his glasses. He did not look up. "You're saying space is compressing."

"I'm saying the signal suggests spatial compression in the Orion sector. The data is—"

"How many times have you run this?"

"Fourteen times."

"Fourteen." Kowalski set down the report. "Reeves. How long have you been here?"

"Six years, sir."

"Six years. You know how this works. You find something interesting, you file it, you move on to the next dataset. You don't—" He waved the report. "—you don't go running to Congress with it."

"Sir, this could be significant. If space is compressing—"

"If space is compressing, it's not our call to make." Kowalski picked up the report and walked it to the shredder. "Delete the dataset. File a null result. Move on."

Daniel stared at him. "You want me to delete it?"

"I want you to follow instructions. Delete dataset 4471-B. File null. Move on."

He did it. He deleted the dataset. He filed a null result. He moved on.

For three days.

On Saturday, he went back to the center. He didn't have to—nobody expected him on weekends—but he couldn't stop thinking about the signal. He logged into the system from his apartment, pulled up the raw data archive, and found 4471-B.

It was still there. He had deleted the analysis, but the raw data remained. Government systems were like that—everything was backed up, everything was stored, nothing was truly gone.

He ran the analysis again. Same result. Spatial compression in the Orion sector. Accelerating.

He ran it a third time. Same result.

On Monday, he took it to Dr. Patel.

Dr. Patel was fifty-two, the center's chief scientist, and the closest thing Daniel had to a mentor. He had come to America from India in 1978, when Daniel's father was in college, and had spent every year since trying to understand the universe. He was also, Daniel had learned, a man who understood bureaucracy.

Patil read the report in silence. When he finished, he took off his glasses and rubbed the bridge of his nose.

"Daniel," he said, "how did you get access to the raw archive?"

"I—I have clearance for analysis. The archive is part of—"

"You have clearance for routine analysis. The raw archive is restricted. If anyone asks, you never touched it."

"Dr. Patel, the data—"

"I know what the data says."

Daniel stopped. "You do?"

Patil looked at him with an expression that was equal parts pity and exhaustion. "Daniel. How long do you think I've been looking at this data?"

The room felt smaller. "How long?"

"Since 2019. We've had four separate reports of spatial anomalies in different sectors. Orion, Perseus, Cygnus, and—this one I haven't filed—Lyra. Each one showed compression. Each one was classified. Each one was deleted from the active database."

"Classified? By whom?"

Patil didn't answer. He didn't need to. The answer was in the silence.

Daniel left the center early. He walked to a diner on Broadway, ordered black coffee, and sat by the window watching people pass. He thought about what Patil had said. Four reports. Four anomalies. Four deletions.

Someone was erasing evidence of something happening to the universe, and nobody was supposed to know.

He thought about deleting the data. Really deleting it this time. Not just filing a null result, but going into the system and scrubbing it clean. It would be easy. He knew the commands. He had run them a dozen times on test datasets.

But he couldn't.

Not because he was brave. He wasn't. He was a data analyst. He filed reports and moved on. That's what he did.

He couldn't do it because the data was true. And truth, once you'd seen it, was hard to unsee.

He went home and made a backup copy of the dataset. Not on the government system—on a hard drive he kept for personal use. He encrypted it. Hid it in a folder labeled TAX RECORDS 2021.

Then he looked up Sarah O'Brien.

Sarah was a reporter for the New York Times, covering science and technology. She was twenty-nine, sharp, and had a reputation for asking questions that made government officials uncomfortable. Daniel had met her once, at a press briefing, when he had been sent to provide technical support. She had asked him a question about deep space monitoring that he had not been able to answer, and he had remembered it ever since.

He called her on a Sunday. She answered on the second ring.

"O'Brien," she said.

"Mr. Reeves. Deep Space Monitoring Center. I know you. You asked me about signal filtering at the briefing last year."

"I remember. What do you want?"

"Can we meet? Tomorrow? I have something I want to show you."

She was at a coffee shop on 43rd Street at noon. She arrived ten minutes late, carrying a notebook and wearing a blazer that looked expensive but had been worn a lot. She sat down, ordered a latte, and looked at him over the rim of her mug.

"Talk."

He showed her the data. Not the full dataset—he couldn't bring the hard drive to a coffee shop—but screenshots, printed copies, analysis summaries. She read everything in silence, her expression changing from skepticism to something harder to read.

"When did you first see this?" she asked.

"Last week. But Dr. Patel says the anomalies go back to 2019."

"Four years. And nobody's gone public."

"Nobody who saw the data was allowed to."

She set down her coffee. "Reeves, do you understand what you're telling me? You're saying that the U.S. government has known about spatial anomalies in deep space for four years, and has actively suppressed the data."

"I'm saying that the data shows spatial compression. I'm saying that four different sectors have been affected. I'm saying that the analysis was deleted from the public database and replaced with null results."

"By whom?"

He shook his head. "I don't know. Dr. Patel won't say. Mr. Kowalski won't say. I'm a data analyst, Ms. O'Brien. I don't have security clearance for chain-of-command questions."

She was quiet for a long time. When she spoke, her voice was different—sharper, more focused. "I need to see the full dataset. The encrypted backup."

"I can't—"

"You can. And you will. Because if you don't, and something happens—and something is happening, Reeves, four years of suppressed data doesn't go away—and you'll carry it for the rest of your life."

He carried the hard drive to the coffee shop the next day. She photographed every page, every chart, every line of analysis. She took three hours. When she was done, she put everything back in the TAX RECORDS folder and handed him the drive.

"Thank you," she said.

"Don't thank me yet."

She published the story on a Thursday. The headline was simple: DEEP SPACE ANOMALIES SUPPRESSED FOR FOUR YEARS, DOCUMENTS SHOW.

By Friday, Daniel was summoned to Kowalski's office.

"Sit down," Kowalski said. He did not offer coffee.

"Mr. Kowalski, I can explain—"

"There's nothing to explain. You leaked classified data. You violated your security clearance. You compromised national—"

"It's not classified. It's not national security. It's data about space."

Kowalski's face was red. "It was classified by order of the Office of Science and Technology Policy. Your data, your analysis, your conclusions—all of it was classified. And you violated that classification."

"I didn't classify it. I analyzed it."

"That's semantics, Reeves. The question is whether you had the right to make it public."

He didn't have an answer.

On Monday, he was terminated. Security escorted him out of the building. He was required to surrender his badge, his access cards, and any government property—including the hard drive, which was taken from him at the door.

Sarah published a follow-up. Then another. The story spread. Congressional hearings were announced. Dr. Patel testified behind closed doors and refused to comment in public. Kowalski retired "in disgrace," though nobody knew what that meant exactly—disgrace for what, exactly? The data had always been there. The anomalies had always existed.

Daniel got a job at a small tech company in Queens. He analyzed user data for a streaming service. It was nothing like deep space monitoring. It was boring, repetitive, and completely unimportant.

He liked it.

Six months after the hearings ended, six months after the story faded from the news, Daniel was sitting in his apartment on a Friday night, eating takeout and watching television, when his phone buzzed.

It was Sarah.

"Reeves," she said. "You need to see this."

"What is it?"

She sent him a link. It was a paper, published in a peer-reviewed astrophysics journal. The authors were anonymous—the paper had been submitted under pseudonyms, and the review process had been expedited. The title was simple: OBSERVED SPATIAL COMPRESSION IN MULTIPLE SECTORS OF THE ORION-CYGNUS ARM.

The abstract said what Daniel already knew. The compression was accelerating. It was not localized to one sector. It was spreading.

And it had been happening for longer than four years.

"Someone published it," Sarah said. "Someone inside the system finally decided to go public the right way. Peer review. Open access. No classification."

"Who?"

"I don't know. But the data is yours, Reeves. Your analysis. Your conclusions. You were right."

He looked at the screen. The data was there, confirmed by independent researchers, validated by peer review. The compression was real. It was accelerating. And it was spreading.

He thought about Dr. Patel, testifying behind closed doors. He thought about Kowalski, retiring in disgrace. He thought about the four years of silence, the deleted datasets, the null results filed in a database nobody read.

Then he thought about the paper. Published. Open access. Peer-reviewed. Real.

He picked up his phone and typed a message to Sarah: Thank you for publishing my story.

He set the phone down and looked out his window at the Queens skyline. The city lights were bright, but above them, beyond the atmosphere, the stars were still there.

For now.

He would keep watching. He would keep analyzing. Not for the government. Not for Congress. For himself.

Because someone had to count the fading stars.

And he was the one who was counting them.