Debts of Betrayal

Chapter Four

It opened.

Blank.

She clicked another. Blank. Another. Blank. She was clicking faster now, folder after folder, document after document, the cursor moving across a screen full of files that looked correct and were completely empty and the architecture of her work without any of the substance, the bones of four years without any of the flesh.

Someone had gone through every file. Every single one. Opened them, stripped the content, left the shell. Left it looking intact so she would not notice immediately. So she would find out at the worst possible moment, when she needed it most, when she had already told a journalist she had documentation and a committee she could prove everything.

They had not deleted her work.

They had hollowed it out.

She sat on the cold stone step outside the home she could no longer enter and looked at the screen full of empty files and felt something happen in the centre of her chest that was quieter than breaking and more permanent. Not grief. Not panic. Something slower and colder and more total than either.

She had nothing.

No home. No access. No documentation. No files. Her name was in fourteen articles tonight with one word beneath it and she had no proof to put against it, nothing to hold up and say here, look, this is the truth.

Her phone buzzed.

She looked at the screen.

A text from a number she had never seen before.

"Give up, Serena. It's already done."

She did not sleep on the step.

The cold would not allow it. It settled into her bones the way cold does when it finds a person who has already been emptied out. Not just the skin, but deeper, into the places where warmth is supposed to live. She sat there until the fourth floor light went out and then she sat there a little longer because getting up meant accepting that there was nowhere to go.

Eventually she picked up the laptop bag and the lighter of the two bin bags and walked.

The diner was four streets away. Twenty four hours. She had been there once before on the best night of her life. The night the first working model ran without error, when she had walked out of the lab at midnight too alive to go home and sat at the counter and eaten eggs she barely tasted with the feeling that everything she had worked for was finally within reach.

She pushed the door open and the warmth came at her face like something she did not deserve.

The man behind the counter looked at the dress and the bag and the torn shoulder and made the quiet professional decision that none of it was his business.

"Coffee?" he said.

"Please," she said.

She took the booth at the back, opened the laptop, and searched her own name.

Twenty six articles.

She counted them. Read every headline. Then she went further. She opened each piece and read the full text, because she was not here to feel. She was here to think. And something in the coverage was pulling at the edge of her attention, a small persistent feeling like a word on the tip of the tongue that will not come.

She found it in the fourth article.

It was a longer piece, the kind that had been written with access to documents, with supporting material, with what appeared to be an official research summary submitted under Celeste's name. The journalist had reproduced sections of it. Quoted from it directly. Presented it as the foundational record of the work that had just won the Halloway Prize.

Serena read it once quickly.

Then she read it again slowly.

Then she sat very still in that booth with her cold coffee and her bin bag under the table and felt something sharpen in her chest that had nothing to do with grief.

There was a mistake.

Not a spelling error. Not a formatting inconsistency. A technical mistake. A description of the interface's signal calibration process that was almost correct and precisely wrong in a way that only meant one thing. Whoever had written that summary understood the outcome of the calibration. They knew what it produced. They had the final result and they had worked backwards from it, filling in the process with logic and inference and educated guesswork.

But they had not been there when it happened.

They had not been in the lab at three in the morning on a Wednesday in February eighteen months ago when the calibration sequence failed for the seventh consecutive time and Serena had sat on the floor for twenty minutes and then stood up and made a decision to change the input sequencing entirely. A decision that should not have worked according to every model she had built, a decision she had made on instinct and exhaustion and four years of knowing this thing from the inside out.

That decision was the reason the interface worked the way it did.

And the summary described it wrong.

The kind of wrong that a committee of reviewers, none of whom had built the system themselves, would read and accept without question.

But Serena had built it.

And she could see every seam.

She opened a new document and started writing.

She wrote from memory and she wrote fast, the way she wrote when the thinking was coming faster than the typing could keep up with. She laid out the actual calibration process the real sequence, the failed attempts, the specific decision that changed everything and why it worked when the models said it shouldn't.

Then she went back to the article and pulled out Celeste's version and placed it beside hers and wrote the comparison with the precision of a surgeon. Point by point. The claim versus the reality. The summary versus the actual development record. She showed exactly where the language slipped. Not in the conclusion, which was correct, but in the journey, which had been guessed at by someone who only ever held the destination in their hands and never walked the road.