A Score That Changes Overnight

Create up to 20 seconds from 0:00. Available for free opening episodes. Preparing a clip downloads this episode on your device.

The result had become forty-one while David slept.

He found it on Tuesday morning while the kettle in his hotel room was beginning to shake. The page looked so much like the one he had closed the previous evening that he first thought he had opened another participant's example. Then he checked the number above the score. His number. His submission identifier. Forty-one, adverse, awaiting validation.

The cost field still said Not determined.

He moved the kettle off its base before it boiled dry. There was water in it, but he could no longer remember how much. For a moment he stood holding it by the handle, performing the safe part of a morning which had suddenly stopped making sense.

The saved result showed sixty-four. He opened it beside the current page. Neither was a memory now. Both gave the same assignment code and the same submitted filenames. The later one had a new revision time, four minutes after midnight. Its explanation field was blank.

He downloaded it without replacing the earlier copy. Then he sent a question attaching both results and asking whether the change reflected a new assessment, a corrected display or a revision to the reference. He did not use the word rigged. It had appeared in his head immediately, which seemed a good reason not to put it in the first message.

At the assessment room, Priya was already waiting beside his desk. She did not look surprised when he told her.

"Mine moved too."

"Down?"

"Down. But that doesn't mean it moved for the same reason."

She had kept her own earlier result. She asked whether he had changed his submission after the deadline. He said no, then checked the actual files rather than expecting her to accept his certainty. His sent message held the original attachments. The copies he had saved the night before were the same. That established what he had delivered; it did not establish what Meridian had processed.

"They could have corrected a mistake," he said.

"Yes. They can tell us which one."

Elise arrived carrying a tablet and a paper list. The tablet made her look available to answer questions; the list made it clear she had somewhere else to go. David asked for five minutes and showed her the two results.

She said provisional scores could change during validation. The programme had never represented them as final. On that point he could not contradict her. The word provisional was present in both copies.

"Then what was validated between these two?"

"I don't have the reviewer notes on this screen."

"Can I have them when you do?"

She looked at Priya, who had not yet spoken. "Each participant needs to raise their own assessment queries. We can't discuss another person's result in a joint conversation."

Priya stepped back half a pace. "I'm asking about the general process. He can ask about his number."

Elise said the technical team would provide an explanation through the portal. She asked them not to disrupt the morning briefing. David wanted to say that a changed score was disrupting rather more than a briefing, but he could hear how that would sound across the room.

He sat down. Lena was setting a reminder on her phone. She glanced at his screen and then at her own, where the adverse label was also visible.

"Did yours do that before breakfast?" she asked quietly.

He nodded. Her mouth tightened, but she turned back to her work. She had a call at ten and could not spend the hour waiting for someone to become available.

The day's assignment required them to explain the sensitivity of their earlier model. David tried to work through it. Every assumption he touched now appeared to be an admission he did not understand. It was a bad way to analyse anything. He closed the result page and wrote the assignment question on paper.

What would change your conclusion?

It was the right question, even here. He listed three things that could alter his model and one thing the supplied data could not answer. When he reached the last item, he added a separate note: unexplained movement in assessment is not evidence that my model is correct.

That distinction cost him some of the comfort anger had provided.

At eleven thirty, a reply appeared. Validation had applied the current reference and standard adjustments. The result was provisional until the participant response period closed. A link offered an assessment query form with space for a single reason.

The choices were Input error, Incomplete submission and Other.

He selected Other and entered the two scores, the unchanged submission identifier and the question about the revision. The form warned him not to include confidential third-party material. He included only his own records. A receipt arrived with another number.

At lunch, he found Priya outside again, this time with a container she had brought from wherever she was staying. He asked what happened if every answer simply sent them to another form.

"Eventually we have a very complete record of asking," she said. "I'd prefer an answer."

She had noticed something he had not. The revised report's footer referred to a reference revision, but the field which should have named it was empty. The downloaded copy was the same. She showed him the empty field on her own report, keeping the score and her number off the screen.

David opened his. The footer had the same gap.

"That could be a template problem."

"It could. We should ask that too."

He felt absurdly disappointed by her refusal to build the larger accusation with him. Then she asked whether he wanted a reassuring theory or a question which could survive an answer, and he stopped feeling entitled to her agreement.

They wrote a short general question about the missing reference label. Each sent it separately, linked to their own report. They agreed not to circulate each other's files without asking. It was a small arrangement, hardly an alliance, but David found it easier to breathe once its limits were clear.

The afternoon briefing was delivered by a recorded presentation. It spoke about resilience after an adverse result. On the screen, a fictional analyst thanked a reviewer for helping him find a weakness. David watched the actor nod and wondered how much of the film had been made before Meridian knew who would be sitting in the room.

When the recording ended, Elise took questions about the next deadline. A young woman near the front asked whether the resilience exercise was mandatory. Her badge said Asha Lin. She had received an adverse label too, she said, but no explanation on which to base the reflection the exercise requested.

Elise offered to arrange a meeting. Asha asked whether she would receive the explanation before the reflection was due.

"We'll do our best."

The young woman looked down, then raised her head again. "Please record that I asked about the order."

David recognised the effort it took to finish that sentence in front of everyone. He did not applaud or turn toward the others. Making her question into a demonstration would have been another way of taking it away from her.

Later, near the lifts, Lena asked him whether he had found the original methodology. He said he had not. Samir Dunn, the man waiting beside her, said his account was showing a validation reserve against the second-stage assessment. He did not know whether it was merely a displayed estimate. It had no invoice number or due date.

They spoke in low voices until someone came out of the meeting room and the conversation broke apart. David disliked how quickly they had learned to look occupied.

He suggested talking outside the building the following morning. An ordinary café, before the first session. Bring only what each person wanted to discuss, no demand to expose their whole agreement. Lena said she could spare forty minutes. Samir asked the price of the coffee.

"Everyone buys their own," David said. "Nobody has to buy anything to talk."

That evening, Meridian sent a generic reminder that assessment materials should remain within approved channels. David read his participation letter again. He could not decide the exact limits of that sentence for himself or anyone else. He changed his invitation to make clear that they would discuss their own experiences first, without exchanging restricted materials.

Then he saved the invitation beside the two scores. One was a record of what had changed without his knowledge. The other was the first thing he had decided to change himself.