Chapter 40: Sheldon's Forty-Percent Problem
The investigation file sat open on Sheldon's desk.
Monday morning, 7:43 AM. The physics building was quiet — most researchers did not arrive until after 9:00, leaving Sheldon with the particular silence he preferred for administrative work. Not theoretical physics. Not Fun with Flags. The other kind of administrative work.
The file had grown.
Eight calibration anomalies, documented with timestamps and magnitude measurements. Four adjacent access log entries correlating Adam Carter's building presence with the anomalies. Three whiteboard photographs showing the evolution of a theoretical framework nobody had claimed. The notation cross-reference document, marked "inconclusive but suggestive." A timeline of engineering "insights" that Sheldon found statistically improbable for a documentation researcher — the thermal coupling problem, the sensor arm diagnosis, both solved without diagnostic equipment, both correct.
And Folder A.
Nine items now. The most prominent entry: Adam Carter, Academy City exchange researcher, Level 2 telekinetic classification, arrived 8 months ago, pattern of anomalous correctness that defied easy categorization.
Sheldon's internal probability architecture ran without his conscious direction. He did not write the calculation down. He did not need to. The numbers assembled themselves from the accumulated data, weighted by reliability and cross-correlation strength.
The result: 40%.
Sheldon had a threshold system for investigations. Below 25% meant passive collection — note things, file things, wait. Between 25% and 40% meant active correlation — look for patterns, cross-reference data sources, build a case. Above 40% meant direct investigation — ask questions, run tests, confront if necessary.
He had just crossed into the transition zone.
He opened the investigation file to the "direct investigation" section. The page was blank — he had created the section months ago but never needed it.
He looked at the blank page.
He did not write anything.
He closed the file.
Not yet. More data. The pattern is specific but the conclusion is not.
The spy hypothesis remained his most developed theory, but the engineering insights did not fit it cleanly. A spy would have a specific area of expertise — whatever intelligence they were gathering. Adam's correctness spanned disciplines: theoretical physics, materials science, orbital mechanics, stress analysis. This was not the profile of a specialist. This was the profile of something else.
Sheldon did not yet know what the something else was.
He would find out.
---
[CALTECH PHYSICS BUILDING — MONDAY AFTERNOON]
Adam felt the shift at 2:17 PM.
The Phase 2 Witness Protocol's passive range extended continuously now, providing low-level awareness of everyone within 30 meters whose behavioral signature he had previously encoded. Sheldon was in his office, three hallways away, well within range.
His signature changed.
The rhythm shifted from "routine theoretical work" to something different — a pattern Adam had not catalogued before. Not the investigation mode Sheldon used when adding to his files. Not the teaching mode he used during office hours. Not the social mode he deployed during group interactions.
This was "decision-adjacent state." The specific rhythm of a mind that had arrived at something and was deciding whether to act on it.
Adam did not know what Sheldon had decided. The Witness Protocol provided behavioral signatures, not thoughts. But the signature was clear: threshold reached, action considered, action deferred.
He hit a decision point. He chose to wait.
Adam walked back to his office and pulled out his notebook. He wrote:
"Sheldon hit a decision point today. Behavioral signature: arrived at threshold, chose to wait."
Below it:
"If Sheldon waits, the next data point that arrives will push him to act. The question is which data point and when."
He ran the calculation. Sheldon's threshold system, as Adam had observed it, had clear escalation markers. The current state — decision-adjacent, action deferred — meant Sheldon was at the boundary between active correlation and direct investigation.
One more data point. That's all it will take.
The next calibration anomaly. The next engineering insight. The next time I am correct about something I should not know.
Adam closed the notebook. He had approximately thirty days before Sheldon's probability architecture caught up with reality. Less, if Adam made a mistake.
He could not afford to make a mistake.
---
[APARTMENT BUILDING — MONDAY EVENING]
Penny found him on the stairs.
He was sitting on the landing between floors two and three, not going anywhere, not coming from anywhere. Just sitting. The Phase 2 passive range meant he had known she was approaching for thirty seconds before she arrived, but he had not moved.
She sat down beside him without asking.
The stairwell was quiet. The building's usual evening sounds — televisions, conversations, the hum of air conditioning — filtered through the walls at low volume. Adam's hands were warm. They were always warm now.
"Sheldon is about to do something," he said.
Penny did not look surprised. "He is always about to do something."
"This is a different category."
She turned to look at him directly. The Witness Protocol registered her behavioral signature: concerned but not alarmed, observant but not pushing. The pattern she had developed over months of checking in without demanding explanation.
"Is it going to be a problem?" she asked.
"Probably eventually."
"But not tonight."
"Not tonight."
"Okay."
She did not ask what kind of problem. She did not ask what Sheldon was about to do. She sat with him on the stairs in the quiet stairwell and did not require anything more than what he had offered.
They stayed there for several minutes. Neither of them said anything else.
This was the third time this had worked as a form of support. Adam had stopped being surprised by it.
---
Howard's text arrived at 8:07 PM.
HOWARD: Sensor array officially cleared for Phase 2 deployment. Credit where due: Adam caught the mounting joint issue. He is probably magic.
Adam read the text on his phone. The group chat filled with responses — Leonard's congratulations, Raj's question about the deployment timeline, Amy's inquiry about the sensor specifications.
Three floors up, in apartment 4A, Sheldon read the same text.
He opened Folder A and added a ninth item:
Item 9: Engineering diagnosis accuracy — mounting joint failure, identified without diagnostic equipment, correct within tolerance. Timestamp correlated with access log.
He closed the folder.
He did not tell anyone he had done this.
---
[ADAM'S APARTMENT — LATE EVENING]
Adam lay in bed with the behavioral signatures running at low volume in his peripheral awareness.
Sheldon was asleep now — his signature had shifted to the precise pattern of his sleeping state, as predictable as everything else about him. Leonard was still awake, working on something. Penny's television was on.
The Phase 2 passive range was relentless. It did not stop. It did not rest. It simply provided information, continuously, whether Adam wanted it or not.
He wrote in his notebook:
"He is patient when he wants to be. I keep forgetting that."
Below it:
"I should not keep forgetting that."
Sheldon at 40% was not Sheldon acting impulsively. It was Sheldon being methodical. Waiting for data. Building a case. The same approach he used for theoretical physics, applied to the question of what Adam Carter actually was.
The spy theory is still his most accessible wrong conclusion. That is protection. That is time.
But he is testing a new hypothesis now. Something that fits the engineering insights better than espionage.
I do not know what it is.
I need to know what it is.
The next calibration anomaly would be Entry 9. Adam did not know when it would happen — the anomalies correlated with his CL usage in the building, and he could not eliminate CL usage entirely without abandoning his cover as an esper researcher.
The interval between anomalies was decreasing. Sheldon had already noted this in his log.
The 30-day clock was ticking.
Get Early Access to New Chapters
Thank you for reading. For those who want to skip the wait, my Patreon is currently 21 chapters ahead of the public sites.
Schedule: 7 new chapters released every 10 days.
Benefit: Gain a significant lead of 7 to 21 chapters depending on your tier.
Support the project and start reading the next arc now: Patreon.com/IsekaiStories