Monday noon.
A new line appeared on the bulletin board.
— Anomaly Correction Class List Released.
No explanation.
No definition.
Only student numbers.
Third Year · No. 24.
Second Year · No. 15.
Fourth Year · No. 17.
The list was posted beside the timetable.
No bold text.
No red frame.
Yet it stood out more than any disciplinary notice.
Above it, the ability training schedule remained unchanged.
Monday to Friday afternoons: Ability Training.
Saturday morning: Ability Training.
The handwriting was precise.
The reformed curriculum had been in effect for a full year.
No further adjustments to frequency.
No changes to rhythm.
No variation in intensity.
Time had been cut cleanly.
Stable. Structured.
—
Third Year: 100 students.
No. 24 stood in the middle.
Neither strongest.
Nor weakest.
He had two recorded peaks.
One during combat suppression, exceeding average output.
One in a simulated pressure field, breaking his limit.
Both recorded.
Neither a violation.
—
Second Year: 150 students.
No. 15 ranked near the top.
Consistently stable performance.
His interference ability ranked among the most efficient in the grade.
But his fluctuation value was high.
The system was more sensitive to "fluctuation" than to "victory."
—
Fourth Year: 80 students.
No. 17 was in the upper tier.
Combat evaluation: never below standard.
Win rate: stable.
Error rate: low.
Yet his long-term trend was a flat line.
No rise.
No drop.
No breakthrough.
No risk.
No memory point.
—
No one knew the selection criteria.
More people lingered in front of the board than usual.
No one spoke.
Eyes paused briefly.
Numbers registered.
Then moved away.
Like checking the weather.
Knowing it would rain—
but never asking where the clouds came from.
—
Monday afternoon.
Ability training proceeded as usual.
Field lines were clear.
Sensor lights activated in sync.
Evaluation screens hung above the arena.
Every release was recorded.
The rhythm remained.
—
No. 24 stepped onto the field.
Opponent: Third Year · No. 37.
Fast initiation.
Compressed air formed visible distortion.
A clear surge peak.
Judgment light activated.
Scores displayed in real time.
Values fluctuated.
Curve rising.
No violation.
No boundary breach.
Only—
Above average range.
No applause from the viewing area.
A prompt appeared below the screen:
Fluctuation amplitude: beyond predicted range.
No. 24 slowed by half a second during convergence.
The system recorded that half-second.
—
No. 15's combat style leaned toward control.
He specialized in disrupting judgment—
forcing opponents into wrong decisions within half a second.
Not offense.
Rhythm manipulation.
In the previous match, he had successfully interfered three times in a row.
That night, his score was lowered.
Reason:
Risk fluctuation.
Excessive interference frequency.
Unpredictable judgment deviation.
Today, he reduced his triggers.
Twice, the opponent should have failed.
He let it pass.
Score: Stable.
—
No. 17's curve was nearly flat.
No bursts.
No errors.
Clean execution.
Stable efficiency.
One term repeated in evaluations:
Replaceable.
His win rate was sufficient.
But he had never once exceeded system expectations.
—
Tuesday.
Basement Level 1 of the laboratory building opened.
Cold lighting.
Electronic signage lit up.
Anomaly Correction Class · Phase One.
No class designation.
No grade division.
No seat numbers.
Three stood at the entrance.
No. 24.
No. 15.
No. 17.
No exchange between them.
The evaluator stood behind the podium.
No name.
Only a code.
The projector activated.
"Your problem is not your ability."
First sentence.
"It is fluctuation."
—
No. 24's records filled the screen.
Ability curve enlarged.
Sharp peak.
Delayed convergence.
System annotation:
Unpredictable.
Prediction model deviation rising.
Risk weight increasing.
No. 24 watched.
No argument.
No explanation.
—
Next: No. 15.
Combat footage played.
Opponent's movement abruptly shifted.
Rhythm misaligned.
Outcome already decided.
Paused frame.
Zoom.
"The opponent did not realize their judgment was interfered with."
"This type of ability, in a stable environment, creates structural risk."
A line appeared:
Threat index rising.
Prediction success rate decreasing.
"The system cannot lock your trigger timing in advance."
No. 15 said nothing.
His fingers tightened slightly.
—
Third: No. 17.
A straight line.
Stable to the point of mechanical.
The evaluator displayed long-term trends.
A horizontal line.
"No rise."
"No drop."
"Stability means manageability."
"But not necessity."
"No trend means no added value."
Silence.
For the first time—
No. 17 understood.
Stability was not safety.
—
The first session was not training.
It was dismantling.
They received printed reports.
Every fluctuation point listed.
Every peak.
Every anomaly.
And one blank column:
Control Strategy.
Only one task.
Write a control plan.
How to suppress.
How to converge early.
How to avoid peaks.
How to become predictable.
Analysis is control.
The moment a person begins to justify their fluctuation—
they have accepted the framework.
Signature required.
Signature meant agreement.
—
Wednesday afternoon.
Training continued.
No. 24 visibly restrained.
Ending bursts early.
Cutting output before peak.
Score: Safe.
Prediction error decreased.
—
No. 15 reduced interference intensity.
Only baseline disruption remained.
Fewer opponent mistakes.
Score: Stable.
Prediction success increased.
—
No. 17 remained flat.
Evaluation:
No trend change.
Maintained.
—
Underground system panel updated.
Third Year fluctuation rate decreased.
Second Year prediction success increased.
Fourth Year risk weight lowered.
Recovery rate slowly rising.
Target population decreasing.
—
Thursday.
Second session of the class.
Anonymous case displayed.
Fifth Year.
ID undisclosed.
Three consecutive breakthroughs.
Evaluation curve surged sharply upward.
Then—
File marked:
Danger assessment error.
Status: Closed.
No explanation.
The screen remained.
Silence filled the room.
Cold white air.
"Danger is not strength," the evaluator said.
"Danger is unpredictability."
"When the model fails—
the system eliminates the variable first."
—
No. 15 looked at the words.
For the first time, he spoke.
"What if the prediction itself is wrong?"
The room grew colder.
No response.
Only a record.
Entry added:
Questioning tendency.
—
Friday afternoon.
The training field felt quieter.
Third Year peaks reduced.
Second Year interference decreased.
Fourth Year curves smoother.
Shorter matches.
Faster conclusions.
No violent fluctuation.
No boundary breaches.
No high-score spikes in the viewing area.
Fewer system alerts.
Prediction success nearing stability.
—
No. 17 did not leave after training.
He stood by the stands.
Watching the field.
Clear lines.
Defined boundaries.
Everything within predictable range.
No surprises.
No risk.
No breakthroughs.
Then he realized—
He had become part of that flat line.
—
Saturday morning.
Training continued.
Same rhythm.
Stable scores.
Fewer alerts.
The model ran smoothly.
—
Underground statistics updated again.
Threat index decreased.
Anomalous fluctuation individuals reduced.
Curves flattened.
Prediction error minimized.
—
The class expanded.
Second batch released.
No explanation.
Only numbers.
More students.
Even distribution across grades.
Lower grades began suppressing peaks voluntarily.
Early convergence became default behavior.
More conservative strategies in combat.
—
In one match, No. 15 could have secured a lock.
He paused.
Abandoned it.
Opponent scored.
Evaluation: Stable.
System note:
Risk decreased.
—
No. 24 reduced output in high-intensity combat.
Win rate decreased.
Risk index decreased.
Prediction accuracy improved.
—
No. 17 remained unchanged.
System evaluation updated:
Maintained.
No risk.
No trend.
—
No one was publicly punished.
No one was taken away.
No isolation notices.
The class was just another course.
Another method.
Another report.
—
But the numbers on the system panel changed.
Peaks disappeared.
Fluctuations suppressed.
Prediction success increased.
The system became more stable.
—
The academy did not punish failure.
Nor reward bursts.
It only recorded.
Compared.
Flattened.
What was erased—slowly—
was not those who made mistakes.
But those who could not be fitted into trends.
—
Anomaly correction was not treatment.
It was adjustment.
Voluntary convergence.
Voluntary suppression.
Voluntary integration into the system.
—
When all curves flatten—
when all peaks are cut in advance—
the academy appears safer.
More stable.
More predictable.
—
And the result of reform—
is not strength.
It is control.
—
The class had no slogans.
No banners.
Only one invisible rule:
Analyze yourself.
Suppress yourself.
Enter the system.
—
The weekend ended.
Monday returned.
Training continued.
Same rhythm.
—
The bulletin board was clean.
The list replaced.
No one remembered the first batch.
Only the data remained underground.
Recovery rate rising.
Threat index falling.
Prediction model stable.
The system continued running.
—
The academy had not changed its appearance.
Only the curves—
were becoming flatter.