The first test of the new system arrived faster than anyone expected.
No warning period. No practice round. No easing into complexity.
At 08:00 sharp, every classroom screen across the second-year floor activated simultaneously.
A single line appeared.
INTERFERENCE EXAM: ACTIVE
Below it, a second line followed immediately.
OBJECTIVE: COMPLETE DECISION CHAINS UNDER DISRUPTION
Then the rules unfolded.
Slowly.
Deliberately.
As if the school wanted every student to feel each clause settle into their understanding like weight.
—
Each class would participate in a multi-phase academic exam spanning three subjects.
During each subject:
Leaders may send one interference signal per question set to another class.
Interference may alter, obscure, delay, or distort communication between opposing leader and their selected representative.
However—
Any interference misalignment resulting in incorrect self-class communication will trigger penalty inversion.
—
The final line appeared last.
Almost casually.
Victory condition: highest cumulative decision integrity score.
—
Silence spread through Class D.
Not confusion.
Not shock.
Something more structural.
A recalibration of assumptions.
Because now the system had clarified its intention:
This was not about knowledge.
Not even strategy in the traditional sense.
It was about coherence under pressure.
—
Sudō broke the silence first.
"…Decision integrity?"
He looked around.
"…What does that even mean?"
No one answered immediately.
Because even the strongest students were parsing the implication in real time.
—
Rei stood at the front of the room.
Her gaze fixed on the screen.
Her expression unchanged.
But internally—
the model had already updated.
This was not a variation of previous exams.
It was a meta-evaluation system.
The school was no longer testing answers.
It was testing whether answers survived transmission.
—
Horikita spoke quietly.
"…They're measuring consistency between intention and execution."
Rei nodded once.
"…Yes."
A pause.
"…And how interference degrades that consistency."
—
Kushida shifted slightly in her seat.
Not visibly distressed.
But attentive in a different way than before.
Less defensive.
More analytical.
—
Rei noticed it immediately.
Kushida was no longer reacting to identity threats.
She was analyzing structure as protection.
That change mattered.
It meant adaptation had already begun.
—
Horikita crossed her arms.
"…So interference isn't optional sabotage."
Rei responded.
"…It's a resource allocation mechanism."
A pause.
"…Limited influence with measurable consequences."
—
Silence followed.
Because that reframing made everything clearer.
And more dangerous.
—
Sudō leaned forward again.
"…So who do we target first?"
Rei glanced at him briefly.
"…No one."
That answer caused immediate confusion.
Even Horikita turned slightly.
"…Explain."
Rei remained focused on the screen.
"…Targeting requires assumption of linear advantage."
A pause.
"…There is no linear advantage here."
—
That statement shifted the atmosphere.
Because it implied something deeper.
This system could not be gamed in the traditional sense.
Only navigated.
—
Horikita lowered her voice.
"…Then what do we do?"
Rei answered immediately.
"…We stabilize internal decision chains."
A pause.
"…And accept external distortion as constant."
—
Kushida spoke quietly.
"…So we ignore interference?"
Rei turned slightly toward her.
"…We account for it, not resist it."
—
That distinction mattered.
Resisting implied loss minimization.
Accounting implied structural integration.
—
Horikita exhaled slowly.
"…Then leadership becomes filtering, not directing."
Rei nodded.
"…Correct."
—
A silence settled.
But it was different from previous silences.
This one carried alignment.
Not certainty.
But shared comprehension.
—
Then the system activated.
—
The first subject began: Advanced Mathematics.
Immediately, the classroom screens split into dual layers.
Left side: questions.
Right side: communication interface.
But now, a third layer appeared.
Interference feed.
A blank stream waiting to be activated.
—
Rei observed it carefully.
Because this was the first operational instance.
And the system always revealed its true structure only under execution.
—
Question one appeared.
Straightforward.
Linear algebra.
Kushida began processing immediately.
Her posture tightened slightly.
But she didn't hesitate.
—
Rei watched both layers simultaneously.
Answer trajectory forming on the left.
Communication signal forming on the right.
And interference capacity sitting dormant above it all.
—
Then the opposing class activated interference.
Class C.
Expected.
—
The communication interface flickered briefly.
Not broken.
Distorted.
A delay inserted between input and output.
Subtle.
But measurable.
—
Kushida paused.
Her eyes narrowed slightly.
A half-second hesitation formed.
That was enough.
—
Rei acted immediately.
Not by correcting.
By stabilizing.
"…Ignore latency," she said through communication.
Kushida blinked.
"…What?"
"…Your reasoning is intact. Only transmission is affected."
—
Silence.
Then Kushida exhaled once.
And continued.
Answer locked.
Correct.
—
Horikita observed quietly.
That intervention had not changed content.
It had changed perception of reliability.
That was critical.
—
Second question appeared.
More complex.
Multi-step derivation.
This time, interference came faster.
Not from Class C.
From Class A.
—
Sakayanagi.
—
The communication interface didn't flicker.
It fragmented.
Messages from Rei to Kushida became delayed inconsistently.
Words arrived out of sequence.
—
Kushida froze.
Not from confusion.
From disrupted temporal order.
—
Rei narrowed her eyes slightly.
This was precise.
Too precise.
Sakayanagi wasn't disrupting content.
She was disrupting temporal trust.
—
Horikita whispered.
"…She's breaking sequencing."
Rei nodded.
"…Yes."
A pause.
"…She's attacking cognition order."
—
Kushida's breathing changed slightly.
Her answer process began destabilizing.
Not because she lacked knowledge.
But because she could no longer trust instruction timing.
—
Rei adjusted immediately.
"…Kushida."
A pause.
"…Stop waiting for confirmation."
Kushida hesitated.
"…But I can't tell—"
"…You don't need to."
—
A beat.
"…Act on initial coherence."
Silence.
Then—
Kushida moved.
Answer submitted.
Correct.
—
Rei exhaled slightly.
Because Sakayanagi's interference was elegant.
But predictable in principle.
It exploited dependency on sequential validation.
Remove validation—
and the structure collapses.
—
Horikita noticed it too.
"…She forces dependency removal."
Rei nodded.
"…Yes."
A pause.
"…Which increases autonomy pressure."
—
That was the real battlefield.
Not accuracy.
Autonomy under disruption.
—
Question three appeared.
And now—
The interference escalated further.
—
This time, the signal came from Ryūen.
—
The interface darkened slightly.
Not delayed.
Not fragmented.
Contaminated.
False prompts appeared briefly inside Kushida's communication panel.
Incorrect "suggestions" mimicking Rei's tone.
—
A direct attack on identity recognition.
—
Kushida's eyes widened slightly.
"…There are multiple messages."
Rei understood instantly.
Ryūen was simulating leadership voice patterns.
Creating ambiguity of authority.
—
Horikita tensed.
"…That's dangerous."
Rei responded immediately.
"…Not if recognized."
—
Kushida hesitated again.
Confusion rising.
Authority dilution active.
—
Rei acted.
Not by overriding.
By anchoring.
"…Only follow structural logic," she said.
"…Ignore linguistic similarity."
—
A pause.
Kushida's breathing steadied slightly.
Then—
She separated content from source.
Answer formed.
Correct.
—
But Rei noted something important.
Her recovery time had increased.
That mattered.
—
Because increased recovery time meant cumulative fragility.
And fragility scaled exponentially under sustained pressure.
—
The exam continued.
Question after question.
Interference layers stacking.
Sakayanagi controlling time.
Ryūen controlling identity ambiguity.
Class C controlling latency.
—
And Class D—
stabilizing internally.
—
Horikita leaned slightly closer to Rei.
"…We can't sustain this for long."
Rei nodded.
"…Correct."
A pause.
"…So we don't."
—
Horikita frowned.
"…Then what?"
Rei's gaze sharpened slightly.
"…We transition phases."
—
Before Horikita could respond, Kushida spoke unexpectedly.
"…I can handle one layer of interference alone."
Silence.
Everyone turned toward her.
—
Her expression was different now.
Not fragile.
Not defensive.
Focused.
Controlled.
—
"…I can separate voices," she continued quietly.
"…If I stop trying to identify them emotionally."
—
Rei studied her carefully.
This was not denial of instability.
It was functional compartmentalization.
—
Interesting.
Very.
—
Rei nodded once.
"…Approved."
Horikita looked at her sharply.
"…You're assigning her solo processing?"
Rei responded immediately.
"…Yes."
A pause.
"…Because distributed interference requires distributed cognition."
—
Silence.
Then Horikita exhaled.
"…Understood."
—
Kushida activated solo mode.
Her communication interface shifted.
Now isolated.
No external guidance.
Only raw problem stream.
—
And immediately—
her stability increased.
Not emotionally.
Structurally.
—
Rei observed it carefully.
So her strength increased under isolation.
That was a key data point.
—
The exam progressed into its midpoint.
Scores fluctuating across classes.
No clear leader.
No stable advantage.
Just shifting integrity curves.
—
And then—
A new interference layer activated.
—
Ayanokōji.
—
Rei felt it immediately.
Not visually.
Conceptually.
Because unlike others—
his interference did not appear as disruption.
It appeared as absence of expectation.
—
Kushida paused mid-problem.
Her expression blanked slightly.
Not confusion.
Not delay.
Something subtler.
—
Rei narrowed her eyes.
He wasn't interfering with communication.
He was removing predictability of response outcome.
—
Horikita noticed too.
"…What is he doing?"
Rei answered slowly.
"…He's collapsing anticipation bias."
—
A pause.
"…He's making outcomes feel uncaused."
—
Silence.
Because that was difficult to counter.
Even structurally.
—
Kushida's answer rate slowed.
Not incorrect.
Just uncertain in selection threshold.
—
Rei evaluated rapidly.
Sakayanagi: sequencing disruption.
Ryūen: identity contamination.
Ayanokōji: causal opacity.
—
Three different attack vectors.
Three different cognitive layers.
—
And Class D was absorbing all of them simultaneously.
—
Horikita spoke quietly.
"…We're reaching limit."
Rei nodded.
"…Yes."
—
Then she made the decision.
"…We shift to internal synchronization protocol."
Horikita turned sharply.
"…Now?"
Rei answered immediately.
"…Yes."
—
Kushida looked up slightly.
"…What does that mean?"
Rei replied.
"…We stop receiving external instruction entirely."
A pause.
"…We solve as a closed system."
—
Silence.
Then realization spread.
—
Sudō whispered.
"…So no leader messages?"
Rei nodded.
"…Correct."
—
Horikita frowned.
"…But that increases individual burden."
Rei responded calmly.
"…It also eliminates interference vectors."
—
A pause.
"…Tradeoff is intentional."
—
Silence.
Then Horikita nodded slowly.
"…Do it."
—
Kushida deactivated external feed.
Sudō followed.
Then the rest of the class.
—
And suddenly—
the room became quieter.
Not externally.
Structurally.
—
The interference field had less to grab onto.
Fewer dependencies.
Fewer vulnerabilities.
—
Rei observed the system response.
Stabilization increase: 18%.
Interference reduction: partial.
Cognitive load redistribution: successful.
—
But she also saw the cost.
Individual uncertainty increased.
Slight hesitation growth across all members.
—
Still acceptable.
—
Because now—
they were no longer a signal system.
They were a processing unit.
—
And that changed everything.