Leaders needed discretion.
But discretion without review became a hiding place.
They redesigned several processes.
Negative career consequences within six months of protected complaints required an independent second-level review.
Not prohibition.
Review.
School removals needed documented reasons.
Counseling language could still address professionalism.
But vague phrases such as “organizational loyalty” required concrete behavioral examples.
Complaint subjects could not control complaint categorization.
Simple changes.
Not perfect.
Harlan objected through written comments.
Even removed from the pilot committee, he remained a colonel with influence.
He warned that excessive review could make leaders timid.
Ava agreed partly.
That surprised Miriam.
“You agree with Harlan?”
“On one risk.”
Miriam raised an eyebrow.
“Systems can overcorrect.”
“If leaders need permission for every ordinary decision, they stop leading.”
“So?”
“We build review around high-risk decisions.”
“Not everything.”
Balance.
Again.
Mia Brooks later attended a policy session.
She challenged Ava.
“You're still designing rules for good leaders.”
Ava frowned.
“What do you mean?”
“Good leaders hate abusing discretion.”
“Bad leaders don't.”
“So every loophole becomes their favorite tool.”
Ava leaned back.
“Then what do you suggest?”
“Make patterns visible.”
That became the next principle.
One denied school might be reasonable.
Five complaint-filers losing schools?
Pattern.
One delayed transfer?
Maybe.
Six?
Pattern.
The system needed to detect clustering.
Not assume guilt.
Trigger review.
Miriam smiled after Mia left.
“You're collecting people who disagree with you.”
“Better than collecting people who don't.”
“That sounded like your mother.”
“Don't.”
Miriam laughed.
Months later, the six-soldier review closed.
Three records corrected.
Two school reconsiderations.
One financial back-pay issue fixed.
Two former soldiers declined further action.
One case could not be fully reconstructed.
The final report did not call the remediation complete.
Ava insisted on different wording:
Administrative correction completed where possible. Individual harm may extend beyond available remedy.
Some leaders disliked the sentence.
Too negative.
Ava kept it.
If the institution could not restore what was lost, it should not pretend paperwork made the person whole.
Colonel Jacob Harlan requested a formal hearing.
Ava expected him to deny everything.
He didn't.
That made the hearing more difficult.
Harlan admitted he withheld the true target of the investigation.
Admitted he added an unauthorized secondary evaluation of Ava.
Admitted he overrode concerns about intervention thresholds.
Admitted he believed the operation required “controlled exposure.”
His defense was not factual.
It was philosophical.
“Without covert observation,” he said, “we would never have exposed Voss.”
The review panel remained silent.
Harlan continued.
“Traditional systems failed.”
“Surveys were clean.”
“Complaints vanished.”
“Records were manipulated.”
“Kael's report alone was insufficient.”
“We needed natural behavior.”
Ava sat in the back.
Not a witness yet.
Listening.
Harlan looked composed.
Reasonable.
That was his strength.
He was not Voss.
He did not retaliate for loyalty.
He believed deeply in institutional oversight.
He simply believed mission urgency allowed him to control how much others knew about the risks they were taking.
The panel asked:
“Why add Captain Mercer as a secondary subject?”
Harlan answered:
“Because concealed authority changes people.”
“Was that approved?”
“No.”
“Why not request approval?”
“Because approval would have expanded knowledge of the test.”
“Meaning?”
“More contamination.”
The panel chair leaned forward.
“So secrecy became self-justifying.”
Harlan paused.
“No.”
But the answer sounded weaker.
Ava testified next.
Harlan watched her.
No hostility.
Almost curiosity.
She described the field operation.
Kael.
Mara.
Intervention concerns.
The tray incident.
Then the moment Harlan referred to soldiers as the operating environment.
One panel member asked:
“Do you believe Colonel Harlan intended harm?”
“No.”
“Do you believe he acted maliciously?”
“No.”
“Do you believe his methods were unethical?”
“Yes.”
“Why?”
“Because he treated informed people as a threat to valid evidence rather than as participants with limits he was required to respect.”
Harlan's jaw tightened slightly.
The panel asked:
“Would telling you Voss was the subject have contaminated your observations?”
“Probably.”
“Then isn't Harlan's logic sound?”
“Partly.”
That answer surprised the room.
Ava continued.
“Not telling us the target can be justified.”
“Not telling us we were also being evaluated for our careers cannot.”
“Those are different layers of secrecy.”
“Why?”
“One protects the integrity of the investigation.”
“The other gives a supervisor hidden power over subordinates.”
Harlan finally spoke directly.
“Captain, if I had told you, would you have behaved differently?”
“Yes.”
He looked almost triumphant.
“Then my concern was correct.”
“No.”
Ava met his eyes.
“I would have consented differently.”
Silence.
That distinction mattered.
Behavioral contamination was not the only value in the room.
People were not data collection instruments.
The panel asked Ava one final question.
“Do you believe the pilot should have been terminated?”
Ava thought carefully.
“The version Harlan controlled?”
“Yes.”
“Could a redesigned version ever be ethical?”
Ava hesitated.
“Yes.”
Harlan looked toward her.
Surprised.
She continued.
“If intervention thresholds are clear.”
“If evaluator conduct is independently auditable.”
“If secondary career assessments require explicit authorization and consent.”
“If the people running the program cannot disable the safeguards.”