And the guilt he'd carried because he hadn't acted sooner.
Now he was standing before other deputies, teaching a lesson he'd learned at enormous personal cost.
It didn't erase what happened.
But it mattered.
After the class, Bennett approached me.
"Director."
I nodded.
"Deputy."
He smiled nervously.
"How was it?"
I considered the question.
Then answered honestly.
"Useful."
He looked relieved.
I continued.
"I especially appreciated that you didn't treat intervention as something that requires a dramatic confrontation every time."
Bennett nodded.
"That's what I keep thinking about."
I waited.
He looked toward the empty classroom.
Then said, "If I'd asked Mercer to explain the stop sooner, maybe things would've been different."
I took a breath.
"Maybe."
He looked down.
I continued.
"We can't know exactly what would have happened."
He nodded.
"But you can teach other officers to recognize when a situation is moving in the wrong direction."
For several seconds, he was quiet.
Then he smiled.
"That's what I'm trying to do."
I nodded.
"Keep trying."
The sheriff's office changed slowly.
Not perfectly.
Not without resistance.
There were still complaints.
Still mistakes.
Still difficult encounters between officers and residents.
And still people who believed oversight was an attack on law enforcement rather than part of making it accountable.
But the department was no longer permitted to measure success simply by how few formal complaints appeared in its records.
The new system tracked how allegations were received, classified, investigated, and resolved. Reviewers examined whether evidence preservation requirements were met, whether supervisors documented their reasoning, and whether complainants received appropriate information about the process.
Training expanded beyond memorizing policies.
Officers discussed real situations involving unnecessary escalation, communication failures, and the difference between lawful authority and personal anger.
Supervisors were expected to evaluate patterns rather than treat each incident as completely unrelated to what came before.
And the county continued publishing aggregate information so the public could see whether the process was actually being used.
Some of the numbers increased.
The first year under the revised system showed more formal complaints than the previous year.
A few critics described that as evidence the department was getting worse.
But the comparison required context.
The earlier system had improperly diverted some complaints into informal categories.
Under the new procedure, more allegations were being recorded as allegations.
That was not automatically proof of increasing misconduct.
It was evidence that the system had begun counting things it previously failed to count.
I explained that distinction repeatedly.
At public meetings.
In written reports.