The conversation began, as conversations with Harlan often did, at dinner and continued into the library afterward, one of those evenings where the initial subject was adequate and the real subject revealed itself only gradually, the way depth in water reveals itself as you wade further in.
The initial subject was historical: a case from three hundred years ago, recorded in one of the books Lira had been working through, in which a provincial governor had been poisoned by his own physician in order to prevent a genocidal military campaign. Harlan had mentioned it in passing; Lira had asked him to stop and return to it, in the way she asked him to return to things that were interesting.
"He broke the law," she said. "The governor was acting legally. His physician had no official authority to stop him."
"No," Harlan said. "But the governor's legal actions were causing mass harm, and the physician had access and capability, and determined that —"
"That his private judgment was more reliable than the legal structure," Lira said. "Which might have been correct in that specific instance. But the precedent."
"The precedent is dangerous," Harlan agreed. "Any system that licenses private judgment over legal structure can justify anything, if the judgment is sufficiently certain of itself."
"So what separates the physician from — anyone who has decided their private judgment is correct?"
Harlan looked at her. She was fifteen, eating her dinner, asking this question with the focused attention that he had come to understand meant she was not asking academically.
"Accountability," he said. "The physician's action was documented. He made no attempt to conceal it. He presented himself for judgment by the legal structure he had circumvented. He did not claim immunity from consequences." A pause. "He believed his action was justified. He did not believe it was therefore exempt from scrutiny."
Lira was quiet for a while. She moved to the library; he followed, because the conversation was not done.
She told him, sitting in the library chairs by the fire with her journal on her knee, that she was building a framework. Not for decisions she planned to make tomorrow — for decisions she would make eventually, in circumstances she couldn't fully predict, and which she needed to have thought through before she was in them rather than during.
"Because during is the wrong time," she said.
"Yes," he said. "During is when the thinking is contaminated."
She had three principles. She read them to him from the journal's back pages, where she had written them in the clear hand she used for things she wanted to return to rather than the compressed operational hand.
First: the target must be actively causing harm that cannot otherwise be stopped. Not merely be opposed to what you want, not merely be inconvenient, but actually causing harm that has no other available remedy. The standard is high because it should be high. If it weren't high, it would be permission for anything.
Second: all non-lethal options must be genuinely considered. Not nominally considered, not considered for the form of the thing while already decided. Genuinely — meaning with real openness to the conclusion that a non-lethal option is sufficient, even if the lethal option is available and easier.
Third: the action must serve a purpose beyond personal satisfaction. Revenge, by itself, does not qualify. The action must produce a consequence that benefits people beyond the actor — must remove harm from those who have been experiencing it, must prevent future harm, must do something that the actor's personal feeling of justice does not, by itself, constitute.
She read them out and looked up.
Harlan was quiet for a moment. Then he said: "These are not easy standards."
"They're not supposed to be," she said.
"The second one," he said. "How do you verify that you've genuinely considered the alternatives? The human mind is very good at the appearance of deliberation while moving toward a conclusion it has already reached."
"You write it down," she said. "Before you decide. You write down every alternative, in detail, with honest assessment of whether it would work. Then you read what you wrote and you try to find the places where the assessment was dishonest." She paused. "It doesn't guarantee correct judgment. But it creates a record that you can examine afterward. And the process of creating the record makes the dishonest assessment more visible, because dishonesty in written form is harder to sustain than dishonesty in thought."
He looked at her for a long time. "Your mother would have liked this conversation," he said.
She absorbed this. She did not cry, which was not because she was unmoved but because she had learned to receive her grief as information rather than as something that required expression. She sat with it for a moment and then returned to the journal.
"Mira's response was interesting," she said.
"What did she say?"
"She said most people who do this work never think to write it down. Then she asked if the writing would change what I was willing to do. I said no." She looked at the principles on the page. "It might change how, but not whether."
"That's an important distinction," Harlan said.
"Yes," she said. "I know. That's why I thought about it for three weeks before I concluded it."
The principles became the back-page document she returned to whenever she was working through a specific assessment. Not as a checklist — a checklist implies that the process can be completed mechanically, which was exactly what she was trying to prevent. More as a calibration: a set of questions that the specific case had to survive before she moved to action.
What she found, over the following months, was that the framework did not prevent action. It made action more deliberate. It made her more confident in the decisions she reached through it, precisely because the process had been rigorous enough that the confidence was earned rather than assumed.
It also made her more patient with the cases that didn't survive the framework — the situations where her instinct said act and the framework said not yet, not on these grounds. There were several of these in the months of her remaining training, and each time she sat with the framework's conclusion and did not act, she experienced the specific discomfort of suppressed instinct, and each time the discomfort passed and she found that the decision to wait had been correct.
She filed this as the framework's most important quality: not what it authorized, but what it withheld.
Mira found this, when Lira reported it, entirely characteristic. "Most people build rules to permit the things they want to do," she said. "You built rules to stop you from doing the things you want to do. That's backwards." She paused. "It's also correct."
Lira closed the journal. "I know," she said.