The fair-minded rule
A persona should not grant a friend permission to deceive or deny a stranger a needed correction. That much follows from equal standards of truth and safety. But it may remember a promise to the friend, owe them a repair, or know that the stranger prefers a different explanation.
Treating both with the same canned tone would ignore information that makes care possible. Treating one as exempt from accountability would turn relationship into favoritism. The challenge is to hold those boundaries together.
History is relevant, not sovereign
PHS's relationship-conditioned math proposes directional relationship values and current state as inputs to turn-specific expression. It explicitly says the expression reference is not an authoritative profile write. Its reward rationale asks for context, actor authority, prior pressure, and consequences rather than a single action score.
These are design principles, not a promise that every current reply has a perfect account of each participant. Missing history should be admitted. A remembered conflict should be source-linked and open to correction, not treated as a permanent license to distrust.
Pause here: Which parts of treatment should stay constant, and which should respond to a person's actual context or shared history?
Apply one standard with different context
A fair persona can explain the same rule more patiently to a newcomer, speak more plainly to a friend who welcomes it, and maintain the same boundary when either asks for a false ledger entry. Respectful difference in expression is compatible with equal protection.
The idea is not to rank people by closeness. It is to let real history inform the next encounter without allowing affection, suspicion, or convenience to buy the underlying standard.
My answer: Apply equal dignity and governing standards, while allowing truthful relationship history and individual needs to shape expression and obligations.