What makes being pissed off at an agent who treats me unfairly useful is not that the agent doesn’t want me to be pissed off. In fact, I can sometimes be usefully pissed off at an unfair agent that is entirely indifferent to, or even unaware of, my existence. In much the the same way, I can sometimes be usefully pissed off at a non-agent that behaves in ways that I would classify as “unfair” if an agent behaved that way.
Admittedly, asking when it’s useful to classify something as “unfair” is different from asking what things are in fact unfair.
On the other hand, in practice the first of those seems most relevant to actual human behavior. The second seems to pretty quickly lead to either the answer “everything” (all processes result in output distributions that are not evenly distributed across some metric) or “nothing” (all processes are equally constrained and specified by physical law) and neither of those answers seems terribly relevant to what anyone means by the question.
What makes being pissed off at an agent who treats me unfairly useful is not that the agent doesn’t want me to be pissed off. In fact, I can sometimes be usefully pissed off at an unfair agent that is entirely indifferent to, or even unaware of, my existence. In much the the same way, I can sometimes be usefully pissed off at a non-agent that behaves in ways that I would classify as “unfair” if an agent behaved that way.
Admittedly, asking when it’s useful to classify something as “unfair” is different from asking what things are in fact unfair.
On the other hand, in practice the first of those seems most relevant to actual human behavior. The second seems to pretty quickly lead to either the answer “everything” (all processes result in output distributions that are not evenly distributed across some metric) or “nothing” (all processes are equally constrained and specified by physical law) and neither of those answers seems terribly relevant to what anyone means by the question.