The problem is not with your actions, but with the actions of all the others (who are following the same general kind of utility function but because your utility function is agent-relative, they use different variables, i.e. they care primarily about their family and friend as opposed to yours). However, I was in fact wondering whether this problem disappears if we make the agents timeless (or whatever does the job), so they would cooperate with each other to avoid the suboptimal outcome. This is seems fair enough since acting “perfectly moral” seems to imply the best decision theory.
Does this solve the problem? I think not; we could tweak the thought experiment further to account for it: we could imagine that due to empirical circumstances, such cooperation is prohibited. Let’s assume that the agents lack the knowledge that the other agents are timeless. Is this an unfair addendum to the scenario? I don’t see why, because given the empirical situation (which seems perfectly logically possible) the agents find themselves in, the moral algorithm they collectively follow may still lead to results that are suboptimal for everyone concerned.
No, but you need some decision theory to go with your utility function, and I was considering the possibility that Parfit merely pointed out a flaw of CDT and not a flaw of common sense morality. However, given that we can still think of situations where common sense morality (no matter the decision theory) executed by everyone does predictably worse for everyone concerned than some other theory, Parfit’s objection still stands.
(Incidentally, I suspect that there could be situations where modifying your utility function is a way to solve a prisoner’s dilemma, but that wasn’t what I meant here.)
The problem is not with your actions, but with the actions of all the others (who are following the same general kind of utility function but because your utility function is agent-relative, they use different variables, i.e. they care primarily about their family and friend as opposed to yours). However, I was in fact wondering whether this problem disappears if we make the agents timeless (or whatever does the job), so they would cooperate with each other to avoid the suboptimal outcome. This is seems fair enough since acting “perfectly moral” seems to imply the best decision theory.
Does this solve the problem? I think not; we could tweak the thought experiment further to account for it: we could imagine that due to empirical circumstances, such cooperation is prohibited. Let’s assume that the agents lack the knowledge that the other agents are timeless. Is this an unfair addendum to the scenario? I don’t see why, because given the empirical situation (which seems perfectly logically possible) the agents find themselves in, the moral algorithm they collectively follow may still lead to results that are suboptimal for everyone concerned.
You don’t follow a utility function. Utility functions don’t prescribe actions.
… are you suggesting that we solve prisoner’s dilemmas and similar problems by modifying our utility function?
OK, bad choice of words.
No, but you need some decision theory to go with your utility function, and I was considering the possibility that Parfit merely pointed out a flaw of CDT and not a flaw of common sense morality. However, given that we can still think of situations where common sense morality (no matter the decision theory) executed by everyone does predictably worse for everyone concerned than some other theory, Parfit’s objection still stands.
(Incidentally, I suspect that there could be situations where modifying your utility function is a way to solve a prisoner’s dilemma, but that wasn’t what I meant here.)