But you can still end up with a “flat” rule for the human art of rationality, when the expected negative utilities associated from biased decisions that “the end justifies the means, in just this one case here”, exceeds the expected positive utilities from cases where the universe really does end up a better place from shooting someone who makes an argument you don’t like after taking all side effects into account including encouragement of similar behavior by others.
That might deserve a post of it own...
There will be no such consequences when no one (including the person attacked) will ever find out that one has responded to an argument with violence.
There are also effects on the attacker—specifically, it may make it easier for them to attack again. It will change their attitude on many issues. You’d need a murder-suicide to pull this off properly...
One can easily think of many theoretically possible circumstances (not involving superintelligence) in which one can prevent immense evils and bring about immense goods by responding to an argument with violence, and yet satisfying the condition above, that no one will ever find out.
Indeed. Eliezer’s rule cannot be a flat-out rule unless it’s the only flat-out rule of moral conduct—otherwise it’s nearly certain it can be contradicted by one of the other flat-out rules (barring strange specific moral rule constructions that are used by nobody).
But if we take “flat-out” to mean “This rule has more weight behind it than you can imagine. Much more. You’re not there yet—you can’t yet conceive how bad the consequences of violating this rule will be”, then it’s acceptable.
But you can still end up with a “flat” rule for the human art of rationality, when the expected negative utilities associated from biased decisions that “the end justifies the means, in just this one case here”, exceeds the expected positive utilities from cases where the universe really does end up a better place from shooting someone who makes an argument you don’t like after taking all side effects into account including encouragement of similar behavior by others.
That might deserve a post of it own...
There will be no such consequences when no one (including the person attacked) will ever find out that one has responded to an argument with violence.
There are also effects on the attacker—specifically, it may make it easier for them to attack again. It will change their attitude on many issues. You’d need a murder-suicide to pull this off properly...
One can easily think of many theoretically possible circumstances (not involving superintelligence) in which one can prevent immense evils and bring about immense goods by responding to an argument with violence, and yet satisfying the condition above, that no one will ever find out.
Indeed. Eliezer’s rule cannot be a flat-out rule unless it’s the only flat-out rule of moral conduct—otherwise it’s nearly certain it can be contradicted by one of the other flat-out rules (barring strange specific moral rule constructions that are used by nobody).
But if we take “flat-out” to mean “This rule has more weight behind it than you can imagine. Much more. You’re not there yet—you can’t yet conceive how bad the consequences of violating this rule will be”, then it’s acceptable.