I’m one of the judges of the competition. My comments shouldn’t be taken as a full review of a post. And, unfortunately, I won’t have capacity to comment on every post or engage with all replies. Thanks so much to everyone who entered!
I think this is a neat and thought-provoking argument that ought to give proponents of cluelessness pause.
My thoughts on this haven’t fully settled, but some reactions:
Suppose that, per the post’s argument, the most choiceworthy actions according to (the most plausible version of) imprecise consequentialism are the somewhere-uniquely-optimal ones. There is still a question of how this view should enter into intertheoretic aggregation. I think it would be odd for imprecise consequentialism to dominate the intertheoretic aggregation, given big disagreement between different elements of the representor about the value of somewhere-uniquely-optimal actions.
Example:
A is great according to common-sense morality, and is nowhere-optimal.
B and C are meh according to common-sense morality, and are uniquely optimal according to p_B and p_C, respectively. Also, B is terrible according to p_C and C is terrible according to p_B.
It seems to me that A should at least be permissible, when aggregating over these two theories.
I’m not sure yet that we can distinguish somewhere-uniquely-optimal and nowhere-uniquely-optimal actions. The space of policies available to us in real life is extremely high-dimensional and it’s not yet clear to me that you couldn’t run a similar argument to your MAWF example to cast doubt on the somewhere-optimality of any policy.
I tend to think that one should never be required to strictly c-prefer a mixture of A and B to C whenever neither A nor B is c-preferred to C. (For one thing, this involves a kind of dynamic inconsistency: I know that, once the coin lands, no matter how it lands, I will no longer have a c-preference between the continuation plan and C.) I tend to think that it’s an unfortunate feature of the dominance rule that this happens, and I would love it if there was some repair to the dominance rule that didn’t have this property. Sadly I don’t have one to give, and I suspect any such theory would have its own significant costs.
But, if there were a way to avoid the conclusion that a nowhere-optimal action is always c-dispreferred to some mixture, then there is only the weaker argument that we ought not choose nowhere-optimal actions. And I’m not yet convinced that nowhere-optimality is a problem.
I’m one of the judges of the competition. My comments shouldn’t be taken as a full review of a post. And, unfortunately, I won’t have capacity to comment on every post or engage with all replies. Thanks so much to everyone who entered!
I think this is a neat and thought-provoking argument that ought to give proponents of cluelessness pause.
My thoughts on this haven’t fully settled, but some reactions:
Suppose that, per the post’s argument, the most choiceworthy actions according to (the most plausible version of) imprecise consequentialism are the somewhere-uniquely-optimal ones. There is still a question of how this view should enter into intertheoretic aggregation. I think it would be odd for imprecise consequentialism to dominate the intertheoretic aggregation, given big disagreement between different elements of the representor about the value of somewhere-uniquely-optimal actions.
Example:
A is great according to common-sense morality, and is nowhere-optimal.
B and C are meh according to common-sense morality, and are uniquely optimal according to p_B and p_C, respectively. Also, B is terrible according to p_C and C is terrible according to p_B.
It seems to me that A should at least be permissible, when aggregating over these two theories.
I’m not sure yet that we can distinguish somewhere-uniquely-optimal and nowhere-uniquely-optimal actions. The space of policies available to us in real life is extremely high-dimensional and it’s not yet clear to me that you couldn’t run a similar argument to your MAWF example to cast doubt on the somewhere-optimality of any policy.
I tend to think that one should never be required to strictly c-prefer a mixture of A and B to C whenever neither A nor B is c-preferred to C. (For one thing, this involves a kind of dynamic inconsistency: I know that, once the coin lands, no matter how it lands, I will no longer have a c-preference between the continuation plan and C.) I tend to think that it’s an unfortunate feature of the dominance rule that this happens, and I would love it if there was some repair to the dominance rule that didn’t have this property. Sadly I don’t have one to give, and I suspect any such theory would have its own significant costs.
But, if there were a way to avoid the conclusion that a nowhere-optimal action is always c-dispreferred to some mixture, then there is only the weaker argument that we ought not choose nowhere-optimal actions. And I’m not yet convinced that nowhere-optimality is a problem.