It’s true that correctness and usefulness of models are both measured by the accuracy of their predictions, however, the weights are different. Usefulness strongly weighs the accuracy of questions that come up often, while correctness weighs all possible questions more uniformly.
Do you think you have enough epistemic rationality to determine when it’s really worth sacrificing epistemic rationality for something else? Better to keep increasing your epistemic rationality, just to be sure.
This is ridiculous. “Sacrificing epistemic rationality” is a risk with uncertain rewards (let us assume that the rewards do exist). It’s not necessarily stupid to take risks. It is stupid to wait until you have perfect information about the reward you would receive, because that will never happen.
Also, there is another issue—converting to a religion doesn’t immediately make you retarded, as you seem to immagine. Religious people are perfectly capable of high instrumental rationality, even if we agree that their epistemic rationality is diminished. Even further, it can be useful, for empathy, to model religious people, and crackpots in general, as perfect bayesians with really weird priors.
It’s true that correctness and usefulness of models are both measured by the accuracy of their predictions, however, the weights are different. Usefulness strongly weighs the accuracy of questions that come up often, while correctness weighs all possible questions more uniformly.
This is ridiculous. “Sacrificing epistemic rationality” is a risk with uncertain rewards (let us assume that the rewards do exist). It’s not necessarily stupid to take risks. It is stupid to wait until you have perfect information about the reward you would receive, because that will never happen.
Also, there is another issue—converting to a religion doesn’t immediately make you retarded, as you seem to immagine. Religious people are perfectly capable of high instrumental rationality, even if we agree that their epistemic rationality is diminished. Even further, it can be useful, for empathy, to model religious people, and crackpots in general, as perfect bayesians with really weird priors.