But we can see why the probability isn’t counterbalanced without having a visceral grasp on the quantities involved. There may be some uncertainty in our view that there aren’t enough counterbalancing forces we’ve taken into account, but in practice uncertainty almost always places you some nontrivial distance away from .5 credence. We still have to have credence levels, even about quantities that are very uncertain or beyond our ability to compute. Metauncertainty won’t drag your confidence to .5 unless your uncertainty disproportionately supports ‘I’m underestimating the likelihood that there are counterbalancing risks’ over ‘I’m overestimating the likelihood that there are counterbalancing risks’.
But we can see why the probability isn’t counterbalanced without having a visceral grasp on the quantities involved. There may be some uncertainty in our view that there aren’t enough counterbalancing forces we’ve taken into account, but in practice uncertainty almost always places you some nontrivial distance away from .5 credence. We still have to have credence levels, even about quantities that are very uncertain or beyond our ability to compute. Metauncertainty won’t drag your confidence to .5 unless your uncertainty disproportionately supports ‘I’m underestimating the likelihood that there are counterbalancing risks’ over ‘I’m overestimating the likelihood that there are counterbalancing risks’.