Robin, that's a fair reply for saving majoritarianism. But it doesn't save the argument from bias-variance decomposition, except in the special case where the loss function is equal to the squared difference for environmental or moral reasons - that is, we are genuinely using point scalar estimates and squared error for some reason or other. The natural loss function for probabilities is the log score, to which the bias-variance decomposition does not apply, although Jensen's Inequality does. (As I acknowledged in my earlier post on The Modesty Argument.)

This leaves us with the core question as "Can you legitimately believe yourself to be above-average?" or "Is keeping your own opinion like being a randomly selected agent?" which I think was always the key issue to begin with.