True, but you should first assign appropriate weights to the two categories you mention based on the expected cost of having an incorrect belief.
This seems obviously correct, but at the same time it seems at odds with the virtue of evenness.
No, the weight factors into an expected utility calculation, it's separate from the probability calculation. Miller didn't say otherwise.
BTW, the opening three comments of this thread would make a great introduction to what the LW website is all about.
Here's the new thread for posting quotes, with the usual rules: