I am not committed in any way to the traditional concept of character - the concept of "character trait" as involving predictable behavior. I am committed to a view in the neighborhood - the view that the moral worth of one's actions depends on the quality of will expressed in them.
I think that if your tenure case depends on your proving what you thought was a mathematical theorem and the proposed theorem turns out to be false just before your tenure decision, and you want to get tenure very badly, there is a sense in which it's perfectly understandable and reasonable of you to wish the proposed theorem were true and provable, even if it's logically impossible for it to be.
Though we think intrinsic desires tend to be pretty stable, we do not think they imply anything like the amount of predictability in behavior that traditional virtue ethics requires for someone to have a one-word-in-English character trait such as "benevolence". Other things being equal, a person with more of a desire for other people's wellbeing will do more for other people's wellbeing, but things are almost never equal.
