> it strikes me as intuitively sensible...

This might be intuitive if A weighted each N of some class equally, and N>>Kevin for each _class_, but observation indicates this is not the case. Also certain N tend to be overweighted regardless of the knowledge class to which they belong.

IRL I have observed that B can work around A's bias by framing knowledge expressed as a question.

> Sounds like narcissism.

I think this is a common bias with implications in the "machine consciousness" question, but I'll have to wait ton finish that point. Look for an edit.