Of course, a "lecture" on Perl is probably as useless as a "practical with computers" on Metaphysics.
A truly great lecture is akin to a piece of theatre, especially in the humanities. Really, that's one of the problems with the methods that Khan promotes: they appear to work very well for quantitative subjects that can be accurately tested, but that doesn't apply to many of the arts.
I don't see why the humanities should be relegated to "ungraded seminars at night on art and literature". And, as others have pointed out here, one of the reasons Nobel laureates who are faculty may have little time for students is because they have a lot of other things to do: the same is surely true for the professionals Khan wants to advise students?
I don't want to sound as if I'm against the Khan Academy - it's fantastic. I think there is real value in implementing many of its ideas into standard education, and I know that's already happening with projects like edX and the like.