I like that idea of actually trying to measure what people know and learn. Makes sense to me. Whenever the discussion of education comes up, I always mention Free to Learn by Peter Gray. It's an excellent way to look at learning and education in the human species.
RE: A testing paradigm for the 21st century