It sounds great - teach ambiguity by introducing complex questions that do not have easy answers.    The problem is that a lot of students aren't going to get what they need out of classes that way, notes Chad Orzel of Scienceblogs.com.   
I quickly run up against two major problems: first, that we have certain material that we need to cover for our own courses, and second that we have certain material that other people expect us to cover when their students take our classes.
Indeed, a number of students taking mechanics aren't going to be physics Ph.D.s in the future but rather practicing engineers.   The hypothesis-confirmation-replication method won't really apply to many who take science classes but a more dynamic way, of how real science works, won't necessarily work in the classrooms that actually exist.

Read about the benefits and perils of trying to teach ambiguity and the scientific method in the world outside humanities-centric idealism.