"What hadn't been realized in the literature until now is that merely to describe how severely something has been tested in the past itself embodies inductive assumptions, even as a statement about the past."