I think these types of articles get written after people realize how massive testing makes real refactoring harder, not easier.

To them the problem must be the way people refactor because it can't be the way things are tested.