I like this line of thought and to an extent, I agree with it pretty strongly.

I also think that if you are adhering to the notion of a "single level of abstraction" per function, then you will naturally need to structure your functions in the way Fowler recommends.