I sense that these posts are written for a specific audience, rebutting a set of arguments familiar to that audience, and that's why they seem so reductive and narrowly-applicable, but I can't quite grasp how much of the argument translates to the rest of the world.
It is aimed at design architectures which seek to separate application level concerns from Rails. The theory being that Rails and your specific application get too tightly coupled. The aspect being talked about here is the testing.
