The author seems like a proponent of functional programming, not necessarily TDD. The input->output testability seems to vary much more than the author describe

The author seems like a proponent of functional programming, not necessarily TDD. The input->output testability seems to vary much more than the author describes depending on system size. I love TDD, but even the most adherent acceptance testing fails to cover the case where you're laying the final brick in the arched bridge, to use his analogy. It was a great article, and the author made very good points.