I think the analogy doesn't quite fit with typing speed in programmer productivity. If anything, it's analogous with time spent thinking on a problem. It's easy sometimes to spend quite a bit of time just determining how to attack a problem.
And often times it's a good thing, but in my experience sometimes it's good to just throw your first instinct at the problem. So instead of determining it doesn't work in my head, I find the problems while running the code. This usually reveals more information than dismissing the idea during the planning process.
I think the author's conclusion was that while being a fast typist doesn't make you a better programmer, being a slow typist will make you a worse one, as it will take you longer to commit your initial mistakes and iterate, akin to playing many games of speed chess to improve at chess.
And often times it's a good thing, but in my experience sometimes it's good to just throw your first instinct at the problem. So instead of determining it doesn't work in my head, I find the problems while running the code. This usually reveals more information than dismissing the idea during the planning process.