What on earth is an "inoperative" bit? That sounds like a model simplification/idealization, not something that has theoretical distinction. It is either a part of the system or it isn't. And the nasty truth that we always trying to drive out of our engineering models is that there is only ONE system, and everything is a part of it.
I'm with you in wanting to look at the whole of "information theory and molecular biology" as a single transaction. That is the point of the term "information" used by Shannon in his theory - it is all about the action of the reduction of uncertainty in the receiver, not the condition of a gain of information content. The value or meaning of the message is outside the scope of Shannon. (Though it becomes relevant in our discussion of abiogenesis.)
With regard to "inoperative bits" - in the Shannon model, that would be noise. To apply "information theory and molecular biology" to evolution, noise would roughly equate to "random mutations". The distinction is between that which was encoded at the source and noise in the channel - both factor into the reduction of uncertainty in the receiver, if not filtered out by decoding, etc.