Oh, Foucault. Why so Concrete?

On page 905 of "What is an Author?", Foucault questions what makes up an author's work. He asks whether "rough drafts" and "deleted passages" qualify. As usual, he doesn't answer these questions. The lack of answer didn't bother me, it was the fact that the questions were asked at all. The entire paragraph I'm referring to (last full paragraph on 905) leads me to believe that Foucault saw an author's work as something concrete and definitive. Was that the accepted assumption of the time? It was jarring for me, because I see text just like any other art form in that it has no definitive conception or conclusion. Basically, my perception of an author's work is more fluid than Foucault's. I began to see some foundation for his questions when, on 913, he brings up "the great danger with which fiction threatens our world". If Foucault was solely referring to non-fiction writing earlier in the piece, I can see more reason for this questions. If there is a solid factual answer offered in a text, than I can see why he would wonder what all constituted as 'work' and therefore would attribute to the solid factual answer. This doesn't completely convince me that the question are necessary though.

Along the same lines, I want to look at Saint Jerome's four criteria for determining a singular author (909). Briefly: 1) the author has a "constant level of value", 2) the author must not contradict himself, 3) the author must maintain a singular style, and 4) the author functions in his time. I found the same fault here as I did with Foucault's questions on an author's work. Initially, I directly applied these four criteria to a human author. They seem ridiculous because authors are like all humans in that they change their mind (number 2), their style evolves or they explore different styles (number 3), and that they don't have complete control over how someone values their work (number 1). I had to go back and reapply these criteria to the author-function instead. I guess only including the similar pieces to compose an author's work would make applying that work to a discourse easier (?). I'm not convinced of this though, because Foucault goes on to discuss how the author-function evolves and changes. If an author's work, which makes up the author-function, is comprised only of pieces which are homogenous, doesn't that skew how the author-function is perceived/used?