Here’s something Foucault wrote (or said and then edited):

Even when an individual has been accepted as an author, we must still ask whether everything that he wrote, said, or left behind is part of his work. The problem is both theoretical and technical. When undertaking the publication of Nietzsche’s works, for example, where should one stop? Surely everything must be published, but what is “everything”? Everything that Nietzsche himself published, certainly. And what about the rough drafts for his works? Obviously. The plans for his aphorisms? Yes. The deleted passages and the notes at the bottom of the page? Yes. What if, within a workbook filled with aphorisms, one finds a reference, the notation of a meeting or of an address, or a laundry list: is it a work, or not? Why not? And so on, ad infinitum. How can one define a work amid the millions of traces left by someone after his death? A theory of the work does not exist, and the empirical task of those who naively undertake the editing of works often suffers in the absence of such a theory.

*This essay is the text of a lecture presented to the Societé Francais de philosophie on 22 February 1969 (Foucault gave a modified form of the lecture in the United States in 1970). This translation by Josué V. Harari has been slightly modified.