The Conclusion is framed as a dialogue between Foucault and a hypothetical critic, in which Foucault answers a variety of broad challenges to his project. For simplicity's sake, I have maintained this structure here in the summary.
Critic: Your method is simply a disguised, twisted structuralism, a structuralism which, in refusing to recognize itself as such, creates a slew of theoretical 'oddities.' Further, you have fantasized a discourse that does not depend on speaking subjects; therefore, you do not take account of the full range of richness and irregularity in discourse. Finally, you have removed discourse from history itself, refusing to acknowledge that it depends on actual things and events that occurred in a historical chronology.
Foucault: My method is not an attempt to move structuralism into a completely isolated region beyond its recognition. 'I never once used the word 'structure' in The Order of Things.' It is, rather, an attempt to outline an analysis that does not follow the principles of structuralism at all. And in any case, the debate over structuralism is dead, carried on only by 'mimes and tumblers.' Moreover, I did not erase the speaking subject, but rather approached the issue at the level of discourse, describing the diversity of positions from which the subject can speak. Neither did I erase history; I simply replaced the blurry, monotone notion of 'change' with a set of specific discursive transformations.
Critic: You are evading the point. Your method is problematic not simply because it resembles structuralism, but because it (like the worst forms of structuralism) attempts to place discourse on a level that totally frees it from its 'constituent activity,' its progression from a real origin and along a fundamental teleology. The subjectivity of discourse means that these factors are essential; discourse cannot transcend them.
Foucault: I'm glad you made that distinction; it is not really structuralism that you're worried about, nor is it the possibility of a transcendence of discourse over the real world of empirical history. You are really concerned with trying to maintain some sort of easy, comforting continuity (through the subject, through teleology, through causality) that structures history. You do this to cover the real 'crisis' we now face as historians: in the face of our decades-old abandonment of a transcendental historical subject, what is the new 'status of the subject?' Archeological analysis offers a new answer to this question, and does so without blending historical difference into broad, unconsidered categories.
Critic: Fine. Let's suppose those are the stakes of our argument, and that it is framed by this 'crisis.' How, then, can you claim any kind of positive truth for your own statements? If 'discourse' includes your own theories, how can you claim to analyze discourses as empirical data? Is your work really history, or is it just philosophy?