Our graduate seminar had a lengthy discussion about agency, and the possibility of taking action in a world where agency is found in chains of iterations rather than human actors.
To wit: the critique of the Butlerian/deconstructive positions holds something like--your view of subjectivity denies the possibility of interventions into the world because while on one hand it eviscerates reference (this evidences all kinds of practical problems) it also elevates these references (names) into a higher order of existence (this would be the nominalist position, that things are their names) by establishing that their naming functions as a kind of overdetermination. Presumably the result would be debilitating for scholars interested in fighting violence and oppression because without being able to refer to categories around which meaningful resistances might be articulated (race, class, gender, national origin, etc.) critical rhetoricians would be locked into pursuing something like a politics of unintelligibility. Or, if recourse to such categories is allowable, it would only be by means of contradiction (i.e. the invocation of these categories is not neutral, they are historically developed, and so how can we ensure that our use is not just another pernicious effect of a chain of iterations?)
A response, following what is here (allowing some Butler and Foucault as well) might hold something like: the recuperative argument is undone by installing initially and taking for granted the demand that words mean something more than "just their name". That is, to yearn for a concept or set of notions under which one might organize some kind of resistance (to say nothing of the rather large debate about whether agency is about making change, and what counts as change in a world where subjects are constantly being remade rather than shuffled around on a chess board) presumes in advance that said concept is meaningful/necessary to produce change. Of course, it is the desire to have some kind of stable footing in a dynamic world that operates as the cause (and I don't mean cause in a strict case/effect model, I mean something more like De Man argues here when he notes that what distinguishes the cause and effect is nothing more than the naturalization of metaphor, a naturalization whose repetition consistently a certain kind of ordering whose rightness can be verified by no external authority) of that very anxiety in the first place--what is displaced is that understanding that the anxiety stems from the absence of a founding point for the reading. Instead, this anxiety of no origins is replaced with an anxiety that stems from the moves away from metaphorized/naturalized understandings whose extensive repetitions secure (to use a bit of debate jargon) the status quo.
If this seems theoretically dense and annoying (guess who's been studying for comps!), an example: if we say that agency is in a chain of iterations, rather than in people, a good democrat might complain that this forecloses the ability of humans to attempt to mobilize causes for justice around the image of "the people", because that understanding will always be purely discursive rather than grounded in some actually existing democratic coalition. My counter would be: the understanding that agency stems from chains of iterations rather than agents themselves is the key condition for mobilizing something like "the people" in a way that is not historically determined. After all, it is the materially grounded understanding of "the people" (the loss of which is anxiety generating for those interested in recuperating a doer behind the deed) that seduces us with its claim to intelligibility, but it is the claim to intelligibility that tautologically prohibits us from questioning necessity of the term itself (because its relevance is naturalized rather than being subjected to critique). For example, the failure of "Joe the Plumber" as a conservative enthymeme in the 2008 campaign was a result of the fact that his deployment assumed a logic of external reference. The McCain campaign assumed that a certain kind of Americanness was available through JTP, and didn't hestitate to push him for a couple weeks. The problem was that the selection of JTP as a campaign avatar was guided by the belief he could function as a synecdoche, but instead he had a life of his own, becoming a kind of allegory for Americanness. He had back taxes, no plumbing license, and whenever he talked his political opinions never seemed quite up to snuff. Confronted with the possibility that JTP showed a kind of absence at the heart of politics, subjects simply began to identify against him--rather that confronting the split between "the people" and politics, he simply represented neither category properly. If politics worked with referents, as my fellow compatriots hold, JTP's failure would have sparked some sort of rethinking. Instead, he was always "in and for himself".
Here we see that the problem is not a slide to "pure nominalism"--the function of langauge to cut us off from external reference, and then its ability to seal us from encountering that fact by having already installed itself as a natural rather than a contingent truth ensures that we are already on a nominalist plane whether we like it or not. Trying to rescue some kind of agency from this fact lies in letting it be an assumption that drives our theorizing--and its meaning as an assumption is evacuated if we alter or forswear it by insisting that things are otherwise, as the call to recuperate agency does.