At Larval Subjects, Levi considers the issue of how an assemblage theory such as DeLanda's accounts for the causal relations that result in the emergence of new assemblages. He writes
the question of causal mechanisms is an empirical question. In other words, there is not one answer to this question as these mechanisms will function in a variety of different ways depending on the assemblage that we’re talking about. The mechanisms that allow an organism to be something more than its subatomic parts will be different than the mechanisms that allow a city to be something more than the people, buildings, infrastructure, etc., that make it up.
I agree with this, though noting two things. First is that perhaps this is a different kind of empiricism. For many I'm sure empiricism invokes its own theoretical perspective quite different from DeLanda's. This issue is also present for actor-network theory. That said, empiricism is likely the right word here as it denotes a clear departure from correlationism. Second, though A New Philosophy of Society certainly focuses on causal mechanisms, DeLanda also notes a virtual-topological dimension to assemblages that is mechanism independent. This does not mean, however, that assemblages are magical (a critique to which Levi is responding). Instead DeLanda is pointing to singularities that operate in a non-deterministic fashion in the unfolding of assemblages. More on this in a moment.
So let me turn to my quotidian situation of students' commenting on their peers' writing in a composition workshop. Millions of people are likely engaged in this activity this week in first-year composition courses around the nation. An almost uniform instruction that those students will receive from their instructors is to not focus on grammar and proofreading. That instead, we ask students to look at larger rhetorical issues such as purpose, audience, etc. Despite these instructions, many students will still focus on proofreading.
Why do they do this?
If we were to examine this behavior from the perspective of assemblage theory, we would, as Levi suggests, approach it as an empirical question. To a degree the answers may be singular to each student, but the fact that so many students from so many different backgrounds perform the same activity, we would be inclined to look for a general answer. The commonsensical (and disciplinary) response is to focus on their largely common experiences in K-12 language arts/English classes, where we hypothesize that many of student attitudes and practices in relation to writing have formed. Undoubtedly these attitudes are reproduced through mass media and in various social situations to the point that if one mentions to a newly met person that one is an English teacher, the response is almost always "I better watch my grammar, then." Because, of course, that's what English teachers care about.
So in an empirical investigation of this matter, one would have to decide how far to follow the network. The idea of teaching the writing process and de-emphasizing issues of proofreading is hardly a radical notion. It's hard to imagine that students haven't encountered teachers who taught from that perspective. Yet somehow it is the proofreading that lives on.
We cannot just dispense this with "society (or English teachers) made them do it."
From either the perspective of ANT or assemblage theory, one would need to begin in the classroom itself and then follow our students through their writing processes into the events of peer review. There are many potential threads, so perhaps we can never track down all the nuances, but we might apply DeLanda's notion of "redundant causality." As such, we might consider which elements are necessary to result in this general practice that we call proofreading.
I haven't performed the empirical investigation, so I can only speculate. Speculating, I would suggest that even though as instructors we provide guidelines, the students are faced with largely the same context they have faced before: a peer's essay, a pen, white space in the margins, the same amount of time to respond, the request to make inline comments, etc. They are also faced with a common social situation that makes critical commentary uncomfortable. And, of course, it is critical commentary with which they are most familiar (from reading teacher feedback on their own essay). As a result they might tend to stick to where they feel authoritative (which is, oddly, grammar) and then some rather tepid commentary. So even when we suggest the classic Peter Elbow advice of writing "movies of the mind" or what not, many of the object/actors are communicating a different message to our students.
I have to say though, I am less interested in exploring how assemblage theory might explain this behavior, as using this perspective to imagine different approaches to peer review. That is, if we change the context, the assemblage/network, in which peer review occurs, we likely change the outcome. So, for example, we could change the technology at work by having students comment online or having them only listen to a paper without reading it. The idea here is not necessarily to ensure "better" comments but rather to force students to approach peer review with a different set of tools. In a related vein, one might ask students to perform a different activity than writing inline comments: e.g., adding or rewriting a paragraph in the essay, writing a one-paragraph summary, comparing two students' arguments.
To veer back toward more theoretical language, by shifting the empirical mechanisms at work in an assemblage, which I would think of as those elements that delimit/enable its possibilities, one also opens new potentialities, shifting its topology. If we can distinguish between possibility and potential, we can see where peer review (and indeed much of composition pedagogy) needs to go. When we read a student's paper within the existing institutional contexts in which we regularly operate, we are faced with a limited set of possibilities, both for how we might respond and how the paper might be revised. In shifting those empirical mechanisms, we not only introduce new possibilities but also might become more aware of potentialities that offer topological mutations. Many of those mutations might not be of interest to us, but our awareness of them can help us understand how the mechanisms at work in a writing assemblage shape the work.