Marc Bloch's The Historian's Craft, especially the section on “Historical Criticism,” provides something of a consensual baseline for history-writing as it was practiced throughout most of the twentieth century. The essential, albeit generally implicit, moment for Bloch is the cleavage between history as actual event and history as it is interpreted by the sources. History-writing here is something like iron-smelting: one must take the raw source-material and process it until the impurities (forgery, embellishment, error) can be filtered out and the only material that remains is a close approximation to “what really happened” (not his words). There are several tools one can use to process the ore, but what they all have in common is their reliance on the broader ecosystem of sources as a check on the possible defects of an individual source. Thus, we determine if a charter is genuine or forged by comparing it with documents from the period that are known to be genuine; likewise, we discover if someone is prevaricating about an episode in the Napoleonic Wars by comparing his account of the events with other contemporary accounts. By a judicious application of these methods, we at last arrive at a kind of arithmetic mean of the sources, which may be assumed to be more or less what took place.
The reason I call this model “naive” is that, in its pursuit of historical fact, it forgets that its sources are more than just pieces of ore. Bloch acknowledges, and then dismisses, the possibility that the source-concurrence approach might generate serious error—if, for instance, the Napoleonic memorialist was the only one to speak the truth and the other accounts were false or erroneous. Put in this way, of course, it looks like only a conspiracy theorist could object. But Bloch himself would surely blanch at taking the concurrence of any number of seventeenth-century doctors for positive evidence that someone's illness was the product of a humoral imbalance. Bloch's standard is thus nothing more than the reflection of the most widespread prejudices of an age, occasionally supplemented or leavened by those of our own. When naive history-writing attempts to partition off “what really happened” from “what is said about it,” it really accomplishes the reverse: it presents as “real” only the “said” shorn of any constitutive individuality.
You might object that if the historian were to be presented with the choice of a consensus-1809 and an embellished-1809, it would be stupid to choose the latter. But in fact both are equally unsatisfactory: an account of the Wars which excludes the lies and distortions around them is no more faithful to their historical reality than an outright fabrication. (Picture a history of Piltdown Man that consists of a single sentence: “Piltdown Man did not exist.”) Naturally, few historians, including Bloch himself, would defend such a hardline approach—which means only that painting “what really happened” as the Holy Grail of history-writing is inadequate.
The speculative realist objection is clear. Lying is one way that an object (a memoir) can interact with another object (the Napoleonic Wars). Accurate reportage is another. A statistical or factual generalization compiled from a collection of primary sources is yet another form of interaction. The historian's objective is no longer to singlemindedly pursue one such form: it is to poke and prod the object from as many directions as possible, to understand in maximum detail how it responds and has responded to other objects. Although the possibility of an exhaustive understanding is obviously denied in advance, the gap is considerably less disturbing than that between “what really happened” and “what is said about it.”
The Foucauldian model, as set forth in The Archaeology of Knowledge, is the “critical” alternative to Bloch's “naivety.” What Foucault understands himself to be doing is analyzing discursive formations. Such a formation operates purely on the level of “what is said”: it is, in his words, “a group of rules for a discursive practice.” In other words, it determines what kind of statements can be made, how these statements are organized, how they can be said to be meaningful or true, and so on. Early eighteenth-century medical discourse is one example of such a formation; political economy is another. The boundaries between discursive formations are fundamentally disjunctive or quantized—that is to say, there are no smooth gradations between them, either chronologically or at any given moment.
In the context of the general reaction to “theory,” Foucault has often been lumped in with Derrida as someone who believes that “there is nothing outside the text”—that there is no space outside of discourse at all. That is patently false; Foucault believes that non-discursive domains exist, though he does not elevate them to any particularly privileged position. Actually, non-discursive domains can and do shape discursive formations: Foucault cites the 1829 cholera epidemic and the French Revolution as events that profoundly affected a variety of discourses. Yet the archaeology of knowledge only studies the non-discursive as it is reflected or as it reverberates in the discursive. This is not necessarily a limitation, since Foucault's objective is to study discourse and not to dig beneath it. The archaeology of knowledge is not a general theory of history-writing but rather a specific methodological approach.
Nonetheless, Foucault is slippery when it comes to defining what exactly the non-discursive can look like. Non-discursive domains, according to him, include “institutions, political events, economic practices and processes” (the list is not meant to be exhaustive). It is clear from Foucault's own work that these domains are not inherently non-discursive; to outline an economic process is obviously a discursive act, and statements about that process are governed by a discursive formation such as Marxist economic theory. Where does the non-discursive end and the discursive begin? He would say, presumably, that it is precisely the task of the discursive formation to figure that out. But that gives us nothing to work with. Either the non-discursive domain is entirely produced by the discursive formation, in which case it is not true that there are non-discursive domains, or the non-discursive domain preexists the discursive formation and exerts a fixed and concrete influence on it, in which case there is no line at all between the discursive and the non-discursive.
The speculative realist view can once again come to the rescue here. By now it is clear what this would be: the relationship between political economy and, say, the French Revolution is a Harmanian relationship between objects. The Revolution exhibits certain properties (for instance, the elimination of the privileges of the nobility) with which political economy interacts. Because of the nature of discursive formations, the interaction between the two is on a discursive level and falls under the ambit of the particular formation in question—but there is nothing that says the Revolution is restricted to exhibiting properties of this kind. It may present other discursive formations with other properties, or it may interact on an entirely non-discursive level (the destruction of the Bastille, for example). The crucial point is that its causal potential is not exhausted by either mode. Foucault, in other words, needs something like a speculative realist (both speculative and realist) understanding of non-discursive domains--or else the subtlety and flexibility of his methodological approach collapses into dogmatism or self-contradiction.
In both Foucault's case and Bloch's, the decisive factor is the reluctance of Harmanian realism to make claims of even potentially exhaustive access to an object. By refusing to concede to either side of the debate, this approach keeps things interesting--after all, there's no reason traditional social or economic history should be deemed illegitimate simply because it relies on an understanding of history that doesn't start from the text. From this point of view, speculative realism begins to resemble Nietzschean perspectivism--which is ironic, but unsurprising, given the pragmatic caveat we made at the beginning of the essay.
Next time: speculative realism as a working methodology; Loyalism as a test case.
[ed.: the book wasn't talking about Waterloo]