Conducted behind their backs

A striking methodological salvo from the opening chapter of James C. Scott, Two Cheers for Anarchism:

The populist tendency of anarchist thought, with its belief in the possibilities of autonomy, self-organization, and cooperation, recognized, among other things, that peasants, artisans, and workers were themselves political thinkers. They had their own purposes, values, and practices, which any political sys­tem ignored at its peril. That basic respect for the agency of nonelites seems to have been betrayed not only by states but also by the practice of social science. It is common to ascribe to elites particular values, a sense of history, aesthetic tastes, even rudiments of a political philosophy. The political analysis of nonelites, by contrast, is often conducted, as it were, behind their backs. Their “politics” is read off their statistical profile: from such “facts” as their income, occupation, years of schooling, property holding, residence, race, ethnicity, and religion.

This is a practice that most social scientists would never judge remotely adequate to the study of elites. It is curiously akin both to state routines and to left-wing authoritarianism in treating the nonelite public and “masses” as ciphers of their so­cioeconomic characteristics, most of whose needs and world­ view can be understood as a vector sum of incoming calories, cash, work routines, consumption patterns, and past voting behavior. It is not that such factors are not germane. What is inadmissible, both morally and scientifically, is the hubris that pretends to understand the behavior of human agents without for a moment listening systematically to how they understand what they are doing and how they explain themselves. Again, it is not that such self-explanations are transparent and nor are they without strategic omissions and ulterior motives—they are no more transparent that the self-explanations of elites.

The job of social science, as I see it, is to provide, provision­ally, the best explanation of behavior on the basis of all the evidence available, including especially the explanations of the purposive, deliberating agents whose behavior is being scruti­nized. The notion that the agent’s view of the situation is ir­relevant to this explanation is preposterous. Valid knowledge of the agent’s situation is simply inconceivable without it. No one has put the case better for the phenomenology of human action than John Dunn:

If we wish to understand other people and propose to claim that we have in fact done so, it is both imprudent and rude not to attend to what they say. . . . What we cannot properly do is to claim to know that we understand him [an agent] or his action better than he does himself without access to the best descriptions which he is able to offer.

Anything else amounts to committing a social science crime behind the backs of history’s actors.

This is an especially pernicious flaw in a great deal of historical writing, where motives and explanations for large-scale social transformations—say, the American Revolution—are often identified solely and unreflectingly with the motives and explanations of elites (intellectual, social, political). Writing history is very hard, at least if one wishes to be truthful. One way to characterize the error Scott discusses is to say that it turns all history into intellectual history: the history of what intellectuals have said and done. (I am reminded that when I expressed an interest in intellectual history to one of my math professors at MIT—a world-famous algebraic geometer—he mused, skeptical of the need for the adjective and with all the false confidence of the naif, “Isn’t all history intellectual?” He meant something banal rather than provocative—isn’t historical work intellectual work?—but I still wince at his reflex to dismiss what he found unfamiliar.)

I am struck by this passage in part because it helps to illuminate one dimension of the academic debate—never mind the more general culture war—about Nikole Hannah-Jones’s claims for the revolution in the 1619 Project (the most contested being that “one of the primary reasons” the United States fought the war was to preserve slavery): there is a historiographic contest over whose reasons for revolution matter—Ben Franklin’s and James Madison’s, your salt-of-the-earth antislavery New Englander’s, or that of the scale-tipping planter Virginian who had not yet gone over to the Patriot cause but was spurred to do so after Lord Dunmore’s Proclamation threatened to free his slaves.

The problem is not always political or ideological, or a residue of Great Man history writing (though it sometimes is, more than it ought to be). At bottom, the temptation to ignore the reasoning and views of nonelites is material and methodological: we typically have more records, more documents, more texts—or even if not numerically more, at least more easily accessible ones—from elites than from the masses. This is a lesson the full significance of which I have learned only gradually myself, not least because of my absorption in intellectual culture and my interest in the history of ideas. It is far easier for the graduate student, the journalist, or the researcher to reach for a popular book still in print, or a volume only an interlibrary loan away, or the carefully preserved archives of prominent or simply well-remembered members of society, than to try, painstakingly, to piece together a more faithful portrait of the full range of opinion from the vastly more unorganized mass of materials that record what the mass of most of us think. There is always more being said than what the most prominent or visible members of a society are saying. (And not only is there more being said; it matters, as a force of history. The conflation of history with elite history both rides on and reinforces the conceit that power only resides with elites—that only their beliefs and actions make anything happen.)

To put it more succinctly, this is just another example of selection bias. More than a half-century after the first broadsides in the new bottom-up social history of the 1960s were published, and forty years after the title of Howard Zinn’s A People’s History of the United States inaugurated a new formula for restoring agency to the Rosencrantzes and Guildensterns of history, the evidence historians choose to adduce remains predominantly elite, and the overblown conclusions they draw remain trained on a remarkably limited set of data. We like to criticize data scientists for analyzing whatever information just happens to be available, however biased, however unrepresentative a sample; we too readily forget that historians are data scientists, too, of a sort. All of them: even the ones who don’t do Franco Moretti–style quantitative social science or digital humanities. This is not to say that to write history one must do statistics. Historians just have a more capacious, less numerical notion of what counts as data. And when they choose to read whatever data is readiest to hand—off the shelf, in popular memory, national mythology, or other received wisdom—they almost inevitably distort the richness and complexity of the period they study. (As Sontag said in an interview for BR, “The main mistake people make when thinking about something, whether an historical event or one in their private lives, is that they don’t see just how complicated it is.”)

I say almost inevitably, because a few conscientious objectors at least take care to cast their conclusions in sufficiently modest language, acknowledging that their limited datasets do not license grand claims. (The price we often pay for this modesty is tedious prose: we sacrifice the stylistic thrill of a breezy narrative, propelled by the easy causation of a story, for more insipid accuracy. The pressure to oversimplify that no writer can escape is perhaps above all stylistic, even more than cognitive. It is not so hard to think about complexity. It is very hard to express it elegantly.) “My goal isn’t to portray the full complexity of this period, but only this very tiny part of it” is always a way out of this trap. But even these more responsible projects—the best kinds of micro-history, now the dominant mode of so much academic work, which has forsaken the writing of Bigger History precisely for some of the reasons I’ve given here—must be subjected to scrutiny: why not do the harder work of writing the history that is not so easily told (whether because it is the history of elites, or because it is the history of terrain so narrowly drawn that only very few readers will be interested)? We must always be on guard against only writing the history that is easiest to write, just as we must be on guard against only thinking about those things that are easiest to think about.