Italo Calvino’s grave and lovely novel, Invisible Cities (1978), ^{Footnote 1} describes a debate between Marco Polo and Kublai Khan about chess and social inquiry. The Khan wants to simplify his vast, diverse, and exasperating empire into something corresponding to the stark simplicity of a chess game, in which the rules that govern the “angular shifts” of the knight, and the “diagonal incursions” of the bishop would somehow capture the hidden logic of the empire and its cities. Polo argues back by pointing to the quiddities and contingencies of the chessboard that the emperor is using:

Calvino’s imaginary argument is one version of the enduring fight between nomothetic approaches, which seek to establish abstract and universal laws, like those that dictate the knight’s hopscotch perambulations across the chessboard, and idiographic ones, which devote themselves to the particulars. ^{Footnote 3} Roy Germano’s argument on behalf of analytic filmmaking wants to discover abstract rules, but through a medium better suited to nuances and contingency. There is a tension between Germano’s aspirations towards nomothetic explanation and the specific advantages of film making, which he suggests are found in the testing of nomothetic arguments, but which are really in public presentation, and in film’s ability to capture and preserve the nuances that are strained out of the mixture by more traditional forms of social inquiry.

Germano’s ambitions are clear. He wishes to contribute to social science as traditional social scientists commonly understand it, that is, to the steady accumulation of law-like generalizations that can be tested using standard evidence. As he describes it, analytic filmmaking emphasizes “nomothetic explanation over descriptive storytelling and character development” (p. 665). At greater length,

The problem, as Germano effectively acknowledges, is that analytic film does not represent any very useful method for testing nomothetic statements. More traditional forms of quantitative and textual analysis are much better suited to this kind of testing. From Germano’s perspective, film still has two benefits. First, raw footage presents a repository of unprocessed information that can be referred to later in order to detect and correct earlier mistakes of interpretation. Second, appropriately edited films can illustrate causal processes “more vividly” than text. While both of these aspects of analytic filmmaking are useful, they are largely ancillary to the scientific project. If analytic films serve primarily to check errors and to present scientific findings to the broader public, they are hardly central to the process of scientific discovery.

There is a broader argument lurking behind Germano’s unnecessarily narrow justification of analytic film. As he says, film can capture information—nuances of facial expression, of tone, of body language—that written interview transcripts and quantitative data have difficulty conveying.

The problem is that such information is typically not the kind of information that nomothetic accounts can make use of. It is difficult to formulate universal laws about what a shrug of the shoulders or a wink means, because a shrug or a wink can mean very different things in different contexts and to different audiences. They provide information about the knots and woodcarvers’ gradients in the chessboard, but not, or at least not directly, the abstract rules of the game of chess. There is, of course, a body of scholarly knowledge that speaks to such nods, shrugs, and winks, but it is explicitly and emphatically not nomothetic. It is the approach of the late Clifford Geertz, who prized idiosyncrasy over universals and who quoted Thoreau to emphasize that we did not go to Zanzibar to count the cats. ^{Footnote 4}

An analytic filmmaking that was strictly nomothetic would do its damnedest to filter and strain out such ambiguous and equivocal information, retaining only the purified element of abstract scientific knowledge. It would fail, of course, since it would necessarily convey more and more ambiguous information than its maker intended it to convey. Such failures might be both self-deconstructing and constructive. A self-aware analytic filmmaking might indeed exploit such contradictions to create a social science equivalent to Rashomon, transforming brute data into a space for Keats’ negative capability, in which “multiple possible causes for an outcome can be allowed to exist alongside each other without being resolved, or even given definitive weights” (Spufford Reference Spufford2012). Robert Irwin’s novel, The Arabian Nightmare, describes a terrifying world of dream where there are always more causes than events; this may be closer to the current state of the social sciences than we usually care to imagine. ^{Footnote 5}

Such radical playfulness is likely to make most social scientists uncomfortable. The analytic narratives employed by Robert Bates, Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry Weingast present a different, and perhaps more practical model for analytic filmmaking (filmmaking is, after all, certainly a form of narrative). ^{Footnote 6} Bates and his colleagues do not begin with the ambition of discovering, or testing, universal laws; instead they start from specific empiric puzzles. They employ the strictures of game theoretic reasoning to build abstract models (although one could employ other forms of reasoning to generate them). The detailed causal narrative is not a simple application of the model; instead, the narrative and the model argue with each other, inform each other, and end up to some degree changing each other. As Margaret Levi argues, some degree of contingency is necessary to make an analytic narrative—if it were simply an illustration of some grand abstract law working out its ineluctable logic, it would lose its complexity and most of its interest. ^{Footnote 7} The tension between the abstract and the particular, the model and the story it is supposed to illuminate, is a useful one.

It is understandable that Germano appeals to nomothetic language in order to justify his claims for the virtues of analytic filmmaking. This language has appeal to political scientists, even if they have tacitly abandoned the grand ambitions of their predecessors to construct a vast edifice of law-like generalizations with universal application. Yet an analytic filmmaking that seeks only to illustrate abstractly hypothesized logics will fail to deliver on its promise. Because films are dense, rich narratives, they can convey an intimacy of detail that standard kinds of social science knowledge cannot. Yet it is exactly the most valuable details that are least likely to fit well with the promised “hypothesized logics.” Films, like all good narratives, will surely escape from the shackles of empiricism that we use to try to subdue them, and we can discern their true worth only if we recognize this.