Week of April 19, 2020
The Curse of Cliodynamicist Thinking
I have been hesitantly skirting a more detailed review of the “cliodynamics” school of quantitative history, which—through its long-awaited database Seshat, journal Cliodynamics, and a range of popular titles—has sought to introduce mathematical modeling into historical theory. Their most prominent author, Peter Turchin, is a retired population ecologist, and his pivotal contribution has been the application of his former statistical techniques to abstract long-run analysis. I am attracted (perhaps enough not to warrant belief, but somewhere close) to the notion that deep patterns and laws underlie macrohistorical processes, and am immediately excited by new efforts to discover them. Economic history offers a distinct, broad, and relatively rigorous method of attempting this, usually from what I conceive to be a synthesis of quantitative empirical microstudies with qualitative macro-narratives and theories. On a granular level, the researcher takes a traditional narrative—say, that slavery undercut an aspect of African development—identifies potential causal variables, and conducts statistical comparisons (from simple eye tests to regressions) to determine whether the model/story and facts coincide. This is structurally similar to the classic mode of historical research, except that theorizing more frequently precedes the data-gathering phase. A diplomatic historian of the English school, a Taylor or Namier, would walk his preconceived narrative into the nearest library to run it by the wealth of documentary files; the difference is that the test is wholly qualitative, typically confirming the presence of an actor at a certain location or the opinions he voiced at a critical juncture.
Nevertheless, both classical and quantitative historical research methods are rife with subjectivity, and I confess that—though, objectively, certain things did seem to happen and cause other things—the extent to which reconstructive interpretation is necessary for locating independent variables is debilitating. I tend to agree with John Lewis Gaddis’s line that “we evaluate our findings by asking how closely our representations fit the realities we seek to explain,” a process that he likened to “tailoring” the narrative to the shape of the perceived events. This descriptive view all but scraps my idealism by actively embracing subjectivity, but there often seems to be no way around it. No matter how much data or processing power we leverage, we must either limit our inquiries to ahistorical questions or limit our explanations to unscientific narratives. Thus I am deeply skeptical of the research program Turchin devises in his 2003 book Historical Dynamics, which evaluates a body of fundamental hypotheses about the determinants of imperial power with quantitative tests on a complex, multilevel historical system. It all seems so wonderfully simple: work out the relationships between events and broad, statistically-grounded processes of population growth and territorial expansion and simulate as one might predator-prey relations. Consider:
An empire is a dynamic entity because various aspects of it (the most obvious ones being the extent of the controlled territory and the number of subjects) change with time: empires grow and decline. Various explanations for imperial dynamics address different aspects of empires. For example, we may be concerned with the interacting processes of surplus product extraction and warfare. Then we might represent an empire as a system consisting of such subsystems as the peasants, the ruling elite, the army, and perhaps the merchants. Additionally, the empire controls a certain territory and has certain neighboring polities (that is, there is a higher-level system—or metasystem—that includes the empire we study as a subsystem). In the dynamical systems approach, we must describe mathematically how different subsystems interact with each other (and, perhaps, how other systems in the metasystem affect our system). This mathematical description is the model of the system, and we can use a variety of methods to study the dynamics predicted by the model, as well as attempt to test the model by comparing its predictions with the observed dynamics. Turchin 2003, 3.
The ideal test would follow a schematic process attractive to the humblest statistician: “define the problem,” “identify the primary data set,” “identify a set of hypotheses” to “translate” into mathematically tractable forms, and “solve”! Turchin himself admits that this is “positivistic” (he’s right, though it’s not always a sin); however, his methods face a deeper problem. Models need not exactly represent reality, and indeed, to draw useful generalizations about the mechanics of a process, they probably should not. This is no excuse for flimsiness, nor an assertion that errors in model construction cannot differ in degree. Turchin’s formula for theorizing is actually less rigorous than the hard-headed qualitative approach. Its redefinition of all relevant parameters and classes as new objects of analysis within a novel “subsystem” permits infinitely subjective characterizations of the facts and actors involved. What is a merchant? Must he abstain from all manufacturing processes, or can he be a putter-out of craft goods, too? Can a peasant be a merchant if he conducts some short-range commodity transport on the side? Has he become a capitalist if he employs a hired freeman or two? Where do city-states and informal political leagues fit on the medieval state spectrum? Do empires as fragmented as the Carolingian deserve the name? Turchin can answer these types of classification questions in any way that seems convenient, and thereafter ignore them (or fail to do them analytical justice) by focusing on the “interesting” and “valuable” results of his hypothesis testing. His p-values will be all but meaningless, however, because he can adjust the data at will to produce a desired arrangement. For all its pretensions to positivism and objectivity, the complex historical system risks epistemological emptiness.
The consequences of this kind of cliodynamicist thinking are obvious and infectious. Turchin helpfully reminded his readers of this peril, however, by following up on his post “Coronavirus and Our Age of Discord” with another attempt at associating disease outbreaks with periods of historical crisis. The former displayed sufficiently loose reasoning, accompanied by a ridiculous graph, that I went back to check on his original methodology in dismay:
[M]ajor pandemics tend to happen during Ages of Discord… There are several general trends during the pre-crisis phase that make the rise and spread of pandemics more likely. At the most basic level, sustained population growth results in greater population density, which increases the basic reproduction number of nearly all diseases. Even more importantly, labor oversupply, resulting from overpopulation, depresses wages and incomes for most. Immiseration, especially its biological aspects, makes people less capable of fighting off pathogens. People in search of jobs move more and increasingly concentrate in the cities, which become breeding grounds for disease. Because of greater movement between regions, it is easy for disease to jump between cities.
Elites, who enjoy growing incomes resulting from low worker wages, spend them on luxuries, including exotic ones. This drives long-distance trade, which more tightly connects distant world regions… As a result, a particularly aggressive pathogen arising in, for example, China, can rapidly jump to Europe.
Finally, when the crisis breaks out, it brings about a wave on internal warfare. Marauding armies of soldiers, rebels, and brigands, themselves become incubators of disease that they spread widely as they travel through the landscape.
His latest post adduces a chart linking the various Ages of Discord with historical pandemics, from the Bronze Age Collapse (of 1177 B.C. fame) to the 1918 influenza pandemic. Considering the outsized historical role that disease plays, as well as the links between globalization, population growth, and some previous pandemics, the argument is actually quite standard. His mechanistic description is not, however. First, what exactly is an “Age of Discord”? In short: anything that Turchin needs it to be, which in this case might simply entail grouping a disturbing event and a plague in the same sequence. His three most recent instances are the Crises of the Sixteenth and Seventeenth centuries and the “long nineteenth century” (1789 to 1919)—in short, periods spanning nearly four consecutive centuries with a seventy-year break in the middle. With such loose bounds and the prevalence of disease outbreaks in preindustrial Europe, he can hardly help but make an “association.” His periodization, however, is both completely meaningless and unhelpful. And lest you object that mainstream historians do something similar, examine the graph. On the y-axis stands an indefinite “pandemic index,” while across the top we see the supposed “Ages of Discord”… which overlap, and have no demarcations whatsoever. The distinctions between plague and cholera are completely obfuscated, as are the locations of incidence (Venice? Genoa? Krakow?). To return to the excellent Carlo Cipolla:
In England between 1351 and 1485, plague broke out in thirty different years and between 1543 and 1593 it broke out in twenty-six years. In Venice between 1348 and 1630 plague broke out in epidemic form in twenty-one years. In Florence too plague broke out in twenty-two years between 1348 and 1500. Between 1348 and 1596 Paris was hit by plague epidemics in twenty-two years. Between 1457 and 1590 Barcelona suffered from plague epidemics in seventeen years. Cipolla 1974, 132.
Disease was ever-present during the late agrarian age, and so were episodes of globalization and population growth—not consistently in one place, but somewhere in the Eurasian ecumene. With sufficiently loose historical thinking and inadequate definitions, one can collate apparently compelling examples into as subjective a narrative as any quantitative historian and produce a grand theory. Considering the ambitions and natural-scientific origins of the cliodynamicist project, this result is profoundly disappointing. I deny neither the failings of qualitative models and historical narratives, but I refuse to remedy overreach with naivete.
If I go missing for a while in the near future, you’ll find me in search of an exculpatory passage in one of Turchin’s books. We do need mathematical theory and grand speculation in history, and now more than ever. But that should be no reason to do either badly.
Grendel (1971), by John Gardner
If Gardner had only engaged the purely stylistic and literary tasks of reproducing Beowulf from an antihero’s perspective, Grendel would nonetheless be a remarkable novel. Savage and pitiable, hilarious and morbid, Gardner’s warped narrator views the world like one of Richard Adams’s rabbits disenchanted, bemused and amused by the bizarre human world just beyond his participation. The legendary prehistory of the already-legendary poem is transformed into a compelling, almost anthropological origin story for both Grendel and humanity itself—replete with political development, religious transformations, and ecological catastrophes. The monster is the focus, but the few substantial characters are compelling, from the hero-turned-cynic Unferth to the barely-anachronistic Hrothgar. Even the intellectual caricatures, like the bloodthirsty radical Red Horse and the devious priest Ork, are engaging diversions. Grendel’s gaze unsparingly defamiliarizes political order, socialism, and religion, but deftly avoids the trap of clumsy misanthropy—recognizing both the idiotic and admirable in our fumbling species. His experiments with linguistic form rarely obstruct the reader from his unencumbered view of civilization, stripped of its pretentious trappings to a rutting, sneaking, boasting mess. Grendel’s narration is primal and beastly, and so are we.
Yet Gardner aimed at much more than a skillful lampoon. Grendel is a half-converted Sartrean, driven by delusion and despair to doubt the efforts of his human neighbors (and his own) at constructing meaning. Their self-assertion and theorizing inherently appear to him empty, banal, especially across the never-defined species barrier that separates him from their rituals. Yet there is something inescapably compelling in humanity’s struggle to wrest value from cosmic materialism, a quest embodied by the persistently attractive Shaper-poet, whose recitations of the text inspire Grendel to fits of adoration, jealousy, and loathing. To believe in heroes and justice is inspirationally, enviably human, endowed with an appeal that frequently defies the materialistic nihilism of the dragon. “Why shouldn’t one change one’s ways, improve one’s character?” he protests. What, wonders Grendel, can one glean from a world of accidents, in which the mind is “merely a new complexity” among a billion others—an event reducible to the ticking of the cosmic clock? As it turns out, the answer is nothing. Arbitrarily standing in the “brute existent” by which man realizes the need for self-definition in the face of inevitable destruction is unsatisfying and useless, as insipid as sitting on a gold-hoard while the stars wink out. Beowulf’s arrival is a blessing for the corrupted Grendel, because prior to his triumphal landing there was nothing left to do. Kill or spare, create or destroy; all impulses and virtues lapse into synaptic equality.
For a deep, explicit battle with Sartre, look elsewhere, but for an entertaining appraisal of Western history and philosophy with a note of real existential dread, dive beneath the shaggy waif on Gardner’s cover.
An original study of hunter-gatherer social networks suggests that settlement clusters, kinship networks, and intercamp mobility helped to overcome geographical barriers to cultural transmission when human societies remained at low densities, accelerating the spread and retention of ideas and practices. Interesting, and related to last week’s essay on a similar theme.
Time Discounting and Wealth Inequality, by Epper et al
Another paper associates patience and saving behavior (“low time discounting”) with higher positions on the wealth distribution, and finds that the relationship both persists across a wide range of prominent inequality drivers and is as influential as years of education. From an economic history perspective, the paper stands as further evidence for the kinds of bourgeois/capitalist behavior (as argued by Greg Clark) that underlie successfully developed economies.
Industrial Espionage and Productivity, by Albrecht Glitz and Erik Meyersson
An amusing review of espionage files in East Germany finds that illicit information flows significantly reduced the TFP gap across the Iron Curtain (by 13.3%)—evidence that Soviet regimes worked best in copying Western techniques and for the importance and existence of catch-up growth in general.
Natural Law and History, by Graham McAleer
McAleer, in a thoughtful and erudite essay, compares the legal-historical theories of Leo Strauss, Thomas Aquinas, and Martin Heidegger in defining the concept of natural law (related to intrinsic impulses, as opposed to historically-defined positive law). Somehow, he preserves a measure of lucidity in summarizing a summary of several different summaries, delivering a nuanced interpretation of the effort to locate a transcendent moral basis of legal action.
Finally, I’ve got a new article out this week in the Berkeley Economic Review. There’s a little more levity than usual, so it might be a nice change of pace.