No Easy Lessons

From the Financial Post:

In an uncertain time, forays into the past for advice are becoming ever more common. They often make for grim reading, like the attempts to harness the “spirit of the Blitz” in the United Kingdom post-Brexit or equating the crisis of American democracy with the downfall of the Roman Republic. Without proper historical expertise, attempts to draw lessons for policy from what happened in the past often end up wandering in the wilds of history without a map. Historical data is not a house cat that purrs on command. It’s more of a wild tiger that will chew you up if you don’t treat it with respect.

To be sure, instrumental use of history goes back a long way. Nowadays, governments in countries as diverse as Poland, Hungary, and India try to write narratives of the past that fit their current goals. Earlier, communist and fascist regimes excelled in weaponizing history to achieve what they wanted. As did Roman emperors, for that matter. But recently, a different form of this phenomenon has appeared. Backed up this time with spurious “data-based” claims, these sweeping statements are often in the service of a fictionalized and self-serving version of Western history. Most of this is coming not from historians themselves but from scientists or pundits who decide that they have discovered a magic key to the past.

Last September, for instance, neuroscientist Lou Safra and her team attempted to analyze which factors contribute to how “trustworthiness” changes over time using machine learning. The authors examined historical portraiture in order to identify facial features that correlate with “trustworthiness” and discovered that it rises over the period 1500 to 2000 and that this increase correlates with “higher levels of affluence.”

Over the span of the last decade, Peter Turchin and his collaborators have championed a new approach in which history as a discipline will be replaced by cliodynamics, a new way of reading the past through discovering great patterns that explain the course of history and can even predict the future. This is not a new idea in itself. The 20th century, especially 20th-century conservatives, had a love affair with using history as futurology, with varying degrees of credibility. Oswald Spengler and Samuel Huntington wished to see patterns in the historical record that could explain not only why things happen but also how and if they happen. This often meant being very selective with said historical record. It also often meant falling for various forms of Western exceptionalism. Playing loose with history gets worse with every such attempt. Recently, evolutionary biologist Joseph Henrich and other proponents of interdisciplinary use of the WEIRD theory—focusing on societies that are “Western, educated, industrialized, rich, and democratic”—reached a new level of distortion and Western exceptionalism, claiming that the rise of the West is attributable to psychological reasons rooted in the way the medieval Catholic Church directed an overhaul of marriage rules in Europe. The prohibition of kin marriages, they argue, broke down the clannishness of Western societies.

But those studies lack an important element: context. The authors of the “trustworthiness” study did not, as they thought, take big data on human depictions and extract patterns out of them. In reality, they took fashions, cultural norms, and power structures of a colonial European polity and put them into predictions about an arbitrary feature they called “trustworthiness.” The correlations that they thought they had seen between the rise of this arbitrary feature and the rise of affluence were based on a false understanding of the societies that created those portraits and ultimately of what portraiture actually is. This led them into the vast and broad desert of racial, gender, and economic bias.

Read the whole thing. Peter Turchin was referenced earlier on this blog.