One of my favorite analysts, Peter Turchin, is one of the few people who loves data even more than I do. He quite usefully chose a pair of opposite predictions concerning the Ukraine war back in 2022, one from Paul Krugman and one from Scott Ritter, and constructed models on the bases of those predictions in order to track the way the war unfolded.
Now, I could have told him that Paul Krugman’s model would be wrong, because Paul Krugman is always wrong. But that’s some high-level UHIQ pattern recognition in action; warning: do not try this at home! In the statistical world, one has to at least pretend to take his predictions seriously and give them a fair shake, even though one has a very high level of confidence that they’ll comprehensively fail.
One of the topics that I wrote about in End Times was Ukraine. After I turned the final version of the text to the publisher in late 2022, I continued monitoring the news about the course of the conflict there, because I was curious to see how well my assessment of the Ukrainian state (a plutocracy) and the war there (a proxy conflict between NATO and Russia) would fare as history unfolded. It was, thus, interesting to see that in the early 2023 the views on this conflict, and predictions about its future course, could be so diametrically opposed, depending on who was writing and what ideological background they came from. The tone in the MSM (main-stream media reflecting the official American position) was quite triumphant. But many American analysts, former military and intelligence professionals, held a very different view.
It occurred to me at that time that this difference in predictions is actually amenable to an empirical test. As long-time readers of my blog (now here on Substack, previous posts archived on my web site) know, I view ability to empirically test predictions from rival theories as key in doing Science (with a capital S). Just search my blog archive using the keyword “prediction” and you will see multiple posts on this subject. So I decided to conduct a formal test.
For concreteness sake, I selected two predictions, both based on an explicitly quantitative argument, but coming from opposite ends of the ideological spectrum. One was from Paul Krugman, channeling the official American position. The other was from another American, who is, however, considered as a “rogue actor” and a “Putin’s stooge”, Scott Ritter. You can read the actual quotes from both in the Introduction of the SocArxiv article, in which I “pre-registered” predictions of my model.
I won’t repeat the details here, because you can read them in the series of blogs I published two years ago, followed by the SocArxiv article that put it all together in a systematic manner and provided R scripts that allow others to replicate all my results.
They’re all well worth reading, although by the middle assessment, it’s already perfectly clear which of the two models, which Turchin labels the Economic Power model (Krugman) and the Casualties Rates model (Ritter), works better, although he combined elements of both into what he describes as an Attrition Warfare Model that appears to outperform both. This makes since, because what really matters most is Industrial Capacity and Male Population Demographics, both of which are presumably incorporated in Turchin’s AWM.
And he explains exactly what his AWM suggests at the moment.
As you can see (the dashed red line “We are here”), we’ve already entered the region where Ukrainian army can collapse at any moment, although this “moment,” according to the model can happen at any point between now and February 2027 (corresponding to 60 months after the start of the conflict). As I explained in my posts and the article, the final outcome is not much in doubt, but the rupture point is extremely difficult to predict. The situation is akin to seismology. For example, the recent Kamchatka earthquake of exceptional power was predicted 30 years ago, except nobody could know when it would actually strike. The Attrition Warfare Model is actually more precise than that. From its point of view, it would be a surprising outcome if Ukraine is still fighting beyond February 2027.
Note that I said, “from its [the model’s] point of view.” I emphasize that the future is unknowable in precise terms. In any case, the goal of this article was not to predict the future, but to use the method of scientific prediction to empirically test between two, or more theories.
The Attrition Warfare Model (AWM) encodes both alternative theories, (1) the Economic Power hypothesis, which predicts a win for Ukraine (Krugman) and (2) the Casualties Rates hypothesis, which predicts a win for Russia (Ritter). It is clear that the first theory will be rejected, no matter when the war ends.
Turchin’s work can be a little wonkish for the average individual to follow, but it’s not as complicated as it might look at first. He keeps things simple enough, and his writing style is clear enough, that with just a little concentration, that it’s both insightful and educational for anyone with the intelligence to be paying attention to these small matters of war, revolution, and societal survival.