I launched this Substack on April 20, 2025 with the post A Chronicle of Revolution: the Inaugural Post. Now that a year has passed, it’s a good time to sum up my experience so far.
As readers I am sure noticed, I post regularly a substantive post on the weekend (preferably on Saturday, but sometimes it gets delayed to Sunday). In the middle of the week I usually publish a less substantive, shorter post (such as this one). So far, I posted 87 articles, which works out to 1.4 article/week. And this is on top of hundreds of posts on my previous blog, which you can browse through in this Blog Archive. It’s amazing to see how, if you work at a steady rate, the body of your writings gradually becomes increasingly more substantial.
Incidentally, while I use A.I. during the research phase, I don’t use it for drafting posts. First, I want this blog to convey my unique style and perspective, warts and all. It’s usually quite obvious when authors rely on A.I. too much — from the bland quality of the resulting texts.
Second, using A.I. for writing actually doesn’t save time for me. If I know what I want to say, I can actually type faster than (or, at least, as fast as) I speak. Incidentally, I wasn’t like that at the beginning of my career. Then, I would often sit in front of the typewriter (yes, I remember the times before computers and word processors) for hours without being able to squeeze out a single sentence. For those of you who suffer from this problem (“writer’s block”), take heart — if you persist, you will overcome it. Perspiration is 90 percent of success.
Which of my posts have had the most impact? Substack provides a variety of statistics, for example, “open rate” and “engagement rate.” But these do not seem to be terribly informative, because there is not much variation and it’s not clear what it reflects. Much more useful are numbers of new subscribers, generated by a post. By far the champion in generating paid subscribers is The Structural-Demographic State of America. It is followed by other installments in this series, by the posts on Japan, and my geopolitical posts (e.g., Sick Man Europe). This makes sense to me. These topics are where I can bring non-trivial insights from Cliodynamics to bear on tectonic shifts affecting us all.
On the other hand, my post on lemmings was largely ignored… Oh well, humans are more interesting than rodents; I should have guessed 🙂.
The absolute record in attracting free subscribers is, surprisingly enough, The French Wars of Religion I. This is also heartening — it appears that many readers agree with me that history matters.
Despite a lot of variability in success of different posts (measured in subscriptions), overall I see a steady rate of increase in the numbers of both free and paid subscribers. Both kinds are important to me, which is why I will continue publishing some substantive posts without paywall.
But of course, paid subscribers are most valued. After all, one of the chief goals of this Substack is to generate income to support researchers in my project (see About). I am happy to report that, thanks to the generous contributions from paid and founding subscribers, I now pay wages of two part-time researchers, a historian/anthropologist (whose job is checking the output from LLMs) and a data scientist (who helps with the Seshat DB maintenance and works on expanding functionality). If the current trends continue, I will be able to hire another researcher in Summer.
Many thanks to all of you who are supporting Cliodynamic research!
Bravissimo! You are a model for us all.
> It’s usually quite obvious when authors rely on A.I. too much — from the bland quality of the resulting texts.
The term I like for this is "semantic ablation". See https://www.theregister.com/2026/02/16/semantic_ablation_ai_writing/ .