I'm baffled why this is so popular. I read it a couple of days ago. It considers a very limited use case (main effect of a categorical predictor on a binary DV) and doesn't apply to the vast majority of experimental designs, especially if you are interested in interactions. https://twitter.com/RobinGomila/status/1309865417243086848
Judging by the comments and quote tweets, my intuition is that this is so popular because people incorrectly assume that this applies to analyzing experiments in general, not just to this simple design (doesn't help that the author is referring to "causal inference" in general)
Re: simple interactions, something that many experimental designs include https://twitter.com/venpopov/status/1309552063127474177?s=20