It seems to me that many of the objections here miss the point of the post — at least, it certainly isn't about rational argument not working on non-rationalists. (Though at least the last paragraph here seems correct and germane.) Tom, would this be a fair summary?

From e.g. the Nazis, we see that rationality is not necessary to win (in some important sense) when you have the right narrow skills and/or luck; and since (at least stereotypically) many of these skills are hard to develop or substitute for using rationality, it's not sufficient either. We should be skeptical of claims that sincere Bayesian wannabes have a justified high expectation of overwhelming general real-world success.