When I was younger and more naive about the ways of the world, I used to be very annoyed by people doing a pattern that for a while I called "recursively wrong."

Like if I talk to somebody and I identify a mistake in their reasoning, it's fairly likely they either "double down" on their mistake (most frequently), or sometimes acknowledge I was right on that specific point but then they claim they were right on the broader point, and then immediately make the same meta-level mistake.

I used to be pretty annoyed by this, and thought that the only reasons someone could be this massively wrong is a combination of bad faith, trying to annoy me specifically, and abject stupidity.

I thought about it since and now I think it makes no sense to be annoyed by this. The simplest and most elegant explanation for why someone's "recursively wrong" is that my diagnosis was correct in the first place!^[1] Of course they'd be "recursively" wrong. If my original diagnosis was correct, that just means they're wrong!

Continuing to be annoyed here is just a meta-modeling failure on my end, and/or double-counting one mistake into several.

Hammering home the same error is also not a correct conversational move, since it's unlikely we can reach convergence on either the object-level or meta-level points here if I try to focus on the error.

I think if I identified an error in their reasoning (eg somebody ignoring base rates, or they have a failure of theory of mind, etc) the epistemically more positive/collaborative thing is to mention it once and then try to steer the conversation towards other lines of evidence.

1. ^{^}

   To be clear, I don't give evidence within this comment why I believe this. You can have a symmetric story where I'm entirely wrong and the other party's being equally rational by being extremely exasperated why I'm denying an entirely valid line of evidence by calling it mistaken.