Thanks for link-checking -- I've fixed the link. Yes, I was referring to spoken languages. That's an interesting point, I cannot definitely say it wouldn't generalize, BUT the converse must also hold true, you can't definitely say that it DOESN'T generalize. I'm by no means an expert in linguistics, and I hesitate to make any further arguments along this line because I just simply don't know enough. I thought ambiguity tolerance was an interesting idea, and I'm glad that my statement did its job -- it's sparking thoughtful conversation between me and people like you! I cannot win the argument of specificity and measurability, but I hope that you won't discount thinking about this hypothesis purely on the basis of lack of evidence. Anecdotally, when I first learned the tidyverse, there was a lot I didn't understand, and I adopted what I later recognized as a high tolerance for ambiguity. I know that the strength of a n=1 case is not compelling, but it goes to show that there is at least one person for whom this applies!

I guess some final thoughts are, my article was not meant to be a point by point rebuttal of Matloff. If it were, I would've made my opening paragraph, and the title of this article to explicitly reflect that. If you have any evidence that compares the base-first or tidyverse-first approaches, I'd love to see it! But until then, opinions remain opinions. And, medium is the place to discuss these opinions IMO! Don't get me wrong -- I'm not so biased that I would blindly recommend tidyverse-first 100% of the time. I know base R just as well as tidyverse. It just so happened that in my case, I learned R through the tidyverse first, and now that I've been using it for years as a data scientist, I'd say I turned out ok!