This was helpful, thanks! goodpractice, rhub, and vcr were new to me.

One package that I find useful that ties many of the pieces together that you've mentioned is precommit: github.com/lorenzwalthe...

I do find Claus’ other arguments quite compelling (suitability of functional programming for data processing, fractured ecosystems in Python, etc.). I also attribute much of it to having the good fortune of having Hadley, Lionel, and others creating an incredible ecosystem in R.

Being able to supply bare symbols as arguments makes for nicely readable code when you are referring to column names only, but the flip side is that your code becomes a bang-annotated mess when you are unquoting variables. I don’t see this a better than col(‘colname’) with no need for unquoting. 2/2

This is a thoughtful pair of articles and Claus certainly has the wealth of experience to draw these conclusions from. I remain unconvinced that NSE is a fundamental reason why R is so nice for data science. 1/2

I always upvote bookshelves! And yes, I do spot a couple of my favorites.

This has inspired me to post a picture of mine as my first ever Bluesky post: bsky.app/profile/dapr...

My prediction is nearly the same as yours - that R will continue to be the place to go for to use advanced stats methods (in academia or otherwise), but Python will claim most of its market share over time for other data science tasks.