A few years after I started programming I used to spend lots of free time tweaking my editor and shell configuration. After many years, I find I'm back to doing that 😕, only I then spend a few hours every few days trying to write up my configuration for imaginary others. This most recent post I'd never have gotten around to if I'd known how much proof-reading it would take. https://akkartik.itch.io/carousel/devlog/686788/lots-of-charts

Nothing entices me to literate programming more than blogging about code. :)

I periodically noodle about a LP setup atop LÖVE. Nothing has come of it yet. The potential hot take here is that the need for LP is proportional to how static the language is and how large the codebase is. If you're in the right quadrant along those axes, you don't need it.

There’s a right amount of dynamism for not needing LP?

Dynamism isn't quite it. Maybe just sophistication? Consider Haskell. I'm not an expert, but I think it's sufficiently loosely coupled that you can get surprisingly far just by reordering definitions. This is not to say we don't need LP in Haskell. We just don't need LP tools, Lhs notwithstanding. Forget the tools, focus on the work of describing codebases. The other example I'm thinking of is JavaScript. Docco is a pretty nice tool. Even though it's not really LP, it's possible JavaScript needs nothing more. You can compose behaviors just using first class functions.