This is the core point. Yes of course the existing d3js docs are better than nothing. That isn't the complaint. The complaint is the current docs are significantly worse than older d3js docs that weren't notebook-based.
My best guess is that you’re referring to bl.ocks.org, which hosted d3 examples as standalone html examples, but this was independent from the d3 docs.
Bl.ocks.org wasn’t officially part of the D3 docs, but it played a critical role: it was the on-ramp. People learned by tweaking examples, not by reading method signatures in isolation. Saying people just want to “copy-paste” kind of misses the point, examples are how many developers build actual understanding. They’re not just blindly pasting code; they’re reverse-engineering patterns, figuring out how the pieces fit together.
We get it — you enjoy the library. But many users don’t appreciate the direction it’s taken. Such a bizarre defense. I guess we are not supposed to look at the examples when trying to grok a new library, especially a library for chart visualizations which is historically not known to be the easiest to work with.