As I read this again, and thanks for putting this back on the table LarryI , I wonder if there is a less resource-intensive way to "expand all" without losing threaded replies.
"Have my context and eat it too."
Like a mini-expand-all to return all (or most?) of the replies in short-form; first 100words, first line, (AI summary? <evil grin>) or whatever that the user may scan to find the best/right place to jump in and do a full expand.
A google search found this useful (yet old) discussion on exactly this where the researcher provides good language.
https://elezea.com/2015/09/how-to-display-threaded-discussions-on-the-web/
Teased, capped, infinitely threaded, etc.
There may be better tech available now.
Cheers.
