@John-1 Thanks for your thoughts.
We discussed this approach many many times over the years, but it comes down to a matter of technical limitation. We designed NodeBB from the start to be linear in format, and so in order to support a threaded model, that would require a significant rework of how we manage the pids in a topic.
That's not to say that it would never happen, but at currently, we feel the approach of exposing one layer of direct replies to be a sufficient workaround.
Keep in mind if we supported both linear and threaded formats, it would essentially double the code and quadruple the complexity. That in and of itself is a reason not to do it