The Production Cycle Does Not Allow for Major Revisions
The tl;dr crowd can breathe a sigh of relief; my fingers and mind grow weary.
The problem with writing the now-standard kind of RPG dialogue is that it takes so long that it is seldom finished far in advance of whatever the final hard cut-off date is (localization, vocalization, shipping, whatever). Only in journalism, I think, is there the same compression of writing and publication. In other media -- in my experience, in legal writing, in fiction writing, and in academic writing -- revision is critical, and often takes as long as the writing itself, if not longer.
One major aspect of revision in other media is that it happens at the end. To be sure, writers revise on the fly. But when the piece is done, editors (including the author) read it from start to finish and make considerable changes. Another draft is produced, and the process begins again. Then you have polishing drafts -- your tool slowly goes from a hacksaw to fine-grained sandpaper. For important pieces of writing, I've had more than 20 or even 50 rounds of revision.
This is simply impossible with RPGs.
There is too much text, it is finished too late, and it cannot simply be "read" because often how it works is not apparent without playing it, and playing it in context. Of course, individual conversations are reviewed as they come in. And sometimes they are reworked quite a bit. There are conversations in TTON that had 40+ versions in Perforce. But I am skeptical that it is possible to undertake major revisions, let alone multiple major revisions, once the whole game is written.
In essence, the nature of the beast is that by the time the writing is done, there is no time to edit it. It would be like tearing down a house after it was built because it turned out to be not quite what the builder wanted.
To some degree this could be dealt with by reducing the amount of text, but I doubt the text could be reduced enough (in the kind of RPG we're talking about) to make such revisions feasible. Thus, I think that an inherent feature of narrative, dialogue-heavy RPGs is that they are immune to the rewrites and structural changes that you would tend to see in other written works, despite the fact that editing and iteration are critical to good writing.
I think there are some people whose response would be that this is an argument to return to the much slimmer text of RPGs before the mid-90s and to focus on other means of story-telling than dialogue. That is a whole other debate, and one that's a little hard for me to wrap my head around because it is at least a little bit like saying, "If you find fantasy novels too long-winded, you should just watch fantasy movies." There obviously is a huge swath of players who like dialogue-tree-based RPGs, and so I think trying to abandon the form altogether is probably not a great idea.