This was bold:
I got to the halfway point on “A Bridge To The Quiet Planet” and realized I had lots of notes, things I wanted to improve and tweak, and so on. In general I wanted to “tune up” what I had and improve my plot outline. So I did something kind of ambitious: I decided to review the entire first half of the book scene by scene, both adding and rewriting, as well as fleshing out the plot outline.
I don't normally recommend doing this, but the key word here is "normally".
At one point when writing Flight Of The Vajra, about a fourth of the way through the story, I ran into a situation that necessitated me backing up and ditching several thousand words. Maybe I could have faked my way through it, but it would have been painful and confusing. It's always a bad idea to write part of a story one way, then pretend for the rest of the story you've written it another way, if only because it forces you to keep track of changes that aren't actually on the page.
Sometimes you can put your head down, bluster through the rest of a draft, and fix everything next time around. Sometimes you simply can't, especially if the psychic pressure created by the need to make those changes in the first place becomes a distraction.
My main objection to stopping and turning around mid-draft is that it breaks momentum. Anything you can do to sustain momentum is helpful. But if it comes at the cost of the overall maintainability of the work, it's not worth it.
New York City
Other Lives Of The Mind