To me, a story is a puzzle to be fit together carefully. This is especially true of a mystery, where you have a lot to weave in about what the audience doesn't know, and what's going to happen, and what's a lie and what's the truth -- but it's also true of other kinds of stories.
For this reason, I do most of my revision as I write. By the time it's complete, the pieces really do lock into place, and they cannot be easily moved. So rewriting pretty much involves fine tuning -- polishing up, expanding a little here, and editing there.
Alas, I have discovered that even my way of working doesn't prevent me from revising myself into trouble. I just do it earlier in the process. Sometimes you just have to trust that Heinlein knew what he's talking about. Sometimes you should not second-guess your instincts. Even when it's a different set of instincts that starts you rewriting.
I've been working on the middle sequence of The Man Who Did Too Much lately. It is a relatively simple development of a subplot, with a comic turn. That turn is important to the overall story of the characters, but as I originally blocked it in, the sequence didn't really advance the mystery plot much. It had to do with the interpersonal drama which was tangled up with the mystery -- a place where we learn some deep background and are introduced to hidden clues -- but it was mostly character development.
At one point I looked at it and thought: "Gee, this is an awful long sequence which doesn't exactly move the story forward in any obvious way. I need to pull it apart and shake it up. Introduce a "man with a gun" or something."
This is a very good instinct -- that a story has to move forward overtly. It's critical. I wasn't wrong to be concerned about this.
So I pulled the sequence apart. I unraveled it and started playing with ways to bring this back to the main story -- to weave in new threads. I moved everything. But then I realized that if I put that incident here, this other one wouldn't work, so I had to move that around. Drop this major thing , and play up that minor thing here instead. Piddle, twiddle and resolve.
It really required re-envisioning the whole sequence, but by golly it was new and exciting and it really worked! And I let it ripen a bit and worked on other things, but it kept getting better and better!
What actually happened was that I kept running across problems with finely tuned things I'd done in the original. If a certain incident happened later, it took all the punch out of a certain joke, and that joke was the energy that drove this other thing. So I kept coming up with those, and finding other ways to handle it. Taking extensive notes and scribbling out bits of new material. (Note: this was all happening in my head or in outline -- I had not actually accomplished pulling it all back together.)
I kept fixing things until it was way way way better than before, and then I got too busy and set it aside completely. And while I was away from it, everything quietly moved back into place.
Where it belonged.
My original instincts were absolutely right. I should have paid attention to the rest of that quote from 1776: Piddle, twiddle and resolve... not a damn thing do we solve.
That joke was critical to the movement of the story. That other incident had to happen when and where it did, and HOW it did. Moving it to later, and making it the "climax" of the sequence just made it the whole darn sequence feel contrived. Even though it seemed to support what happened next better, that wasn't the natural place for it to happen, so it didn't help the next bit. It just made it feel contrived too.
Heinlein was right. Sometimes we're just too smart for our own good.
On the other hand, Heinlein was wrong in this: In my efforts to stretch this rubber sequence around to fit a different mold, I created a lot of new pieces, and explored a few more things. I also figured out that I didn't need to change the flow of the action, or even the timing, but I did have to move around bits. Jokes, information, turns of conversation, even clues -- a dozen little things which support the action. This joke is not needed here, but if I put it there, it actually becomes a really good reason for the change in mood which happens there.
Without getting into any spoilers, here is one detail that may illustrate why revision was a problem:
George is drunk. Not plastered, but buzzed. And one of the things that happened with the misbegotten revision was that he sobered up to deal with some things in the middle. And that means the emotional trajectory of the sequence changes. Once he sobers up, he's done being drunk, and done with everything associated with it. It doesn't matter if it's logical for him to continue to drink and get drunk again later -- that's DONE dramatically speaking. Even if he doesn't actually sober up -- if he just acts more sober -- getting him drunk again means starting a new sequence. And a repetitive one at that.
It wasn't obvious, though, because the drunkenness was just a background bit. It wasn't overtly important to what was going on. It's like any character mood. Like being depressed or euphoric. Or afraid. Or brave.
Emotional state is trajectory.
A character's emotional state is everything. It affects not only how the character behaves, but how others behave toward that character. And when it changes, it changes everything -- you can't just go back and forth.
So if George being drunk is the trajectory of the scene, it's actually the front story. I don't need to interject more front story. I need him to get more drunk. Suddenly everything works when I do that. Even the new material -- it all gets woven neatly into the sequence where the headline is that George is drunk. (Furthermore I don't lose anything by it, because, contrary to appearances, George is NOT the brains of the outfit. Karla is.)
And, just as a general rule, letting the characters spin more out of control is a good thing, dramatically speaking. It pushes the envelope, and takes you places you might not have thought of.
True, it is also risky to go with instincts all the time. You can ramble off into nowhere, fly off into space. But that's why you do it early in the process. What I'm learning with this sequence is that the key is to respect the trajectory, and if it doesn't take you where you want, maybe you should take it further. Fiddle with it, but the key word there is with.
Otherwise you're fighting with the forces of nature.