This week I took Three Cages Arc I, the first nine chapters, the Cage of Protocol, back through a full editorial pass, one chapter at a time. The goal wasn’t new prose. It was verification: make sure the hand-edits I’d been making chapter to chapter actually flowed, didn’t leave artifacts, and hadn’t quietly introduced new problems while fixing old ones.
Nine chapters. Two passes each: once to read and flag, once to verify the fixes landed clean. Here’s what came out of it.
What the continuity pass caught
This is the unglamorous work and it’s most of the job. A handful of small contradictions that no single read would catch but a cross-referencing reader absolutely would:
- A nightmare-journal entry quoted with different wording in two chapters that were the same calendar day. Rather than flatten them, I reframed it: the entries are re-reads of the same longer log, and they read differently each time because the dreams are either getting clearer or she’s getting worse at pretending they’re only dreams. Both versions true. The contradiction became characterization.
- A character’s shrine age drifting between “twelve” and “thirteen” across chapters, which matters because the timeline hangs on it. I searched the whole book to confirm I’d killed the last instance.
- A day-count off by one against the two chapters around it.
- A misquoted “exact phrase”: a character claiming verbatim recall of a line that didn’t match what was actually said earlier.
- A point-of-view leak: a first-person narrator describing a gesture she wasn’t in the room to see.
None of these are dramatic. All of them are the kind of thing that, left in, makes a careful reader trust the book a little less.
The real find: the density tell
The most interesting pattern was about prose tics under pressure.
Two chapters read heavier than the rest. They turned out to be the two highest-intensity chapters in the arc, a major plot turn and the arc climax. Same culprits in both: a single simile construction inflating to fifteen-plus uses where it should appear maybe once, and a verb triad running triple what the prose can carry before they start to show.
The lesson going in the notebook: my crutch constructions spike under set-piece pressure. When the scene gets dense and the stakes climb, the prose reaches for the same comparison frame and the same handful of verbs over and over. The quiet chapters stayed clean the whole way through. So the fix isn’t a blanket rule. It’s knowing to vary the vocabulary going into the next big scene, not cleaning it up after.
Reworking those two chapters, substituting verbs Kuronami’s hands-on-stone magic actually needs, converting the repeated simile into direct statement, is what tightened the arc most.
Where I deliberately stopped
The most useful decision of the week was not over-sanding the prose.
I could keep planing down the repetitions across every chapter. But a lot of what’s left isn’t sloppiness, it’s the voice. Signature words. Recurring phrasings that are doing real character work. A climax that needs its repetitions because the contrast is the whole point: routine measured against catastrophe.
Strip all of that out and the prose gets smoother and reads more airless, not less: sanded flat, no texture, no fingerprints. A book should sound like a person wrote it under pressure, because someone did. Past a certain point, “cleaner” and “better” stop being the same direction.
So I called it. Arc I is sitting where the voice wants it. Done is better than buffed.
Next
Arc I is locked. A couple of small forward notes parked for later: a reference doc that needs to catch up to a manuscript edit, and one character beat to keep an eye on as it pays off down the line. Then on to Arc II, the Cage of Flesh.