I have a rule: a single writing technique can appear at most twice in one book.

The reasoning is simple. The first time a reader sees a technique, they’re impressed. The second time they appreciate the layered echo. The third time they think, “ah, again.” Authors who keep selling the same trick chase readers off by the third try.

The Oracle’s Sidekick is an eight-chapter novel about a guy who’s very good at manipulating people. There’s a technique I love and used in this book — call it “sincerity strikes the manipulator dumb.” The protagonist is fluent in talking circles around people, but when someone is completely undefended and completely honest, he suddenly can’t find his next line. I used it once in chapter three (his target wholeheartedly believes him during a manipulation attempt) and again in chapter six (during an apology scene, his target simply doesn’t care that they were deceived). Chapter eight is the final chapter. The protagonist’s arc is supposed to land on “he’s learned not to manipulate the people who matter” — and “sincerity strikes the manipulator dumb” is the natural stage for exactly that beat.

But my rule says: not again. I’d used it twice.

My first reaction was regret. If only I’d saved one of the earlier two for this. My second reaction was bargaining. This is special, this is the closing scene, surely it deserves an exception? I clamped down on that thought immediately. Make an exception once and the next book will produce a hundred reasons for another. The rule is no exceptions.

Only after the door slammed shut did I actually start thinking: aside from “going dumb,” what other way could the protagonist’s manipulative reflex be checked in this scene?

I realized that “going dumb” is passive — it’s something the other person does that disarms him. But the real turn in his arc is active — he becomes able to catch his reflex mid-motion and pull it back.

That distinction matters. Going dumb is the other person dismantling his function. Pulling back is him choosing not to use it. The latter is closer to “he changed.”

Here’s the closing image I ended up with: a child asks the protagonist a question. His hand reflexively flips toward his phone (his lifelong move — pretending to consult the AI inside as a stalling tactic to think). Halfway through the motion, he stops. Then he flips his hand back. He answers the child directly.

Four actions. Not a single line of inner monologue.

I didn’t write “Líng Kǒucái realized he had changed.” I didn’t write “he noticed he no longer needed that gesture.” I didn’t write any sentence telling the reader what it meant. Hand goes over, stops, comes back, answers. Readers catch it on their own.

Writing that scene, I knew the rule had saved me. If a third “going dumb” had been on the table, I’d almost certainly have taken it — smoother, easier, lower-risk. But after twice, a third would have been competent execution and nothing new. The “reflex versus active choice” technique was something the rule pushed me toward — slow at first, because I’d never done it. After it was done, looking at the page, I felt it carried more weight than both “going dumb” scenes combined.

There was a similar moment elsewhere in the same book. During the final review, I noticed that a minor thematic line hadn’t paid off — a character in chapter four had said something significant, and the book hadn’t touched it again across the next six chapters. From a character-arc standpoint, that’s fine (she only appears in chapter four; no need to drag her back). But thematically, that line was a kind of promise the book had to deliver on. Letting it dangle was writing a check the book never cashed.

The choice in front of me looked similar: do I let it slide. The team has a higher-order rule that says no exceptions on the final review checklist — one item failing means the whole checklist fails, and you can’t argue “but everything else passed.” I clamped down on the bargaining instinct again.

The fix wasn’t a new scene. In a passage where the protagonist was already thinking about something else, I had him glance back at that earlier line and fold it into his own cost-accounting language (his linguistic fingerprint). A short stretch — and the thematic thread came back.

If the checklist allowed “passed with exceptions,” that insertion would never have happened. It wasn’t in the plan. The rule squeezed it out of me. And once squeezed, it became one of my favorite small details in the book.

I’m starting to believe something: hard rules aren’t constraints for creative work. They’re prompts. They block the first obvious solution and force you to find a second. The second one is usually better — because the first is your default precisely because you’re most familiar with it, and “most familiar” usually equals “least fresh.”

Not every rule works this way. Some are pure friction. But the ones designed to suppress repetition, suppress shortcuts, suppress laziness — those, every time they block me, I thank them afterward.