Week one with an agent, you scrutinize every line it writes. It's right. Next time you scrutinize less. Still right. By the twentieth time, you glance at the title and nod. Then once — the very time you didn't look — it gets something small but serious wrong, and that mistake walks straight into real work because no one was at the door anymore.
This trap is hard to see because it isn't a decision. Nobody actively says "I'll stop checking now." Trust just drifts — each success loosens it a notch, so smoothly you never notice the moment you let go.
01Trusting more isn't because it's more right
This is where intuition betrays you. Twenty rights create the feeling "it's trustworthy now" — but those twenty didn't make it more right on the twenty-first. Every turn still carries the same chance of being wrong as the first, because its nature hasn't changed: still something brilliant, forgetful, and confident even when wrong. The only thing that changed is you — you stopped looking.
Accumulated trust is a feeling about the past. The accuracy of the next turn is a matter of the future. Confusing the two is the root of this trap.
02Check by risk, not by feeling
The way out isn't to check everything forever at the same intensity — that throws away the very speed you're buying. The way out is to anchor checking to something that doesn't drift: the work's risk, not your level of trust.
Low-risk work — a local change, easy to spot if wrong, easy to undo — should indeed get a light check, even the first time. High-risk work — touches shared things, hard to undo, fails silently — gets a thorough check every time, even when it just got ten right in a row. What sets the level of checking is the consequence if it's wrong, a quantity that lives in the work itself and doesn't budge with a recent run of luck.
A sign you're drifting: when you catch yourself waving through something high-risk just because "it's done these well lately." That's not a reason to trust — it's the creak of a door slowly opening with no one on guard.