The Cost of Under-Trusting

Under-trusting happens when you rewrite everything because it's not exactly how you would have done it. This defeats the purpose of direction. You end up producing the work yourself, just with an extra step of delay. You've saved no time. You've gained no leverage. And you've signaled to the executor that their work won't be trusted anyway, so why should they care about getting it right.

Under-trusting looks like this: you get output back, you read it carefully, and instead of evaluating it against your outcome standard, you evaluate it against how you would have written it. The phrasing is different. The organization is different. The emphasis is different. None of these are wrong — they're just different. But you rewrite because you want your version. You've confused "different from what I would have done" with "actually insufficient."

The cost is multiplied with AI. AI can produce many iterations quickly. If you under-trust and rewrite, you're essentially saying "don't use the executor in any meaningful way." You're using AI as a research tool or a first-draft generator, not as a capable executor you're directing. That's fine if that's what you want to do. But it's not direction, and it won't scale.