In computer science, a policy is a rule applied in all scenarios. It’s a consistent behavior we can test and evaluate over a long horizon. We don’t judge policies by whether they work in one case, but by how they perform across thousands.

By that measure, honesty really is the best policy.

Dishonesty often looks efficient in the short term. You can close someone on a false pretense. You can persuade with selective truths. You can defer hard conversations that might otherwise slow things down. It works, for a while.

But every lie creates a debt. When people’s expectations don’t match reality, trust erodes. For them, it feels like betrayal. For the liar, it’s exhausting. You have to remember what you said, manage around the contradictions, and live in the gap between what’s real and what you’ve claimed. Over time that gap becomes resentment — toward the people you misled and the situation that forced the dishonesty in the first place.

Honesty isn’t just virtuous; it’s scalable.

A culture that defaults to truth avoids the compounding drag of suspicion, double-checking, and post-hoc justification. It creates alignment not because everyone agrees, but because everyone is operating from the same reality.

That’s what makes honesty a true policy. It isn’t the best because it always feels good in the moment. It’s the best because it performs best over time.

“Every falsehood creates a future obligation to the truth that will eventually be revealed, often with high consequences” - Chernobyl