How do you develop architectural judgement in an engineer, as opposed to reviewing their designs?

What is being tested

Whether you understand that transferring judgement scales and correcting output does not.

Why it matters more than reviewing

An architect can be involved in a limited number of decisions. An architect who has developed judgement in ten engineers influences every decision those ten make, indefinitely, including the ones made when they are unavailable.

The arithmetic is not close, and reviewing is the default because it is immediate and feels productive.

What to teach: the questions, not the answers

• "What happens when this dependency is slow?"
• "What is the cost of being wrong here?"
• "Is this reversible?"
• "What would make this the wrong choice?"
• "Who else is affected by this, and have you asked them?"

Someone who has internalised those will handle problems you never anticipated. Someone given the correct answer to this design has learned one answer.

How to do it

Ask before telling. Let them reach the conclusion; correct the reasoning rather than the output. A conclusion someone reaches themselves is owned; one supplied is followed.

Explain your own reasoning aloud, including the uncertainty. "I think X, mostly because of Y, but I am not confident about Z." This teaches how to hold a position provisionally, which is most of what judgement is.

Let them be wrong where the cost is bounded. Recoverable mistakes are the most efficient teaching mechanism available, and preventing all of them prevents learning. The skill is knowing which mistakes are recoverable.

Give them the decision, with support. Ownership accelerates development more than advice does.

Work through real decisions, not hypotheticals. The context is where the learning is.

What makes it fail

• Reviewing rather than developing — correcting output without transferring the reasoning.
• Answering too quickly, which is efficient once and unhelpful repeatedly.
• Only mentoring people who already agree with you, which narrows the organisation's thinking.
• No time allocated, so it is displaced by delivery every week.

The uncomfortable part

Accepting a decision that is 80% as good and owned by the engineer over one that is 100% and imposed. That trade is the core of the role, and it is genuinely difficult for people who became architects by being right.