"Expert judgment requires accepting that some uncertainty is irreducible.
Knight separates risk from uncertainty: some things can be priced, some cannot be known yet. In engineering this shows up in roadmaps, estimates, and dependency promises that feel like they need a percentage attached. You can model it, but the model will not settle it.
False precision buys temporary calm at the cost of credibility. When a leader speaks like the uncertainty is gone, the team stops surfacing bad news early because it sounds like incompetence rather than reality.
Make uncertainty explicit without becoming vague. Name what you will not pretend to know, commit to the next action anyway, and choose the signal that will force a revisit. That lets you be decisive without pretending you are certain. Put it in writing so others can hold it.