The mayor of a small Idaho city sits in a meeting about a wastewater treatment upgrade. The plant supervisor and a consulting engineer are discussing membrane bioreactors against a conventional activated sludge expansion. Roughly eight million dollars separates the options, the decision will bind the city for thirty years, and by the fourth minute the mayor has lost the thread entirely.
She will have to decide. Neither man in the room can do it for her, and both of them understand the technical question far better than she ever will.
The two ways this goes wrong
The first is performed competence. The leader learns eleven pieces of vocabulary, deploys them at intervals, and begins making calls on the technical merits. This is visible to the specialists within about a meeting and a half. What follows is worse than embarrassment: they stop giving her the real picture, because explaining anything now requires first navigating around a confidently held misunderstanding. She has made herself expensive to inform.
The second is abdication. The leader concludes that since she cannot evaluate the engineering, she cannot evaluate anything, and defers wholesale. The technical recommendation becomes the decision. She retains the signature and has surrendered the office, and if the choice turns out badly there is nobody who was actually accountable for it, which is a thing the city will discover in year seven.
Both failures rest on the same assumption: that the only way to assess a recommendation is to understand the domain it came from.
Assess the reasoning, not the answer
You cannot judge whether the membrane sizing is correct. You can judge, with no engineering background whatsoever, a set of things that turn out to predict a great deal.
Whether alternatives were genuinely considered, or whether one option was assumed at the start and the rest assembled afterward to justify it. Whether the confidence expressed matches the evidence described, or exceeds it by an obvious margin. Whether the person can explain the tradeoff in language you understand, which is a real test rather than a courtesy, because someone who cannot describe why the other option is attractive has usually not examined it. Whether they have been right before, in ways you were able to verify. Whether they update when a fact changes, or defend the original position with new arguments.
None of this requires domain knowledge. All of it is assessable by an attentive generalist, and together it produces a reliable read on whether a recommendation should be trusted.
Four questions do most of this work. What would have to be true for the other option to be right. What is the strongest case against your recommendation. How confident are you, in a number. Who in your field would disagree with this, and what would they say.
The last one is the most revealing and the least asked. A specialist who cannot name a credible dissenter either does not know the state of their own field or is managing you.
The part that remains yours
Here is the thing the deference framing gets wrong, and it is a category error rather than a matter of degree.
The engineer’s question is what is true within engineering. The mayor’s question is what the city should do given engineering, and finance, and the rate impact on households in a town where median income is what it is, and the twenty-year growth assumption that may not hold, and the fact that the public works director retires in three years and the membrane system requires an operator the city cannot currently hire.
Nobody in that room is positioned to weigh those against each other. The specialist optimises within a boundary; the leader owns the boundary itself and everything outside it. This is not a lesser task performed by someone less qualified. It is a different task, it is genuinely difficult, and there is exactly one person whose job it is.
Which is why the honest formulation is not that the leader defers to the expert. It is that the leader takes the expert’s answer as one input, correctly weighted, and then does something the expert was never asked to do.
The expert who has learned what works
One hazard deserves naming, because it is common and it is invisible from inside.
Specialists who have spent years presenting to non-specialists learn, without necessarily intending to, that certainty is persuasive and hedging is not. The ones who express doubt get cross-examined. The ones who say a thing flatly get approved. Over enough cycles this shapes how they present regardless of what they actually believe, and the result is a room in which the most confident recommendation is not the best-supported one.
The correction is to make hedging safe and to say so explicitly. Ask for the confidence figure and then do not punish a low one. Ask what would change their mind and treat a good answer as a mark in their favour. A leader who rewards calibration gets calibration, and a leader who rewards certainty gets certainty whether or not it is warranted.
What the specialists are actually watching
People who know more than you about their work do not expect you to know it too. They have generally made peace with that and often prefer it.
What they are watching for is whether you will pretend, and whether you will decide. Pretending costs you their information. Refusing to decide costs you the job, because a leader who defers on every technical matter has quietly reassigned the accountability without moving the authority, and the specialists know they now carry a weight that was never theirs. Saying plainly that you do not understand the chemistry and asking to be walked through the consequences instead loses you nothing at all. It is the pretence that costs, not the ignorance.
Edited by Patrick J. Wolf, PhD