An expert asked to advise on policy is being asked two questions and usually answers them as one.
The first is what the evidence shows, and the second is what should therefore be done.
Only the first lies inside the , and the second requires a judgement about values that a does not confer.
Blurring the two is the most common failure and the most damaging to public trust.
A scientist who says a measure is necessary has moved from description to without marking the step.
When the public disagrees with the value judgement, it rejects the description along with it.
That is a rational response to an argument presented as a package.
is the second difficulty and is handled badly in both directions.
Communicating a range invites the reply that the expert does not know, and communicating a invites a charge of dishonesty when it turns out wrong.
Audiences handle ranges better than the advisers expect, provided the range is given first and not as a later .
A revised upward is heard as new information; a quietly widened is heard as an excuse.
statements carry a cost that is rarely weighed.
They compress genuine disagreement into a single line and make the field look more certain than it is.
A reader who later finds the disagreement concludes that it was , which is worse than having seen it.
Fields that published their internal disputes have suffered less from this than fields that presented a front.
The advisory role works best when the expert states the range, states the assumptions and hands the decision back.
Politicians dislike that arrangement because it returns the responsibility to them.
Experts dislike it because it reduces their influence, and they are usually the ones who describe it as unrealistic.