On Expertise: Who Should We Believe?

We live inside a paradox. Expert knowledge has never been more available — the consensus of entire fields sits in every pocket — nor, by every measure of trust, more widely doubted. The epidemiologist and the entertainer arrive in the same feed, in the same typeface, asking for the same faith. It is not surprising that many citizens, asked to choose between them a hundred times a day, have begun to decline the choice altogether.

Begin by being precise about what expertise is. It is not, in the first place, a credential. A diploma is a claim; expertise is what a domain builds underneath the claim: a machinery of correction. Peer review, replication, professional ruin for the fraudulent and the sloppy — these unglamorous systems are the actual product. A field deserves trust roughly in proportion to the strength of its machinery for proving itself wrong. This is why the consensus of climate science and the consensus of a marketing department are not the same species of thing, however similar their press releases look.

The decline of trust has real causes, and honesty requires listing them. Experts have been spectacularly wrong in public — about weapons, about diets, about the stability of financial systems their own models blessed. Institutions, caught erring, have too often protected themselves before correcting the record, teaching the public that the machinery of correction stops at the institution's own door. Distrust, in other words, was not manufactured from nothing. Some of it was earned, and any defense of expertise that skips this paragraph is propaganda.

But the fashionable opposite — to dismiss all expertise as a costume and treat every loud voice as an equal authority — commits the mirror error. It flattens the only distinction that matters, between claims that have survived a domain's machinery and claims that have merely survived its marketing. The citizen's task is not obedience and not refusal, but audit: to ask of any confident claim, expert or insurgent, what would change its author's mind, and whether anything ever has. Humility is demanded of the credentialed. It is demanded, equally, of the rest of us, who must know the borders of our own competence as carefully as we police the experts'.

Trust, then, should be provisional, specific, and revisable — granted to a field's consensus because of its machinery, withdrawn when the machinery is shown to be captured, and never owed to a title alone. Believe, but keep the receipts. The alternative to audited trust is not freedom of thought; it is the rule of whoever performs certainty best.