Authority bias

How it works

In 1961, Stanley Milgram asked ordinary people in New Haven to give what they believed were electric shocks to a stranger, on the instructions of a man in a grey lab coat. In the best-known version, 65% went all the way to the highest switch, labelled 450 volts. Nobody was actually shocked, but the participants did not know that. The coat and the calm “the experiment requires that you continue” were enough.

The study has been picked over for sixty years. Archival work found that the experimenter often went off script, that many participants suspected the shocks were fake, and that obedience varied hugely between Milgram’s twenty-odd versions: it collapsed when the orders came by phone, or from an “ordinary man” instead of the scientist. A 2009 partial replication, stopped at 150 volts for ethical reasons, found similar rates. The honest takeaway is not “people are sheep”. It is narrower and more useful: the costume of authority buys a lot of compliance, and the costume is cheap.

Manipulators know the costume is what counts. So you get the “institute” that is one man and a website, the doctor speaking confidently about economics, the “former NATO general” quoted by a friendly outlet, the AI-cloned TV host recommending an investment platform. EUvsDisinfo’s database is full of “experts” whose only qualification is agreeing with the message. Fake experts and deepfakes are the industrial versions of a lab coat.

Example

@geopolitics_today, 7h A retired general with 30 years in NATO says the sanctions have already failed and the West knows it. But the media won’t show you this interview.

@joe_1975, 6h Finally someone who actually knows. Not some journalist.

@footnote_person, 3h Looked him up: retired in 2004, has given the same interview to the same three outlets since 2022, no data in any of them. Could still be right, but “general” is not an argument.

Notice how the second user’s entire reasoning is the rank. Nobody has asked what the general’s evidence is, or whether a career in logistics makes someone an economist.

Why it works on you

• Deference is usually efficient. You cannot verify everything, so trusting pilots, surgeons and electricians is sensible. The shortcut fails when the signal of expertise, the title or the uniform, is faked or borrowed from another field.
• Status triggers obedience before thought. Milgram’s participants argued, sweated and still pressed the switch. The reaction to authority is partly automatic; the reasons come afterwards. The same thing happens when a confident face in a suit appears in your feed.
• It lets you stop thinking. “An expert said so” closes the question, which is a relief. Lobby-funded studies and the tobacco industry’s old playbook of hired scientists were built on exactly that relief.

What to do

When a title is doing the work: ask two questions. Is this person in-field? A virologist on viruses is an authority; a virologist on monetary policy is a guy. What is the evidence? A real expert can point to it; a costume can only point to itself. Then search the name. Lateral reading takes a minute and is what fact-checkers do first.

When you catch yourself deferring: notice the moment you stopped asking. If you would have demanded a source from an anonymous account, demand it from the professor too. Expertise is a reason to listen carefully, not a reason to stop listening.

In a comment section: do not sneer at credentials in general. That is the other failure, where everyone’s opinion counts the same and nothing can be known. Ask narrow questions instead: what is their field, which study, what data. Narrow questions expose a costume fast.

Don’t confuse it with

• Trusting real expertise. Relying on in-field experts with evidence behind them is not a bias; it is literacy. The bias is when the title replaces the evidence.
• Checking a source’s credibility. Asking “who is this person and what do they know” is exactly right. Authority bias is skipping the question because the answer sounded impressive.