The skill that makes all the others work. Everything in the catalogue at the bottom of this page had a moment where somebody could have updated and did not, and the reasons people fail to update are well studied and depressingly consistent.

You are not evaluating the evidence. You are defending the conclusion.

Motivated reasoning is the finding that people apply far more scrutiny to evidence they dislike than to evidence they welcome, without any awareness of doing it. The tell is asymmetric effort: noticing that you went looking for the flaw in one study and not the other. Intelligence does not protect against this and may make it worse, because a better arguer builds better defences for whatever they already believe. motivated reasoning

State the other position better than its holder can.

Steelmanning is not politeness, it is a diagnostic. If you cannot construct the strongest version of the argument you are rejecting, you do not yet understand what you are rejecting, and your objection is probably aimed at a version nobody holds. The uncomfortable part is how often the exercise changes your own position partway through. steelmanning

Put a number on it, then check the number.

Saying you are fairly confident commits you to nothing and can never be wrong. Saying seventy per cent can be scored. Tetlock's work on forecasting found that the people who improved were the ones who made explicit numeric predictions, wrote down their reasoning, and went back to look — and that the biggest single predictor of accuracy was willingness to update in small increments rather than defend or capitulate. calibration and forecasting

The less you know about a field, the more confident you feel in it.

The Dunning-Kruger effect is widely misquoted as stupid people being confident. What it actually describes is that the skills needed to do something well are the same skills needed to recognise doing it badly, so incompetence is partly the inability to perceive itself. It applies to everyone, in every domain where they are a beginner, including you and including this page. Dunning-Kruger

Semmelweis was right, and it destroyed him.

In 1847 he showed that doctors washing their hands in chlorinated lime cut maternal deaths on his ward from around 18 per cent to about 2. The profession rejected it, partly because it implied doctors had been killing their patients. He was dismissed, deteriorated, and died in an asylum at 47, of an infection. Germ theory arrived a couple of decades later and proved him right. The reflex named after him is the reflex to reject a finding because of what accepting it would mean about you. the Semmelweis reflex

Sometimes the establishment is wrong and the fix is a decade of patient evidence.

Barry Marshall argued that ulcers were caused by bacteria rather than stress, against a settled consensus and a large antacid market. Unable to get an animal model, he drank a culture of Helicobacter pylori himself, gave himself gastritis, and documented it. He won a Nobel Prize in 2005. The lesson people take is be bold; the better lesson is that he won by producing evidence that could not be argued with, not by being certain. Barry Marshall

Say the words out loud: I was wrong about that.

Not I misspoke, not to be fair at the time, not the situation has evolved. Organisations where senior people visibly do this are organisations where a junior engineer will tell you the deploy looks wrong at 4pm on a Friday. Every case in the record below happened somewhere that had, one way or another, made that sentence expensive to say. intellectual humility