Experts frequently reach correct conclusions without being able to say how, and this is neither mysterious nor uniformly reliable.
is pattern recognition acquired by repeated exposure to a .
Whether it works depends on two properties of the rather than on the person.
The first is : the situation must contain patterns that are genuinely there to be learned.
The second is : the expert must find out, reasonably soon, whether the judgement was right.
Chess supplies both, and chess intuition is accordingly excellent.
Long-term political forecasting supplies neither, and forecasters perform at about the level of chance.
Their confidence, however, is from that of the chess player.
That is the finding with practical consequences, because confidence is what organisations act on.
A can also supply , which is worse than none.
A recruiter who never learns how rejected candidates performed elsewhere receives only on her successes.
She will become steadily more confident and no more accurate, and the two are commonly confused.
The remedy is not to distrust experts but to ask about the before trusting them.
Two questions do most of the work: are there stable patterns here, and does anybody find out the answer?
Where both answers are yes, to the expert and do not require an explanation.
Where either is no, require the reasoning and be prepared to find it thin.
Experts in low-validity s are not frauds and have simply learned patterns that were not there.
The learning process is identical, and only the determines whether it produces skill or .