What Does It Take to Be an Expert?

It seems to me that a useful definition of an expert is somebody who, within their domain of expertise, makes predictions which are consistently more accurate than chance. Experts can be quantified by the z score of their predictions. A 2 sigma expert is 97.7% accurate. When hiring a structural engineer to calculate whether your building will collapse, you want a 5 or 6 sigma expert. Many domains do not permit accurate computation of correctness, or the event rate is so low and time lapse so long a statistically significant value cannot be computed. In these fields, there are no experts, just people with opinions and various kinds of credentials. In other fields, nobody significantly betters chance expectation. There are, thus, no expert roulette players, stock pickers, or market timers.

One of the biggest rackets in academia and policy making is attributing “expertise” to a person or institution based upon knowledge of, experience with, or investment in a topic rather than quantified performance in predicting events related to it.


I did not watch the video – the cover picture failed to pass my common sense screen. 50% of experts are white women of a certain age? None of those competent & intelligent Chinese or Japanese ladies need apply?

That is a key problem with trying to quantify expert performance. And then we have to remember one of the precepts of Decision Analysis – Good decisions can have bad outcomes, and bad decisions can have good outcomes. Statistics rescues the expert from her (usually her, per the picture) incompetence.


With all due respect, Gavin, that’s an extremely petty dismissal. The video was made by an American for an American audience. 30% of Americans are white women. A random sample of 8 Americans will have 4 or more white women 19.4% of the time. No one is under obligation to you to choose a globally racially representative sample of humanoid figures for a YouTube thumbnail.


It is not petty to detest Political Correctness in all its many guises – it is common sense, as you know full well.

There is a big statistical issue with your critique – your assumption that the distribution of experts in a population is identical to its distribution in the general population. If we look at expert basketball players, for example, the evidence is rather clear that reality does not match that assumption.


You will notice that complaining about the over-representation of white women is politically correct, and to a greater degree, than complaining about the under-representation of Chinese women is politically incorrect. You’ve chosen a mixed position on the PC-PIC axis, but overall I rate your position as PC.

If you go to Somalia and Ethiopia you will probably find the actual composition of experts, doctors, international businessmen, and so on, are highly skewed toward Chinese and Lebanese nationals. But if you go there and complain about the lack of Chinese representation in generic media, you will probably look like a bit of a nut. The same applies here.

1 Like

Huh? In the normal sense of the term, Political Correctness demands over-representation of certain favored identity groups.