[parsadotsh]

JSMP: "...this example is a theoretical use case where using IQ testing is not so useful. To look at something a little more realistic, let’s say a company wants to avoid people with a performance more than 2 standard deviations below the mean. (Perhaps such employees have a risk of causing large harm, which could for instance be an issue in the military.) And we again compare admitting people at random vs only taking applicants with above average IQ."

[for reference, if using IQ as an indicator for performance, 2 standard deviations below the mean would be 70 IQ]

Taleb: "The argument by psychologists to make IQ useful is of the sort: who would you like to do brain surgery on you/who would you hire in your company/who would you recommend, someone with a 90 IQ or one with 130 is ...academic. Well, you pick people on task-specific performance, which should include some filtering. In the real world you interview people from their CV (not from some IQ number sent to you as in a thought experiment), and, once you have their CV, the 62 IQ fellow is naturally eliminated. So the only think for which IQ can select, the mentaly disabled, is already weeded out in real life: he/she can’t have a degree in engineering or medicine. Which explains why IQ is unnecessary and using it is risky because you miss out on the Einsteins and Feynmans."

[in3d]

“Which explains why IQ is unnecessary and using it is risky because you miss out on the Einsteins and Feynmans."

Einstein and Feynman would certainly not be rejected if you automatically rejected people with below average IQs, so I don’t really get this part of the argument. We have no number for Einstein and a supposed 125 for Feynman but he did extremely well (the best in the nation) on Putnam, so his nonverbal (or at least “quant”) score was likely very high.

[nullc]

You do risk weeding out those parties: The tests aren't perfect and can give low results from people who are otherwise intelligent. 125 wouldn't pass the above conjectured 130.

A piece of intuition is that if you select a population based on a criteria, the correlation between that criteria and the target in the selected subpopulation is usually greatly diminished, and sometimes reversed.

Imagine you know that being very tall is very good for basketball... so you select very tall people to be players. You will likely find that among the selected players height is not very correlated with performance, or even inversely correlated because your initial selection wiped the correlation out and potentially noise or other effects (like agility) which are correlated with being short begin to dominate the differences within that population.

This also holds for correlated traits-- e.g. if you select very tall players it may wipe out or reverse the correlation shoe-size has with performance.

As a result if you first select people for an intelligence requiring task specific capability and also select for IQ you're likely selecting for noise, and potentially selecting for people who are just really good at taking tests and which weren't actually as good at the task specific capability as a randomly selected person in your first selection would be.

[in3d]

125 on a high school-administered test might have just meant average verbal ability and a perfect score on nonverbal questions. Presumably you would count the nonverbal section much more when picking candidates for a job in physics.

I agree overall that task-specific selection is better, I was just questioning the example given.

[username90]

When Feynman did his IQ test they still used Age Scale, meaning IQ of 125 meant you are mentally 25% older. So if a 8 year old knows as much as a 10 year old you have IQ 125. It is a pretty bad measure which is why we no longer use it.

So when people say "Feynman had iq of 125" it is very different from someone having 125 IQ today.