>>13117151The main issues with psychometrics is that IQ has no proven construct validity to speak of and most of its "predictive power" is just correlational relationships to other highly dynamic social variables, making it highly circular.
Physics can make predictions for the past, present, and future behavior of physical objects in a space, provided you know all the variables. If I drop a bowling ball from a certain height and I know its weight, all other variables controlled, I can predict its velocity down to a precise amount. This is the predictive power that psychometrics like to believe IQ has but in general a level that social sciences is not really close to approaching.
Another problem with IQ tests is that the tests themselves literally have to be literally updated every year to keep the average mean score at 100, making the tests essentially a non-constant variable determined and constructed by human test makers to a large degree, thus where the Flynn effect comes from. An average teenager in the U.S. from 1910 would likely score on an IQ test today in a range that would classify them as mentally retarded. There may be something there being measured but that's a lot of noise that has to be accounted for.
I am not as bombastic as Taleb but I agree with a lot of his more pointed criticisms that about its statistical validity. I would still like to see it used a measure of development though.
