I'm as surprised as you and I'm glad someone took the time to write that up (I was about to when I saw your comment). I don't understand why someone would call IQ "unscientific" without addressing research showing it to be a good predictor of future performance. Note that this still doesn't invalidate the point that people can be smart in many different ways (yet have the same IQ), which seems like a reasonable position.
Just because something is a good predictor doesn't make it scientific. There could be a third variable like wealth which needs to be controlled for, but is rarely done in those studies with few exceptions.
In those exceptions, it was generally found that wealth of parents played a much bigger role in future success than IQ.
> Just because something is a good predictor doesn't make it scientific
Perhaps, but Taleb’s charge of it being nonscientific is based on it not being a good predictor of success except at the low end, from which he concludes it is not a valid measure of intelligence. This is wrong both on the facts (it is a good predictor across the board), and the underlying logic (intelligence and propensity to succeed are two different things, while it turns out that IQ is a good predictor of the latter, that actually is not logically necessary for it to be a good measure of the former.)
> There could be a third variable like wealth which needs to be controlled for, but is rarely done in those studies with few exceptions.
I've read lots of studies of the relationship of intelligence to various success measures, and controlling for other variables (especially various socioeconomic ones) is the norm, not an exception.
> In those exceptions, it was generally found that wealth of parents played a much bigger role in future success than IQ.
While that's true for some success measures (for a number, parental educational attainment is actually a better predcitor than parental wealth, which mostly seems to be a predictor as an imperfect proxy for parental educational attainment), that doesn't invalidate IQ as a scientific measure of intelligence or intelligence as a factor in success, since the intelligence as measured by IQ predicts variation not explained by parental wealth. Your argument is like trying to invalidate a measure of speed because it is a worse predictor of winning a race than which starting line a contestant is allowed to use.
To be scientific would require proving a definite causal relationship between IQ and intelligence, while also convincingly proving that intelligence can be measured with a number
Why does it have to be causal? A variable can be a real predictor of another without any of them causing the other one (for instance, they may have a common cause). That doesn't mean the predictor is of any less value.
I would say that results on certain cognitive ability tests (i.e. "IQ tests") being at least a moderately good predictor for future performance in certain human activities is a rather well proven fact. For instance, the correlation between an IQ score and future job performance on a job with complex requirements is about 0.5.
Of course, that does not mean the relationship is 1-to-1 and that it is a perfect predictor (which it obviously isn't).
> To be scientific would require proving a definite causal relationship between IQ and intelligence
That's not how science works. A scientific theory includes a hypothesis with a causal explanation, but science doesn't deal in “definite causal relationships”, but in empirically useful models which only get viewed as approximating definite causal relations with age and lack of a better (simpler and/or with broader explanatory power) model being discovered.
