Not to mention, the assumption is that if IQ is not useful for predicting success regardless of circumstance (i.e. whether somebody is employed as an astrophysi

Not to mention, the assumption is that if IQ is not useful for predicting success regardless of circumstance (i.e. whether somebody is employed as an astrophysicist or as a sweatshop seamstress), it must be not useful in all cases. The fact that IQ is a useful metric at the low end of the scale is exactly the hint as to why it's not useful when misused at the high end. That is, the difference between being able to return 100% of balls in tennis at, say, one return every 800ms, is not any greater help when your opponent can only return every 1400ms anyhow. In order for a difference of general cognitive ability to be useful, a cognitive task of sufficient complexity over a sufficiently short period of time must be presented.