IQ metrics are continually damned. There is still no real consensus that they are useful at all. But the social sciences is filled with a lot of soft metrics like IQ because studying complex human behavior with hard science is very hard, and probably impossible.
Good, calibrated IQ tests measure something, and that something has a positive correlation with work efficiency in at least certain types of work.
That said, there are many types of IQ tests and some of them are in my opinion rather weird, for example anything where your score depends on your knowledge of English language or English history as I hear has traditionally been the case in US.
Around here it seems they try to make the tests as little dependent on language, history and even math as possible and the ones I have taken has largely been about:
- pattern matching (of these 6 squares with various patterns, which one doesn't fit in?)
- sequences (based on these 3 squares with patterns, which is the next one?)
- and that kind of stuff
These skills are things that are very real, very measurable and often linked to work throughput. Yes, I rarely visualize boxes in my head and rotate them and I rarely sift through through black and white squares with alternating patterns at work, but it seems the ability to do so is often linked to the ability to mentally single step code or spot patterns in bug reports.
It doesn't mean that you are good person if your score is high (or low) and it doesn't mean that you will be a good employee and not waste your time on HN.
The tests that work in measuring human behavior are fairly simply and very repeatable. Think Fitts's or Hick's Law, which involve hit testing and searching through an unsorted list. I can imagine pattern matching being fairly repeatable as well, but I'm not sure how it really relates to real problem solving. Then again, medical doctors are basically pattern matchers (and why they are fairly excited about AI assistance in...matching patterns).
