[mod50ack]

As an aside — intelligence is not the same thing as accomplishment. A lot of people do well in school (and do well on IQ tests) and do not achieve anything of note. The people we remember are those who achieve great things, not those who do well on puzzle-solving tests. Those aren't the same thing, and this is probably why Stephen Hawking said that caring about IQ is for losers.

Hawking was undoubtedly a smart guy, but that fact alone did not make his career. He did a lot of hard work in a field he was passionate about. Why would you tell Stephen Hawking how good you are at solving puzzles — why would he care about that? I'm sure he would have found news of some finding relevant to his research interests much more compelling.

Now, are IQ-type tests useful? Yes. They are quite good to administer in school to gauge people's reasoning abilities — to a certain point. The point of the tests was never to rank the smartest people, and to think about these edge cases — the ones tests are worst at measuring — is pointless. There are better things to concern yourself with; life is not an IQ leaderboard.

[lifthrasiir]

The first ever intelligence test (Binet-Simon) was designed partly in order to identify children who have intellectual disability and give a better treatment for them. The test itself was crude, but I believe that should be the only meaningful use of IQ tests because overly low or high IQ does predict certain kinds of complications. Otherwise they are easy to game, they only measure a particular slice of human intelligence, and their error bar is large enough that even a unit difference in SD can be not meaningful [1]. IQ is just a meaningless number that whoever have a higher one tends to have unjustified superiority over others.

[1] Depends on the particular test of course, but I can safely guess that +2 SD and +2.5 SD are not statistically distinguished by most tests.

[mod50ack]

Caring too much about IQ is not exempt from Goodhart's law.