I'm curious about your claim that GI is hard to define (true) and measure (not true?).

GI is whatever the standard iq test measures. Higher numbers seem to correlate with better outcomes in certain settings. Right? What's controversial about that?

IQ is far from comprehensive, the aspects of the mind it measures are discrete and relatively low level, this is why two people with the same IQ can have different intelligence as perceived by others. As you traverse the higher level of abstractions it becomes impossible to compare, some just don't exist in other peoples minds, they are not a fundamental part of the biological structure (subjectivity starts to define the quanta, not biology, so how do you compare something that does not exist in all individuals, not to mention the number of possible abstractions multiply as the level of abstraction increases).

As I said before IQ is an indicator of potential at best (the potential for the abstractions built on top of it), even of the low level abilities it does not include all of them, we cannot externally measure all of them.

Here's another way of looking at it: if IQ _was_ comprehensive, the number of individual questions would be astronomical, it would be impossible to construct or take such a test because it tries to enumerate and quantify all of the different learned abstraction above the inherently low level ones across all of humanity.

Taking that fictitious test a bit further, note that if we imagine all humans actually taking it, their score will be extremely low, because it's impossible to have 100% (you would need a massive brain that could simultaneously learn all of the _different_ ways of thinking).

> GI is hard to define (true) and measure (not true?)

Doesn't this seem like a contradiction to you? how do you empirically measure something that you cannot define?

Certainly one number is vaguely ridiculous (yet effective for the original purpose), but you are conflating learned models (which are content), as opposed to capacity (such as raw short term memory).

> you are conflating learned models (which are content), as opposed to capacity (such as raw short term memory).

There is no correct definition, however most people agree that intelligence is more than innate ability. If you are focusing on the later as a definition then you can indeed measure it, but consider that there are many people with an IQ as high as Einstien's, but no such deep understanding in quantum mechanics or even any subject.

I would argue that casting out "learned" as something superficial is incorrect, learning is not simply acquisition of knowledge (as education systems have very slowly begun to understand since the mid 20th century)

Hmmm. Yes to your P1 and maybe to your P2.