Not only is IQ a poor measure that's perhaps best at capturing near average performance, "intelligence" is an attribute that's not readily ranked in a linear manner as height can be.
IQ isn't a linear scale [1][2]. (It is ordinal.)
You said "'intelligence' is an attribute that's not readily ranked in a linear manner" while complaining about IQ. IQ doesn't rank intelligence linearly.
What is your issue here?
IQ isn't linear because it is not superposable. A 100 IQ person isn't twice as clever, within the definition of cleverness IQ measures, as someone with a 50 IQ. (An 8' man is twice as tall as a 4' child.) If you need to force it into a geometric analogy, the distribution IQ rests on cannot be described by a line.
You appear to be thinking of a uniform distribution as compared to a normal distribution.
Nominally both height and IQ approximate a normal distribution (sans the infinite tails, leaving aside discrete ordinal bucketing).
"Linear" (see the O.E.D. for example) is a word with many meanings, mostly pertaining to lines and things resembling a line.
You can line up people by height or by IQ score, you cannot line up people by "intelligence", it's a partial order at best.
> the distribution IQ rests on cannot be described by a line.
You're the only one of us that connected the scalar IQ distribution to a line .. I've been aware of the central limit theorem for 45+ years now.
The geometric analogy is that mapping people by intelligence to a line is a poor fit, to a plane or higher space would make more sense.
As evidenced by this exchange.
Ranking people by "intelligence" is compounded by it being multifactored, language skills and numerical skills can differ independently, spatial skills are a third vector unaligned with either.
people who perform well on intelligence tests may not present as generally intelligent