[FredPret]

A 200 IQ is not possible at current population levels.

It’s a statistical measure comparing the test taker against the average, much like percentiles.

At a certain IQ score, somewhere in the 170’s I think, the expected number of individuals with that IQ is about 1.

If we had absolute measures of intelligence (that would be a breakthrough for the ages), then we could say “A is twice as smart as B” and award A twice the points of B. In such a system, the sky is the limit for the number of points.

EDIT: If/when we build a human-level AI, perhaps we could use the number of transistors / artificial neurons involved as a proxy for an absolute measure of how difficult it is to answer some problems. This would be imperfect but better than nothing.

[naveen99]

It’s close though. Iq sd is 15. 200 is about 6.7 sd above mean. Odds of being 6.5 standard deviations away is 1 in 12 billion.

https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rul...

[vitus]

Don't forget you also have to divide by two (so, 1 in 24 billion) if you only want to consider the upper end of the range, since that number is essentially "fraction of the population that doesn't fall within mean +/- 6.5 stdev".

[FredPret]

Sure, but we can’t say someone has a 1-in-12B intelligence when we only have 8B (or whatever) people. We can only go as high as 1-in-<current_pop>

[n4r9]

That sounds wrong. Perhaps you're confusing frequency with likelihood?

IQ is standardised so that population scores on a standard test have a mean of 100 and standard deviation of 15. It's possible to obtain this with as fewer than 100 people, distributed as:

* One "genius" who scores 200

* One "dumbass" who scores 0

* 87 "everyman" who score 100.

The mean here is clearly 100, and the variance is sqrt(20000/89) = 14.99.

Of course this is very contrived and doesn't look much like a Bell curve in the first place. But with say a million people it wouldn't take much to come up with a more realistic looking example.

[dleeftink]

Matter of perspective; one in the 100~150 billion sapiens (or more) that have gone before and the rates go up. It's also possible rates are underestimated, as we have only tested a relatively small sample compared to N=all

[FredPret]

If we test every living human, we can figure out who is the very smartest one. That person has a 1-in-<current_pop> intelligence, from which we can calculate their IQ. That IQ is nowhere near 200.

It would be handy to have standardized test answers from every human ever, but sadly most are dead as you point out.

[willvarfar]

we can compare their intelligence to those who have previously lived?

A quick googling gives estimates of ~117B humans have ever been born.

So if you were the cleverest person on the planet, ever, you'd have 1-in-117B intelligence?

[jll29]

I was thinking along similar lines, bit it becomes a theoretical question when you cannot test/evaluate/interact with subjects anymore.

Also, I'm wondering whether a difference of 1 IQ point is even noticeable, and if not, what's the smallest noticeable increment or faction of a unit.

[FredPret]

A difference in 1 IQ point in the 100-101 range might be a difference in absolute problem-solving ability of x units, while than the difference between 170-171 is y units.

[entropicdrifter]

>EDIT: If/when we build a human-level AI, perhaps we could use the number of transistors / artificial neurons involved as a proxy for an absolute measure of how difficult it is to answer some problems. This would be imperfect but better than nothing.

Wouldn't that be roughly equivalent to equating brain size/volume to intelligence? I know there's a decent correlation between intelligence and head-size, but it's not that consistent. Some brains just work better for their size.

[FredPret]

That’s why it’ll be a proxy that will get better over time as AI architectures converge towards optimality.