Re: the ways of child prodigies

From: Michael Vassar (
Date: Tue Nov 08 2005 - 19:39:45 MST

>For the rest of this post, I'm going to assume that child prodigies are
>talented kids with quantitively more talent, not something qualitatively
>different. I don't really know, I've never met (to my knowledge a prodigy
>{not counting Eliezer, whom I've never met in the flesh}, but I went
>through an accelarated learning programme at high school, so I know a bit
>about how talented kids develop and end up

Ollie, an "accelerated learning program" tells you nothing at all about how
relevantly talented kids develop. Among other things, accelerated relative
to IQ 100 still probably means within the range where everything correlates
pretty well with 'g', while beyond about 2 SD most correlations fall
dramatically. Second, prodigies are not just talented kids only more so, or
at least, they aren't generated by the same normal distribution that
generates talented kids. The article I cited indicates a normal
distributino SAT score of 520 SD 140 for 12 year olds, a value strongly
consistant with they hypothesis that SAT performance is an approximately
linear function of age (consistant with the mental over physical age
definition of IQ) Based on that distribution, Eliezer's score would be 6.93
SD from the mean. Scores like that don't occur in normal distributions.
Such a score might occur in a log normal distribution, which is actually a
better fit to IQ data than a normal distribution, one time a few million,
but Eliezer is not the most extreme case of high performance. Sho Yano,
currently 13 or 14 and a MD/PhD at the University of Chicago, scored 1500
(post 1996, similar to Eliezer's 1410) at age 8. Even a log-normal
distribution predicts no such data points, even after regressing to the
mean. Empirically, mice genetically engineered to express more of the
enzyme CREB learn about ten times faster than ordinary mice but their
learning levels off at a level achievable by normal mice. Prodigies might
have some comparable neurochemical characteristics, combined with parents
who enable them to use their abilities to learn material appropriate to
their abilities rather than to their age (far too many parents and most
schools would not enable this). The above model suggests little benefit in
seeking out prodigies, but lacks sufficient evidence to be held as anything
but a hypothesis. The truth is, we don't know that much about prodigies,
probably not enough to know if they could be useful. If we had unlimited
resources, finding out would be worth while. With current resources it
still might be, but probably not.

>SIAI needs persons with expertise in computers and cognitive studies
>(roughly). Persons with expertise in pure maths or physics could be
>useful, but not particularly.

SIAI needs geniuses in pure logic and applied rationality. The computer and
cog-sci stuff are only needed at the ordinary "expert" level, though more is

>However, because my brain is unwilling to remember any word that's not
>English, I can't learn complete languages to save myself.

Funny. I learn new vocab easily, but can't keep languages seperate in my
mind. Talk to me in Spanish and I'll probably answer in Russian. Also, I
can't learn accents *at all*.

It doesn't really matter what things ordinary adults might incorrectly
expect someone to excel in. All that matters is "can someone develop FAI
theory". So far, we know of one person who can. He's sort of an ex-prodigy
(depending on the definition) and his childhood ability set was definitely
not the result of a normal distribution. Not a lot of data to go on, but at
least suggestive enough that given time following that line of thought might
be wise.

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:53 MDT