The ways of child prodigies

From: Eliezer S. Yudkowsky (
Date: Mon Nov 07 2005 - 10:13:26 MST

If I might offer something of an inside view on all this...

For a start, don't hold it against Song Yoo-geun that he wants to build
flying cars based on string theory. Being a child prodigy in physics is
not like being a college student in physics, even if both know some of
the same information. When I scored 1410 on the SAT at age 11, I was
not anything like a high school graduate who scores 1410 on the SAT. In
some ways I knew more and was smarter, in other ways I knew less and was
dumber. At the age of eleven I was not an ordinary eleven-year-old, but
I was not an ordinary bright high school graduate either.

In particular, you should realize that child prodigies don't have the
same kind of absurdity-detectors. In your dream you may see an elephant
in your bedroom without noticing anything wrong. As a child some parts
of your mind have not awakened. This may be startling when the dreamer
knows quantum physics, for we expect physicists to be awake; yet still
the dreamer dreams. Back in my child-prodigy days I also dreamed of
building flying cars, based on cold fusion (which was just then being
announced). The same level of physics understanding will give rise to
very different conceptions of one's own ability to build flying cars,
depending on whether that understanding exists in a college student or a
child prodigy.

As for whether Yoo-geun will go on to contribute anything to science,
the key concepts here are "diagnosticity" and "reversion to the mean".
For example, let's say that I take a group of student recommendation
letters. First I ask subjects to rate the recommendation laters on a
scale of 1 to 10 for how glowing the recommendation was. Then I ask
subjects to predict the student's performance during the freshman year
of college, using only the recommendation letters as their data. Then I
ask subjects to predict the student's performance during the senior year
of college. The consistent result, which makes no sense statistically,
is that all three judgments are equally extreme. If a letter of
recommendation is not a *perfect* predictor of performance during the
freshman year, then your prediction of freshman performance should be
less extreme than your evaluation of the recommendation letter. If the
correlation is not *perfect*, then your prediction should revert to the
mean. Similarly, since the senior year of college is farther away than
the freshman year, there's more opportunity for other factors to come
into play. Thus the recommendation letter will be less *diagnostic* of
the senior year than the freshman year. Therefore your prediction of
the senior year should be less extreme than your prediction of the
freshman year. A lot can happen in four years!

Ever since I was old enough to know about reverting to the mean, I've
been repeating to myself: "I will not revert to the mean. I will not
revert to the mean." Even before then, I'd heard of Sidis too. I was
determined that I would *not* fade away, I would *not* burn out, I would
*not* become one of those prodigies that excel in childhood and then are
never heard from again. I have always disliked being called a "young
genius", because the amazing thing about a dog that can write is not how
well the dog writes, but that the dog can write at all. I wanted to do
things that would be impressive in their own right, not impressive
because I did them when I was young. Whenever someone complimented me
on being a prodigy, it drove me forward to keep doing more, to keep
improving further, until my deeds stopped being remarkable *for their
youth*. I suspect that whether a child prodigy grows into an adult
genius depends, among other things, on the strength of that prodigy's
determination not to waste his potential.

Another strong factor is probably rationality. I can't Google this
story, but I once read about a child prodigy who was being admitted to
college study of mathematics at age 11, or something like that. And it
was recounted in this magazine article how the prodigy had apparently
found a disproof of some famous mathematical theorem or other, so that
it took the startled teacher a few moments to find the flaw. There was
a quote from the prodigy - I don't remember it exactly - where he
wistfully says, "It would have disproven the whole foundation of the

I read this and smiled. Back in my child-prodigy days, I think at
around the age of 13 or so, I thought I had found a disproof of Cantor's
Diagonal Argument. Ah, the dreams of glory that danced in my head! A
week later I found the blatantly obvious flaw in my disproof. I was
initially a bit disappointed. I thought to myself: "I'll get that
theorem eventually! Someday I'll disprove Cantor's Diagonal Argument,
even though my first try failed!" I resented the theorem for still
being obstinately true, and I started looking for other disproofs. And
then I realized something. I realized that I had made a mistake, and
that, now that I'd spotted my mistake, there was absolutely no reason to
suspect the strength of Cantor's Diagonal Argument any more than other
major theorems of mathematics. I saw then very clearly that I was being
offered the opportunity to become a math crank. I did not wish this to
be my future, so I gave a small laugh and let it go. I waved Cantor's
Diagonal Argument on with all good wishes, and I did not question it
again. Later I would read that magazine article about that other young
prodigy and smile quietly, saying to myself: "He has not quite let go
of his dreams of glory."

If you are smart enough to have bright ideas as a child, you may fall
into a child's trap when your bright idea is disproven. I had to think
all those rational thoughts *at the age of 13*, or fail. Of course
adults fall into this trap too, especially if they don't have many
bright ideas, so that their bright idea is precious to them. I had the
advantage of being confident that I would come up with other bright
ideas if I let go of that one. If Yoo-geun cannot give up his dreams of
antigravity after his absurdity-detectors awaken, he will become a
crackpot, as other child prodigies have done.

There's also the claim - I forget who made it - that the number-one
thing a genius needs is an important problem to solve. I can't say any
deliberate planning went into that one, since I was catapulted into my
present existence when I randomly picked Vernor Vinge's "True Names and
Other Dangers" off a library shelf. Good luck, Yoo-geun! And beware
that you will not get far by saying to yourself: "I am a genius, now I
need an important problem." People who think such thoughts fill in
"important problem" by automatic word-association from magazine
articles, and they go off to study some cliche "important problem" that
ten thousand other scientists are already working on. If you are
horrified at the world you find yourself in, and you wish to remake it
for good, then you may recognize an important problem when you see one.

Another key variable is whether Yoo-geun's accomplishments so far have
been driven by his parents' ambitions. I was exceptionally lucky on
that score. My parents never once pressed me to be any kind of genius.
  Instead they tried to keep it from going to my head, and wisely
admonished me that intelligence wasn't the critical thing in the world -
social skills and life experience were much more important. The effect
of this was to fix firmly in my mind the sovereignty of intelligence.
(My parents were especially strident about the value of life experience
when I tried to point out some of the logical flaws of the Jewish
religion; they said to me, "You'll understand when you're older." It
wasn't until much later that I realized that rationality, not
intelligence, had been the critical variable in that test - that I had
not outthought my parents; rather they had outthought themselves.) I
cannot conceive that if my parents had tried to cram my head full of
impressive knowledge and proudly exhibited me to the media as a genius,
I would have grown into anything like my present form.

Even if Yoo-geun doesn't fade, I wouldn't count on him having ideas
impressive in their own right (rather than being impressive in one so
young) until, oh, at least the age of 16, and more likely his twenties.
  Relative accomplishment is relatively easy. It may mean you have the
*potential* for absolute accomplishment, but absolute accomplishment is
still a lot harder.

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

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