From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Tue Dec 10 2002 - 16:50:30 MST
Bill Hibbard wrote:
>
> The fact that IQs have a normal distribution doesn't imply
> anything about whether they are a linear, logarithmic or other
> scale of intelligence. IQ is defined as a linear measure
> of mental development rate (as ratio of mental to calendar
> age), but of course that doesn't make it a linear measure
> either. And the fact that a person with an IQ of 200 can solve
> problems that 1000 people with IQs of 100 cannot is due to the
> difficulty of collaboration on certain problems, rather than
> any logarithmic scaling. This is nicely discussed in Fred
> Brook's The Mythical Man-Month.
IQ is no longer defined as a ratio of mental to calendar age, and hasn't
been for quite some time. Today IQ is actually *defined as* a normal
distribution, with, if I recall correctly, 16 IQ points being equal to one
standard deviation. That is, if you have an IQ of 132, it *means that*
your test scores are in a percentile two standard deviations from the mean.
A person with an IQ of 200 can solve problems that 1000 people with IQs of
1000 cannot due to "difficulty of collaboration", but it should be noted
that this is the kind of "difficulty of collaboration" that would require,
at least, computer-mediated telepathy to solve. I suspect a single
computer-mediated dual human could outthink Einstein - he was smart, but
he didn't have twice as much brain mass as one human - but I could prove
to be wrong about that. On the other hand, six hundred chimps
collectively have a hundred times as much prefrontal computing power as
one human, but even a computer-mediated 600-chimp Beowulf cluster might
prove unable to outthink a human, the complex adaptations necessary being
absent. But the "g-factor" in variation in intelligence between
individual humans should not be confused with "general intelligence",
i.e., the factor that separates humans from chimps, as opposed to Einstein
from Joe.
-- Eliezer S. Yudkowsky http://intelligence.org/ Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:41 MDT