Re: Wil McCarthy on Singularity

From: Dani Eder (
Date: Tue Apr 05 2005 - 13:43:23 MDT

> >
> Or not. Depends on what he means.

He defined the 'inflection point' as the midpoint
of an 'S' shaped growth curve. At that point growth
is linear. Prior to that point it was accelerating,
and after it is decelerating to an asymptote.

Human population from about 1600 to 1960 is best
described as a hyperbolic. If you graph 1/population
you get a nearly straight line. The growth rate
peaked in the 1960's at just over 2% per year. The
absolute change in population per year is what peaked
in the late 1980's at about 87 million per year.
At that point population growth was linear. Today
it is about 74 million per year, so we are now
sub-linear, i.e. slowing down. The census bureau
projects the world will reach replacement fertility
in 2050. That means just enough babies (2.1 per
woman) to replace the previous generation. Zero
population growth would occur about 2090.

I expect the census bureau's projections will turn
out wrong for any of a number of reasons:

(1) A singularity of the type we mean on this list
could occur before 2090. In that case any number
between zero and 10^50 could be right.
(2) A major war or pandemic could occur in this time
period. That includes AIDS remaining out of control.
(3) The undeveloped world will catch up economically
with the developed world in the next 50 years, which
would likely have a downward effect on population.
(4) Some ecological or resource limits (for example
today's high oil price affecting fertilizer and
tractor operating costs) causing problems. These
problems will eventually be overcome, given human
inventiveness and economic forces, but if it takes
decades for things to adapt, there could be large
impacts during the transition.
(5) Sufficiently cheap space transport bleeds some
population off planet.
(6) Perfect robotic or virtual reality playmates
replacing real sex.

One of the definitions of the Singularity is that
time in the future beyond which you can't make
useful predictions because too much has changed.
In population projections, the uncertainties reach
a factor of 2 range from high to low in about 50
years. That's projections by sober demographers
at the census bureau and the UN, not including
items like the ones I listed above. Allowing for
some measure of the unexpected, I would put the
useful prediction limit in this field at 50 years.

In semiconductors, the useful prediction limit
may be as short as 10 years - we don't know enough
to say if we will hit a limit on silicon based
semiconductors in that time, or whether any
alternatives will work out.


Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT