From: Damien Broderick (damien@ariel.ucs.unimelb.edu.au)
Date: Mon Oct 06 1997 - 06:09:06 MDT
At 03:20 PM 10/5/97 -0700, Hal wrote:
>Actually I find that the Singularity stays about 20-30 years in the
>future. [...]
>Personally, I think we are going to hit a big wall in computer technology
>in the next ten years. [...]
>It's our old friend, the S curve.
Here's a relevant bit from THE SPIKE (pp. 37-8):
============================================
Today's cutting-edge components are not just small, they are very, very
small. Sooner or later, one is bound to run out of profitably accessible
space, down there on the etched circuits. Even if there's ample unused
room at the nano level, will we be able to push the atoms around quickly
enough, and cheaply enough, to continue our dizzying plunge downward into
micro- and nano-chip utopia, and upward toward the Spike?
Certainly, Hans Moravec assured me. Relax. `The engineers directly
involved in making ICs [integrated circuits] tend to be pessimistic about
future progress, because they can see all the problems with the current
approaches, but are too busy to pay much attention to the far-out
alternatives in the research labs. As long as the conventional approaches
continue to be improved, the radical alternatives don't have a competitive
chance. But, as soon as progress in conventional techniques falters,
radical alternatives jump ahead, and start a new cycle of refinement. When
short wavelength ultraviolet is no longer good enough, there are X-ray
synchrotrons, electron-beams and even scanning tunnelling microscopes.
Existing ICs are still so far from the 3D, quantum electronic possibilities
already demonstrated in the lab, and the economic incentives to keep the
race going are so huge, that I see no reason to expect things to start to
slow down for several decades - and then we have a whole new regime, that
will make our current time look like the Stone Age. (And if chip plants
cost ten billion dollars, well, that's partly because they have to be so
big to make the quantities that the market demands.)'
The latest news from Moravec is even more astonishing. By 1997, according
to his analysis of computational bang for your buck, the Moore's Law curve
returned to its original doubling-every-year, and has now swept onward into
even swifter acceleration. He comments: `By 1993 personal computers
provided 10 MIPS (million instructions per second), by 1995 it was 30 MIPS,
and in 1997 it is over 100 MIPS. Suddenly machines are reading text,
recognizing speech, and robots are driving themselves cross-country.'
================================================
Every curve saturates. Will new curves emerge (or be pushed) to take up
the slack? Seems plausible to me.
Damien Broderick
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:45:00 MST