[ExI] Ethics and Emotions are not axioms
eugen at leitl.org
Wed Jun 6 07:33:15 UTC 2007
On Tue, Jun 05, 2007 at 01:47:46PM -0700, Lee Corbin wrote:
> I based the 10^33 uploaded humans eventually running on/in the Earth
Since you need 10^17 bits to just represent the brain state (and 10^23 ops
to run it), that is a cm^3 just for storage, using Drexler rod logic
memory. No computing yet. Just about three orders of magnitude away from
the real wet thing.
And it is really not prudent to argue about how much bits a
human equivalent needs. Because we just do not know yet, apart
from a (rather impressive) upper bound.
If you want to run a more meaningful benchmark, let's assume
#1 of Top 500 (a 64 kNode Blue Gene/L) is a realtime mouse,
and just scale up the mouse brain volume to 1.4 l.
> (just for the sake wanting to know a good upper limit) on Drexler's
> conservative rod-logic. An account can be found on pages 134-135
> of Kurzweil's "The Singularity is Near".
I'd rather not repeat my opinion about Kurzweil here.
> "Neuroscientist Anders Sandbert estimates the potential storage capacity
> of a hydrogen atom at about four million bits (!). These densities have
> not yet been demonstrated, so we'll use a more conservative estimate..."
In practice, you need about 10^3 atoms to store a random-access bit in 3D, give
or take some order of magnitude. (No, atoms in cubic carbon lattice do not
really qualify as random-access).
> and then later on p. 135
> "An [even] more conservative but compelling design for a massively
> parallel, *reversible* computer is Eric Drexler's patented nano-
With prior art going back to Leibniz, or so.
> computer design, which is entirely mechanical. Computations are
> performed by manipulating nanoscale rods, which are effectively
> spring-loaded.... The device has a trillion (10^12) processors
> and provides an overall rate of 10^21 cps, enough to simulate
> one hundred thousand human brains in a cubic centimeter."
I don't think so. Ops and bits are apples and oranges, and you
still need 10^23 apples, according to my estimate.
> So then I took the volume of the Earth (6.33x10^6 meters) ^ 3
> times 4pi/3 = 10^21 cu. meters x 10^9 cubic millimeters/
> meter^3 x 100 (human brains) = 10^33 humans.
> (Since this was the second time I did the math, it's probably right.)
Your math might be right, but it doesn't have a lot of meaning.
Even assuming 1 m^3/person (because you need power and navigation),
not all these atoms in there are equally useful.
> > But I don't see that as a limit. Since a nearly arbitrarily small computer
> > could run a human process (assuming we knew how to do it, until which even
> > Jeff Davis and Ray Charles would agree it is hard) then we could run a human
> > process (not in real time of course) with much less than six micrograms of
> > stuff.
> Yes, the rod-logic is very conservative, to begin with.
Rod logic is certainly quite conservative, but every other assumption
you rely on is not.
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE
More information about the extropy-chat