From: Eugene Leitl (Eugene.Leitl@lrz.uni-muenchen.de)
Date: Mon Jan 06 1997 - 12:44:27 MST
On Mon, 6 Jan 1997, John K Clark wrote:
[ 50 bit/synapse ]
> Too high! .001 bit/synapse would be closer to the mark, LOTS of redundancy.
> See Science January 28 1994 page 466.
John, don't you think, that once in a long while a honest researcher should
use realistic, not ever worst-case (ask perfidous Penrose for these)
numbers? We have seen a decade-long gallery of countless earlier
estimates, by the most distinguished researchers, proven wrong, haven't
we? What is the purpose of a discrete, modular, pretty precise structure
if all it is good for is storing just .001 bits. Evolution doesn't
produce complex circuitry for no good reason, if a simpler one would
have been sufficient, no way.
Ok, it has been in Science. I am not convinced, as a number of wet
neuroscientists as well. While my estimate may be too high yours is
certainly drastically too low. Future will tell.
> >0.5 Mbit/neuron, let's say 1 MBit/neuron (10^ 6 bit/neuron)
>
> I would say 5 bits/neuron or less, that's bits not megabits.
I wish you were right. Unfortunately, I think this is orders of magnitude
too low. Reminiscient of earlier estimates "1 transistor=1 neuron". Ask
Joe Strout, he burns any number of MFlops for hours to simulate just _one_
biologically realistic neuron, and far from running in realtime.
> >10^17 bits in toto, human brain representation storage
> >equivalent
>
> 10^11 or 10^12 bits.
No comment.
> >nanotech (here comes Dr. D. to save the daaay) is uncertain
> >at best, as no physical feasibility proof exists.
>
> Life is a feasibility proof, but as long as something does not violate the
Wrong. Life is a feasibility proof of a wet, folded linear biopolymer
complementary autoassembly/autoreplication nanotech, _not_ of Drexlerian
dry mechanosynthesis-based nanotech, a wholly different kettle of fish.
We don't know whether mechanosynthesis-based autoreplicators are feasible.
> laws of Physics I don't see why we need one, it's your responsibility to
We don't know whether it violates the laws of physics. It may, it may not.
We can't apply the laws of physics we know to say whether a certain
complex structure is consistent with its advertised novel modus operandi. QM
numerical runs of mechanosynthetic reactions are hazy at best. Whoever
says the simulations have been validated is not telling the truth. They are
not. Movies do not lie, lyers render movies.
> prove it's impossible not mine to prove it's possible. Nobody has made an
Wrong. Science works differently. _You_ have to prove it, not vice versa.
It this attitude which allows science mainstream to go on ignoring
Drexler's ideas, even if they do not deserve it. The Moon is made from
green cheese. Prove it. A case of speculative engineering needs extremely
convincing evidence for the other guys to listen at all. (And without the
other guys nanotech will be impossible to do).
> airplane that weighs 1000 tons, the biggest is about 300. Would I have to
> physically present you with the finished airplane to convince you it was
> possible?
No, but man-made things already fly. No man-made mechanosynthesis works. A
STM demonstration of basic set of mechanosynthesis reactions, validating
computer runs is sufficient for _me_. We don't have such evidence, that's
why I rage against radiating smug certainty. We have just hints, no
certainty. Don't give wrong signals.
> >Anybody just said "cryonics"?
>
> Absolutely, right now it's the only game in town, but if I didn't think
> Nanotechnology would work I wouldn't waste my time with it.
I don't know whether Drexlerian nanotech will work. Fortunately, it's not
necessary. I know that a certain brand of nanotech can work (because it
is based on virtually unmodified live-derived machinery), which is not
sufficiently strong to repair damage yet is sufficiently strong to grow
molecular circuitry fit for running uploads. While destructive scan of
vitrified cerebral tissue_is_ simpler with strong nanotech, it is not
absolutely necessary. It won't be quick, it won't be pretty, but it will
probably work.
ciao,
'gene
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:43:58 MST