From: Charles Hixsn (charleshixsn@earthlink.net)
Date: Wed Aug 06 2003 - 09:16:26 MDT
Anders Sandberg wrote:
>On Tue, Aug 05, 2003 at 02:18:34PM -0700, Robert J. Bradbury wrote:
>
>
>>>Wouldn't it be possible to create a quantum computer based on
>>>spin-coupling as a communication "backplane"?
>>>
>>>
>>Yep, you could propose such a computer. But Robert will jump
>>...
>>
>...and some error correction). I have the feeling (like Robert) that the
>original proposal suggested some entanglement communication, and that is
>magical physics. If we had such FTL communication, then M-brains could
>be designed much more efficiently by keeping remote parts linked, and I
>think the analysis of size and movement would become very changed. ...
>
>
If we had entanglement linkages built into M-brains, then there would be
no practical limit (foreseeable from here) to the degree of
intelligence. Actually, a lot of things would depend on the exact
nature of the link. If it had a long start-up delay, or was expensive
to use (say, emitted a lot of heat, or had a slow bit rate), then there
could well still be speed limits in how fast the final
M-brain-with-quantum-link could think, but I can't see any limit on the
size of problem that it could solve. Merely on the speed at which it
could be solved. I suppose that one could say that the same is, in
principle, true of even a simple Turing machine, but this isn't the same
kind of limit. The separable parts could be handled separately...
If the link were fast and cheap enough it could replace everything else
for linking components. This is quite unlikely, but who can say. It
would be equivalent to seeing the universe as it exists as a computer
that one could program.
If the link were somewhat expensive, but had a fast bit-rate, then
M-Brain modules would get, say, as large as a desk, or perhaps a shoebox
and these modules would be linked using some development from, perhaps,
Beowulf technology (or Globus?). Thus each M-Brain module would have
perhaps 1000 times the computing capacity of a human brain (or more, I'm
being a bit conservative here) and these modules would be linked
"instantaneously" to a net which could be spatially distributed
throughout the universe. One has to hope that there are a large number
of "frequencies" available. I find myself believing that were this to
be possible, the first civilization to discover how to do it would
saturate the available frequencies, and latecomers would only find a
channel too noisy to use. (The earlier developers would be frequencies
so high and spread across the available spectrum that those coming later
wouldn't find an available channel, or a readable signal.)
If the link were fast and cheap, but short range (say an exponential
decay), then everything depends on the available distance. If it's
across the solar system, then M-Brains could be a lot more efficient
than we have believed. If it's across the hall, then it wouldn't have
much effect at all, but it might up the rate of bit transmission a bit.
Simply having the possibility of a quantuum linkage doesn't tell one
enough about the engineering trade-offs. And since we don't yet know
how to do it, we don't know what the constraints and limitations would
be. But unless we're the first it seems to range from "interesting but
not really worth the effort" up to "quite handy, and well worth the
cost". Only towards the high end of the scale do we start noticing
significant performance changes in the M-Brains. At the nearly top end
(practical signalling extends out to around 8 light years with a high
bit rate) we get a dramatic change where the universe becomes accessible
for total conversions into a single intelligence. (This says nothing
about whether such a choice is wise or ethical.)
This archive was generated by hypermail 2.1.5 : Wed Aug 06 2003 - 09:26:24 MDT