From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Sun Sep 08 2002 - 07:43:03 MDT
This is in part a response to Eliezer's comments as well as Anders:
On Sat, 7 Sep 2002, Eliezer S. Yudkowsky wrote:
> An engineered galaxy, if posthumans bother engineering galaxies,
I'm not sure how strong your "if" is here. MBrains with solar
power sources radiating at the CMB temperature are several light
years in size (but need multi-solar systems worth of material).
Frank has also speculated on the sizes of reversible computers
that would trump all non-reversible computers and they are larger
than solar system sized.
> is one that (a) has been transformed into a very dense clump of
> computronium
Yes, we know from Anders and Lloyd's work that neutronium brains
or black-holeium brains would be the densest computronium --
but that doesn't indicate in any way shape or form that you
can get the information *out* of that computronium to do
something useful in the rest of the real universe.
This raises an interesting question about extropic vectors.
Is it extropic if you morph yourself into a computronic state
that cannot *ever* communicate your discoveries to other
computronic states. Raises some very sticky issues since
a post-singularity AI may never be able to communicate its
rationale to a pre-singularity "human" and thus would seem
to be a pretty exclusive vector (perhaps in violation of some
of the proposed statements the ExI board have been exchanging
offlist recently regarding racism issues).
> and (b) has sent out further probes as soon as the first replicators
> arrived, which (friendly or unfriendly) would reach our planet before
> the light from the final fully transformed galaxy arrived.
Hoag's Object is a different galaxy. I don't know how far it is from
our galaxay but even if they have sent out replicators they may well
not have arrived here yet. If replicators travel at 0.1c and light
travels at c there is ample opportunity for one to view a transformed
galaxy before the replicators arrive.
Then we get into the debates about *why* should one waste valuable
material on "seeding" distant galaxies that can never provide a
useful ROI to oneself and are likely to be competitors in the
long run. There is also the fact that unless you can launch
an overwhelming amount of mass and energy at a specific target,
e.g. Earth, the replicators may not win the day (see the Freitas
Ecophagy paper) and may instead produce one really annoyed species
with a very long memory (e.g. humanity). So replicating across
intergalactic distances is a *very* questionable strategy IMO.
The Hoag Object has several interesting features. Most of the
stars are concentrated in a central area. This is good for
the minimization of communications delays. Since there still
appear to be stars this is a JBrain architecture rather than
an MBrain architecture (unless the MBrains are in the region
between the central core and the outer rim of massive stars).
Massive stars are dangerous. They produce lots of radiation
which is bad for nanocomputronium. So generally one wants to
stay away from such. But massive stars have the nice property
of using gravity to breed large amounts of heavier elements
which may be essential components of "designer" computronium
(the unique magnetic properties of gandolinium come to mind).
So one may want to breed large amounts of heavier elements and
massive stars may be the best way to do that. Presumably one
wants to do it on the outskirts of ones neighborhood however.
I think it is a mistake to assume that post singularity
humans/AIs can discard the laws of physics as we now
understand them. If one has that perspective, then placing
bets on the outcome seems to be an exercise of ones fantasies.
Robert
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:16:50 MST