[extropy-chat] Singularity heat waste

[Robert Bradbury]

George,

The simple answer revolves around what kind of an answer you want out of the
singularity?

If the answer is "42" (assuming Adams is correct) and you compute it
entirely using revsersible computing and you are storing the result into
memory which is entirely suboptimal (i.e. !42) you would have to flip a
minimum of 6 bits.  So the cost could be very low (in terms of waste heat
produced).  If you can reduce the answer to 0 or 1 then you only need to
flip a single bit and the cost is at the limit of what is possible.  Now of
course somebody somewhere has to know *what* that 0 or 1 (or 42) actually
means and it is difficult to say how much the knowing of that might cost.

At the other end of the spectrum you get into the points Keith was making
about being limited by the surface area that you are using to radiate the
wasted bits.  The larger one is the more bits you can throw away per unit
time.  So singularity development is fundamentally limited in part by the
volume in which it is taking place.

In practice there are probably several plateaus (Kurzweil S-curves) in the
singularity.

1) Limit of easily available energy (sub KT-I civilization).
2) Limit of heat capacity of the Earth (~KT-I civilization).
    This is the Frietas "hypsithermal limit" which equates to about 10kg of
operating nanotech per person for a population of ~10 billion.  This limit
can be transcended but it requires significant planetary reengineering.
3) The level of a relatively optimized inner solar system (KT-II
civilization).
4) The level of an entirely optimized solar system (KT-II+ civilization).
5) The heat capacity of an expanded highly evolved Matrioshka Brain (several
light years in size).

At each of these transitions there tend to be fairly significant shifts in
the architecture of the computronium.  How fast those shifts occur tend in
large part to involve questions regarding how much energy and/or matter one
is willing to "throw away".  These are economic questions related to how
much computational and/or memory capacity one wants available at time T,
T+1, T+2, etc.   Since we don't know *what* MBrains would want to think
about it is difficult to imagine what the cost/benefit tradeoffs might be.

Regarding the host bio-sphere question, one possible solution to minimize
effects on the current highly sub-optimal biosphere is to develop the
singularity from the asteroid belt outward.  This would have minimal impact
on Earth and provide most of the capacity that even a highly evolved MBrain
would have.  It would however develop more slowly than a "burn the bridges
behind us" strategy of full development of the inner solar system first.
But since "slowly" in singularity time would be measured in spans of decades
to centuries it may be entirely acceptable if the bio-compuntronium is not
overthrown by AI nano-computronium based entities.

In answer to Chris's question, yes the problem has been thought about (it is
a crucial aspect of MBrain evolution) and the singularity doesn't have to
generate waste heat that would be considered by some (many?) to be
detrimental unless it is allowed to evolve along that path.

Robert
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20060714/5d81e671/attachment.html>