From: Eugen Leitl (eugen@leitl.org)
Date: Mon May 13 2002 - 09:17:13 MDT
On Sun, 12 May 2002, Robert J. Bradbury wrote:
> > There's enough fuel on the surface and the few m of humus layer to
> > transform Earth's surface (strictly) radically in a course of a few days.
>
> If unopposed. You are also assuming an AI bright enough to be able to
We're talking about technologies developed in a critical Singularity hot
spot. The technology delta between the hot spot, and the periphery is so
large that the hot spot completely rules the place.
> develop the means of turning raw materials into computronium optimized
> for the AI -- that is hardly something that occurs in a few days.
Computronium is not really that difficult to do, even in absence of
machine-phase nanorobotics. It's just hierarchical complementary
purification and assembly. We've come a long way towards the goal of
engineering macromolecular self assembly, including crystal growth.
> There are 10E148 patterns of atoms that can occupy a cubic nanometer
> [Drexler, '92]. Even the most advanced AI can't explore a fraction of
> that in a few days.
It doesn't have to be optimal to be very useful. Plus, the amount of
constraints present almost completely spell out the solution for you. We
can do the thing with current technology, after ~decade of R&D.
> > I'm not sure how soon you could leap into space, but it would be probably
> > also on hour range.
>
> Since we can leap to space the ability to do so would presumably
> a very early capability. Thus it makes much more sense to leap
Assuming, you hijack some launching facilities. It might be easier to
assemble stuff from scratch, though.
> to space (an environment humans can't easily adapt to) and optimize
> yourself from there.
Humans do not register in an ecology build on postbiological principles
(unless protected, most people will die in the course of hours/days).
Going into space is absolutely required in your quest for lebensraum,
since you're bottlenecked by your rate of growth on a planetary surface.
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:02 MST