Self-replicating factories [was Re: Our rocky solar system may be rare]

Robert J. Bradbury (
Sun, 12 Sep 1999 11:58:48 -0700 (PDT)

On Sat, 11 Sep 1999 wrote:

> I had commented in my comments on Robin's Great Filter paper:
> > Even without nanotechnology, Von Neumann type factory operations
> > would allow asteroid and planetary engineering. If current growth
> > rates in silicon wafer production continue, we could construct a
> > Dyson Shell with microprocessors within 300 years.
> My response:
> If we extrapolated the use of horses in the 19th century we could build a
> Dyson Shell out of horseshit in a few hundred years.
I hope there is a smiley after this Hal. Because in fact you can't build strong structural stuff out of horseshit (unless you have a trick I don't know about). You can however build computers *and* relatively strong structures out of Silicon. We use aluminium instead of silicon in outer space because it is lighter. Since Si is quite a bit more abundant than Al in the sun, planetary crusts and solar system overall, we *will* be using Si as a a building material once we get it out of the gravity wells. Of course in the solar system overall, Carbon is more abundant than either, so in the long run much of the construction will be done out of diamond. The problem is we have to dismantle all of the outer planets and comets (where most of the CO2/CH4 is). That is going to take 12+ years (Uranus, Neptune & Pluto) if we are willing to use all of the available power to do it. Since we probably won't be willing to do that (unless we can tap the H2 in Jupiter & Saturn using thermonuclear reactors) we will probably take much longer to shift from the Fe/Si/Al Age to the Diamond Age.

The main point I was trying to make was to counter Robin's statement that we don't know if we can do hard nanotechnology. Robin is correct in that there is no "evidence proof" as there is with wet(bio)-nanotechnology. However, molecular assembly per se *is* not nessessary for a Dyson Shell full of computronium in a short time. What *is* necessary is self-replicating factories to build the parts that (a) dismantle planets; and (b) build the Dyson Shell satellite subunits. The NASA 1980 Study that Robert Freitas was the co-author of showed that self-replicating factories could be done with macroscale components, the problem is that the doubling time is long. I'm moderately confident that using MEMS technology and/or nano-imprint lithography you could get the building blocks small enough (but still not atomic sized) that you could construct a building-block factories factory. I think with small universal building blocks, you significantly decrease the doubling times to something closer to nanotechnology based doubling times.

It would be interesting to determine what the functional requirements would be and how small you could make nano-legos that had (a) 4-bit CPUs, snap-together velcro hooks, internal plumbing to pass fluids (with nanoparticles) back and forth between the functional blocks and some means of self propulsion (fluid/air drive?). If you get something like that in the micro-to-milli-meter scale, then you have what you need to build the evergrowing factories. It doesn't require nanotechnology, it simply requires clever design.

The other main point to the statement was that even if current growth rates slow, we have one good thing - a lot of computer power and one bad thing - we run out of silicon.

If your point was simply to point out the problems with extrapolations, I'm aware of them. However, I feel there is a difference between how much "transportation" an individual can consume and how much computer power we can use. We seem to have more computational capacity than we know what to do with now (individually) but at the same time people seem to be getting more creative about interesting things to do with it.