Re: Roaming Assembler Size/Replication Rate?

From: hal@finney.org
Date: Mon Jan 15 2001 - 19:20:38 MST


John Marlow writes:
> Maximum practical number of "arms" on a free-roaming assembler...

Maybe you could clarify what you mean by free-roaming. Roaming in what?
An assembler that roams in water might be different than one which roams
in air, and one that roams in vacuum would be more different yet.

As far as the number of arms, I suppose you could have larger or smaller
assemblers with different number of arms. At a minimum you'd probably
need two arms. But of course you could have a much larger assembler
with more arms. I don't see why there would be any particular limit to
the number of arms you could have on a large assembler.

> Approximate diameter (or largest cross-sectional measurement) of such an
> assembler, including "arms..."

I can't find my copy of Nanosystems, but Nanomedicine describes a minimal
two-armed assembler with an external power supply and control unit that
would be about 200 nm long by about 50 nm wide.

> Approximate maximum replication rate for each individual assembler of the
> type specified above, i.e.--how many copies of itself it can construct per
> second, assuming an environment rich in raw materials...

Freitas estimates that the assembler can work at the rate of 1 million
atoms per second (not to imply it necessarily places one atom at a time).
The simple assembler above has about 16 million atoms so would take 16
seconds to replicate. A more complete assembler with nanocomputer and
instructional tape would take more like 100 seconds.

> References to earlier speculations by others on these matters are also
> welcome. I'm aware of some, but surely not all. I would guess the most
> practical shape for a free-roaming assembler would be spheroid.

Again, it depends a great deal on what it is going to do. If it is
supposed to engage in locomotion then we need to have room for propulsion
units, etc. Similarly, what are the raw materials for the replication?
If it is given a "molecular soup" of tagged reactants designed to be used
by the replicator then its job is relatively easy. If it is sitting in
the ocean then things are harder. If it is supposed to build copies
of itself out of bare rock then it probably needs a lot of components
dedicated to dissolving the rock and making it useful.

It may also be that the arms are not pointing outward, but rather inward
so that the work area is clean (i.e. hard vacuum). In that case some
other shape might be best.

As another data point you might look at Freitas' respirocyte design,
http://www.foresight.org/Nanomedicine/Respirocytes.html. This does
not replicate, it just stores and releases oxygen and CO2, but it is
a one micron, spherical machine built from about 27 billion atoms.
If you had an assembler+extras of similar size with only the two arms
inside it could self-replicate in 27000 seconds or 8 hours. I suspect
you could put many more arms in something of this size though and get
replication time down to several minutes.

Hal



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:04:55 MST