Re: Looking for Info. on Self-Replication/Microprocessor Design

Hal Finney (
Sun, 11 Oct 1998 09:32:34 -0700

At 09:56 PM 10/10/98 -0400, John Heritage Jr wrote:

>Sorry for the wasteful post here, but I've exhausted
>my methods of search and am still looking for more information
>on what is required to create self-replicating biological and
>mechanical (and combo) devices.. Any URLs, Books, contacts,
>phone #'s, or anything would be appreciated.

The big question with a self-replicating device is, what are the raw materials it is allowed to use?

At one extreme, you could require that it be set down on an asteroid, which may or may not have differentiated ores and such, and using only sunlight for power, set out to identify elements, refine them, and build copies of itself.

At another extreme, you can provide it with raw materials in already processed form: computer chips, motors, structural elements. It needs to build a copy of itself from these pieces. There has been some discussion of trying to build a robot from Lego pieces which could build more copies of itself given piles of sorted Legos (including motor and computer Legos).

>I'm also wondering if anyone has any experience in
>designing full microprocessors (ie: x86 processors, RISC
>chips, or other similar large-scale devices). I'm wondering
>what kind of educational background was required to get to
>that stage, and what I need to do to either find out or go
>down that path.

Trying to create silicon based integrated circuits is going to require a huge infrastructure. You might want to consider other alternatives for the basic technologies that the selfrep system will need, fundamental characteristics like motion and control systems.

Instead of silicon chips, consider older technologies like electronic relays, or even mechanical punched-card or punched-tape systems. It would be great if you could avoid electricity altogether, so that you don't have to create batteries, wires, etc. But you need some kind of power source.

An interesting possibility would be to use fluidics, a mechanical analog of electronics based on fluid flow. When I was in college we did a lab where we built the equivalent of several kinds of electronic ciruits using fluidic equipment. You could make oscillators, flip flops, etc., using nothing more than tubing and some special "fluidic transistors". These are purely mechanical devices which use the Coanda effect to allow a small fluid flow to control a much larger flow.

Do a dejanews search on sci.nanotech for a post called "Partial design for macro-scale machining self-replicator" where there is a proposal for a fluidic, punched-tape based self-replicator built from a fluid which self-cures to plastic under ultraviolet light. There are a lot of problems to be solved but it sounds easier than trying to build a chip fabrication factory.

(BTW, I did a web search on the Coanda effect, and for some reason the whole topic has become associated with pseudoscience, from UFOs to anomolous water to cleansing auras. Apparently Coanda was quite a character. Here's a page on what appears to be a quack approach to life extension through modified water, based supposedly on some claims by Coanda himself. It also mentions pre-extropian G. Harry Stine.