[extropy-chat] Self replicating computer programs ?

[Brett Paatsch]

Eugen Lietl wrote:

> >Brett Paatsch wrote:
> > Hmm. Maybe I am overlooking the bleeding obvious
> > but I can't think of a single fully self-replicating 
> > computer program - one that does not require
> > impute from outside itself to kick off the duplication. 
>
> I do not understand what you're getting at. 

Sorry I wasn't clear.. I elaborated a bit more in my reply
to Emlyn. 

I was doing two things and so did neither well. 

First I was not doing heavy thinking at all just sort of 
musing, so my chances of error are higher.

Second, I am wary of the term self-replicating. I don't think
its wrong necessarily but I don't think its as intuitively obvious
as most folks seem to think it is. I've got a bit of a word 
fixation at present that seems to have come to me as a result
of watching a lot of political discourse. I think that some large
nuggets of truth may be staying hidden or have been passed over
by the general tendency of people to speak and thus think 
loosely. 

When I see words like self in self-replicating I know that there
are substantive discussions taking place as to what "self" maps
to cognitively even in people. The pattern-identity question is
not satisfactorily resolved yet for me. I don't accept that I can
be reduced down to what others see as my pattern-identity. I
can for them of course. But they are not mapping my "self:" 
they are, rather, mapping one of their "others". Just because
others are happy with the accuracy and veracity of their model
of them and of me doesn't mean I am. Or that the model is
identical rather than merely equivalent to the real thing. 

If self isn't a simple or clear concept philosophically then
- "self"-replication may be a messy thing for engineers to 
be trying to produce in their designs. 

> <http://necsi.org/postdocs/sayama/sdsr/java/> (will need
> Java to run),
> being an near-optimal supportive context. Corewars is only
> marginally less supportive, and many real systems can be hit 
> by pathogens fitting inside a single packet 
> (404 Bytes UDP packet):
> http://www.cs.berkeley.edu/~nweaver/sapphire/

Thanks for the links

> Self-replication in physical reality is harder (viroids and
> viruses use other, more complex self-replicators as substrates),
> but  some extremophiles have very small genomes and 
> overall complexity.
>
> Similiarly, artificial self-replication is far easier in artificial,
> very rich environments offering optimal support. A successful
> grey goo weapon is considerably harder to design.

I found I started to design one (high level only of course) in my reply
to Emlyn and stopped typing it as it seemed imprudent to just barff
than stuff out and post it to a public list. 

My interest in the self-replication concept with respect to nanotech
was/is a political interest. I thought that if it could be shown  that
there was no class of "self-replicating" device that could be made
that would be both interesting to-a-designer (even say a terrorist
designer) and really autonomous (not needing further instructions
from outside) then I might be able to do an end-run around the
whole grey-goo political concern by showing it to be not just hard
but logically impossible. 

I don't think I can do that end-run today. I doubt I will be able
to do it tomorrow either :-) Hence my reluctance to wax
philosophical with an engineering bent into how to design nano-
machines for war.

It seems imprudent to propagate how-to instructions for the
making of weapons or poisons until one has a *very* good 
antidote. Maybe.

I think I'd rather test myself against someone else's "aggressive"
weapon design then put my mind to the engineering problem of 
making and discussing a  better weapon myself. - Until I can find
a way of doing a logical end-run around the whole grey-goo 
scenario. 

Sorry if that's clear as mud.

Regards,
Brett