 
Message-ID: 
Date: Thu, 14 Jul 1994 11:28:31 -0400 (EDT)
From: Faisal Nameer Jawdat 
Content-Type: X-BE2; 12
If-Type-Unsupported: alter
To: corporations-ate-my-cat Mailing List
    
Subject: Re: electric evolution
Cc: 
In-Reply-To: <774122676.6308.0@borman.res.andrew.cmu.edu>

\begindata{text, 1637823366}
\textdsversion{12}
\template{messages}
> so there's all this bizness of doing "genetic" things on computers,

> in software. i wonder if it would be possible to make some model

> of some mechanical/electrical/whateverical system, stuff it into

> a computer, and let it "evolve" into something more efficient?


    We were toying with this in house at my company (QKS).

We make a pmt multi-threaded Smalltalk implementation for

the Mac, and we were wonderring how we could do this.


    What we found was that, although a Smalltalk like model

of data-space and object interaction is idea for modelling

such a system, the way Smalltalk is implemented for application

type design is not ideal for something where you want to create

little objects and have them all operating asynchroniously.


    Basically you need the sort of data and process abstraction

you can get in Smalltalk, but the system has to be designed

with the idea of giving each object its own thread.  In thinking

about this, you could probably do this as a set of programs

running in mach.  It would be more interesting on a massively

parallel box, in my opinion.


    I'd be interested in seeing it.


    On another note, Spafford wrote a paper on viruses as a form

of artificial life, but I can't remember where you can ftp it from.

I'll post if I find it.


    -fiz


\enddata{text, 1637823366}

 prev message 
 next message