Copy Link
Add to Bookmark
Report
Alife Digest Number 025
ALIFE LIST: Artificial Life Research List Number 25 Wednesday, June 20th 1990
ARTIFICIAL LIFE RESEARCH ELECTRONIC MAILING LIST
Maintained by the Indiana University Artificial Life Research Group
Contents:
thanks!
more vamv
Re: Artificial Life Digest, #24
----------------------------------------------------------------------
From: upheisei!rick@attunix.att.com
Date: 1990 Mai 25 Fri 4.11.99 EMT (90/06/19 Tue 09:53 GMT)
Subject: thanks!
Thanks to David Chess for comments on VAMV. I'm now looking further
down the road...my current time constraints have made it impossible to
get the next projected version very far.
Sexual reproduction is, of course, a necessity, but I'd not really
considered it actively at this point (one step at a time). They do
have some "cross over" in the sense that there's a "join" instruction that
allows them to trade genetic material directly, without reproduction.
> The developing Common Wisdom suggests that asexual reproduction with
> [searches]
> less effectively than sexual reproduction that allows crossing-over
Thank you for pointing this out. It means I should definitely do
something about it.
The tuning space. I had not considered using another GA to search the
tuning space. Thank you for the idea. What I consider "interesting"
populations are ones that use the largest variety of instructions, and
have the largest number of different types. That's about exactly the
opposite of what *they* think is the ideal way to live in the environment.
One other trick that I've thought about to get diversity to develop
is to have different parts of the environment act in different ways.
Since, as you said:
> The things that the critters come to use are going to be
> exactly those things that are useful
it seems that by having an "environment" that's non-uniform, more
global diversity would tend to develop as populations develop to take
advantage of local niches. I have noticed that they tend to produce
waves of "consumers" and "producers" depending on local food
abundances. With abundant food, consumers can reproduce very quickly,
but when food runs out, they're out of luck if they can't also
synthesize new food...
Thanks for the thoughts!
Rick
------------------------------
From: upheisei!rick@attunix.att.com
Date: 1990 Mai 25 Fri 4.51.70 EMT (90/06/19 Tue 10:50 GMT)
Subject: more vamv
The last reply got me thinking... what's the point of the VAMV
exercise?
Eventually, what I want (ho ho ho don't we all) is to have a very
large, unpredictable genetic space with an immense range of possible
behaviours and a rich and varied environment in which amusing critters
could contentedly cavort in complex ways...kinda like Earth, only
smaller. Then, create a few random hopeful monsters and turn them
loose with no guidance to limp along for a while.
It's the "no guidance" part that interests me, to see what develops in
a gigantic space without "telling" the creatures what to do, or how to
do it, or even what I'm looking for (and what I'm scoring them on).
VAMV is a small part of this, developing some "behavioural machinery".
I'd rather not have a limited and pre-specified 'size' of the
machinery, and rather not tell them about such things as sexual
reproduction, but let them figure those things out for themselves. (I
realize this is anthropomorphizing, sorry.) The other side of the VAMV
coin is a system I have (currently overhauled all over the garage
floor, so to speak, and unlikely to come flying together of its own
accord) that doesn't have such constraints. It uses linked lists of
variable length for chromosomes, but the machinery and the behavioral
code isn't really there yet...that's why I went back to the drawing
board and started with something simpler...
The fundamental problem with this sort of thing (little environments
of screen-critters), as I see it, is to devise something that results
in a seemingly endless range of "fascinating" behaviour (both in the
individual and en-masse) and a bizarre range of captivating visual
forms, yet is "fast, compact, simple, easy to program, and easy to
understand" on the lowest levels. DNA, for example, is fascinating
because, in one sense, it seems so stupidly simple (like, any ol' god
could do it over lunch) yet results in what we see around us, in all of
its infinite tear-jerking variety. I haven't yet come up with anything
easy to program, requires "no input from the operator" that results in
anywhere near the long-term formal and behavioural complexity that I
desire. (The "fast" part is tough too: creatures that take up several
thousand bytes of core can't be replicated as fast as creatures that
take up a mere hundred bytes, so things run hideously slowly...)
I guess the problem (except for the speed part) boils down to
devising a simple "code" that can be programmed in six days at most,
that has enough potential to remain endlessly fascinating enough
to keep me amused for a thousand and one nights...
(I trust you'll forgive a non-biologist from being in awe of the
universe here... :-) Simple "micro" codes with near-infinite potential
seem rather rare and frustratingly difficult to devise. More common
(and easy to implement, it seems) are relatively complex "macro" codes
-- like the one used in VAMV -- that have much less potential for
variety. Makes me appreciate DNA all the more...and wonder just how it
is that Life goes from such molecular interactions on the one hand to
things like daffodils and herds of elephants on the other...
Unfortunately, I haven't got billions of leisure years to figure it
out...
Rick
------------------------------
Date: 19 Jun 90 11:36:18 BST (Tue)
From: Harold Thimbleby <hwt@compsci.stirling.ac.uk>
Subject: Re: Artificial Life Digest, #24
Article on ALII:
Harold Thimbleby, Artificial Life, Computer Bulletin, pp22-23, May 1990.
PS The editorial process stripped the references and the introduction!
References should appear in a subsequent issue...
------------------------------
End of ALife Digest
********************************
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=---=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
= Artificial Life Distribution List =
= =
= All submissions for distribution to: alife@iuvax.cs.indiana.edu =
= All list subscriber additions, deletions, or administrative details to: =
= alife-request@iuvax.cs.indiana.edu =
= All software, tech reports to Alife depository through =
= anonymous ftp at iuvax.cs.indiana.edu in ~ftp/pub/alife =
= =
= List maintainers: Elisabeth Freeman, Eric Freeman, Marek Lugowski =
= Artificial Life Research Group, Indiana University =
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=---=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
End of Alife Digest
********************************
