Copy Link
Add to Bookmark
Report
Alife Digest Number 035
ALIFE LIST: Artificial Life Research List Number 35 Wednesday, August 29th 1990
ARTIFICIAL LIFE RESEARCH ELECTRONIC MAILING LIST
Maintained by the Indiana University Artificial Life Research Group
Contents:
polyandry in Nepal, modeling conflict in belief systems using alife
----------------------------------------------------------------------
Date: Tue, 28 Aug 90 05:57:16 -0500
From: Marek W. Lugowski <marek@iuvax.cs.indiana.edu>
Subject: polyandry in Nepal, modeling conflict in belief systems using alife
Hello, soc.culture.nepal! I am very happy to see this group take form;
we just received it here today.
I am copying this to alife@iuvax.cs.indiana.edu for reasons that will
become apparent shortly.
As a child (4-6) I was taken all the way to the border on Nepal, on
the Indian side, but I don't know offhand if we ever crossed over.
I'll have to ask my parents. Oh well, there is always future tense...
I remember the very very very steep streets and worrying about falling
over into the valley if I did not watch my step.
This note is inspired in part by an anthropology text I have here at
home, a fascinating study of the Nyinba, an ethnically Tibetan group
living in Nepal. The book is by Nancy E. Levine, a UCLA anthropology
associate professor:
_The Dynamics of Polyandry: Kinship, Domesticity, and Population
on the Tibetan Border_, ISBN 0-226-47569-7, University of Chicago
Press, 1988 (5th printing). It's also available in hardback,
ISBN-0-226-47568-9.
I'm copying this to iu.general because we happen to have here at Indiana
a significant presence of scholarship on both Tibetan people and languages,
as well as gender studies and on anthropology.
My first question is to ask for any information that any of you may
have on polyandry in Nepal. I might continue with some specifics from
Nancy Levine's work at a later date.
I am after a machine-based ethical calculus such as may be do well at
representing a detailed clash of belief systems, not so much
representing the content of the beliefs themselves. I am also interested
in modeling metaphor as a structuring property of perception.
I am also interested in the economics of polyandry, albeit Nancy
Levine's central thesis is that polyandry amongst the Nyinba cannot be
explained by economic arguments or changes in demographics alone, with
the evolved kinship structures as the sustaining reason. I am
interested in possibly modeling such emergences using a particular
type of computer, described in detail below.
------
For the curious, the rest of this message describes my computer
modeling approach, which itself is one of the strands of the new
methodologies of the new field of artificial life. For a large
(688pp.) introduction, reference and annotated bibliography (and
index) to the field of artificial life, see Christopher Langton's
_Artificial Life_, available from Addison-wesley's toll-free 800
number, from the continental USA. ISBN-0-201-09356-1 for paperback,
0-201-09346-4, hardback.
I am a computer science graduate student working in the field of
artificial life. I wish to use a type of a data-parallel encoding in
conjunction with non-zero-sum game theory as modeled on my simulated
tile automaton, Computational Metabolism (ComMet). I have already
axiomatized in ComMet the Prisoner Dilemma (PD) strategy TIT-FOR-TAT
(TfT) with strictly local rules on tiles that displace each other and
change their algebraic tile color. Now I am looking forward to
extending the dynamic to the iterated TfT and other iterated
Tft-derived variants.
In 1983 Douglas Hofstadter and I staged here at IU one of those
University of Michigan's Robert Axelrod's (_The Emergence of
Cooperation_, I don't have a copy here to give more info) computer
tournaments for iterated PD strategies. Although we never actually
bothered to published the results, an interesting outcome did occur:
The pure TfTs strategies, and there were 4 copies of them in a field
of 63, all came out together at 19th ot 22th place. They were
displaced by other more suited to that particular environment
strategies, including non-TfT variants. My own a priori entry,
MAssively_REtaliatory_striKe ("cooperate until the opponent defects,
henceforth defect") finished 6th.
I am now in a position to model such curious dynamic effects systematically:
The nice thing about my machine is that it can be characterized as
Lindenmayer-system grammars and growths. This makes it very easy to
custom-code PD strategies and other intended "genetic" unfoldings.
Indeed, ComMet has been likened by a developmental biologist as an
incubator for studying the mathematics of development. We shall see if
this optimistic characterization pans out in practice. At least we
are in a position to state what our experiments are to be and what they
are meant to achieve. Much of my recent quick progress has been due to
having had to prepare for qualifying examinations that I just took, as
well as by all the added insight the tools I was learning about gave me,
enabling me to make links (or finally understand the links others have
made well before I noticed!) with other artificial life results:
See Richard Dawkins' biomorphs in C. Langton and the late (last
October) Aristid Lindenmayer's and Przemyslaw Prusinkiewicz's
L-systems chapter there for these very ideas in other settings.
In my case, the same formal grammars, Lindenmayer-systems (L-Systems),
are simply rewrites of tile neighborhoods into tile neighborhoods
(a contest sensitive grammar, by the argument from the length of the
Right Hand Side (RHS), a tile's neighborhood before the fact -- and
the Left Hand Side (LHS), a tile's neighborhood after the fact:
The LHSs must be at least as short as the RHSs; in fact, so far this
year our LHSs are exactly of length 1, which means that our grammars
to date are context free, as well. The rule looks awfully like a
restricted Post Production Systems production.
As far as putting this computer science stuff to use, I am hoping to
combine my insights with that of a mathematically inclined economist:
Barbara Hardy, graduate student in economics at IU, and I would like
to explore the space of evolving non-zero-sum games, but we only want
to look where interesting things happen, of course! The space is
complex, in the chaotic sense, and enumerating complex spaces makes
pretty pictures but does not cover usually cover a lot of the space.
So instead of enumerating the space, we would like to follow some
"dreaming tracks" in it, and these dreaming tracks would be encoded by
grammars that we write down.
We wish to look in the space of iterated tournaments as well as the
space of "strategies": behaviors not frozen (like TfT is: "Cooperate
first, then always reciprocate") but which exist as flux: think of
them as conformal homomorphisms in the tiling itself, where tile
position relative to those of other tiles in the set defines the
semantics of the entire set at that moment.
These conformal mappings are chaotic behavior but are on the level of
super-set membership ....constant! So, my merely shifting point of
reference, one is dealing with chaos, or one is dealing with order:
In fact, chaos in ComMet drives the deterministic simplicity of the
emerging patterns (I call the family of these lovely patterns
"Emergent Mosaics"). Having such fluctuating geometric niches means
that we have computed a geometrical mechanism that traps and molds a
chaotic enumeration of space -- a first step in the goal to program
with chaos, which is my stated objective (see my chapter in C.
Langton, above), pretty much to replace altogether logic and
arithmetic that are the backbone of today's computer science.
Before Barbara Hardy and I get started shaping these contained captive
whirlwinds, I am happy to report that these conformal mappings have
been generated already, in truth, by naive accident, not design"
This took place on a University of Maryland's Institute for Advanced
Studies Connection Machine (CM) and its associated two output buffers,
the weekend of 13 May of this year, which I spent camping out right in
the carpeted framebuffer room, much to the merriment of the rest of
the Indiana U. weekend blitzkrieg party (they went out to have a
great time; I stayed in and had a time of my life: 2 CM frame buffers
and no one else using the machine! I managed amongst other things to
run a 2 million-generation ComMet computation and study it for hours.
Alas, no video tape machine was available there at the time.
Also unfortunately, I do not have access to a CM-machine framebuffer
at Indiana. We don't have a CM. A frame-buffer and its 80
mebatyes/second cable to the CM CPU is needed in order to produce
output to make a video tape of this but I am hoping to travel
elsewhere and take care of archiving this result and, hopefully,
taping future ones that we actually design as opposed to stumble upon.
However, the overall flavor of programming these tasks has much of the
flavor of devising an experiment that grows something, usually then
turning out not exactly as the dreamed up thing that was expected. A
refinement follows, and the cycle continues. This is natural
selection at work.
Thus before we even get started, Barbara Hardy and I already have "on
the shelf" the class of behaviors of the complexity we need.
Now, we will explore the space of PD "strategies", in quotes for
reasons discussed below. To do so, we will discuss and design, then
write down the L-system grammars that would interact in such a way as
to search through the space of possible "strategies" along interesting
paths. Keeping to interesting paths is critical here, and grammars
are devices for, well, grammatical behavior, or keeping on paths.
We wish to keep on our paths for we certainly do not intend to
enumerate chaotic space! Finding the paths of interest by deforming
grammars and observing how the deformations play out in the tiling is
the way to do this, I am hoping. Barbara will bring to this what
wisdom economics has to offer as a pursuit of paths. We are very
excited about the possible implications of our putative success on
both computer science and economics, taken jointly.
Incidentally, the implementation of ComMet we are using here at
Indiana (actually, using from here in Indiana while telecomputing at
Los Alamos, Maryland, NCSA and Thinking Machines, Inc. -- we thank
them all!) was written by my fellow researcher Eric Freeman, also
graduate computer science student at Indiana and member of the
Artificial Life Research Group. Both Eric Freeman and I, as well as
Beth Freeman (also CSCI graduate student at IU) have spend a good deal
of time since December pondering the various aspects of ComMet. It is
doubtful that any of this work would have taken place had we not
become best friends on arrival at IU last fall and decided to first
implement ComMe,t and then to find something about it to study! Well,
it is funny to put it that way, now with the benefit of hindsight.
There's just a lot to shake a stick at, here.
------
Well, please forgive me the detour into artificial life, which is not
exactly a Nepalese phenomenon per se.
I invite your comments and insights, and of course, in particular,
more information on Nepal and on the Tibetan people Nyinba who live
there. I recommend Nancy Levine's book as superb reading even if
modeling conflicts on computers that look like bathroom floors that
jump is not your morning cup of tea...
-- Marek Lugowski
Artificial Life Research Group
LH 101, Indiana University
Bloomington, IN 47405
USA
marek@iuvax.cs.indiana.edu
marek@well.sf.ca.us (<-- permanent address,
forwards to iuvax)
------------------------------
End of ALife Digest
********************************
