 |
Virtual
Ecosystems -- Self Organization
|
|
"Each
individual neither intends to promote the public
interest, nor knows how much he is promoting it...he
intends only his own gain, and he is in this...led by an
invisible hand to promote an end which was no part of
his invention"
Adam Smith -- The Wealth of Nations 1776 |
Self-organization is the evolution of a system behavior
into an organized form from an apparently random state
in the absence of external influence or management. The
self-organized structures and dynamics at the ecosystem
level are caused by many individual members of the
population all following a set of simple local rules.
Any system that displays a pattern not imposed from the
outside (e.g. by walls, a leader, machines, or natural
forces) can be said to be self-organized.
The idea of self-organization is not intuitive. Indeed
we expect that, left to themselves, things become
disorganized. We expect that an external force (or
manager) is needed to restore and maintain order. But,
for some things, this expectation is wrong. Certain
systems start from a very random state and, without any
help or management from the outside, become organized.
Self-organizing systems must be constantly moving (i.e.
dynamic) to maintain their order. This means that the
individuals in the system must be constantly interacting
with each other and operating on the local rules. This
dynamic quality requires that there be a flow of energy
to and from the system. This is consistent with the
notion that ecosystems are thermodynamically open
systems.
It is important to note that self-organization must
usually be coupled with some positive feedback mechanism
if the system is to come together in the first place.
Birds, fish, and penguins must first be motivated to
come together before self-organization can occur. That
joining mechanism is usually positive feedback coming
from such genetically induced factors as protection,
foraging, or a sense of community.
In addition, a system does not live in isolation.
Physical and biological influences from outside the
system serve to shape individual rules and, in turn,
system patterns. Ambient temperature, wind or current
movements, and the presence of predators are examples of
external influences. The regulatory responses to these
influences from individuals in the ecosystem are
examples of negative feedback.
The system behavior that results from the actions of
individuals following a set of local rules is known as emergent
behavior . A system's emergent behavior is always
greater than the sum of the individual behaviors because
the interaction between individuals is also a part of
the system's behavior.
Fish
schooling is a well studied example of emergent
behavior. Here, there are no leaders. Instead, each fish
locally applies a set of simple rules that govern his
speed, distance, and direction with respect to his
nearest neighbors. It is the actions of each individual
with respect to his neighbors that result in the
behavior and organization of the school. What is so
intriguing is that the simple local rules result in
system-level behaviors that are quite complex.
Suppose you have a friend who is visiting a planet in
another galaxy where sending telegrams is prohibitively
expensive. He forgot to take along his trigonometric
tables and he has asked you to send them. You could
simply translate the numbers into a binary code and
transmit them directly. But even the most modest tables
have a few thousand digits. The cost of transmission
would be very high. A much cheaper way to convey the
same information would be to transmit the instructions
for building such a table from an underlying
trigonometric formula such as Euler's equation ( e =
cos(x + i)sin(x) ) -- an equation with 16 characters.
Inherent in this formula is all the information
contained in even the largest tables.
This little tale illustrates the value of a rule or
algorithm. For the price of 16 digits, we can provide
information that represents thousands of digits. All we
need is a machine to do the computing. Our genetic
structure operates in much the same way. Genes do not
carry all the information necessary to create and
operate an organism. But, they do carry the set of rules
that are needed to generate the information. If genes
didn't carry the rules, genes would be required to carry
millions of explicit codings for all the chemicals,
antibodies, etc. needed for life to exist. This would be
an impossible task.
This idea of information economy carries into
self-organizing ecosystems. Simple rules at the level of
the individual eliminate the need for massive amounts of
information at the system level.
The scientific study of self-organizing systems is
relatively new, although questions about how system
organization occurs have been raised since ancient
times. Many natural systems show organization (e.g.
galaxies, planets, chemical compounds, cells, organisms
and societies). Traditional scientific methods attempt
to explain system organization by studying the processes
applicable to a system's component parts. This worldview
is called reductionism. But, reductionism fails to
account for the relationships between the components in
a system.
Yet system studies can be approached in a very different
way by examining the rules that govern individual
behaviors and observing the emergent system behavior
that results when these rules are applied. It is here
that modern computers prove essential by allowing us to
simulate the dynamic changes that occur over vast
numbers of time steps and with different rule sets or
rule values. The values used for a simulation can be
collected field data, postulated data, or a combination
of both.
It is important to note that a system pattern alone does
not prove that a system is self-organizing. An
understanding of the underlying mechanism for pattern
evolution is essential. This means determining if the
individuals in a system are operating under a set of
local rules. |
|
|
|
|
|
