Agents, Emergence, Evolution, and SD

Locked
Justin Lyon justin1028 yahoo.com
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Agents, Emergence, Evolution, and SD

Post by Justin Lyon justin1028 yahoo.com » Tue Nov 22, 2005 11:27 am

Posted by Justin Lyon <justin1028@yahoo.com>
Hi everyone,

As part of Simudyne's focus on the convergence of ABM
and SD in Simulation Science, I've been thinking about
the linkages between evolution by natural selection
(Darwin), agent-based modeling and emergence
(Kaufman), the game of life (Conway) and system
dynamics (Forrester).

We believe this convergence has tremendous commercial
value.

I would value your feedback.

on p644 of _The Origin of Order_ by Kaufmann, he says
that ""Evolution is not just 'chance caught on the
wing.' It is not just a tinkering of the ad hoc, of
bricolage, of contraption. It is emergent order
honored and honed by selection."" This would seem to be
a refutation of evolution by natural selection via the
power of self-organization, but it is not. Emergence
does not refute evolution by natural selection, nor do
I think (assumption here) that is what Kaufmann is
saying? Does anyone know his views on this as I'm not
that familiar with his thinking?

Dennet argues in _Darwin's Dangerous Idea_ on p226
that Kaufman is saying that ""there are principles of
order that govern the design process, and that force
the tinker's hand. Even a blind tinker will find the
forced moves; it doesn't take a rocket scientist, as
one says."" But, is that what Kaufman is really saying?

In Conway's game of life, we can see the emergence
from very simple rules what 'looks' like purposeful
movement (gliders, eaters, spaceships and many others
in the simulated world) but this emergence is clearly
not 'purposeful'. Yet, these simple automata are just
like the collection of automata that make up humans
and snakes and hawks and doves in the real world.

In biochemistry we have the repair of DNA by
'proteins' or 'enzymes' or 'agents' or 'automata' or
'robots' or 'macros' - choose your jargon. These
proteins, like DNA glycosylase and DNA polymerases,
follow simple 'rules,' much like cellular automata in
the Game of Life, and in following these simple rules
they repair large numbers of errors in DNA on a daily
basis (something like 20,000 bases are damaged every
day in human DNA). And from this (and a huge host of
other processes) emerges the complicated process of
DNA repair that keeps us alive and kicking and
choosing to dance the night happily away or spending a
quiet evening alone reading.

Or, consider the 'emergence' in bacteria of stopping
tryptophan production (trp) when it 'finds' trp
present in the environment (because it can 'eat' trp
from the environment). If trp is not present it starts synthesizing it from chorismate. This process is mediated by represession, transcription attenuation and feedback. So, you can have simple rules (synthesize trp when none available in the
environment) that emerge into something 'greater' –
apparently purposeful bacteria that start producing
trp when it finds none to eat. Yet, bacteria are no
more purposeful than the 'robots' or 'agent's that
make us up as humans. You can model trp production
kinetics using a set of differential equations to
describe the trp operon behavior over time and solve
the equations using the fourth-order Runge-Kutta
algorithm. It's been done in Mexico by Moises
Santilla´ and published in the Journal of Theoretical
Biology. Classic system dynamics.

Or, you can also model enzyme kintetics using
agent-based modeling where you decompose the system
into 'agents' or 'actors' or 'capsules' (in a rational
rose real-time model) – choose your jargon.

Whether you use agent-based modeling or system
dynamics does not really matter -- it just depends on
the level of emergence that you are interested in
studying.

And, interestingly for the manipulation of high-level
emergent behavior in human systems like businesses,
you can use System Dynamics to model (1) oscillations
in supply chains, (2) limits to growth, (3) success to
the successful, (4) fixes that fail, and on and on . .

Finn Jackson finn.jackson tangle
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Agents, Emergence, Evolution, and SD

Post by Finn Jackson finn.jackson tangle » Thu Nov 24, 2005 2:43 pm

Posted by ""Finn Jackson"" <finn.jackson@tangley.com>
I'm not quite sure what question Justin was asking here. But in response to his specific ""Does anyone know [Kaufman's] views on this?"" I would say that to me the quote:

""Evolution is not just 'chance caught on the wing.' It is not just a tinkering of the ad hoc, of bricolage, of contraption. It is emergent order honored and honed by selection""

chimes exactly with Gregory Bateson's view (expressed in ""Steps to an Ecology of Mind"") that it is the whole ecology that evolves, not just a species. For example, taller giraffes encourage taller trees, and faster leopards result in faster antelope, and so on. This is a classic systems view, combining randomness with directionality. Hence ""emergence"".


Justin lists several 'metaphors' where this pattern applies. To me the missing pattern is commerce, where each 'agent' is a business. For an examination of how the basic interactions between these agents result in evolution and in 'meta-evolution', see my book ""The Escher Cycle"".


Regards,
Finn Jackson
www.theeschercycle.com
Posted by ""Finn Jackson"" <finn.jackson@tangley.com>
posting date Wed, 23 Nov 2005 17:40:03 -0000

Justin Lyon justin1028 yahoo.com
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Agents, Emergence, Evolution, and SD

Post by Justin Lyon justin1028 yahoo.com » Fri Nov 25, 2005 1:22 pm

Posted by Justin Lyon <justin1028@yahoo.com>
Hi everyone,

I've read Finn's book and it is excellent. Highly
recommended. To quote from the epilogue: ""People who
'get' the Escher Cycle will implement it. Those who do
not will not.""

Well, Finn, thanks to your thinking and our
conversations over time, Simudyne gets it and we're implementing The Escher Cycle!

Let me clarify my post.

Does it not seem clear to other people that System
Dynamics is a meta-language for describing Daniel
Dennet's 'principles of order' and the appearance of
emergence in systems (whether natural or created by
humans)?

Are not the 'archetypes,' that is, the 'behavior over
time' as a result of 'structure' of system dynamics,
the key to unlocking and describing the seeming
'mystery' or 'magic' of emergence?

Is the concept of 'emergence' related or perhaps even
the same as the SD concepts of 'behavior over time' as
a result of structure?

Is it possible that system dynamics is the 'physics of biology;' the 'meta-engineering' of life through the wonderful, yet blind and simple algorithmic process of evolution by natural selection of agents that replicate?

Best,
Justin
Posted by Justin Lyon <justin1028@yahoo.com>
posting date Thu, 24 Nov 2005 10:23:59 -0800 (PST)

John Gunkler jgunkler sprintmail
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Agents, Emergence, Evolution, and SD

Post by John Gunkler jgunkler sprintmail » Wed Nov 30, 2005 12:00 pm

Posted by ""John Gunkler"" <jgunkler@sprintmail.com>
Emergence is a tricky concept in the philosophy of science, albeit quite attractive intuitively.

Carl Hempl, in his ""Aspects of Scientific Explanation,"" wrestled with this concept. Here's what he said (in my words unless quoted):

1. The intuitive notion of ""emergence"" includes not merely being unexpected, but also being ""unexplainable, or unpredictable, on the basis of information concerning ... constituents of the system in which the phenomena occur ...."" And he provides the typical example of the characteristics of water (such as ""its transparence and liquidity at room temperature and atmospheric pressure, or its ability to quench thirst have been considered emergent on the ground that they could not possibly have been predicted from a knowledge of the properties of its chemical constituents, hydrogen and oxygen.""

2. This is appealing -- but, as you can see, requires quite a few assumptions about (a) what theories we have to predict physical characteristics of chemical compounds, (b) what ""constituents"" or ""parts"" are to be considered, and (c) what class of attributes of those constituents we are to infer from.

To illustrate point (c), it most definitely IS a characteristic of hydrogen that it, when combined suitably with oxygen, will form water (and we know the properties of water.) So, unless NOTHING could EVER be ""emergent,"" we need to find a way to rule out such characteristics of constituents, and Hempl suggests we do so by enumerating or otherwise defining a class of attributes for the constituents and then say that a certain characteristic of the ""whole"" be emergent with respect to this class of constituent attributes.

3. Hempl suggests a definition for emergence that doesn't have these obvious problems:

The occurrence of a characteristic, W, in an object, w [note: w is chosen to suggest ""whole""], is emergent relative to a theory, T, and part relation, Pt, and a class, G, of attributes if that occurrence cannot be deduced by means of T from a characterization of the Pt-parts of w with respect to all the attributes in G.

In this sense, then, I question whether there are any ""emergent"" events that arise from SD models. We can, and do, predict the behavior of systems from their constituent parts (structure.) What may, perhaps, be said is that SD provides explanations of events that are ""emergent"" with respect to other theories that have previously been invoked.

In fact, isn't that the main point of using system dynamics -- to explain what has previously not been well understood using previous theories and methodologies and constructs? We substitute our theory, our particular brand of mathematical modeling methodology, and the key construct of the feedback loop (and their interrelationships with other feedback loops) for the theories (etc.) of, say, economics or biology or ecology. Posted by ""John Gunkler"" <jgunkler@sprintmail.com> posting date Tue, 29 Nov 2005 11:18:09 -0600

Jean-Jacques Laublé jean-jacques
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Agents, Emergence, Evolution, and SD

Post by Jean-Jacques Laublé jean-jacques » Fri Dec 02, 2005 12:01 pm

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr> Hi John I agree with you that a closed system is the best solution to explore its dynamic as there is no foreign, independent and unexplained influence on it (the outside word). The problem is whether it is always possible to do it with practical cases. I have not the expertise to answer the question.

Concerning my experiences, I have been always been obliged at a moment or another to figure out an external world, otherwise I would have been obliged to expand the boundary and the complexity of the model. Whether it was possible to avoid that trap, I am not sure. I recognize that it may have come from my inexperience. But the software I use, has implemented the possibility to inject outside data not modelled in the model on a period by period basis. It may prove that it is sometimes useful to do it. But advices about that question from experts should be interesting to know.

I would like to know the exact definition of the qualitative behaviour of a variable. If there is one, I will re-examine some models that I was never able to translate from a qualitative description to a quantitative one, and never could find any use, not being able to find any behaviour from them. Regards. J.J. Laublé Allocar Strasbourg France Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr> posting date Fri, 2 Dec 2005 11:16:08 +0100

Locked