Posted by
glen ep ropella on
Mar 12, 2008; 9:44pm
URL: http://friam.383.s1.nabble.com/Agent-Based-Modeling-s-Role-in-Understanding-Complexity-tp525934p525935.html
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
admin at wkbank.com wrote:
> -new understanding in the way say dreams and dreaming allows us to explore
> and understand ourselves and each other in a way we wouldn't be able to do
> otherwise
>
> or
>
> -does ABM allow us to see something more like the numerical result of p
> divided out for 10 years. That is does ABM allow us to do things we don't
> have the interest or patience to spend our time doing.
Both, though the tendency would be toward the latter because that's
exactly what computers are good at, doing things we don't have the time
or interest to do.
Ultimately, however, you have to remember that the "M" in ABM directly
implies that an ABM (because it's a model) is MERELY a form of rhetoric,
more specifically, it's a form of deductive _logic_. And in that sense,
all ABMs, as rhetoric, commit the fallacy of "begging the question"
(a.k.a. "assuming one's conclusions", petitio principii). Nothing can
come out of it that you didn't already program into it.
So, in that sense, an ABM won't tell you _anything_ you didn't already
know in some form or another.
And that means that the only thing an ABM is good for is taking your
premises and munging them back and forth (using logic of some sort) to
spit them out at you in some other form. (And, yes, this is true even
if there's outside input from, say, a human or some external stochastic
process, at various points in the execution.)
Now, many people use this deductive machinery to estimate or predict the
consequences of some initial conditions. And where the internal logic
of the model is acyclic (no feedback), that's a valid use. But, regular
old models (not ABM) are just as good or better at that sort of thing
than ABM.
ABM is particularly good at munging your premises and spitting them out
in different forms when the logic is (or can be) _cyclic_. I.e. when
there are feedback loops or self-reference inside. The canonical
example for such cyclic logic is in a system where the environment for
some "agents" is defined (collectively) by another set of "agents" and
vice versa.
And if seen in that light, regular old modeling (non ABM) _could_ help
find new understanding by exploring dreams (or any psychologically
relevant acyclic construct). But, because dreams (and psychological
constructs in general) are social constructs developed in cycles with
many dreamers as well as a semantic grounding in a cyclically
constructed ecology, I'd say that of all the types of modeling one might
use, ABM will be much more helpful in exploring ourselves than other
forms of modeling.
> Both can be useful but what is it we are looking for? Can ABM bring us
> closer together in understanding each other? How? In what way?
Respectively: Cycles, specifically co-evolution and stigmergy. Yes.
By helping us to formulate, execute, and criticize those hypothetical
(generative and phenomenal) cycles.
Note, however, that the TENDENCY is to use ABM in the same old way
traditional modelers use computers, to do work we're too lazy or
ill-equipped to do, like counting to high numbers and such. My guess is
that if you examine 1000 ABMs, you are likely to find only a handful
that explicitly study causal cycles. In other words, most ABMs should
not be ABMs at all. And in the handful of cases where cycles are
studied, they are probably studied from a reductionist point of view,
wherein herculean attempts are made to _isolate_ various pathways
(cyclic or not), which defeats the very purpose of an ABM. After all,
it's the _braided_ or woven nature of causal networks (in contrast to
causal _chains_) that gave rise to ABM to begin with.
Anyway, the domain is ripe with abuse. But don't let the poverty of
most ABMs convince you that ABM is vacuous as an approach/perspective.
You just have to view each individual ABM with a critical eye and think
to yourself, "What cyclic reasoning does this model reify?"
- --
glen e. p. ropella, 971-219-3846,
http://tempusdictum.comAnd therefore the victories won by a master of war gain him neither
reputation for wisdom nor merit for valour. -- Sun Tzu, "The Art of War"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla -
http://enigmail.mozdev.orgiD8DBQFH2E7WpVJZMHoGoM8RAm0UAKClIqyMV8esDF4b3bITkEHJ+Qs2JACfYpe9
lfPuvILFsb+xUCWyy37ST9I=
=+DrA
-----END PGP SIGNATURE-----