Don Mikulecky wrote:
> Don Mikulecky replies:
>
> Ricardo Ribeiro Gudwin wrote:
>
> > I have some important (well, ... , even to me) questions regarding building
> > systems ala Rosen.
> >
> > 1) Are we ABLE TO fully understand a complex system ? This is the
> > same as saying that we may find a complete analog for a complex systems that
> > can be represented in some way inside our brains. Or either in other words
..
> > are we able to build a "conceptual model" for a complex system ? Or are we
> > condemned to know it only partially ?
>
> If we are using rosen's definition, we can only know it partially.....
Yes, this is the answer I was expecting !
> > 2) As I could understand from past messages and from Rosen's interview,
> > a complex system is a system where, depending on the observation we do
> > on it, it shows a different perspective. It should be compared to a diamond
> > with an infinite number of faces. And this is what concerns me. These
> different
> > aspects in which we can "see" a complex system are NECESSARILY INFINITE ? If
> > there
> > is a finite (it may be huge, but finite) only number of different
> perspectives,
> > then
> >
> > we could acquaint for a mechanistic implementation of it. So it seems that
> > necessarily
> > it have to be infinite. Then, how to guarantee that we can find a perfect
> analog
> > for
> >
> > a given complex system ? Or in other words ... are each complex system
unique
> ?
> > If yes, and if it is not possible to make a perfect analog of it, then
> > necessarily
> > it is
> > really impossible to know it in its entirety !
> >
>
> Rosen believes that an infinite number of formal descriptions are needed...the
> uniqueness is an important point...since the material aspects do not map 1:1
> into
> the functional aspects, each will be unique......a class of material
> realizations
> may fit all of our limited knowledge about the system....
This observations leads to some interesting speculations ... I would try to
explain.In control theory, we have what we call Frequency Domain Analysis, where
we
convert (e.g. by means a Fourier Transform) a signal in time to its
corresponding
signal in the domain of frequency. This is interesting, because e.g. periodic
signals
do have a finite number of components in the frequency domain, so it is easier
to
understand them in the frequency domain, where instead a signal that is infinite
in time, we have only the coeficients for its finite frequency components. A
non-
well-behaved signal may have though an infinite number of components in the
domain of frequency. BUT, what is interesting from this view is that normally
we may have one or more FUNDAMENTAL components, with a larger intensity,
and the other components just complement it to fully describe the signal.
Now back to our complex systems domain and the metaphor of the diamond
of infinite number of faces. Supose that these faces are not of of equal size,
but
there are a finite number of major faces, and all the other (infinite number of)
faces
are also infinitelly small. Then, if we look at this diamond without care, we
would
think that it have a finite number of faces.
A complex system is a system where we may have an infinite number of views of
its behavior. But, from these infinite number of views, maybe we may get some of
them that are the fundamental, ... , exactly those that are captured by a
mechanistic
system. Then, what it would be a mechanistic system ?? ... just a signal in
frequency
that was TRUNCATED to a certain number of components, and analysed only
through those components.
This would eventually work with most of the artifacts that humans are able to
build and use in its daily life. As the small components don't play that much
(sometimes it does, e.g. when our computer has a memory parity error or in
situations like that), we are able to model the fundamental component part (the
mechanistic model) and predict its behavior, UNTIL some of the smaller
components
take place in some particular situations (e.g. when our machines doesn't work as
we
predict). This allowed the emergence of the reductionistic view of nature by
science.
Conceptually, the problems start in some pathological cases, where this
arragement
of a set of fundamental modes with greater coeficients is not allowed. Then, we
may have infinite components, all of them important for explaining the behavior
of
the system. In this case, the reductionistic view is not going to work, because
if
we truncate the number of components, we are severely prejudicing the work of
the system. These are the complex systems analysed by Rosen !
> > 3) Talking about emergence. Don said that others used emergence to explain
> parts
> > of complex systems as if they were mechanistic systems. But we may have the
> > phenomenum of emergence in purely mechanistic systems where two simple
> > systems coupled to each other form a complicated systems (just to make a
> > difference from "complex" system, as Rosen did), exibiting a behavior that
can
> > not be seen by each simple system in particular. It is a true case of
> emergence
> > of
> > a behavior, dictated by the cooperation of the two simple systems. The
> question
> > here is ... if we have an INFINITE number of cooperating mechanistic
systems,
> > do we reach a complex system ?
>
> emergence is the result of error in our formal description...once we realize
the
> limitations of our formal descriptions, emergence becomes less
> important.......and
> as I interpret the question, ....an infinite number of formal descriptions
will
> not
> even capture the entire complex system (Goedel).
Hmmm ... aren't you contradicting yourself with this ? (see your answer to
myquestion 2)
> ....you envision an infinite number
> of mechanisms giving something other than a mechanism, but I think that would
> not be the case...just an infinitely complicated mechanism.....and yes,
emergence
>
> (error) might come with that
> > 4) Other question comes from the discipline of Artificial Life, as we know
it
> > today.
> >
> > Regarding complex systems, are all the research in Artificial Life (most
> > backgrounded
> > by computer simulation) only rubish ?
>
> no, I think they are fascinatingly complicated mechanisms
>
> > Does it have NOTHING to do with REAL life
> > ?
> > Or is there SOMETHING in life that IS mechanistic, which is captured by the
> > current
> > studies on Artificial Life ?
> >
>
> they are, in the sense of the modeling relation, often nice METAPHORS for
living
> things....realize that among the troubles with computer simulation is the mix
up
> between the computer's hardware and software and the counterparts in the thing
> being
> simulated. unless one is simulating a Turing machine, this is all awry!
there
> are
> many more, equally fundamental problems with the causalities.
We may see it as the TRUNCATION of a potentially infinite number of
components.Then,
Artificial Life would be an "approximation" of real life, that would be as good
as the number of components we include into the model.
> > 5) Can we understand complexity by induction ? We start with simple
> mechanistic
> > systems,
> > build complicated systems with the interaction of them, over and over
> > complicating
> > the system and then, ... , by induction (this incredible human artifact for
> > understanding the
> > infinite) we reach a complex system ? (I'm thinking here in a mental model
for
> a
> > complex
> > system)
>
> no they form disjoint categories...really mutually exlusive.....as soon as
you
> reduce a formal description of a system to a mechanism you loose semantics
which
> you can not recapture
I will try to better explain myself here ... humans have the capacity of
generalization(induction) as a way of understanding infinite things. For
example, I
can give you
two numbers a and b, and say that they are the limits of an interval [a,b] that
has
an
infinite number of points. And with only those two numbers, we mean the whole
set
of numbers that are between them. So, humans have ways of compressing knowledge,
and acomodate infinite things within finite discrete things. The process that
leads
to such
amazing capability is being studied for a long time, but scientists doesn't have
still a
complete understanding of it, which would allows an artificial implementation
of
it.
Maybe, this capability would explain why we can understand complex system !!!
> > 6) Do purelly mechanistic system really exist ?
>
> no they are the creation of mechanistic/reductionist science, especially
> physics
> which is so special as to only study the mechanistic aspects of nature
Hmmm ... would you say that it is a CREATION of science or that sciencesimply
DISCOVERED a resource used by our mind in explaining nature ?
The question is ... before Newton and Descartes formally posed reductionism
as a method, don't you think that humans have prior used it as a mode of
thinking ?
In this case, don't you think that a (pretense) intelligent system SHOULD USE
reductionistic techniques in order to understand its environment ?
> > It seems to me that EVERY real
> > system that is materially implemented is not a mechanistic system anymore,
but
> a
> > TRUE COMPLEX SYSTEM.
>
> As does Rosen
It seems that I am starting understanding Rosen .. :-)
> > It leads to the question that we look to this complex
> > system, but what we see is a mechanistic system So, true mechanistic system
> do
> > exist only in our thoughts. Only models are mechanistic. And this provokes
the
> > following
> > question: Are we able to have thoughts that are NOT of mechanistic systems ?
>
> yes...wwe can entertain notions of non-computable qualities which are
> complex.....Rosen's best illustration is the number system!!!!
You are saying here ... e.g. the understanding of what is the number 4 ?
(asBertrand
Russel had explained, a natural number is the understanding of the
relation that exist among all sets with the same number of elements ... what
obviously can only be understood by induction). Can you explain better
what you mean with number system ?
> > If we observe well, we will see that it is exactly the question 1) above:
are
> we
> able to
> > understand, conceptualize, model a true complex system ? Or are we able only
> to see a system and say ... well, THIS IS a complex system, without
understanding
>
> > it at all ?
>
> we are more able to "understand" on some level than predict or control.....
So it is not exactly an understanding, but an ability to differentiate it from
otherthings, besides not having a full comprehension of what it is !!!
> wow!
> heartfelt thanks!
>
> You got to the meat of a lot of issues. I think I have given you the kind of
> answers Rosen would have , but I am far more limited in my understanding.
Keep
> asking!
> respectfully,
> Don
My interest in complex systems regards the synthesis of true intelligent
systems. I
am
developing an approach, that uses semiotics in order to model intelligent
behavior.
I am interested in how to include complex system concepts within my approach, in
order to better explain the phenomenum of intelligence. And, if I discover that
it
is not
possible, then it would be important to me to measure how much am I from a true
intelligence, using only my approach. It is interesting to know how far can we
go
using only mechanistic ideas, and what would be the improvement in including
complexity in its conceptualization.
Thanks for the answers !
Ricardo
--
//\\\
(o o)
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-oOO--(_)--OOo-=-=-+
\ Prof. Ricardo Ribeiro Gudwin /
/ Intelligent Systems Development Group \
\ DCA - FEEC - UNICAMP | INTERNET /
/ Caixa Postal 6101 | gudwin@dca.fee.unicamp.br \
\ 13081-970 Campinas, SP | gudwin@fee.unicamp.br /
/ BRAZIL | gudwin@correionet.com.br \
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
\ URL: http://www.dca.fee.unicamp.br/~gudwin/ /
/ Telephones: +55 (19) 788-3819 DCA/Unicamp (University) \
\ +55 (19) 254-0184 Residencia (Home) /
/ FAX: +55 (19) 289-1395 \
+-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+