Re: The value of Rosennean Complexity, applied...

[Tim Gwinn <***>]

JohnK,

I think we perhaps roughly agree, but are using terminology (such as
'structure') differently.

How does this sound?:
What sits on the left-hand side (the natural system) of the modelling
relation are the percepts (observables) and the relationships we perceive to
exist between them. Based on those percepts and their relations, we can
build models of the structural organization that we percieve them to
represent, and, for appropriate natural systems, we can also build models of
the functional organization that we perceive them to represent. Both types
of models are equally valid, equally "real". Further, the mechanistic
assumption that all functional models can always be reduced to structural
models is incorrect. Sometimes, both types of models are needed.



Regards,
Tim

> -----Original Message-----
> From: ROSEN Forum [mailto:*** Behalf Of John
> Kineman
> Sent: Monday, March 15, 2004 9:43 AM
> To: ***
> Subject: Re: The value of Rosennean Complexity, applied...
>
>
> Tim,
>
> Your example of functions cutting across multiple structures is an example
> of the first structure-function diagram that you disagreed with. As I have
> argued before, that's how it works when you have a modeling relation that
> is complex. The two sides of the relationship have multiple
> possibilities -
> as demonstrated in the flying maps animation. The point is that happens in
> the other direction too. If you have a defined function and ask how to
> realize it, you would get multiple possibilities on the left. Now, yes,
> those realizations would be whole systems as Rosen described them
> ("Natural
> System"), not just structures, but they can be uniquely identified and
> labeled by their structure, which is what we all do in practice. A simple
> example is a spoon (it is in reality a natural system as Rosen said, but
> here we identify it by its structure). It can function for many things -
> eating ice cream, opening cans, toys, etc. Now take a related function -
> eating ice cream. How many structures (which are also natural systems,
> otherwise we could not find their structures) are there that would match
> the function of eating ice cream? Spoon is one, but there are many others.
> This is how you apply the theory. Don't get hung up on the word
> "structure"
> - it is a label derived from a particular modeling perspective (mechanism)
> and the diagram is an analysis or how one model relates to another. You
> have to attach it to something when you apply it. I assure you this is
> entirely Rosennean.
>
> While it is true that the "modeling relation" itself describes a relation
> between whole natural systems (not just their structure) and models
> (without structure), that needs to be understood as an analytical
> division.
> Any analysis, as we know, is an approximation. So the modeling relation
> itself is an approximation to the real situation. That is why each side
> requires some interpretation in practice. In practice we relate parts of
> systems necessarily because we cannot represent the whole. For example, a
> scientific model does not exist without a structure, even though the
> "formal system" side of the diagram suggests that we consider just its
> formal aspect. A scientific model, at the very least, requires a brain, so
> in that case you would have to say that "formal system" is wrong
> or that it
> is synnonymous with "natural system" and then you have natural system
> relating to natural system (which is the philosophical reality, not the
> analysis).  We have discussed elsewhere at length how the
> modeling relation
> can apply to relations between models. "Structure" is a
> particular model of
> a natural system. So you have two cases: one where you think of the
> relation as between whole natural systems, and the other where
> you think of
> it as between models. Both are valid. The later is the practical
> application.
>
> So, I would have to say that the relation I've drawn is quite valid and
> useful - but it is a distinction we have not agreed on before. The case
> against it should be made with examples and logic, not just a series of
> quotes which then have to be extracted from their own contexts. These are
> the ideas themselves, in practice.
>
> Thanks for the correction to Alfred! I had in my head that Alexander was
> wrong but didn't remember to check it.
>
> JJK
>
> At 09:53 PM 3/13/04 -0500, you wrote:
> >JohnK,
> >
> >Thanks for the link to the presentation. I have a couple of
> remarks. First,
> >I would disagree with the diagram on Slide 2, comparing structure and
> >function. There is not necessarily a correlation between structure and
> >function, such that one would have a modeling relationship between the
> >physical structural organization and a model of functional
> organization. To
> >use an example from Rosen, metabolism, as a function, cuts across myriad
> >structures in an organism.
> >
> >On Slide #4: I believe the name is "Alfred Korzybski", not "Alexander
> >Korzybsky". Also, Korzybski had a concept about language and
> "maps" as being
> >"self-reflexive", which I think may akin to what you are
> describing in that
> >slide, and might possibly be useful. (I haven't read Korzybski for a long
> >time, so I'm kinda hazy on that.)
> >
> >I really liked the animation in Slide #6, with all the maps
> flying across.
> >:)  In your text on #6, you say "the best we can hope for is to
> have a way
> >to relate the many maps needed to describe a complex reality...". I think
> >you bring up a central problem: the problem with a complex system is that
> >the many models are incommensurable. If we could relate them formally,
> >algorithmically, we could have a single "largest model". But, we can't do
> >that. Even using the planar maps analogy from Rosen for switching between
> >multiple dynamical models (a kind of "data-driven" process) as a way to
> >relate simple models for the system, I think we still cannot
> (or, at least,
> >don't yet know how to) relate the noncomputable (e.g.,
> relational) models to
> >that picture.
> >
> >Regards,
> >Tim
> >
> > > -----Original Message-----
> > > From: ROSEN Forum [mailto:*** Behalf Of John
> > > Kineman
> > > Sent: Friday, March 12, 2004 8:51 PM
> > > To: ***
> > > Subject: Re: The value of Rosennean Complexity, applied...
> > >
> > >
> > > Sory for all these versions trying to get it right. You can also go to
> > > www.nexial.org/Rosen and get an index of the two files.
> > >
> > > -------------------
> > >
> > > Well, even the .pdf didn't make it throught the size filter,
> so you have
> > > to go to the URL to see these slides. I put it there in powerpoint and
> > > in .pdf so you can read the words more ealisy in the pdf but still see
> > > the animations in powerpoint.
> > >
> > > ------------------
> > > Hi all, this didn't get through the size filter because of the
> > > powerpoint attachement. I produced it as a pdf so you can read the
> > > words. The powerpoint is located at:
> > > www.nexial.org/Rosen/mapping_complexity.ppt
> > >
> > > just go there and open it or download it, then you can see the
> > > animations in the slide show format.
> > >
> > > -----------------------
> > >
> > > Tim & list,
> > >
> > > Its amazing how close these comments by Tim are to my own
> thinking. Take
> > > a look at the powerpoint slides in the attachment. I extracted these
> > > from a recent talk to USGS. It was how I decided to
> communicate the idea
> > > of complexity to a group of GIS people who are used to making
> maps that
> > > they consider rather definite representations of reality.
> Please excuse
> > > some of the liberal application of the modeling relation idea for the
> > > purposes of communication. I believe the relation can be
> constructed in
> > > many ways, depending on how one defines a system, so the
> variants of the
> > > idea I hope are philosophically acceptable. In any case, it
> told a good
> > > story that I think many in the room of 150 people heard.
> There were two
> > > main points. One was that "functions are real" - see the
> example of the
> > > US-Mexico border. The second was just the same as what Tim
> says here, we
> > > can start thinking about the entire modeling (or mapping)
> enterprise in
> > > a different way. Instead of cranking out definitive representations of
> > > things we design for adaptability and multiplicity, and then make the
> > > system conform to the many different ways one may need to
> represent the
> > > system. This would directly support Holling's adaptive management
> > > strategy too. So, much of what I'm promoting these days is "adaptive
> > > mapping."
> > >
> > > Note: I couldn't get the file size down any more in the
> attachment it is
> > > 3.5MB so it might take some time to download if you don't have a
> > > fast line.
> > >
> > >