> Don Mikulecky,MCV/VCU, Mikulecky@gems.vcu.edu
> As I promised, here's a section from "Anticipatory Systems" by Robert Rosen,
> Pergamon, 1985..pp83...
> "One further aspect of [the modeling relation] may be mentioned briefly here.
> We are subsequently going to relate our capacity to produce independent
> encodings of a given natural system N with the COMPLEXITY of N. Roughly
> speaking, the more such encoding we can produce, the more complex we will
> regard the system N.
"Roughly speaking" is right here. I will argue that this is related
to complexity but does not qualify as a definition.
> Thus, contrary to the traditional views regarding system
> complexity, we do not treat complexity as the property of some particular
> encoding [note..here's where I am afraid Jeff has led us off the path a bit] ,
> and hence indentifiable with the mathematical property of a formal system
> (such as dimensionality, number of generators,or the like). Nor is complexity
> entirely an objective property of N, in the sense of being itself a directly
> perceptible quality, which can be measured by a meter. Rather, complexity
> pertains as much to US AS OBSERVERS as it does to N; it reflects OUR ABILITY
> TO INTERACT WITH N in such a way as to make its qualities visible to us.
This is (now) a view well accepted, and repeated by many thinkers
about complexity. I agree with it but do not think it goes quite
far enough. I propose that complexity is not only relative to the
OBSERVER but relative to the many languages within which that single
observer might seek to construct the model. This is implicit in the
top section of the quote from Roasen, but not so evident here.
> Intuitively speaking, if N is such that we can interact with it in only a few
> ways, there will be a correspondingly few distinct encodings we can make of
> the qualities we percieve thereby, and N will appear to us as a SIMPLE system;
> if N is such that we can interact with it in many ways, we will be able to
> produce many distinct encodings, and will correspondingly regard N as complex.
These last two phrases, show the weakness of this
definition/approach to 'complexity'. The sheer numerosity of the
encodings (or corresponding ways of interaction) is insufficient to
imply 'complexity' (depending a bit on what we mean by 'distinct'),
because (although they are formally essentially distinct) on a
meta-level they may be but trivial extensions of each other.
Numerosity of distinct encodings is a necessary condition for
'complexity' but not sufficient. It is unfortunate that Rosen
blurred what was a basically good insight by prematurely applying the
terms of 'simple' 'and 'complex' here.
(Note that here (at least) Rosen is definitely allowing for more than
a simple complex-simple dicotomy).
> It must be recognized that we are speaking here complexity as an attribute of
> NATURAL systems; the same word (complexity) may, and often is, used to
> describe some attributes of a FORMAL system. But this involves quite a
> different concept, with which we are not presently concerned."
This is a useful point. Here Rosen does not ascribe complexity to
the whole modelling realtion but clearly to the "NATURAL systems".
This is beacuse he is implicitly quatifying over possible encodings
in his definition (that this is utterly impractical - even
theoretically - is passed over in this quote).
Here Rosen is being clear as to his application of the concept - this
is good. We get a different use of the word if we apply it to the
whole "modelling relation" - it is in this context that I find it
difficult to totally exclude the effect of the complexity of the
"FORMAL" system on the whole.
> This seems to be as near self contained as possible, but in fact is very
> context dependent still. Chapters are devoted to preparig us for this. The
> book "Fundamentals of Measurement" lays the grounwork to understanding the
> modeling relation in terms of measurement. Others have written lots on the
> problem of measurement. And, most clearly, Rosen is challenging our notion of
> SCIENTIFIC objectivity without delving into the deeper questions of
> subjective/objective human perception at this time. We seem to be going
> afield here too.
I have read these. Please do not merely try and "immunise" Rosen
from criticism by saying I have to study him more. If you think
previous context is releveant - quote it.
> Be reminded that Rosen has experienced some of the worst of scientific
> sociology. His reasons for asking and trying to answer these questions are
> bringing us necessarily into the COMPLEX nature of science and epistemology.
This is one of Rosen's great strengths. It is unfortunate that he
sometimes chose to use a slightly inappropriate terminology of
'complexity' to do this.
> Any attempt to "sterilize" this is a reduction of the worst kind and misses
> the point entirely.
If by this you wish to _effectively_ judge any criticism as mere
sterlization, then what we say won't matter - you will have done the
job already.
> In this sense, Rosen is the ULTIMATE pragmatist as well
> as an exquisite theoretician.
This seems to be yet another attempt to put Rosen's views beyond
argument and criticism? I hope not. To me having a high regard for
someone's viewpoint is closely tied up with it being able to stand
its own ground against criticism. I regard Rosen's views highly -
they are worth criticising and impoving. This can be furthered by
intelligent dialogue which, alas, will not be possible if we have to
take it as read that he is already CORRECT.
----------------------------------------------------------
Bruce Edmonds
Centre for Policy Modelling,
Manchester Metropolitan University, Aytoun Building,
Aytoun Street, Manchester, M1 3GH. UK.
Tel: +44 161 247 6479 Fax: +44 161 247 6802
http://bruce.edmonds.name/bme_home.html