>I am concerned with the identification of "noise" in patterns, derived
>from measurements of (for example) natural systems. A scientist may
>seek to model this pattern. In doing so she typically ascribes some
>part of the pattern to "noise", uses some technique to filter this out
>(or ascribe it away) and finds a model for the rest.
>
>The question is on what basis should she do this in the absence of other
>information about the nature of the source system? Is there any
>rational basis for such a task?
>
>If she does have information about the source system, exactly how should
>she use this in the identification of such "noise".
Gary Boyd makes the point that:
> Just as a sustem is in the mind of the the
> system modeller, so are its goals/desired states/process/...
Bruce E. responds (Mar 12):
>In the case I detailed there is no control over any of the pattern.
>Does that mean I have to call the whole pattern mere noise? What if the
>pattern was the Mona Lisa? ...my question is more specific ...
>"Given this goal (and the lack of other information
>about the originating process, is there any rational or principled
>approach to indentifying noise and/or modelling the pattern?".
In his book on The Origins of Knowledge and Imagination, Jacob Bronowski
sets out the basis for his view that the concepts of science, as based on
human perception, have the provisional nature of solutions to a
cryptogram. The solution teases out the elementary symbols - entities and
relationships - and constructs the most useful language to describe these.
People tend to polarize over the views that scientists simply discover what
is there, on the one hand, or that the results of analysis comprise a human
imaginative creation. Bronowski provides considerable evidence for his
view that creativity in art and science have deep commonalities, and there
are no definitive truths of ultimate reality we can know. Any decoding we
can make is provisional.
Applying this principle to Bruce E.'s question lends support to the
cybernetic perspective offered by Gary. It also leads to the answer that
there is indeed no *absolutely objective* principles or rational approach
to modelling patterns found in nature about which nothing is known. It is
precisely the task of scientific inquiry to try to ascertain patterns which
might be consensually validated, and seen as rational and objective to that
degree. Moreover, whether the *whole pattern* of possible intelligibility
is included within the sample selected for analysis cannot be known.
So I do not see how one can escape the conclusion to which Gary's comments
point: that noise is relative to one's orientation and perceptions about
the system under consideration.
This seems a very fundamental point for sociocybernetics, that would insist
upon considering the observer as part of the processes of observation. It
is a very fundamental point for any science, well worth further discussion
IMHO.
Cheers.
Bruce B.
_____________________________________________________________
Bruce H. Buchanan, M.D. buchanan@hookup.net
4690 Dundas St. West (416) 231-6235
Etobicoke, Ontario M9A 1A6 CANADA
"We are all in this together!"
"Science may be described as the art of systematic
over-simplification." (Karl Popper 1982)