Just catching up on the discussion between Alfred and Steve (having read it
once quickly), I thought some extracts from a recent paper I wrote with
Francis Wilson might be of interest.
The basis of the paper is a critique of an analysis by Checkland and Holwell
of the
significance of an information system in determining the outcome of the
Battle of Britain in World War 2 based on the use of certain concepts from
Marxist philosophy. Their view is based on soft systems theory (SSM), though
I think the comments on boundary drawing apply to systems theory in general
and also, it occurred to me afterwards, the notion of boundary implicit in
AT. (The unit of analysis used in SSM is typically that of the 'human
activity system'.)
We start from Marx's distinction between abstract and concrete and indicate
that the boundaries of totalities (in dialectical terms, similar to
Checkland's 'holons' or the more general idea of system boundaries) cannot
be given a priori but emerge instead from a dialectical process of
investigation in which the partial, one-sided and abstract starting point is
overcome and the concrete reality reconstructed as a totality (the movement
from abstract to concrete).
This relates to Alfred's point:
<<" Indeed, systems are mostly defined, i.e. the notion implies system
boundaries; what belongs to a system, what not? So we are in a similar
situation as with objects: we have to define them in advance, based on how
they effect us, rather than how they buils themselves under given
circumstances. >>
My response would be that they must be defined by means of a process which
uncovers the real logical interconnections which are relevant in determining
the answer to a particular question / how to reach a given goal. Thus the
boundaries used (which are not the only possible ones) flow out of the
nature of the raw material under consideration - including its historical
development and the conditions for its coming into being i.e. process and
structure or perhaps better put, structure in process.
After a long historical analysis of the deficiencies of C&H's assessment of
the Battle of Britain, in which we use Bertell Ollman's notion of 'modes of
abstraction', we examine the fundamental distinctions between SSM and
dialectical materialism, one of which relates directly to boundary drawing.
Here is that section of the paper:
"The notion of totality in dialectical thought embodies the idea of drawing
a boundary within which the relationships that define the essence of the
subject under investigation are structured. In moving from the abstract to
the concrete, we discover such a boundary in the course of analysis and
reordering of the empirical matter that provides the basis for the
investigation. This is an iterative process in which the adequacy of an
abstraction is tested against this material and for its own ability to lead
us towards a concrete reproduction of it. Thus, unlike Ulrich (1983, p.282),
we do not see boundary drawing as a question of judgement that 'cannot be
validated either logically or empirically' but only polemically. Rather
boundaries must be real - insofar as they reflect properties that can been
identified in the world outside thought - but cannot be absolute. They can
never be complete or eternal and will differ according to the level of
generalisation and vantage point appropriate to the problem under
consideration.
"Ulrich (1987) notes that systems science has traditionally either totally
ignored the question of boundary drawing or determined it by the
availability of data and modelling techniques, which does not tell us
whether the boundaries have been adequately chosen. SSM gives several
definitions of how the boundaries between system and environment should be
determined and they are not all consistent (von Bülow, 1989). The Human
Activity System (HAS) is, according to von Bülow (1989), `the
operationalisation of the systems idea' in SSM and its scope underlies that
of the POM model.
"As we have seen in our analysis of the Battle of Britain, to evaluate the
impact of an information system and the consequences of its implementation,
it is necessary to draw boundaries more widely than those provided for by
the framework of the HAS or organisational boundaries. These units are
presumably chosen on the basis of Checkland's distinction between system and
environment: `an environment may hopefully be influenced, but cannot be
'engineered', whereas a wider system can ... be engineered.' (Checkland,
1981, p.174) This is an example of how, as Levins (1998b, p.558) argues,
`The choice of boundary between what is the system and what is outside is
usually a consequence of the history of each field and especially of the
division of labour between disciplines... Traditional boundaries between
disciplines act to restrict models of problems to include the acceptable
pathways of intervention.' Churchman (1979, Ch.5) has also pointed to this
distinction in his dialectic between the viewpoints of the real and ideal
systems planner, the former believing that "all systems have real
boundaries; it is not necessary to investigate beyond these boundaries in
designing the system", while the latter believes "that there are no real
boundaries of social systems."
"Checkland and Holwell's narrow vantage point is that of the action
researcher as organisational engineer and thus limited to the scope of what
might be susceptible to change, which contrasts with that of the historian -
who cannot change the past, only perceptions of it. What is considered to be
`environment' is not considered part of the necessary causal relationships
of a systemic explanation. Accordingly, the POM model relegates causal
elements outside its narrow focus to the status of `external changes', only
represented insofar as they cause changes to the perceptions of those
involved in organisational change. This builds one-sidedness into the model
from the start, both in its organisational focus (narrow extension) and its
emphasis on meanings and sense-making (partial view of the relationship
between subject and object). [snip]
"This contrasts with a dialectical approach in which no a priori boundaries
are set to totalities, as the boundaries we draw never exhaust or match
exactly the real world interconnections we are investigating. Both the
nature of the subject matter and the stance of the observer will define the
boundaries best suited to a particular goal. Levins (1998a) suggests that
the tendency towards narrowness, towards only seeing parts even when one
aims at holistic explanations, comes from the impossibility of seeing the
whole (i.e. everything in all its interconnections) and a resulting
overeagerness to deal with what appears tractable. As a counter-heuristic,
he suggests that `at some stage of an investigation we should examine a
larger system than is thought relevant', both because it may provide more
adequate explanations not previously considered and because interconnections
will be seen in a broader context. This corresponds to Ulrich's (1987,
p.106-7) principle that 'aspects that are not well understood ought to be
considered as belonging to the system… at least until their significance has
been studied.' Setting our boundaries widely allows us to see the necessary
internal dynamics of the system under consideration, rather than requiring
the deus ex machina of `external changes.'"
Bruce Robinson
This archive was generated by hypermail 2b29 : Sun Dec 01 2002 - 01:00:07 PST