Collaborative exploration of spatial problems
Paul Hendriks
Dirk Vriens
University of Nijmegen
Business School
Email address: p.hendriks@bw.kun
Position paper for the NCGIA I17 Initiative on 'Collaborative Spatial
Decision Making'
The complexity of spatial problems
Spatial problems are usually complex problems. If we are to offer adequate
support to deal with spatial problems, we have to get to grips with the
nature of their complexity. There are a number of elements that contribute
to this complexity. The first is the relation between problems and their
solution space: getting a sufficient understanding of both the problem
space and the associated solution space may suffer from what we might call
a 'technical' complexity. It may, for instance, be unclear which criteria
are relevant, how these are to be combined, what measures are feasible and
what their results may be, how alternative solutions may be scaled as more
or less preferable, etcetera. A second source of complexity is introduced
when more than one goal is associated with the problem. This will lead to
the competitive existence of more than one problem space and more than one
solution space at the same time. A next element to be considered is the
fact that goals can be stated at various levels of abstraction, in the
sense that every solution may be seen as a problem at a lower level of
abstraction, and every problem may be seen as an alternative solution at a
higher level of abstraction. The situation is further confused by what
might be called the social context of problems: the existence of different
and possibly conflicting goals is usually associated with the existence of
different parties, with different interests, different positions with
varying degrees of power within the decision making process, different
access to information sources, etcetera. Finally, a fifth complication is
made up by the fact that goals may vary over time as may our understanding
of alternative solutions, thereby causing a shift in the problems to be
solved as well as their solutions. An example of a spatial decision
problem borrowed from Reitsma & Behrens (1991) may help clarify these
various, what might be called, 'domains of complexity'. The example
describes the case of river basin management in the western part of the
United States. A multi-faceted setting for water management is defined
here by the great variety of factors at play, such as shortage of water
supply, the occurrence of flooding, the use of water for such widely
diverging purposes as power generation and rafting. Many of the management
problems involved fall within the first category, for instance perceiving
which technical measures may help control the water flow and what effects
they will have: what (and when) is the effect of closing or opening dams
up the river for the downstream area, how do effects of measures at
individual dams combine, etcetera. The situation is complicated by the
fact that apart from water control to prevent floods, the waters also have
to be managed for the generation of hydropower, meeting urban and rural
demands for water, maintaining an economic viability in fish hatcheries,
etcetera (second source of complexity identified before). In the third
place we may identify these goals and objectives at various levels of
abstraction. For instance: the use of water as a source of power cannot be
studied in isolation, but should be related to the fact that the overall
goal of power supply can also be attained from alternative sources, and
that the more abstract goal of power generation may have competitors (for
instance: energy saving) for its own higher order goals. In the fourth
place, the decision process in the case of river basin management is, as
Reitsma & Behrens (1991, p.33) explain, not something that can be easily
pinpointed to a number of clearly identifiable meetings in some management
office. There are many parties involved, including local and federal
government, environmental pressure groups, individual consumers and
consumer groups, firms, etcetera, all spread out over a wide decision
network with more or less clearly identifiable cross links. A final
complication stems from the fact that neither these parties, nor their
goals remain stable over time, thereby making the river basin management
problem a highly dynamic one.
The case for collaborative decision support
The case of river basis management is clearly a case in which a group
decision support tool may prove fruitful: the complexity of the situation
consists among other things in the presence of various interest groups. It
will be clear that when looking for tools to support such a complex
process of decision making from the multi-party perspective, our main
concern should not focus mainly on ways to improve cooperation, but to
address the various sources of complexity at the same time. If we fail to
do so, and if we instead concentrate on solving the complexities of only
one source (for instance the 'technical' source) for the various parties
involved, we run the risk of providing the right solutions for the wrong
problems. The question may then be asked what the goal is of designing
tools in the give situation. Three alternatives have been discerned
(Reitsma & Behrens, 1991, p.34):
a. aim at solving the problems, that is design tools in such a way that
they will allow the decision makers (DMs) to relate problems to solutions;
this approach is, for instance, taken when models (such as MCA) are made
available to the DMs; underlying assumption is then that the use of these
models, for instance by allowing variations of the model parameters, may
then prove helpful to the DMs to find their way through the solution
space;
b. aim at satisfying the participants, for instance by exploring ways to
reach consensus with other parties involved about alternative solution
paths; this second alternative may build on the first, for instance by
providing means for participants to have their individual modeling
outcomes combined with these of other participants;
c. build the group decision support system as an information generating
tool that will help participants to gain more insight into how the
proposed decisions will affect their own particular situation. Reitsma &
Behrens identify this as 'the informative GDSS'.
Common to these approaches is that, to a different degree, they all
converge around problem solutions. The paper tries to elaborate a fourth
approach, an approach that concentrates on problem exploration instead.
Basic idea is that the combination of different sources of complexity as
sketched before should be integrated and addressed by the DM as much as
possible. Before thinking in terms of alternative solutions to these
various aspects of complexity, it is seen as essential to explore the
nature of complexity of the problem at hand as widely as possible. The
approach therefore shares with the third alternative described before (the
informative GDSS) the concern not to strive for consensus too soon, it
differs from this approach mainly by its problem orientation rather than
solution orientation.
A formal basis in systems theory
Key issue in a collaborative problem exploration approach is finding a
formal representation that will allow all sources of complexity to be
represented. In a recent paper (Vriens & Hendriks, 1995) we have indicated
that the theory of adaptive systems may serve as a basis that will allow
the introduction of dynamic aspects (the fifth source of complexity as
described before). An adaptive system is basically a system that can show
behaviour aiming at "maintaining the essential variables within [...]
limits" (Ashby, 1960, p.58). Systems theory offers the tools to provide a
general model for problem situations, both at the conceptual level and at
the level of an actual tool to be used to model all relevant aspects. When
put in systemic terms, a problem can be said to occur when a system, in
the cybernetic sense, does not manage to keep its essential variables
within certain limits. At this stage it becomes vital for the system to
adapt in order to reach a new state of equilibrium. In order to do so a
match has to be found between the variety of the environment causing the
problem situation and the variety of possible actions. Here the GDSS comes
into play, as it is conceived here as a means to relate as many
alternative actions as possible to the perceived goals (a more elaborate
description of adaptive systems and how they help address the various
sources of complexity shall be given in the proposed contribution). In
that paper, however, we did not address the social context of decision
making, that is the explicit recognition of the fact that conflicting
goals are usually linked to opposing parties in the decision process.
There are basically two alternative ways to do so: the first is to
introduce a model of the opposing goals into the "single explorer"
situation, the second is to model every distinct goal situation as a
system in its own right and establish conflicts and overlaps between these
individual situations in terms of the actions conceived within each of
them. In the contribution this second approach will be elaborated, as it
is superior in terms of allowing individual parties to explore their own
problem space independently, and identify conflicts and overlaps with
other problem spaces as a separate step. Central in the approach is its
focus on actions that are feasible within each individual context and the
fact that it stimulates the participants to come up with as many
alternative actions as they can conceive. The task of the CSDM-tool is
both to help participants define their private problem spaces, and to
suggest matches between the exploration outcomes of participants with
interests that appeared as opposite, as well as to identify situations
where no such match has yet been reached. It should be stressed that in
the approach as advocated the focus is not on finding consensus or
starting negotiations, but on as wide a problem exploration as possible,
in order to better the chances for overlaps in actions. A simple example
may help clarify this: imagine two people wanting to go to the movies
together but having different preferences as to which movie to pick. It
may be suggested to each of them to contemplate on what they hope to gain
by going to this specific movie, and the outcome for both parties may be
something like 'recreation and relaxation'. It may then be suggested to
them to seek for alternative ways to satisfy this objective, and at the
end of the process we may see them going out to dinner and live happily
ever after. Another - classical - example given by Ackoff illustrates the
same point: in a multi-storied office building firms occupying the upper
floors complained of the long waiting time for the elevators, and three
lines of action were suggested to solve the problem. The first was to
introduce a computer system to manage the, what we might call, ups and
downs of the elevators more intelligently (a sort of decision support
tool, though not a collaborative one), the second to increase the number
of elevators, and the third to reserve certain elevators for the higher
floors. None of these appeared to solve the problem, complaints persisted.
These, however, stopped when someone came up with the idea to put mirrors
up in the elevator hall, giving the persons waiting the opportunity to
check their ties and make-up, and to spy on their fellow waiters. Problem
solved. The waiting time in terms of minutes and seconds had not changed,
but its perception had. As in the previous example, a creative problem
exploration not aimed at consensus but at as wide an search for feasible
actions as possible, proved to be far more rewarding than a conventional
problem- solution centered approach. Our elaboration of adapted systems
theory with collaborative elements may be seen as an attempt to provide
this creative problem exploration process with the formal basis necessary
for defining collaborative decision tools.
References
Ashby, W.R. (1960) Design for a Brain. London: Chapman and
Hall.
Reitsma, R.F. & Behrend, J.S. (1991) Integrated river basin management: a
decision support approach; in: Klosterman, R.E. (ed.) Second International
Conference on Computers in Urban Planning and Urban Management,
Proceedings; Oxford, July 1991; pp.29-41.
Schrage, M. (1990) Shared minds: the new technologies of
collaboration; New York: Random House.
Vriens, D. & Hendriks, P.H.J. (1995) How to define problems: a systemic
approach; in: Timmermans, H. (ed.) Decision support systems in
architecture and urban planning; London: Chapman & Hall (in press).