Session3: on systems thinking (long
post)
Design / systems /
complexity / evolution / etc.
Some considerations regarding old-fashioned concepts in design,
mainly based on Jonas (2003).
Design process models
Stated in the most general manner, a design task consists in
transferring an existing state of a real world "system" into
a preferred one, whereby "system" will normally be
considered as some kind of complex "whole", consisting of
elements and relations between these elements. The preferred state can
be defined as an optimal "fit" of the system or artefact and
its environment. The artefact is what designers design, whereas the
environment consists of the given requirements that have to be met and
which cannot be directly controlled by design. The "interface"
region between the artefact and the environment is the "location"
of design activities.
The systems concept in design appears to be rather simplistic indeed.
There is hardly any reference to the elaborate thermodynamic and
biological theories of open / dissipative systems, which explain how
systems are able to keep a state of high order far from equilibrium,
thus overcoming the 2nd law of thermodynamics. Systems concepts in
design are mainly based on simplified applications of Wiener's
cybernetics (Wiener 1948), dealing with feedback processes
in goal-oriented processes, and Weaver's concept of "organized
complexity" (Weaver 1948), which filled the obvious gap
between the classical concepts of "problems of simplicity"
and "problems of disorganized complexity".
This happened in the 1940s, and Operations Research (OR) can be
regarded as the first application of systems thinking in
"designerly" processes, such as planning and engineering,
from the 1950s onward. The problem-solving process in OR consists of
the definition of the solution space, the formulation of the measure
of merit, the fixing of constraints, and the optimisation process,
leading to a local or global optimum.
The design methods movement in the 1960s adopted and developed these
approaches. Symbolic models of the design problem have to be built,
consisting of factors, which describe the problem situation and causal
relations between these factors. Ideally, the solution criterion is
given in a quantitative manner, as a measure of merit function. Even
aesthetic criteria have been treated in this way, as we know.
Numerical optimisation methods based on closed mathematical calculus
or iterative heuristic algorithms can be applied to this problem type.
The problem space has to be limited and well-known, and if the problem
is properly stated, then the solution is just a re-formulation of the
problem, or a change of representation, which can be carried out by
means of the same symbolic language that was used for the problem
definition. But unfortunately, this has never been the normal
situation in design.
Design problems may be categorized according to the parameters
problem / solution space (elements + relations available), which
can be either limited or open, and solution criterion (measure
of merit), which can be either quantitative or qualitative, which
means influenced by ethical and aesthetic factors. Entirely numerical
solutions are possible, if the solution space is limited and
the solution criterion is of quantitative nature, which is the
case, for example, in a chess problem or in the optimisation of a
streamlined shape according to aerodynamic criteria. In all other
cases we have value-laden solutions of ethical or aesthetic nature.
Even the apparently highly quantitative problem to bring a man to the
moon has a large number of qualitative subtasks, as for example the
interior of a vehicle. Value-based decisions of minor or major impact
have to be taken at various moments during the solution process.
As soon as the relevant environment of a design problem is no longer
natural, but influenced by psychic or social aspects, then the concept
of time in the process is changing. Time is no longer a linear
parameter, the "fourth dimension", but the source of
uncertainty. The future can be conceived as a projective space,
determined not only by natural trajectories, but by plans, wishes,
hopes, fears, decisions, etc. In other words: it is a space of
imagination. The development of social situations is proceeding in
highly unpredictable ways; the fit between the artefacts and the
environments will probably disappear before long. Nevertheless people
cannot stop asking: What about remaining prediction capabilities for
the future fit of solutions in non-natural
contexts?
The question comprises the issue of "how do we want to
live?", and marks the shift from "first-generation"
to "second generation" methodology, which is closely
connected to Horst Rittel (1972). In his view, first-generation
methods seem to start once all the truly difficult questions have been
dealt with already. He introduced the notion of "wicked problems"
and tried to denote the limits of rationality related to this
kind of problems. Rational behaviour means the attempt to foresee the
consequences of intended actions, which results in 4 paradoxes:
- Tracking down the consequences of actions in advance has
consequences (i.e. it costs time, money, effort, etc.) etc. This
means, one cannot start to be rational.
- Actions have consequences, and so forth. This means, one cannot end
to be rational.
- The more rational one is (in discovering the causal chains of future
consequences), the more one is disabled to act. This means, the
possible alternative paths become unmanageable.
- A model of the future consequences of actions has to include, as
most important component, itself.
According to Rittel, these dilemmas have to be overcome by opening up
the closed algorithmic problem solving process and initiating a
process of argumentation and negotiation among the stakeholders
instead. In other
words: he suggests a change from 1st order cybernetics to 2nd order
cybernetics: not systems are observed, but systems observing
systems. This introduces, as
a central new part, the design of the "problem" itself.
Under conditions of second order observation we have to account for
the fact, that the problem itself is not "given", but has to
be constructed by the stakeholders. In consequence, problems are
changing their character in the course of the solution process. No
information is available, if there is no idea of a solution, because
the questions arising depend on the kind of solution, which one has in
mind. One cannot fully understand and formulate the problem, before it
is solved. Thus, in the end, the solution is the problem. Therefore
Rittel argues for the further development and refinement of the
argumentative model of the design process and the study of the logic
of the designers' reasoning, where logic means the rules of asking
questions, generating information, and arriving at judgements.
In view of this situation Rittel (Cross 1984: 326) states in his
slightly ironic manner:
"All of which implies a certain modesty; while of course on
the other side there is a characteristic of the second generation
which is not so modest, that of lack of respect for existing
situations and an assumption that nothing has to continue to be the
way that it is. That might be expressed in the principle of systematic
doubt or something like it. The second-generation designer also is a
moderate optimist, in that he refuses to believe that planning is
impossible, although his knowledge of the dilemmas of rationality and
the dilemmas of planning for others should tell him otherwise,
perhaps. But he refuses to believe that planning is impossible,
otherwise he would go home. He must also be an
activist."
John Chris Jones (1970) puts it more general and metaphoric, when
emphasizing the necessity of designing the design process itself. A
considerable part of the design capacities has to be re-directed from
the problem to the process. The designer as "black box" (the
artist) as well as the designer as "glass box" (the follower
of 1st generation methods) have to change their attitude towards a
self-conception of designer as "self-organizing system", who
is observing the evolving artefact plus himself observing the evolving
artefact.
This might already be taken as the ultimate conclusion. But I want to
go further, trying to denote the problem more exactly. The systems
concepts and models still seem to be deficient and sub-complex with
respect to the task we are facing.
Inherent patterns - circularity and autopoiesis
Circularity as a characteristic of problem-solving and
design processes is showing up. We know DO - loops as instructions for
iterative processes in formal languages in software-programming. We
know the TOTE - scheme (Test - Operate - Test - Exit) from cognitive
psychology (Miller et. al. 1960) as the prototypical pattern for
dealing with iterative heuristics and feedback in design methods. Most
of these design methods consist of linear sequences of steps of
specific subtasks plus TOTE cycles for the necessary feedback. Opaque
systems, called "black-boxes" are rendered manageable
by means of circular feedback-models. Human agents act as detached
operators of these "machines". Thus systems have been
typically treated mechanistically as open (for matter, energy and
information), and in interaction with their context, transforming
inputs into outputs as a means of creating the conditions necessary
for survival. Changes in the environment are seen as input stimuli, to
which the system must respond in defined manners.
The concept of autopoietic closure in living and meaning-based
systems is of utmost importance for the further argument concerning
design processes. Autopoiesis characterizes the logic of
self-(re)producing systems. Maturana and Varela (1985) argue, that
living systems are organizationally closed, i.e. without any input or
output of control information. Operations only refer to themselves and
the system's internal states. The impression, that living systems are
open to an environment, results from an attempt to make sense of such
systems from the perspective of an outside observer. The aim of
autopoietic systems is ultimately to maintain their own identity and
organization. Change in one element is coupled with changes elsewhere,
setting up continuous circular patterns of interaction, which are
necessarily self-referential. A system cannot enter into interactions
that are not specified in the pattern of relations that define its
organization. In this sense the system's environment is really a part
of itself. The theory of autopoiesis thus admits that systems can be
recognized as having "environments", but insists that
relations with any environment are internally determined;
systems can evolve only along with self-generated paths.
The theory of autopoiesis encourages us to understand the
transformation of living systems as the result of internally generated
change. Rather than suggesting that the system merely adapts to
an environment or that the environment selects the system
configuration that survives, autopoiesis places principal emphasis on
the way the total system of interactions shapes its future. It is the
pattern, or whole, that evolves. Autopoiesis presents a modification
of Darwinian theory: while recognizing the importance of system
variation and the retention of "selected" features in the
process of evolution, the theory offers different explanations as to
how this occurs. Changes are eventually induced, but not directed by
means of perturbations from outside. The emphasis is shifting from
adaptation of a system to its environment towards co-evolution of
autonomous systems.
Morgan (1986: 245) was one of the first to apply the biological
concept of autopoiesis to a design-related field, namely organization
theory:
" When we recognize that the environment is not an independent
domain, and that we don´t necessarily have to compete or struggle
against the environment, a completely new relationship becomes
possible. For example, an organization can explore possible identities
and the conditions under which they can be realized. Organizations
committed to this kind of self-discovery are able to develop a kind of
systemic wisdom. They become more aware of their role and significance
within the whole, and of their ability to facilitate patterns of
change and development that will allow their identity to evolve along
with that of the wider system."
This is a very positive interpretation of autopoiesis, and
probably a step forward with respect to the problems of organizations.
But it still neglects the fact that the environments of autopoietic
systems consist of various other, equally stubborn autopoietic
systems. This is essential for design, because design is not a
single system and does not face a single environment, as for
example an organization in a market.
Luhmann (1984) has formulated this radical generalization of
biological autopoiesis. He extends the auto-poiesis concept of living
systems for the purpose of describing mental and social sy-stems. His
theory of social systems provides more delicate instruments for an
identification of the problem and a composed deconstruction of
unfounded expectations in design theory. Organizations, as described
by Morgan, are one of several sub-categories of communicative / social
systems, all of which are operationally closed, autopoietic systems.
Living systems act in the medium of life, mental systems in
con-sciousness, and social systems in com-muni-cation. Both mental and
social systems operate with language and meaning.
Com-munica-tion cannot happen without presupposing consciousness and
vice versa, nevertheless both are closed, without any transfer of
information. Language, which Luhmann calls a "variation mechanism
of socio-cultural evolution", is the ultimate instrument for
coupling mental and social systems. This coupling seems to be the most
powerful driver of human evolution and learning.
Sociocultural evolution - application to design
Autopoietic systems show a high independence from internal and
external perturbations (negative feedback compensates for the
irritations). On the other hand it is one of the insights of chaos
theory, that circularity in simple mathematical models, can cause
so-called deterministic chaos. Minimal differences in initial
conditions of the system parameters can cause completely different
outcomes, so that predictability of final states is lost (positive
feedback amplifies perturbations and triggers evolutionary change).
Natural evolutionary patterns of development, with their sequence of
stable phases and sudden variations seem to be based on an interplay
of negative and positive feedback mechanisms.
The evolution of artefacts shows similar patterns. It is not really
new to describe artefacts as entities struggling for the survival of
the fittest in the hostile environment of the market (Hybs and Gero
1992). But the approach is still sub-complex. We (seem to) know where
we come from, but we do not know, where we are going. At least we know
the ancestors of our current artefacts, which means some
interpretation capacity for design history. Nevertheless we normally
do not know the influences that acted upon the bifurcation situations
and resulted in exactly this and no other development.
Also representations of design processes reveal these patterns (see
e.g. Roozenburg /
Eekels, 1991). The nicely
cut branches after the bifurcation points suggest, that there is a
rational means to overcome the indeterminacy, to take a decision,
which provides more than a random chance, that the decision is viable
in the future. Rittel (1971/72: 48, 54, translation W.J.) comments
this laconic:
"Constrictions are not 'natural conditions' but deliberate
restrictions of the variety of solutions, mostly implicit signs of
resignation. Š
... In reality there is no opposition / sharp conflict between an Š
intuitive approach to solve a problem and Š a controlled, reasonable
and rational approach. The more control one wants to exert, the more
well-founded one wants to judge, the more intuitive one has to be.
The endpoints in the more and more ramifying tree of causal
explanations are always spontaneous judgements."
These evident analogies in the processual patterns of natural and
artefact evolution suggest the application of evolutionary concepts to
the design of artefacts. (There are those generative approaches that
apply evolutionary algorithms to produce topologies and geometrical
forms. This path will be neglected here, even if one could argue, that
the approach might be able to evaluate the theory presented here.) If
we are aiming at new descriptions for the design process, we have to
identify the elements and processes of natural evolution, which can be
transferred to the evolution of artefacts. We should focus on the
problem of increasing the probability of success with respect to a
decision to be taken.
Luhmann's theories are closely related to evolutionary epistemology.
In his main oeuvre (1997) he has started to work out the concept of
social evolution. Evolution theory is based upon the system /
environment distinction; it is this difference, which enables
evolution. Evolution theory does not distinguish historical epochs,
but the circular sequence of variation, selection,
and re-stabilization. It serves for the unfolding of the
paradox of "the probability of the improbable".
Re-stabilization is essential, because it is the condition for
variation and selection being possible at all. Evolution theory thus
explains the emergence of essential forms and substances from the
accidental, relieving us of attributing the order of things to an
form-giving telos or origin. It simply turns the terminological
framework of world-description upside-down. Evolution theory is not a
theory of progress, and it does not deliver projections or
interpretations of the future. Autopoiesis, as outlined above,
enforces a revision of the concept of "adaptation".
Adaptation is a condition, not the goal or outcome of evolution: on
the basis of being adapted it is possible to produce more and more
risky ways of non-adaptation - as long as the continuation of
autopoiesis is guaranteed.
The three separated processual components of evolution can be related
to the components of society, conceived as a communicative / social
system:
- Variation varies the elements of the systems, i.e.
communications. Mainly variation means deviating, unexpected,
surprising communication. It may simply be questioning or rejecting
expectations of meaning. Variation produces raw material and enables
further communication with more open connections than before.
- Selection relates to the structures of the system,
i.e. the creation and use of expectations that control
communication. Positive selection means the choice of meaningful
relations that promise a value for building or stabilizing structures.
Selections serve as filters to control the diffusion of variations.
Religion has been such a filter. Truth, money, power, as symbolically
generalized media serve as filters in modern societies.
- Re-stabilization refers to the state of the evolving
system after a positive / negative selection. It has to take care
of the system-compatibility of the selection. Even negative
selections have to be re-stabilized, because they remain in the
system's memory. Today stability itself becomes a more and more
dynamic concept, indirectly serving as a trigger for variation.
The problems of control and prediction - causality gaps
The previous findings allow us to summarize as follows: Designing
consists of interacting and co-evolving autopoietic systems and
artefacts. Random mutations in nature plus deliberate decisions and
accidental events and connections in social life initiate open-ended
processes of self-organization, in which positive and negative
feedback interact and produce changing patterns that may at some point
assume relatively stable forms, called fashions or trends. This kind
of mutual causality implies, that it is not possible to exert
unilateral control over any set of variables; interventions are likely
to reverberate throughout the whole. Though it is often possible to
spot an initial "kick" that sets a system moving in a
particular direction, it is important to realize, that to our
understanding such kicks are not the cause of the end result. They
merely trigger transformations embedded in the logic of the systems
involved.
We can identify two problem areas: (1) control, due to the
system / environment distinction, and (2) prediction, due to
the variation / selection / re-stabilization distinction.
(1) The problem of control:
Luhmann' systems theory provides a map of the possible gaps
related to these interventions, called design activities. We have the
following combinations to denote the locations, where the gaps are
manifest:
- artefacts / organisms
- artefacts / consciousnesses
- artefacts / communications
- artefacts / organisms / communications
- artefacts / consciousnesses / communications
- artefacts / organisms / consciousnesses, and
- artefacts /
organisms / consciousnesses / communications.
Artefacts as artefacts are
assumed to function; this is not the primary task of designing. With
respect to the autopoietic systems, I introduce the following gaps,
which are always occurring in interaction with different shares,
according to the specific design task:
- organisms
ý the function gap,
which indicates, that it is not a trivial (Š) task to adapt an
artefact to an organism, for example, because bodies cannot
speakŠ
- consciousnesses
ý the taste gap,
which indicates, that it is not a trivial (Š) task, to coordinate
individual consciousnesses, for example to optimise a solution for the
80 million consumers of the German market. They are all different, and
they cannot speak about their taste in clear and distinct
mannerŠ
- communications
ý the fashion gap,
which indicates, that it is not a trivial (Š) task to generalize a
variety of information gathered from individual consciousnesses and to
transfer this into the shape of an artefact, for example to plan a new
collection of household goods for the Turkish marketŠ
(2) The problem of prediction:
- Variation is aiming at alternatives. This is no problem, since
consciousnesses provide abundant "creativity", which is
essential for increasing the variety of choice. This is the
"timeless" task of designing
artefactsŠ
- Selection is aiming
at the fit of alternatives into structures. This is a problem indeed,
because communicative structures are detectable, but not their future
stability. To a certain degree, at least, design research can
examine existing structuresŠ
Single aspects can be tackled by isolated approaches: organism -
artefact gaps by means of ergonomics, consciousness - artefact gaps by
means of cognitive ergonomics, communication - artefact gaps by
means of market research, etc.
- Re-stabilization is aiming at the integration of selected
alternatives into the system. There is hardly any predictability,
because this is a question of long-term viability of selected
alternatives within communicative systems. Futures studies
and scenario planning are dealing with evolving
systemsŠ
Returning to design: The present does not at all mark the "wave
front" of progress, but merely consists of what has remained from
the past. And so it happens, that we do not live in the best of all
possible worlds. Harmony, if at all, is "post-stabilized"
harmony, created in our narratives. The study of failed innovations
("floppology") might be a promising approach to improve
designing: the "dark side" of the field is probably much
richer than the "best practice" view. Design activities
happen "in-between", they intervene into the relations of
co-evolving autopoietic systems by means of creating artefacts that
pretend to improve those relations. The basic problem is neither
lacking individual creativity nor insufficient planning, but the
uncontrollable and unpredictable nature of communication in
the environment of the artefacts. The most developed instrument for
bridging this kind of causality gaps between psychic systems is
language, which enables communication. Functioning communication
is highly improbable. Functioning design is even more
improbableŠ
To sum up: there are two basic problems related to systemic gaps:
(1) The gaps between autopoietic systems involved in designing.
This is fundamental systemic "obstinacy", which is labelled
or covered with the nice and common, but fuzzy terms
"creativity", "subjectivity", "values",
"trends", Š
(2) The gaps between the evolutionary mechanisms involved in
designing. Or: the future orientation of design activities. The
artefact, once released, remains as it is. The environments of the
artefact change in manners, which are in principle
unpredictable.
Conclusion
At this point I have reached
the limits of my argument. Even a perfect language could only bridge
one single gap: the interface between a thought, which is an element
of a consciousness, and the communicative offer produced by this
psychic system. And this kind of ideal language would have to be a
private language, which would probably fail with the addressee. A
functioning language has to be a deficient compromise, a
medium. And design is a medium as well, but a considerably
less universal one compared to language. Language is deficient, but
nevertheless optimal. Language is the problem and the
solution.
Finally, following Horst Rittel, I want to ask for some more modesty
and irony, and less arrogance in dismissing approaches to design.
References
Hybs, Ivan; Gero, John S. (1992) "An evolutionary process
model of design", in: Design Studies Vol 13 No 3 July 1992
pp 273-290
Jonas, Wolfgang "Mind the gap! - on knowing and
not-knowing in design", Proceedings of EAD 2003,
Barcelona
Jones, John Christopher (1970), 1992) Design Methods. Seeds
of human futures, London,
John Wiley & Sons, second
edition, Van Nostrand Reinhold, New York 1992
Luhmann, Niklas (1984), Soziale Systeme, Frankfurt /
Main, Suhrkamp
Luhmann, Niklas (1997), Die Gesellschaft der
Gesellschaft, Frankfurt / Main, Suhrkamp
Maturana, Humberto R. (1985) Erkennen: Die Organisation und
Verkörperung von Wirklichkeit, Braunschweig, Vieweg
Miller, George A.; Galanter, E.; Pribram, K. (1960) Plans
and Structure of Behaviour, Harvard Center for Cognitive Studies,
New York
Morgan, Gareth (1986) Images of Organization, Newbury
Park, London, New Delhi, Sage Publications
Rittel, Horst W. J. (1971/72) "Zur Planungskrise:
Systemanalyse der 'ersten und zweiten Generation'", in: Ders.
Planen, Entwerfen, Design 1992 S. 37-58 (Original
1971/72)
Rittel, Horst W. J.
(1972) "Second-generation Design Methods", in: Cross, Nigel
(ed.) Developments in Design Methodology, John Wiley,
Chichester 1984 pp 317-327 (Original 1972)
N.F.M. Roozenburg, J. Eekels, (1991) Product Design:
Fundamentals and Methods, Chichester, Wiley
Weaver, Warren (1948) "Science and Complexity" in:
American Scientist 36 (1948) pp 536-544
Wiener, Norbert (1948) Cybernetics or control and
communication in the animal and the machine, Cambridge, MIT
Press
--