Systems Change, Concept Maps and Computer Models

From SystemsWiki

Jump to: navigation, search

System is derived from the word for composition and can be thought of as a collection of interactions among component parts. There is usually a reason or purpose for the system. Sometimes this purpose is just making sense of the world. We usually try to restrict our systems with boundaries, called closed systems, so we can limit the number of things we need to think about. Systems are usually maintained by flows of energy and materials between the system and its surrounding environment. Parts of a system can be a mixture of objects in the real world, ideas in our mind or constructs in the social world. We have spent most of our lives trying to make sense of several overlapping systems, health systems, human systems, engineering systems and information systems.

Systems are described by what they are and what they do, their structure (properties) and behaviour (methods) due to the interactions among their component parts. Sometimes a part of a system can also be a system. Therefore a system has parts, and can also be part of a system. This recursive nature of systems leads to a hierarchy of levels, which will be discussed further in a later section. For now, we are interested in how the structure of a system determines its behaviour over time. There has been much discussion about how systems change over the centuries. For instance organizations have a lifecycle of development and change, which can be described as a pattern of events. We can think of the organization as a living body which develops, grows and ages and decays over time, or as a social construction which seeks goals and consensus among members. Or its behaviour can be explained by describing the interaction among its members that cause it to evolve over time or the conflict and synthesis among the ideas and rules of engagement that drive it.

Contents

Models of Systems as Theory and Data

Based on our experience we build mental models or simplified ways to describe and explain what we see and to anticipate the outcomes of what we do. These mental models are a mixture of observed data and building of theories to explain the past in order to guide our future actions. Data belongs to the past, while action occurs in the future.

Models can be considered as a mix of theory from the conceptual domain and data from the empirical domain.

We can share representations of our mental models by telling stories, drawing pictures or maps or making scale 3D models. We can represent how things change over time by using metaphors and successive snapshots or storyboards, or use mathematical or computational models to describe behaviour over time. In this collection we will use qualitative concept maps and quantitative system dynamics computer models.

In general we adopt the viewpoint that zooming out on a problem may be more helpful than getting lost in the minute detail of structures and detailed processes at the immediate “scene of the crime”, namely the space and time where the symptoms of the problem show themselves. We also try to explain the world in terms of the general pattern of events rather than the details of individual events.

Concept Maps

Over the past four decades concept and mind maps have been increasingly used to organize the knowledge we use to answer focus questions about the world. There are no separate systems. The world is a continuum. Where to draw a boundary around a system depends on the questions we want to ask. Concepts are perceived patterns or regularities based on past experience that we use to describe objects or explain events. These concepts can represent the external physical world, our mental world of ideas or our social world of relationships. Concepts are linked to form propositions that describe structures often in terms of parts and relationships. They can also explain observed patterns of events or change in terms of deeper understanding of how things work or why things change.

A weaker form of explanation is describing the probable association of related events. The Babylonians described the heavens using careful observations that could predict celestial events. In contrast, the Greeks developed theories of causation to explain the movement of the heavenly bodies. This reflects a deeper level of understanding based on patterns and causal explanations rather than associations. Levels of explanation or abstraction can always be improved, moving to a succession of more detailed answers to repeating the question, “But why?” Mind maps are a simple tree of linked concepts where the links are usually unlabeled. They highlight hierarchies rather than networks of concepts and linking words.

Why model?

Our mental models are the way we package past experience and apply it to future actions. These are personal and depend on our unique past. We can also learn from the experiences of others through stories and writings. (see Korzybski’s time-binding and general semantics). Formal representations in maps make it easier to share and improve mental models. This becomes important when we are faced with surprises that can’t be explained well by our current mental models. A related problem, common in healthcare, is disagreement among groups about the causes and solutions to chronic persistent problems. Methods and tools to surface our mental models and locate the differences in our perceptions, beliefs and values offer a way to improve our social, technical and personal systems of health and healing. We could consider models as arguments, used to persuade people to take common effective action to improve or sustain the future. Climate change modelling offers a cogent example of this process and difficulties in modelling to produce effective actions of groups with differing beliefs values and incentives.

To summarise, we model to avoid making big mistakes, so our actions are more likely to be effective.

  • Don’t solve the wrong problem
  • Don’t apply the wrong solution
  • Don’t cause worse problems
  • Avoid unintended consequences
  • Provide a safe place for experiment & discussion

Why computational models?

Computer models take our conceptual map structures and convert them into behaviour. We move from static snapshots to videos of changes over time. This provides a logical consistent framework for translating structure into behaviour. Rather than playing out mental simulations in our heads, we use the computer as a prosthesis to be able to share our foresight about the consequences of future actions.

These logical constructs can be built up piece by piece from simple propositions or hypotheses to a more complex theory of multiple propositions. We can then run sets of virtual experiments which play out the results of future actions based on our theories of causation and our perception of the current situation and context.

Rather than just predicting a train wreck, our models help us to decide how to intervene to avoid or mitigate the train wreck. This is the difference between scientific predictive models and models for designing and performing interventions that shape the future.

What is a good model?

A good model is one that fits the purpose it was designed for. Common uses of models are

  • To show insights,
  • To build theory,
  • To predict the consequences of action or inaction,
  • To work out what data to collect in information studies,
  • To analyse data (including parameter estimation)

A good model has the right breadth (Scope or boundary) and depth (Level of detail) for its intended purpose. Sterman describes common features of good models of persistent problems that produce effective action as:

  • Structures which include all relevant causes within the model
  • Dynamic models which explain change over time
  • Include human behavioural responses
  • Show important differences (acknowledging heterogeneity)
  • Are grounded in empirical testing in many settings, including learning laboratories and the real world.
  • Combine both quantitative and qualitative knowledge and concepts
  • Have a broad boundary, a “zoom out” view
  • Integrate perspectives from multiple disciplines

Modellers keep asking “Why don’t you use our models?” Policy makers respond, “Why don’t you make models we can use?” – Dana Meadows, The Electronic Oracle, Computer Models and Social Decisions.

“So little is understood about social policy, so few of the necessary facts about any policy process can be assembled, so many information sources are subjective and self-serving, that no perturbation of the social system, whether computer-generated or not, can be fully traced and evaluated. It is because social systems are so bewilderingly complex that mathematical models are turned to in the first place.” -Dana Meadows and Jennifer Robinson The Electronic Oracle,1985

Questions & Comments to Geoff McDonnell
Personal tools