1.5. Using Dynamic Modeling to Generate Consensus
13
the past. The set of experiences and the patterns they form provide the
basis for generalizations that then influence our decisions. The interpolation and extrapolation among patterns helps form mental models that are
often inadequate to provide a comprehensive perspective on the many interrelated aspects of systems and to anticipate their behavior-especially
when we encounter novel situations. There is often no strong logical base
for mental models . This is why we need to develop formal models to complement our thought processes, and why we need to reflect upon the workings and outcomes of formal models to sharpen our thinking .
In contrast to mental models , the scientific models of marine systems are
usually highly formalized-using controlled experiments to shed light on
individual issues, and mathematics and statistics to express the relevant relations. However, the precision of scientific studies comes at a price: most
environmental conditions are held constant in order to carefully study a
small number of variables, and the temporal and spatial scales over which
investigations are made is limited by budgetary constraints . As a consequence, scientific models may be of limited use to fishermen and conservation professionals because the temporal and spatial scales they deal with
are far larger than those typically covered in scientific studies , and most environmental conditions are not constant over those scales.
This does not mean that scientific studies are useless for marine conservation . On the contrary, they must provide starting points and building
blocks of any serious effort that seeks to generate insights into marine conservation issues. Computer models can help explore uncertainties associated with stretching scientific studies beyond the spatial and temporal
scales to which they apply . We shall return to this issue in Section 2.4. By
using computer models to combine scientific insight of one discipline with
the insights from another, or with less formal knowledge-such as the
knowledge held by stakeholders-we can begin to formally explore uncertainties, feedback and lags to a previously unknown extent.
Whether and how much to base a computer model on scientific studies
and field or lab data depends on the purpose of the model. If the purpose
is predictive modeling, striving for a high level of accuracy is important.
In contrast, if the purpose is descriptive modeling it may be sufficient
to initialize a model with hypothetical data and play out the dynamics
under the assumptions embedded in the model. In this case, the modeler
would concentrate on the trends and perhaps the relative orders of magnitude of changes in system components, rather than on their precise numeric
values .
The boundary between predictive versus descriptive modeling is rather
fluid. Since models are abstractions of reality, they inevitably contain simplifying assumptions, and since parameter values and initial conditions to specify the model are typically not known with 100%certainty, model results will
not exactly match observations. Prediction is only possible up to some
point. The models of this book span this range of descriptive to predictive,
13
the past. The set of experiences and the patterns they form provide the
basis for generalizations that then influence our decisions. The interpolation and extrapolation among patterns helps form mental models that are
often inadequate to provide a comprehensive perspective on the many interrelated aspects of systems and to anticipate their behavior-especially
when we encounter novel situations. There is often no strong logical base
for mental models . This is why we need to develop formal models to complement our thought processes, and why we need to reflect upon the workings and outcomes of formal models to sharpen our thinking .
In contrast to mental models , the scientific models of marine systems are
usually highly formalized-using controlled experiments to shed light on
individual issues, and mathematics and statistics to express the relevant relations. However, the precision of scientific studies comes at a price: most
environmental conditions are held constant in order to carefully study a
small number of variables, and the temporal and spatial scales over which
investigations are made is limited by budgetary constraints . As a consequence, scientific models may be of limited use to fishermen and conservation professionals because the temporal and spatial scales they deal with
are far larger than those typically covered in scientific studies , and most environmental conditions are not constant over those scales.
This does not mean that scientific studies are useless for marine conservation . On the contrary, they must provide starting points and building
blocks of any serious effort that seeks to generate insights into marine conservation issues. Computer models can help explore uncertainties associated with stretching scientific studies beyond the spatial and temporal
scales to which they apply . We shall return to this issue in Section 2.4. By
using computer models to combine scientific insight of one discipline with
the insights from another, or with less formal knowledge-such as the
knowledge held by stakeholders-we can begin to formally explore uncertainties, feedback and lags to a previously unknown extent.
Whether and how much to base a computer model on scientific studies
and field or lab data depends on the purpose of the model. If the purpose
is predictive modeling, striving for a high level of accuracy is important.
In contrast, if the purpose is descriptive modeling it may be sufficient
to initialize a model with hypothetical data and play out the dynamics
under the assumptions embedded in the model. In this case, the modeler
would concentrate on the trends and perhaps the relative orders of magnitude of changes in system components, rather than on their precise numeric
values .
The boundary between predictive versus descriptive modeling is rather
fluid. Since models are abstractions of reality, they inevitably contain simplifying assumptions, and since parameter values and initial conditions to specify the model are typically not known with 100%certainty, model results will
not exactly match observations. Prediction is only possible up to some
point. The models of this book span this range of descriptive to predictive,