A good model will indicate the normal fluctuations in a complex system. Such
variations are sometimes the cause of great alarm and much unneeded change.
Conversely, observation of variation that is unexpected from real world experience could signal the need for action. A good model should be able to indicate
the results of these corrective actions.
• Third, a good model can serve as a thought-organizing device. Sometimes, most
of the value in modeling comes from a deeper understanding of the variables
involved in a system that people are routinely struggling to control. Modeling
requires that you assemble the group of experts on the various parts of the system
to be understood. Each group member gains a better understanding of the system
and of the skills and knowledge of their colleagues. Good modeling stimulates
further questions about the system behavior and in the long run, the applicability
to other systems of any newly discovered principles.
• Fourth, a good model is a growing storage device for data and ideas that the
human enterprise has struggled long and hard to find and learn. Most often such
data and insights are left to gather dust in printed form or reside quietly in some
distant computer. An ever-developing model should capture the knowledge that
we have gained, and document those lessons learned through the years about
how the system actually works.
1.8 Model Confirmation
When do you know that you developed a “good” model? Giving an answer to this
question often is rather difficult. By definition, all models abstract away from some
aspects of reality that the modeler perceives less relevant than others. As a result,
the model is a product of the modeler’s perceptions. Consequently, one model is
likely not the same as the models developed for the same system by other modelers
who have their own, individual perceptions. Plurality of, and competition among,
models is therefore required to improve our collective understanding of real-world
processes. The more open and flexible the modeling approach, and the more people
are engaged in the specific modeling activity, the greater the chance of important
discovery.
By enclosing a selected number of system components in the model, and
determining the model-system’s behavior over time solely in response to the forces
inside the model, the model becomes closed. Real systems, in contrast, are not
closed but open, allowing for new, even unprecedented development in response to
highly infrequent but dramatic changes in their environment. It is therefore not
possible to completely “verify” a model by comparing model results to the behavior
of the real system. There may have been extenuating circumstances that led the real
system to behave differently from the model. The model itself may not have been
incorrect, but just incomplete with regard to those circumstances. Such circumstances will always be present, precisely due to the fact that it is the modeler’s goal
to capture only the “essentials” of the real system and abstract away from other factors.
24
1 Modeling Dynamic Biological Systems
variations are sometimes the cause of great alarm and much unneeded change.
Conversely, observation of variation that is unexpected from real world experience could signal the need for action. A good model should be able to indicate
the results of these corrective actions.
• Third, a good model can serve as a thought-organizing device. Sometimes, most
of the value in modeling comes from a deeper understanding of the variables
involved in a system that people are routinely struggling to control. Modeling
requires that you assemble the group of experts on the various parts of the system
to be understood. Each group member gains a better understanding of the system
and of the skills and knowledge of their colleagues. Good modeling stimulates
further questions about the system behavior and in the long run, the applicability
to other systems of any newly discovered principles.
• Fourth, a good model is a growing storage device for data and ideas that the
human enterprise has struggled long and hard to find and learn. Most often such
data and insights are left to gather dust in printed form or reside quietly in some
distant computer. An ever-developing model should capture the knowledge that
we have gained, and document those lessons learned through the years about
how the system actually works.
1.8 Model Confirmation
When do you know that you developed a “good” model? Giving an answer to this
question often is rather difficult. By definition, all models abstract away from some
aspects of reality that the modeler perceives less relevant than others. As a result,
the model is a product of the modeler’s perceptions. Consequently, one model is
likely not the same as the models developed for the same system by other modelers
who have their own, individual perceptions. Plurality of, and competition among,
models is therefore required to improve our collective understanding of real-world
processes. The more open and flexible the modeling approach, and the more people
are engaged in the specific modeling activity, the greater the chance of important
discovery.
By enclosing a selected number of system components in the model, and
determining the model-system’s behavior over time solely in response to the forces
inside the model, the model becomes closed. Real systems, in contrast, are not
closed but open, allowing for new, even unprecedented development in response to
highly infrequent but dramatic changes in their environment. It is therefore not
possible to completely “verify” a model by comparing model results to the behavior
of the real system. There may have been extenuating circumstances that led the real
system to behave differently from the model. The model itself may not have been
incorrect, but just incomplete with regard to those circumstances. Such circumstances will always be present, precisely due to the fact that it is the modeler’s goal
to capture only the “essentials” of the real system and abstract away from other factors.
24
1 Modeling Dynamic Biological Systems