Information Technology Reference
In-Depth Information
making it table driven. System dynamics, on the other hand, would use
a model that would show that the level would have to be adjusted based
on dynamic factors such as the speed and the type of the car one is
driving (sound insulation), the traffic on the freeway (external noise), or
the condition of the road (e.g., asphalt versus cement surface).
If one is attempting to understand the basic structure of a working
system and its associated behavior in its entirety, system dynamics is a
good approach. In fact, Forrester postulates that most real-world systems
are nonlinear with multi-loop feedbacks, which is why models based on
system dynamics can be better representations of real systems.
You can use the concepts of system dynamics in building software in
the following ways:
1.
Identify the customer problem or situation that needs to be
changed.
2.
Ascertain the behaviors of interest in the problem area and the
associated information base (policies, structure, regulations, rules,
users).
3.
Develop a hypothesis based on the dynamic conditions of the
environment that explains the situation.
4.
Build a computer model (prototype) that best reproduces the
behavior seen in the real world.
5.
Create alternative behaviors (information base changes), and use
the model to see how it affects the situation.
6.
Propose the solution that best matches the desired situation.
These steps probably do not look very different from the software life-
cycle process. Closer examination draws attention to the following:
System dynamics teaches one to use feedback loops to adjust one's
progress for each of these steps. A model that “best” reproduces
real-world behavior can only be determined after multiple iterations
of coding and testing.
One continuously needs feedback across steps. For example, while
building Step 3, one needs to keep an eye on Step 2; for Step 4,
one must constantly reevaluate Steps 2 and 3.
System dynamics offers a few unintuitive suggestions:
When a solution does not meet the requirements, it is advisable
to invest in reviewing the hypothesis on which the solution was
built, rather than retaining the existing hypothesis and jumping in
to tune the solution to match it. This is an important point.
