Measure twice, cut once applies to more than
carpentry. Creating a scientific model that simulates
a variety of possible scenarios is an excellent way to test
a business or policy decision before it’s set in place.
The problem is that our mental models are simplistic:
■ ■ there are important causal variables that are neglected.
The use of automobiles, for instance, depends on factors such
as the price of gasoline, availability of public transportation,
and population density.
■ ■ the interaction of variables is ignored. New roads increase the attractiveness of driving and decrease the appeal
of public transportation. As trains and buses carry fewer and
fewer passengers, authorities might decide to reduce the
frequency of operations or cut routes. This phenomenon, in
turn, reinforces the attractiveness of driving even as commuting times increase.
■ ■ time and space delays are ignored. New highways encourage housing construction in the suburbs. Although it will take
years before people start commuting from their new homes,
they eventually will, thus contributing to traffic congestion.
History shows that there is no shortage of well-intended
initiatives—in both the public and private sectors—that have
produced suboptimal outcomes or that have worsened the
problem. Poorly designed policies are often the result of not
understanding the situation at hand or not appreciating the effect of actions taken to solve it—or both. Decisions in business
and policymaking clearly can be improved by using models
that capture important aspects of reality and that are capable
of simulating scenarios.
Enter system dynamics, a field of study dedicated to model-
ing the structure and evolution of complex systems. Although
the models can be intricate, their conceptual building blocks
are few and, in their basic form, relatively easy to understand.
The ability to create useful models depends not only on criti-
cal thinking but also on the capacity to identify key elements of
reality—a capacity that stems from multidisciplinary thinking
rather than narrowly focused knowledge.
For the purposes of explanation, I’ll assume that the problem
has been defined, that all relevant key variables have been selected, and that the time horizon has been determined. What,
then, are the building blocks of system dynamics?
Causal diagrams show how variables are related from cause to
effect, including feedback loops (see Figure 1, Page 38). Causality is indicated with arrows and polarities with a positive (+) or
a negative (–) sign. A positive polarity means that an increase
(decrease) in the cause is followed by an increase (decrease) in
the effect above what it would have been otherwise. A negative
polarity similarly means that an increase (decrease) in the cause
DEFINING SYSTEM DYNAMICS
DEvELoPED IN ThE 1950S b Y JAY FoRRESTER, a professor at
the Massachusetts Institute of Technology, system dynamics is
a field of study with a rich history of successful applications. The
tools necessary to model complex situations are surprisingly few
and conceptually easy to understand. While the mathematical
formulation can be intricate, it lends itself naturally to numerical
implementation. System dynamics, through computer simulation,
has been used to create “microworlds” or “flight simulators”
that let the user test managerial or policy decisions without
annihilating a department or crashing the business.