To begin tackling a complex problem, we need to be able to represent it. An everyday example of a complex problem is traffic congestion. (At this point let’s not worry about an exact definition of complex.) The simple answer to congestion is to build more roads.[1] If we represent this thinking using a diagram it might be:

How do we read this diagram? Without being too rigorous:

  • “If there’s more congestion, more roads are built than would be the case if congestion hadn’t increased.” The “+” on the arrow says that the “change” is in the same direction.
  • The second part may not be stated, but the thinking is“If more roads are built then congestion will be less than if roads had not been built.” The “-“ on the arrow says that the “change” is in the opposite direction.

When a variable has an impact that connects “back” to a variable that is “earlier” in the logic, the connection is called feedback, and the structure is called a causal loop. (“Earlier” only makes sense for a new venture or action where there is a meaningful starting point. In many cases there are continually interacting players in established relationships.) A benefit of modeling is identifying unknown/influential feedback structures.

The “B” in the centre of the image indicates a balancing loop. If we have an increase in one variable the other will tend to decrease it. More congestion leads to roads, which lead to less congestion. It needs to work in all possibilities: if for some reason I have less roads then I should get more congestion which should lead to more roads. In this case it’s hard to see how less congestion would directly lead to less roads, so we can see there is need for some refinement.

This is called a causal loop diagram (CLD). As a representation and method there are lots of extensions and limitations in general, which we will visit as we need to.

A Better Understanding

But now let’s specifically consider the diagram and the thinking behind it. Is it correct? Under what conditions? What’s missing? A couple of steps toward improving the diagram are shown below.

We can walk through the diagram starting at the top right:

  • There is a pressure to reduce congestion if the actual travel time is more than the desired travel time. (There are other possible logical structures – it might be the percentage of time spent in free-flowing vs stop-and-go traffic – and knowing how it is perceived by people and how it drives decision-making is important in coming up with useful models for real improvements.)
  • As that pressure increases, after a delay not shown, road construction will begin
  • As road construction is completed, after a delay not shown, road capacity increases.
  • Road construction could be negative, and would then be road destruction, suggesting an opportunity to improve the naming.
  • As road capacity increases, travel time decreases.
  • As travel time decreases, attractiveness of driving increases.
  • As attractiveness of driving increases, traffic volumes increase through many mechanisms not shown.
  • As volumes increase, travel times decrease and the cycle repeats.

Note that this is a one way trend since road destruction is rare. Road construction leads to road capacity leads to decreased travel time which leads to attractiveness of driving which leads to traffic volume.


  1. Traffic is a complex problem that is embedded in a large system. There are many other aspects that are connected: alternative transportation; location of housing, shopping, schools, work; commercial and economic impacts, government capital and long-term operating budgets; energy and resource consumption, waste, and pollution; environmental impacts such as micro-climates, flooding, habitat segmentation and loss; and on and on. For a given situation, it is important to define the scope of the model to appropriately meet the needs.
  2. A causal loop diagram offers a way to start to represent the connections between different elements in a system. We have not examined all their possibilities: they have many limitations, and as they become larger and more complex the relationships between the elements become very difficult to understand. We have introduced it because it is has value as an introduction to this type of thinking, and it is commonly in use[2].
  3. It may be easy after developing a diagram to view the results as obvious. There’s a strong human tendency to say “I knew that!” But I didn’t. Maybe I knew bits and pieces, or the specific causation in this particular case, but unless I’ve had specific training and have gone through the process, I did not have a precise and clear understanding of how things are related, a method for thinking through any kind of dynamic problem, and a notation for representing it.
  4. This process is part of what may be called modeling for insight.[3]


  1. This discussion is based on a more extended model in Sterman, John D., Business Dynamics: Systems Thinking and Modeling for a Complex World, pp 177-190, McGraw-Hill, Boston, 2000.
  2. A type of causal loop diagram was popularized by Peter Senge in The Fifth Discipline: The Art and Practice of the Learning Organization and The Fifth Discipline Fieldbook: Strategies and Tools for Building a Learning Organization.
  3. Lyneis, James M. System dynamics for business strategy: a phased approach. System Dynamics Review. Vol. 15, No. 1. 1999.

An Introduction to Representing Complex Problems

Chapter 1 Simple Solutions

Copyright 2017 Thinkitation Inc. – For Educational Use