"Simulation is impressive, but do I really need to use it?" That's a common concern as people struggle with the time constraints of their job and the challenge of creating and simulating models. Yet Dr. John Sterman states flatly in his 2002 Jay Wright Forrester Award Lecture, "Simulation is essential for effective Systems Thinking."

What are we to do if time is scarce, money is tight, and the people most involved don't have the software or the skills to do a simulation?

There are options.

First, no one simulates on a computer before every decision. We rely on our mental models to assess how the situation we face may resemble other situations we think we understand, and we make the best decision we can.

Unfortunately, our intuition may be flawed, so we need to learn from the best sources we can. Good management flight simulators and computer simulation models based on systems thinking concepts can help, but only if we use them well. I find it far too easy to treat management flight simulators as video games, poking a slider here, twisting a knob there, pressing run, and watching the pretty lines evolve, and that's not the way to learn.

If you're going to learn from a model, try the classic action research approach: repeated cycles of action, critical review, planning, and action. Start where you want in that cycle, although I find it most helpful to start with planning. Pick a goal and make a plan on how you'll attain that goal. Write it down! (That may be the single most important part, for, without committing it to writing, it's all too easy to say the result was obvious, even if it was diametrically opposed to your initial intuition.) Draw a graph of the expected behavior. Set up the model parameters according to your plan, and run the simulation.

Then write down what you got, and compare your plans with the model results. What matched? What was different? What can you learn from this? Based on what you've learned so far, what revisions would you make to your plan? Then repeat the process.

While no one simulates before every decision, as Dr. Sterman noted, effective systems thinkers simulate before more decisions than the average person does.

Second, you don't have to use a computer simulation program to do a simulation. Some people use manual methods. The MIT Beer Game, which you may have played, is a popular supply chain board game. If you're facing a particular issue you'd like to understand through simulation but don't own a simulator, you might consider creating a board game to step through the model over time. Be careful to define policies (rules for decision making) carefully so that there's little room for creativity during the simulation. Use the same action research approach for learning from the board game. Be sure to include the feedback loops and stocks and flows you'd incorporate in a computer simulation.

There can be benefits to this approach. For one, you've likely involved a larger number of people in this work. That both gets you more insights to apply to the problem and gives you a broader group of people who understand the eventual conclusions.

There are also disadvantages. As you'll soon discover, it's not easy to create a board game that is both representative of the issue you're creating and reasonable to play. It takes time to run through even a simple manual simulation, and that time will likely reduce the number of different strategies you try. It will also likely limit the testing you do of the model, and that increases the risk you'll be making decisions based on poor modeling. After you've done manual simulations a time or two, you might well decide that it is indeed better to use a computer.

Third, perhaps you decide you do need a computer model, but you don't own a simulator. There are multiple ways to create a system dynamics computer model. The basic ideas underlying system dynamics implementations aren't that challenging, if you can program. My first models were done in Logo on a Commodore 64 computer in relatively few lines of code, and I've coded other models in (at least) C and Pascal. That said, it's awfully nice to have all of the built-in support a professional simulator offers for accepting user input, tying the model diagram to the model equations, producing easy-to-use graphs and tables, and automating such processes as sensitivity testing.

Fourth, if you aren't yet an accomplished modeler, perhaps you'll benefit from some help. You can certainly find a modeler who will be glad to work with you. Besides figuring out who to bring in and working to explain the issue to that person, I encourage you to learn about modeling yourself so you can engage in true collaboration. Take a class, read a book, or explore resources on the Web. Whatever you do, engage in experimentation to exercise your growing insights.

Whichever approach you take -- learning from management flight simulators, doing manual simulations, rolling your own software, or finding someone to help you, use simulation as a way to test and refine your mental models and your ability to solve the tough challenges you face at work or elsewhere.

------------------------
Bill Harris is principal and founder of Facilitated Systems.

Bill Harris
Facilitated Systems
Everett, WA 98208
USA
+1 425-337-5541
bill_harris@facilitatedsystems.com
http://facilitatedsystems.com/