Many people have been asking me how to use systems thinking, how to make models, what are the steps that I follow, how does the development process unfold? To be honest, there is not set template that one can follow and reach the end. As John Sterman says, “It is inherently a creative process.” But there are steps which can guide the model building process. Sterman in his book, Business Dynamics, mentions these steps in detail and I recommend everyone interested to go through them. I myself has used them to write proposals on research using systems thinking and modeling. But over time I realized that applying them required some degree of adaptation of these steps. Since in India the degree of complexity is relatively more as compared to other developed countries where data is available and reliable under most circumstances, we need to be very creative while trying to understand complex systems using systems thinking. To give an example, while working in rural areas one realizes how poor is the data availability even for critical resources such as ground water. I remember once on a water wind mill assessment trip we were struggling to identify water hand pumps which are dry and which are still pumping water, what is their depth, when they were installed etc. People in villages had some vague idea about their date or year of installation, depth etc. But all that was not good enough for us to take a call on whether to implement a water windmill or not. The only parameters for which data was available and reliable (to an extent) was the meteorological conditions i.e. the wind speed, direction, precipitation etc. That too because WOTR had Automated Weather Stations installed there. So this goes to show that while working in India data availability and reliability is a big constraint. The second constraint is the information and knowledge available to understand the local dynamics. Another example being, people in villages find it very difficult to estimate their household numbers be it water, finance, energy or livestock productivity. For eg. their income and expenses figures rarely tally. Giving them the benefit of doubt that they take short term loans, receive some money through family, friends etc. even then all these numbers would often not add up. I have had a first hand experience on this issue too. So modeling social, ecological and economic systems in informal setting in India is a hard, hard work. Lot of assumptions and (educated) guesstimates have to be done to complete any sort of modeling process.
All this relates to how one then applies systems thinking and modeling in dynamically complex situations and systems in India? Does John’s modeling process and steps help our cause? In my opinion they do, albeit to a limited extent. One needs to creatively adapt them and use his/her sensing to move ahead. In my experience serendipity also plays an important role. Almost every time I have been badly stuck in some of the most challenging modeling projects and then I have found a way out either by someone willfully joining the work or through some support. All of it being not part of original plan.
So after doing some modeling work in very un-ordered settings I have adapted John’s modeling steps and have penned them down. The pre requirements before these steps are that the problem has been identified, modelers have familiarized themselves with it and proper scoping exercise has been done with the end user and experts. The modeling steps then are:
- Define purpose of the model and read about the issue or system of interest
- List down sectors, sub systems and determine model boundary
- Create Dynamic Hypothesis (DH)
- Develop questions based on DH
- Show DH to community (end user) and experts
- Review the questions and DH after step 5
- Collect information and data using questions developed in step 6
- Develop initial versions of simulation model
- Review DH based on simulation model
- Show revised DH to community (end users) and experts
- Show initial simulation results to community (end users) and experts
- (Incorporate suggestions) Revise the model, its boundaries and develop questions to fill in information and data gaps
- Collect data and information to fill the gaps
- Develop progressive versions of model (eg. v2, 3…… v17…. and repeat step 13
- Perform model calibration, sensitivity runs, extreme conditions test and finalize the model (sometimes back-casting also helps in model evaluation)
- Share the results with community for mobilization and aiding action
- Make model readable and fit for publication and dissemination
- Get the model reviewed by an expert and incorporate suggestions
- Publish the model and results
- Plan for next phase and how to go deeper into the issue to initiate change process, then start from Step 12 or any other appropriate step as suited
The most important thing to remember is to celebrate learning at each step and document the learning process. You would often be surprised that the learning process is an equally important and influential outcome of the modeling process as much as the end simulation results are.