Subject: Systems Language - closed loops

Posted by Warren Farr on 2/7/2010
In Reply To:Systems Language - closed loops Posted by Niall Palfreyman on 2/7/2010

Message:

>I am looking for a language, and,
>yes, also many examples, which open people's eyes to the possibility of
>seeing their neighbour and their environment not as a 'thing', but as a
>process which is always moving, always changing, and with whom one can
>consequently always make a fresh start at any moment of any day.

I like this vision of SD as a story-telling language. It provides me motivation to learn it, regardless of how difficult.

Perhaps one of our challenges (and strengths) is that we have more than one tool to communicate basic SD concepts such as: delay, momentum, endogenous feedback, etc... Causal loop diagrams are great for brainstorming the elements involved in a system. Implying causality with arrows is especially useful when encouraging students to close loops. Because units (equations) and the specification of stock or flow are not explicit with CLDs, the added flexibility allows inclusion of many elements. I like to use CLDs when I want people to really think about the "process which is always moving" and influencing itself.

Stock/Flow diagrams, on the other hand, demand more rigor. Stocks can only be influenced by their flow values which must have the same units as the stock per unit time. If both males and females need to be represented, then two stocks are required. As a result we end up with "structures" such as co-flows and aging chains. While the language of stock/flow is more demanding, it also has the advantage of being interactive once a model is successfully built. Even more impressive is the size and complexity that can be represented in a single stock/flow model, if required.

Both of these tools help to define system boundaries (what should be included? What are we really interested in?). And both of these tools is subject to error if critical elements are left out (such as including only the birth rate and not the death rate in a population model: How can birth rates be falling, but the population growing?)

My experience is that CLDs help teach the closed feedback nature of systems as well as help to capture all ideas that are offered (without needless rigor). Simple stock/flow diagrams (single stock) are a great way to accurately represent all of the necessary components of a non-linear system (population growth). This simple stock/flow may not be obvious, or intuitive to the student at first, but the stock/flow model will not let the teacher down by mis-representing the "real" process. More complex stock/flow models may have less value in the classroom, unless they are built into a game. Models turned "flight-simulator" can be a great way for someone to gain insight and intuition for real systems. This is because, while the game is based on a complex model, real-life noise like birth rate seasonality or changes in health care policy can be removed leaving only the "core" dynamics of interest.

Thank you for the new found inspiration. I am struck by the idea of a collection of well-structured examples.


Warren Farr
WPI Masters Candidate