System Dynamics Conf. – I – Overview

By | July 26, 2009

So, I’m at the 27th International System Dynamics Conference in Albuquerque, New Mexico, at the moment and, like any wannabe writer with a blog, I’m going to make an attempt at live blogging it.

My flights were nice and comfortable, though the second one, from Phoenix to Albuquerque was seriously delayed. At first, this was due to the plane lacking some internal power unit necessary to start the engine, thus requiring the use of some special plane starting device on a truck. Then, once that was sorted, we still had to wait for the plane to cool down – apparently sitting in the sun for a day in Arizona means an internal temperature of something like 55 degrees Celsius (130 F). As a result, I didn’t get to Albuquerque till about 1:30am, well after the office closed at my hostel. By the time I’d woken one of the staff, found my key, and all that, I didn’t get to sleep till about 2:30am.

Figuring that I wouldn’t get much out of the day if I didn’t have at least 7 hours sleep, I skipped the first couple of presentations, and arrived there a little after 10:30, just in time for the end of a presentation on some kind of network modelling.

Before I start talking about specific presentations, though, I should first provide a brief overview of System Dynamics for the uninitiated. Basically, it’s a method of modelling the dynamics within complex systems by specifying mathematical relationships between specific quantities both abstract and concrete. It’s different from a lot of mathematical models because it focuses on the relationships within whole systems and the simultaneous dynamics of multiple variables, rather than focusing on specific individual variables in isolation. There’s a number of relatively famous books either about or employing System Dynamics that you may have heard of:

System Dynamics is related to the work of Norbert Wiener, the General Systems Theory of Bertalanffy, and the cybernetics employed by James Lovelock in his Gaia hypothesis. It’s variously called systems thinking, industrial dynamics, business dynamics, and many other things. There are, apparently subtle differences between all of these, but I’m too new an initiate to know them. Working with it leaves you talking a lot about positive and negative feedback loops, stable and unstable systems, death spirals, exponential and goal-seeking growth, and exogenous and endogenous variables. If any of those phrases sound familiar, you can probably work out approximately what this is all about.

I’m hoping this description isn’t just making you more confused, but I know it’s only just scratching the surface; for a better introduction, check out the article on wikipedia.

Like any modelling methodology, the predictions of made by models in this field are by no means perfect, but it has the advantage that much of its discourse involves explicit acknowledgement and discussion of the sources of model error. Errors often come from boundary conditions (which factors do you model, and which do you leave out?), nonlinearity and discontinuity (how reliable is our knowledge of the relationship between correlated variables?), and input data (how reliable are the statistics you’re putting into your model?). Even ignoring the precision or accuracy of its predictions, System Dynamics offers a way of examine the relationships between the parts of one’s conceptions of a real world system, bringing up inconsistencies and fallacies that would otherwise have gone unnoticed.

That’s enough for this post. Hopefully, tonight I’ll post highlights of the first day of the conference, otherwise I’ll wrap it in with tomorrow’s events.