In a breakout in the student colloquium at ISDC 2022, we discussed the difficulty of getting a paper accepted into the conference, where the content was substantially a discrete event or agent simulation. Readers may know that I’m not automatically a fan of discrete models. Discrete time stinks. However, I think “discreteness” itself is not the enemy – it’s just that the way people approach some discrete models is bad, and continuous is often a good way to start.
On the flip side, there are certainly cases in which it’s sensible to start with a more granular, detailed model. In fact there are cases in which nonlinearity makes correct aggregation impossible in principle. This may not require going all the way to a discrete, agent model, but I think there’s a compelling case for the existence of systems in which the only good model is not a classic continuous time, aggregate, continuous value model. In between, there are also cases in which it may be practical to aggregate, but you don’t know how to do it a priori. In such cases, it’s useful to compare aggregate models with underlying detailed models to see what the aggregation rules should be, and to know where they break down.
I guess this is a long way of saying that I favor a “big tent” interpretation of System Dynamics. We should be considering models broadly, with the goal of understanding complex systems irrespective of methodological limits. We should go where operational thinking takes us, even if it’s not continuous.
This doesn’t mean that everything is System Dynamics. I think there are lots of things that should generally be excluded. In particular, anything that lacks dynamics – at a minimum pure stock accumulation, but usually also feedback – doesn’t make the cut. While I think that good SD is almost always at the intersection of behavior and physics, we sometimes have nonbehavioral models at the conference, i.e. models that lack humans, and that’s OK because there are some interesting opportunities for cross-fertilization. But I would exclude models that address human phenomena, but with the kind of delusional behavioral model that you get when you assume perfect information, as in much of economics.
I think a more difficult question is, where should we draw the line between System Dynamics and model-free Systems Thinking? I think we do want some model-free work, because it’s the gateway drug, and often influential. But it’s also high risk, in the sense that it may involve drawing conclusions about behavior from complex maps, where we’ve known from the beginning that no one can intuitively solve a 10th order system. I think preserving the core of the SD genome, that conclusions should emerge from replicable, transparent, realistic simulations, is absolutely essential.
Related:
By model free you mean without a simulation model — i.e. a causal loop diagram — right? I have a lot less confidence in simulation models and their analyses than many in the field, so to me a CLD isn’t that much of step down. Also a CLD is endogenous, has feedback as a source of dynamics, and sometimes has stocks and flows too. Yes, we can’t figure out a 10th order system by intuition, but how many people who build a tenth order simulation models actually know all the behavior patterns it can produce?
I actually do have a question about allowing games in system dynamics. Most games don’t have interesting dynamics on their own. And most people who play SD games never grasp the feedback and Stocks & Flows that give rise to the behavior. Today’s interfaces often seem to be a set of levers, that are connected to behavior, but not (so far as the player is concerned) to the structure. Games are also a way to convince people of something, when they shouldn’t be convinced. They’re also a way to make a living. One place where games may be useful is as a way to ferret out how real people make decisions , when the people can’t otherwise describe their decisions/policies. The (inferred) policies can then discussed and be put into an SD model.
Right – I was thinking of CLDs when I wrote that, but there might be other conceptualization methods in the same bucket. I think “knowing all the behavior patterns” is definitely a challenge, but on the other hand, if you have a model, at least you know _one_ pattern, which is better than none. OTOH if you have a 1000th order model, there’s a good chance you haven’t tested it adequately. So perhaps we should ban both tails of the distribution: model free and large models. (I’m almost half serious.)
With a 1000th order model, there’s also a good chance it has errors. In the one case I pointed this out, the model builder, truly one of the best in the field, said that it went to show that all big models contain errors. I have to say say, though, that in this case fixing the error made no difference to the behavior.
Games are an interesting case. If most people really don’t understand what they’re doing, I question why we’re building them. But is it really “most”?
I think convincing people of things when they shouldn’t be convinced is a real problem. It’s actually surprising that there aren’t more nefarious games promoting climate denial or such things. I think the answer may be that ordinary BS is cheaper than games, and reasonably effective.
Most of the games I’ve seen in the conference lit are in the spirit of ferreting out dynamic decision making heuristics.
Very interesting piece. As a practitioner, I would argue that CLDs are very useful in making maps of mental models, especially when we are doing so in participatory/collaborative settings (GMB, rapid SD, etc). In my experience, they surface mental model paradigms and remove the blame element from the conversations. They become super useful to bare the very bones of the biases people in the room have, and therefore, make it easier to collaborate from thereon. I presented my work from one such workshop last year (virtual settings make it difficult for me to gauge if it was received well). I guess this conversation says something about the models within the SD modellers community itself?
Tom, this sounds related to “System Dynamics and Systems Thinking: It Takes All Kinds” by Alan Graham and Sharon Els. It also sounds related to old discussions with Geoff Coyle on the SD mailing list IIRC. For those new to the ideas you presented, those might also be of interest.
Incidentally, that Graham and Els paper is available at https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.609.1974&rep=rep1&type=pdf .
Good thoughts.
Graham & Els:
https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.609.1974&rep=rep1&type=pdf
Early Coyle on discrete events:
https://www.jstor.org/stable/2582417
I recall discussions from the mailing list, but couldn’t immediately locate one. But there’s tons of interesting stuff in the archive:
https://www.ventanasystems.co.uk/forum/viewforum.php?f=15
Oops – the SD archive is
https://www.ventanasystems.co.uk/forum/viewforum.php?f=34
I never found the Coyle material I recall, either.
Those who might not recognize the spectrum of alternatives between pure CLD and full model and who have a bit of study time on their hands might be interested in Alan Graham’s dissertation, “Principles on the relationship between structure and behavior of dynamic systems” at https://dspace.mit.edu/handle/1721.1/16421 .