By Gerrit Van Wyk.
Some define systems as entities consisting of interactions between multiple components, from which properties emerge that are more than the sum of the parts, for as long as the components and interactions remain relatively stable.
From another perspective, the same definition describes complex phenomena. There are multiple components, that interact, the interactions lead to emergence, which is a core component of complexity, and there is what Vickers called a regulator, or Kauffman attractors, maintaining what emerges. The definition also implies a network effect, that initial conditions and the past matters, and that small changes can have big consequences, often removed in space and time, all of which is associated with complexity.
That creates a problem for most systems methodologies founded on mechanistic-Cartesian principles, and specifically system dynamics modelling (SD). Virtually all big problems, or wicked problems if you want, of our time, involves humans, who logarithmically increase the complexity of the problem and its possible solution. Some systems thinkers try to model human behavior, others treat it as a black box and ignore it, both of which is a gift that keeps on failing. The only way to manage human interaction, behavior, emotions, etc., is as the complex phenomena they are, which requires an understanding of complexity and its principles.
Other than West Churchman, in response to a question during a talk, and Edgar Morin, who remains a virtual unknown in the systems thinking community, I’m unable to find any systems thinking literature explicitly making the connection. Churchman’s understanding of systems, and thinking in terms of systems, is based on a comprehensive philosophical foundation and argument, which is absent in other systems methodologies, who all assume a mechanistic reality without making it explicit. Doing that matters a lot.
Your departing ontological foundation, or perspective of reality, determines what truths you’ll find, and the shape of the methodology you’ll use to go about doing so. A mechanistic ontology implies that systems are real, and like the components of a machine, consultants can identify, describe, analyze, modify, optimize, and make them more efficient from the outside, like scientists in a laboratory.
Based on a complexity ontology on the other hand, and note I’m talking of complexity here as a perspective, not a mechanistic reality, there are no systems, but a notion of systems can be useful for gaining insights about a problem space. It follows one cannot know in advance what the best method or methodology is for approaching a complex problem, what you need to do emerges from interacting with the problem, as Rittel and Webber pointed out. Churchman is the only systems thinker to never commit to a specific methodology, and, although he didn’t make it explicit, the interconnected nature of his Leibnizian inquirer suggests the notions of systems and complexity.
As an example of how taking a mechanistic perspective of systems is a problem, let’s look at system dynamics. SD was originally known as industrial dynamics, and is based on an expression of simultaneous non-linear integral differential equations, and therefore on a mathematical approach to feedback control of socioeconomic systems.
The problem for SD, and many systems methodologies, is calling something a system draws a boundary around it. From a mechanistic perspective, what’s inside the boundary is a thing and real, but from a complexity perspective, the boundary is a choice, and that choice has many problems associated with it.
For example, at business school they taught us an organization is a thing, in other words has a boundary, and consists of boxes: human resources, information technology, operations, marketing, accounting, and a management function overseeing everything. When you connect these boxes and the subcategories we were taught, in a SD model, it looks like this.
Most people switch off when I show diagrams like this, but there is much to be learned from it. To begin with, plugging this model into a computer program and running it as a simulation is not useful, there are too many missing components that may impact on the outcome, and adding or subtracting any is a choice with consequences. Secondly, such simulations are stopped at some decision point, complex phenomena don’t stop, they continue evolving. Thirdly, notice the absence of human systems which not only turbocharges the complexity of organizations, they are also fundamental to how they work. And, finally, organizations are interconnected to and interact with many other components in their environments, such as other organizations, resources, competitors, political issues, and many more, which impact on it, and cannot be ignored. Most SD models and simulations, by treating systems as entities, are disconnected from that reality.
We talk a lot within the business world and systems community about things like leadership, as if it is a mechanistic entity, that is easily defined, making it amenable to analysis and manipulation, but from a complexity perspective, it is an emergent phenomenon, as follows.
Or organizational culture, which, using Schein’s description, ends up looking something like this.
In short, the mechanistic-Cartesian perspective of system is not very useful in a complex real word, which requires a different take on reality, and approaches to that reality.
I’m not saying systems methodologies such as SD is not useful, they can be (see above). Rittel and Webber’s wicked problems nowadays regularly crop up in systems conversations, but we grapple with the fact neither Rittel and Webber, nor we, know how to approach them. The reason is what we are talking about are complex problems, and we are approaching them from the wrong direction, because we start in the wrong place.
Circling back to the definition of systems in the introduction, it means there is value to looking at “systems” as complex phenomena, which requires a basic understanding of the principles of complex phenomena, applying those principles to better understand the complexity of human interactions and interconnections, and using those insights as a basis for approaching complex, or wicked problems.
With that I’m not throwing out the baby with the bathwater. It is useful to look at aspects of the problem space not as systems, but as if they are systems, from which useful information may emerge for dealing with the problem situation.
Churchman said politics, dogma, esthetics, and morality are enemies of reason, but with a knowledge of complexity, and complex human social interaction, one may look at politics, for example, as if it is a “system”, and gain insight about it. It doesn’t mean a solution, because, as Flood and Jackson pointed out, when politics, coercion, or violence, irrespective of whether it is physical, social, or cultural is part of the problem space, there may be no solution, or the solution may be unpalatable or immoral.
Another lesson from complexity is one’s insight is contextual; what applies in one context will not apply in another, which is why one can use complexity principles, but not generalize it in the manner of mechanistic-Cartesian methodologies, into a boilerplate methodology for every or even similar contexts. That, in effect, is the value of Churchman’s approach.
My message is systems thinking is impoverished by not acknowledging its connection, or affinity to complexity. When you make that connection, it significantly expands the utility of thinking in terms of systems, and approaching complex problems as if they are systems.
Which leaves the systems community with a choice; either continue pretending there are systems working like machines, which seems to be the dominant perspective, or abandon that project in favor of the fact our world is complex. To me, systems thinking has been stagnant for a long time, and this would present an opportunity to grow the way we think about and approach “systems”.