On systems and systems thinking

By | October 25, 2014

Some time back, in connection with a class I was teaching, I posted a question to the
Socio-Technical Systems Roundtable (STS RT) discussion board on LinkedIn asking about what kind of developed procedures there might be for conducting socio-technical design studies. This provoked a vigorous discussion with a goodly number of participants, and generated a lot of useful advice. Toward the end of the discussion, the tone turned distinctly theoretical, however, and led to my formulating what I think is a useful statement on the nature of “systems” and “systems thinking” in general. So I thought that I’d share a distilled version of these thoughts with you.

The Real World (RW) is made up of a practically infinite number of phenomena. We perceive these phenomena through senses, and we use our senses and our brains to organize these perceptions into understandable chunks. While in principle there is an almost infinite number of ways that they might be organized, fortunately for human society there seems to be a sort of “consensus reality” that keeps most of us more or less in line with the overall perceptions and organized chunks of others. Language is a key tool in this organizing and categorizing process.

A system is an abstract entity defined by human beings; a mental construct; an idea. Its constituents are objects and relations among them. Usually the elements represent chunks of the RW that are perceivable and measurable, at least in principle. The objects and relations are always vastly fewer than might be perceived in the RW, let alone what might exist there. Any “system” that is abstracted out of this Real World will consist of a finite number of these objects and relations, and is thus incomplete. Large portions of the RW will be represented in this abstraction either explicitly or implicitly by “black boxes” within which things operate in unknown fashion, although inputs and outputs may be defined for them.

Any discussable “system” is thus a human creation, defined by humans for human purposes. Its boundaries and level of abstraction are defined arbitrarily, in order to help people think in an organized way about the phenomena it purports to represent. No mental model (system) is ever “correct”, any more than any idea is “correct” – there is no Platonic “system” out there to be discovered.

Some systems consist entirely of entities that are so intangible that there are no conceivable measurements that can be identified for them. These systems – philosophical or theological – have conceptual value; they may or may not have practical value in terms of guiding human behavior.

Systems thinking is a way of contemplating the RW in terms of its objects and relations. At the most general level, it contemplates the fact that the number of possible connections between phenomena that exist in the Real World is essentially infinite and cannot be represented. But most systems thinking is reductionist – that is, it attempts to simplify the phenomena in question and tends to assume, in practice if not principle, that the whole is equal to the sum of its parts. In practice, we understand that the whole is in fact more than the sum of its parts. Some phenomena we term “emergent” appear only at certain levels of aggregation and abstraction, and some phenomena essentially disappear as our level of abstraction changes.

A practical system, often termed a model, is a representation of some organized activity defined for a particular purpose. Like any system, it exists only in conceptual space. It is incomplete and abstract, and generally assumes the aggregation of certain sets of phenomena into usable chunks. It is used to highlight features of the RW that can potentially be manipulated. Those features are tangible and relations among them can be measured and analyzed.

Bounded rationality and multiple perspectives characterize any system construction. Bounded rationality implies that any practical system will be an incomplete representation of the RW, and that many if not most relations will remain as black boxes. Multiple perspectives imply that any two individuals trying to construct a practical system representation will probably define it in somewhat different ways. This is not only acceptable but valuable, since each representation may highlight somewhat different aspects of the RW and suggest somewhat different points of manipulation.

The most important thing to know about any “system” under discussion is the level of aggregation of the phenomena that it purports to represent. There are fundamental differences between a model of the international economy, in which the units are national government and international corporations; a model of line-worker satisfaction, in which the units are individuals, their interactions, and their states of mind; a model of how four fundamental organizational demands compete for organizational resources; and a model of how different ideas interact to form syntheses and further antitheses. All of them are “system models” and share the properties of abstraction and aggregation, but they do so in very different ways and for different purposes.

An organization (however it may be bounded) can be usefully modeled as a system. But even small groups and organizations are very complex, and any representation simple enough to be understood is probably too abstract to provide useful policy guidance to those operating the system. So organizational analyses tend to need simplifying assumptions, that allow concentrated attention to a limited range of phenomena. Socio-technical systems thinking – an approach that abstracts the key parts of the organization into two distinct but interacting sub-systems – a technical system and a social system – has proven to be particularly useful.

Socio-technical relations in the RW are unbounded and undefinable. On the other hand, the relations within a socio-technical system – that is, a human representation of some aspects of those unbounded elements and relations – are bounded, definable, and measurable. The virtue of good socio-technical thinking is that it helps concentrate attention, separate the relevant from the irrelevant, aggregate and abstract only as necessary, and in general derive more helpful ideas for improvement than just contemplating the wheels whirling. One can essentially ignore fairly large swaths of the organization (or more precisely, relegate them to black boxes) and concentrate on two (by contrast at least fairly simple) sub-domains called a “technical system” and a “social system”.

These sub-systems are defined by analysts for a specific purpose, usually related to a proposed change; no one defines them simply for their inner beauty or because they don’t have anything better to do on a Saturday night. Thus, neither has to include all the technology or all the social behavior in the organization; only those parts of it relevant to the particular change process. It is useful that there are accepted and replicable standards for how technical analyses at least are conducted (those for social analyses are more diverse, but still describable). This provides credibility to the findings and recommendations derived by the analysts, even if the actual development of recommendations and even more so, the effective implementation of them, remains more of an art form than a hard science.

The most important characteristic of a model is its usefulness – does it or does it not allow us to manipulate certain aspects of the RW to improve things (however we wish to define “improve”). Socio-technical language and vocabulary are tools that help us construct rather effective systems (models) in conceptual space that will identify for us things in the RW that we can measure and analyze, and thus potentially improve. Like any system approach, its value is assessed strictly by its practical utility, not by its inner beauty or theoretical consistency. But that is enough.

What do you think?