As a matter of definition, systems are composed of a set of entities. These entities are the constituents of the system.

In general, each of the entities of the system will also have form and function. We will sometimes use the word “element” (whose synonyms include “part” and piece”) when we want to emphasize some aspect of the form of the system. The word “entity” (whose synonyms include “unit” and “thing”) tends to more generically evoke both form and function.

The two columns on the right side of Table 2.3 show the decomposition of the form of our example systems into its constituent elements of form. The amplifier circuit decomposes into the two resistors and the operational amplifier; Team X decomposes into its members, and so on. Reading right to left, breaking a system into smaller pieces of form is called decomposition. Reading left to right, collecting pieces into the form of the system is called aggregation.

The middle two columns of Table 2.3 indicate the mapping of the form of the entities onto the function of the entities. Resistors 1 and 2 set the gain. The heart pumps blood. Each of the entities of the system also has form and function. Note that the function of each entity is written as a process and operand.

The two columns on the left side of Table 2.3 show the way in which the function of the system and the function of the entities are related. Reading left to right, breaking the function into constituents is called zooming. Reading right to left, when the functions of the entities combine to produce the function of the system, we find the sought-after property of emergence.

Function is a quasi-static view of the process acting on an operand. When a number of functions act in a sequence of operations, a more dynamic behavior emerges, as will be discussed in Chapter 6.

Sometimes in system thinking, it is useful to think just about function and zooming. For example, to think about the amplifier circuit in terms of amplification and gain setting. This kind of functional thinking is often used early in analysis and design. On the other hand, sometimes it is enough to reason about form and decomposition, such as when you are developing a “parts list” (amplifier, resistor 1, resistor 2). Reasoning about form or function separately is a convenience: it does not imply that both are not ultimately present or that they are not linked.

Our initial reference for “the system” is arbitrary. We could have started higher up with the human body as “the system,” which could then be decomposed to find the circulatory system, the digestive system, and so on. Or we could have started lower down, defining the heart as “the ­system,” which is composed of chambers, valves, and so on. This leads us to a generalization: All systems are composed of entities that are also systems, and all systems are entities of larger systems.

Because the initial choice of “the system” in these hierarchies is arbitrary, all systems must be made up of systems, which are made up of smaller systems, and so on. There are limits: the cosmos on one end (but is there really only one cosmos?) and the quarks on the other end (but is there really nothing smaller than quarks?). What matters is that we choose a system boundary that is useful, so that we train our system-thinking lens on the most important part of the problem.

In practice, definition of the entities and boundary is important and challenging. There are five issues the systems thinker faces:

• Defining the initial decomposition into entities

• Identifying the potential entities using holistic thinking

• Winnowing down to the consequential entities using focus

• Creating abstractions for the entities

• Defining the boundary of the system, and separating the system from context

The remainder of Section 2.4 addresses these five issues.

The level of difficulty encountered in defining the entities, and therefore the internal ­boundaries, of the system depends on whether the system is made up of distinct elements, is modular, or is integral. Sometimes the system is made up of clearly distinct entities, and the decomposition is obvious. Team X is made up of three people. Any other decomposition of the team would not make sense. Likewise, the solar system is clearly made up of the sun, the planets, and the smaller bodies (of which there are a multitude). Unambiguous decomposition into entities is a trait of systems that really are made up of discrete entities brought together and defined as a system (a fleet of ships, a herd of horses, a forest of trees, a library of books, and the like).

For systems that are fundamentally modular, the decomposition is more challenging but still relatively clear. Modules are relatively independent, especially in function. Internal relationships are dense within a module, and relationships between modules are weaker or less dense. For the amplifier circuit system, there are inputs, resistors, amplifiers, internal nodes, and connections. There may be some fuzziness about where the resistor ends and the connector begins, but it is largely clear.

Integral systems are the most difficult to decompose. Integral systems cannot be easily divided with their function intact. They are often highly interconnected systems, such as the components in the steering mechanisms of a car (tires, wheels, suspension, steering gears, column), some of which are simultaneously also components of other systems (ride quality, drive). Truly integral mechanical elements (such as complex forgings and machined parts) and integrated circuits are examples of integral elements. Many information systems are highly integral.

Holism insists on the intimate interconnection of things—on the idea of the whole—and to think holistically is to think deliberately about the whole. Holistic thinking seeks to identify all of the entities (and other issues) that might be important to the system (see the Principle of Holism in Box 2.5). We think holistically in order to bring into view all aspects of the system at hand, ­taking into account the influences and consequences of anything that might interact with the ­system. We use holism to expand our thinking about the problem or issue at hand.

By thinking as widely as is feasible about what might be important to the system, we increase the chances that we will move something into consideration that will ultimately be important. Holistic thinking gets issues onto the “radar screen.”

There is a distinction between the known-unknowns, and the unknown-unknowns. A known-unknown is something that you know is there but don’t know much about. Its presence is known, but its features are unknown. However, you know that you should know more about it. An unknown-unknown is something that you don’t even know is there, so you have no way to evaluate its importance. Holistic thinking works to identify as many potential unknown-unknowns as possible so that their potential importance can be considered.

There are various methods to help stimulate holistic thinking, including: structured and unstructured brainstorming (Chapter 11); the development of frameworks to ensure that relevant issues have been considered (Chapters 4 through 8); thinking from various perspectives (Chapter 10); and thinking explicitly about context (Chapter 4).

As a running example in this section, we will apply the five issues facing a systems thinker to Team X. In a team with discrete individuals, identifying the entities at first looks easy. As indicated in the first column of Table 2.4, let’s assume that we first looked narrowly at design as only concept generation (John) and design approval (Sue). The initial consideration was then expanded by holistic thinking into a longer list of potential team members, including requirements analysts, finance analysts, team coaches, and experts in marketing, manufacturing, and supply chains, as indicated in the second column.

The desired outcome of holistic thinking is a longer list of all the potentially important entities to consider in defining the system and its context. This list is then narrowed by focus.

The next issue that the system thinker faces is focus—that is, to identify what is important to the question at hand (see the Principle of Focus in Box 2.6). This means separating the wheat from the chaff. It means cutting down the list of everything generated in holistic thinking to a shorter list of things that are truly consequential.

The pivotal step in focusing is defining the question, circumstance, or problem at hand and articulating what is important about it. More specifically, what is important to you and your stakeholders? What outcomes are important? Is it the emergent behavior of the system? Is it satisfaction of some specific set of criteria?

Then you can begin to reason through the entities in the whole and ask a simple question that is very difficult to answer: Is this entity important in determining the outcome and the emergence that I am interested in? We could make a very long list of things, but the human brain can only reason about a finite number of things simultaneously, while remaining able to understand their interaction. This manageable number is conventionally thought of as seven +/– two. [2]

What we are doing by focusing is being aware of a longer list of things that are potentially important, and then “swapping in” up to seven of them at any time to really focus on. When the circumstances change, we will swap in another set of issues to reason about.

Returning to the Team X example in Table 2.4, we might reason that the main outcome of the team is a good design, the inputs are the requirements, and that the supporting entities include an understanding of the supply chain and manufacturing. Therefore, we would keep under consideration the three members of the team (Sue, John, and Amy), as well as those who determine customer needs, those who analyze the competitive environment, and the experts on manufacturing and supply chain, as shown in the third column of the table. Based on our focus analysis, the experts on finance, team dynamics, and regulation would therefore be omitted from today’s system thinking.

This leads to the final part of the focus issue: performing a sanity check to make sure that the entities still under consideration are broad enough to cover the important question, circumstance, or problem, but small enough so that they can be carefully examined with the resources at hand.

Once you have a sense of what is important to the question, circumstance, or problem (the outcome of the holistic thinking and focus), the next issue is to define or recognize the appropriate abstractions to represent the entities in the system. An abstraction is defined as “expression of quality apart from the object” or as a representation “having only the intrinsic nature rather than the detail.” Many problems come with predefined abstractions (people, layers, control volumes), which can either enable or disable your reasoning. Creating useful abstractions means bringing to the surface important details about the entity and hiding, within the abstractions, any details and complexity that you do not need to consider.

Let’s look at some abstractions in our four running examples. In the amplifier circuit, we abstracted the operational amplifier into a device that has an inverting input, a non-inverting input, and an output and that amplifies the difference between the inputs; see Figure 2.4. The actual ­circuit to do this is shown in Figure 2.8. The abstraction hides all of this detail and allows us to reason about the function on the “surface”—amplification. In Team X, we abstracted a ­physiologically and psychologically complex person into a “team member” who could create concepts. For the circulatory system, we abstracted the complex organ called the heart into a simple pump. In the solar system, we abstracted the entire solid mass, ecosystem, and population of the planet Earth into a sphere.

From these examples, we can generalize to the following guidelines on creating abstractions:

• Create abstractions of form and function with the important information represented on the surface, and with less important details concealed.

• Create abstractions that allow for representation of appropriate relationships (see Section 2.5).

• Create abstractions at the right level of decomposition or aggregation.

• Create the minimum number of abstractions that will effectively represent the aspects of the system at hand.

It is possible to create less-than-useful abstractions. For example, we would violate the first guideline above if we abstracted an operational amplifier as a heat source. Doing so would be technically correct, but it would not bring to the surface the entity’s important role in amplification. We could violate the third guideline by representing too much detail on the components of an operational amplifier. Again, this might be true, but so much detail may not be necessary to understand the operational amplifier’s role in a circuit.

When creating abstractions, you would naturally loop back many times to the focus issue to ensure that you were creating abstractions that captured the important issues. You might even loop back to the holism issue if you were reminded of something that was missing from the ­holistic view.

Note that abstractions are not unique, and there may be other abstractions of the same entities that are also completely valid. Which abstraction is the right one to choose depends on the nature of the question, circumstance, or problem at hand. You usually cannot make universal abstractions.

To return to the Team X example of Table 2.4, we have maintained as abstractions the three individuals John, Sue, and Amy, but we have abstracted Heather and Chris into “Marketing” and likewise their functions into “market analysis.” Similarly, we have abstracted Karen and James and their functions into “planning manufacturing and supply chain operations.” Although this reduction from seven entities to five seems trivial, we will see in Section 2.5 that the relationships among the entities scale like N² (N-Squared). By defining this smaller set of abstractions, we have probably reduced the possible relationships from 49 to 25, a significant improvement!

The outcome is a set of abstractions that are important to the system but have not yet been defined to be in the system. In other words, we have not yet drawn the system boundary.

In defining the entities of “the system,” it will often be necessary to define a boundary of the system. The boundary makes it clear what is in “the system” and what is outside it. All systems, perhaps short of the cosmos, have boundaries. When we examine systems, we always define them to be of limited extent, either because we are simply not able to consider a more extensive set of entities (a  human capability limitation) or because we believe it is not useful to do so (a human judgment).

In defining the boundary of the system, we separate the system from its context. Context is what surrounds the system. It is the entities that are “just on the outside of the system” but are relevant to it.

Between the system and the context sits the system boundary. In drawing the system boundary, we might consider

• Including the entities to be analyzed (if the goal is understanding)

• Including what is necessary to create the design (if the goal is design)

• Including what we are responsible for implementing and operating (if the goal is ­delivery of value)

• Formal boundaries, established by law, contract, or other legal regime

• Traditions or conventions that distinguish the system from context

• Interface definitions or standards that we must respect, including supplier relationships

When a relationship crosses a boundary, it defines an external interface between the system and the context. These external interfaces are critical for the system and will be discussed in Section 2.5.

Table 2.4 represents the outcome of Task 2 of System Thinking for Team X. If we think that the job is to produce a design, then a logical place to put the system boundary is with John, Susan, and Amy within the system and to place marketing and operations outside it. The system boundary will be consistently shown in this text as a dashed line.

As we conclude the discussion of Task 2 of System Thinking (identifying the entities of the system, their form, their function, and the system boundary and context), we are left with the information shown in Figure 2.9. The entities are shown in the boxes, with the form and function described in the text. The boundary, indicated by the dashed line, separates the ­system from the context.

In summary:

• All systems are composed of entities, which have form and function and are themselves likely to be systems.

• Defining the composition of the system as entities is easy for a system made up of distinct entities, is moderately difficult for a modular system, and is quite difficult for integral systems.

• Thinking holistically helps to identify all of the entities that might be important to represent in the system, but it often yields too many entities to usefully consider.

• Focus helps to reduce the entities to those that are consequential to consider at that moment, but the set may change with time and activity.

• Creating abstractions helps bring to the surface the essential details of an entity, while hiding the rest of the complexity.

• Defining a boundary separates the system from the context.