Copyright 2017 Graham Berrisford. A chapter in “the book” at https://bit.ly/2yXGImr. Last updated 31/05/2021 12:27


Reading online? If your screen is wide, shrink its width for readability.


If system architecture frameworks and systems thinking approaches are to advance, separately or together, then ambiguities in them must be exposed and resolved. This chapter disambiguates some of the term often used by systems thinkers, by defining a handful of general principles related to wholes and parts, holism and emergence.

Holism: wholeism or interaction?

“Our first impulse is to point at [some physical entity] and say "the system is that thing there". [However] every material object contains an infinity of variables and possible systems. Any suggestion we study all the facts is unrealistic [in practice we] pick out and study the facts relevant to some main interest already given.... there can be no such thing as the unique behaviour of a very large system... for there can legitimately be as many [systems] as observers" Ashby 1956.


Some systems thinking discussion is a naive glorification of holism, which is often misinterpreted to mean we must consider the whole of an entity or situation, or zoom out to consider every way in which its parts can relate to parts of its environment.


Holism is not wholeism. When modelling an entity or situation as a system, we exclude almost everything knowable about it. An activity system is soft in the sense it is an observer's view of how selected parts interact holistically to produce one or more emergent properties (effects or outputs) of the whole.


Consider a ridden bicycle as an entity in the world. The system of interest – its emergent properties and the atomic parts that interact to produce them – is determined by the interests the observers bring to their observation.

·       Forward motion emerges from interactions between the riders’ legs and feet, the pedals, the rotating parts of the drive mechanism, the wheels, etc.

·       Steering emerges from interactions between the riders’ arms and hands, the handlebar, the shaft to the axle of the front wheel.

·       Balance emerges from feedback loops between the left-right lean of the rider, the direction of the handle bars, and when in motion, the centrifugal force produced by rotating wheels.

·       Comfort emerges from the rider’s interactions with the saddle, the suspension, the wheels’ tyres and spokes, the handlebars and the pedals.

·       Warnings emerge from interactions between the bell on the handlebars and one thumb of the rider.


System designers may well:

·       Replace one rider by another with no effect on the primary functions above.

·       Remove the warning bell with no effect on the primary functions above

·       Usefully attend to one emergent property and one subsystem at a time.

·       Trade-off between properties, such as comfort and speed.

·       Take the internals of what they see as atomic parts for granted.


The "whole" entity is not is "wholly" knowable. Systems designers will likely pay no attention to, and may be completely ignorant of, a rider’s cardio-vascular system, or the internal structure of a ball bearing.


Note that the scope of a whole and of a part are choices made by observers/describers. The performance of a whole can be measured in many different ways. There are many things a whole can do with only a subset of its parts. The parts of one whole may be more or less autonomous and may also act in other (possible conflicting) wholes.

Ackoff on change

In the works mentioned earlier, and in this article, Ackoff says things disputed here. He characterized a system as a whole in which two or more parts interact, and parts as elements that interact to produce the behavior of the whole.


He wrote: "A part is never modified unless it makes the whole better, that is a systemic principle, you don't change the part because it makes the part better without considering its impact on the whole, that is systemic thinking"


A difficulty with this is that the granularity of a whole and its parts are whatever the observer decides. Systems can be nested, one a part of another, several times over. So, if you took Ackoff’s advice to its extreme, you might never improve a system, for fear its impact on one or more wider systems (of which it is a part) cannot be predicted. (Holism is not wholeism; you can never consider everything.)


Ackoff has been quoted as saying "Improving a part does not necessarily improve the whole". Nevertheless, improving a part on its own is a reasonable way to improve the performance of the whole. Attending to the highest cost part, and removing the largest bottleneck (one at a time) are recommended practices for improving a system. Such incremental development is a feature of biological evolution and agile system development.


Moreover, contrary to some interpretations of Ackoff, you can indeed remove parts from a whole without affecting its ability to meet its main aim. E.g. You can remove the spell checker from a word processor. You can remove the glove compartment, airbag, safety belt, carpet, arm rests and radio from a motor car, and still drive from A to B. You can remove the spleen, gall bladder and appendix from a human body with no significant effect on the functioning of the body.

Holism is not mereology

 "Rather than reducing an entity to the properties of its parts or elements, systems theory focuses on the arrangement of and relations between the parts which connect them into a whole.” Principia Cybernetica Web


The study of whole-part relationships is called mereology. Mereologists see an entity – be it a chair, a machine or an organism - as a structure composed of component parts. A part is a structural element inside a whole, be it an active structure (a subsystem or actor) or a passive structure (material or data). A whole is a structure that “contains” two or more parts, which are connected in some way.


The scope of a whole, and the granularity is parts, are entirely in the gift of describers. Observers bound the whole and select the parts of interest to them. They speak of some parts, and ignore others. They speak of parts as though they are atomic, ignoring what they contain. Clearly, two observers, looking at the same entity, may describe it in terms of different parts and at different levels of granularity.


Often, a system is defined mereologically, as a structure of two or more parts, which are related directly or indirectly. One system thinker proposed “by understanding its structure, we can understand the states a system will exhibit.”  To the contrary however, the organization chart of a business does not explain the processes it performs, or their results.


Find and watch a video of a “double pendulum".  Its structure is very simple - just two components - connected at one point.  Understanding its structure does not help you predict its behavior or the states it will exhibit. The double pendulum looks to be a very simple machine. Yet it displays what some might call "complex" or "chaotic" behavior.


Structural mereology is insufficient as a basis for understanding the activity systems of interest to us. The systems of interest are not merely passive structures that organize or connect things - be it

·       a classification hierarchy like the Dewey Decimal System,

·       a matrix like the chemist’s periodic table,

·       network structure,

·       a timetable for work to be done, or even 

·       a management structure (as in an organization chart).


Rather, the systems of interest are dynamic. They are are describable in terms of actors, activities and the effects or results they produce. An actor is a physical entity (animal, mechanical or social) that can perform one or more activities. An activity is an action, in which one or more actors participate. In thinking about these systems, it helps to recognize and acknowledge some general ideas up front.

Holism is about interactions

“The principal heuristic innovation of the systems approach is what may be called ‘reduction to dynamics’ as contrasted with ‘reduction to components’ ” Laszlo and Krippner.


A system is found in the interactions between things, in how actors interact in the performance of activities. The interactions between systems and parts within systems are central to system dynamics in general and to the success of business operations in particular.


Core concepts: Interaction: an activity involving two or more actors or subsystems. In a physical system, the interaction may be by force, matter or energy. In a social and business systems, interactions involve the communication of information in data flows or messages and via shared data stores or memories.


To think holistically is to think about how effects are caused by interactions some particular things we are interested in. For example, any holistic model we make of a biological ecology excludes almost everything knowable about the physical reality it models.


Holistic thinking can involve zooming in or out.


Zooming out. We may look outside the boundary of a given whole for the cause of an effect, and then widen the scope the whole to include that cause. Consider the dramatic flexing of the Tacoma Narrows bridge. At first, we might describe the flexing as an emergent property of the bridge. Later, we realize it is an emergent property of a wider whole in which some parts of the bridge interact with a part of its environment - the wind.


Zooming in. Designers look first at a required system from the outside. Taking a "black box" view, they identify the system’s inputs, outputs, and effects or results of value to external actors. Then, they look inside the system and divide it into two or more parts (subsystems or actors) and design how those parts interact – holistically - to produce the required state changes or outputs. If any part of the system could produce these “emergent properties” on its own, the rest of the design would be redundant.


Core concepts: Holism: thinking either synthetically (zooming out) of the effect produced when given things interact, or analytically (zooming in) of the things and interactions needed to produce a given effect.


By the way, the idea of zooming in and out is not specific to systems thinking. Basic thinking tools include a) expanding and/or shrinking the boundary of what is being studied, b) testing a proposition or argument by taking it to an extreme, c) making a case for the opposite of what you believe (cf. Shannon’s Test: invert the logic of a sentence to see if the ‘negative’ has more impact and information value).

Reductionism is the flip side of holism

"Were the engineer to treat bridge building by a consideration of every atom he would find the task impossible by its very size [therefore] studying very large systems by studying only carefully selected aspects of them is simply what is always done” Ashby 1956.


It is naive to promote holism and deprecate reductionism, since they are two sides of the same coin. Systems can be nested, such that one person’s whole is another’s part. But you can't zoom out forever (else you’ll find the root cause of all is the big bang at the start of the universe). And when you stop zooming out, and identify how two things interact to produce some effect or result, you are thinking in a reductionistic way about a whole you have just drawn a boundary around.


Whether a system designer approaches the design task by top-down system decomposition, or by identifying atomic parts to be assembled (bottom up) into a system, there is always an atomic level of system definition. The atomic parts are readily obtainable, or another designer's problem, or indivisible agents such as human actors.


Reductionism 1: studying one or more parts of a thing on its own. This is common practice, because one person’s part is another’s whole. A heart surgeon may see the regular beating of the heart as an “emergent property” of particular muscles and valves interacting inside a heart. A general practitioner may see a patient’s heart as an atomic part of a whole body. People zoom in and out; they decompose and compose systems as they see fit. There are times when looking studying a part on its own, and fixing or removing it, is necessary and beneficial. In a social system, you might even have a good reason to study the psychology of an individual human actor!


Reductionism 2: looking for the part responsible for some behavior of a whole. Consider a word processor; the interface to the whole includes interfaces to its parts, such as a spell checker, which fulfils just one function of the whole. Obviously, that part is not responsible for all the functions of the whole, else the rest of the whole would be redundant. Moreover, it can be removed with little or no impact on the whole. (By way of removable parts, consider also the spleen, gall bladder and appendix of a human.)

Emergent properties can evolve or be designed

that a whole machine should be built of parts of given behavior is not sufficient to determine its behavior as a whole: only when the details of coupling are added does the whole's behavior become determinate.” Ashby 1956


With reference to the levels in the table above, we can identify three kinds of emergence. The last is the primary meaning in systems thinking discussion


The emergence by evolution of higher levels over time. This idea is widespread outside of systems thinking.

"Today, it’s a real intellectual deprivation to be blind to the marvellous vision offered by Darwinism and by modern cosmology – the chain of emergent complexity leading from a ‘big bang’ to stars, planets, biospheres, and human brains able to ponder the wonder and the mystery of it all." Quoted from this essay on science.


The everyday emergence of higher-level phenomena from lower-level phenomena. Somehow, conscious thought emerges from electrical activity in the brain. However, we don’t attempt to explain thought in terms of electrons, or explain a baseball game in terms of molecules, or explain Facebook in terms of the hardware components and radio waves it depends on. We usually model a system at one level of thinking, or adjacent levels.


The everyday emergence of effects from interactions between things at one level.

Consider the effects or results that emerge from interacting things in the examples below. In each case, interactions produce an outcome the things cannot produce on their own.


·        The force produced by a wind passing over a sail.

·        The progress of a rider on a bicycle.

·        The V shape of three geese in flight.

·        The shimmering of a school of fish.

·        The price of fish that emerges when customers and suppliers strike a deal.


Not only is emergence is used with various meanings above, it is also wrongly presumed to imply a multitude of actors or agents, or unpredictable or complex behavior.


Emergence does not require a system to have many actors. Two actors are sufficient to produce emergent properties, as in the progress of a rider on a bicycle.


Emergence does not mean a system behaves in a surprising or unpredictable way. Natural systems often produce results or effects that appear surprising or mysterious. At least, they appear so, until you know how they work. By contrast, designed systems are intentionally designed to produce specified results or effects. And when a designed system produces unexpected effects, we call them "unintended consequences".


Emergence does not mean a system is complex in any normal sense of the term. Because in systems thinking, we necessarily ignore the internal complexity of the “atomic parts”. In studying a rider riding a bicycle, we ignore the complexity in the biology of the rider, and the motions of the ball bearings in the wheel hubs. In the discussing the orbits of planets in the solar system, we ignore the composition and atmosphere of each planet.


Core concept: Emergence: the appearance of properties in a higher or wider thing that emerge from coupling lower or smaller things.


If system architecture frameworks and systems thinking approaches are to advance, separately or together, then ambiguities in them must be exposed and resolved. This chapter disambiguates some of the term often used by systems thinkers, by defining a handful of general principles related to wholes and parts, holism and emergence.