Copyright 2020 Graham Berrisford. A chapter in the book at https://bit.ly/2yXGImr Last updated 31/05/2021 11:43


If system architecture frameworks and systems thinking approaches are to advance, separately or together, the many ambiguities in them must be exposed and resolved.

This chapter resolves about 20 ambiguities recognized in previous chapters.




Two kinds of holistic thinking (analytical and synthetic) 2

Two kinds of organization (structural and behavioral) 2

Two kinds of part (passive and active structure) 2

Three kinds of emergence (evolution, hierarchy and interaction) 3

Two kinds of actor (automaton and agent) 4


Two kinds of system (abstract and physical) 5

Two kinds of state (physical and conceptual) 5

Two kinds of behavior (process and state change) 6

Two kinds of state change (homeostatic and progressive) 6

Four kinds of causality. 6


Two kinds of social system (actor centric and activity centric) 7

Two kinds of freedom (to select and to invent) 7

Two kinds of social entity (purposive and purposeful) 7

Two kinds of purpose (intent and outcome) 7

Two kinds of organization (order and social structure) 7


Eight kinds of change. 8

Two kinds of "anti-fragility". 8

Two kinds of agility. 8

Three kinds of self-organization. 8




Two kinds of holistic thinking (analytical and synthetic)

Wholeism (considering every conceivable aspect or element of a thing or situation) is impossible. We can never consider every element or interaction that could be identified in a given system or situation. For example, any holistic model we make of a biological ecology excludes almost everything knowable about the physical reality it models.


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


Holism in system analysis: Holistic thinking can mean looking for causes and rescoping systems. It can mean looking outside the boundary of a given system for the cause of an effect. And then rescoping the system of interest.


Holism in system synthesis: System design is holistic by definition. The requirement is to produce emergent properties or effects. Designers must both a) divide the to-be system into parts and b) design how those parts interact to produce the required emergent properties.

Two kinds of organization (structural and behavioral)

Part: a structure inside a whole, be it an active structure (subsystem or actor) or a passive structure (material or data). To organize some parts into a whole can refer to a structure in which parts are connected, or refer to a process those parts play roles in.


How actors connect in a structure is one thing. How actors interact is another. Systems thinking is more about the latter than the former.

Two kinds of part (passive and active structure)

Structures can be classified in many ways; we must at least distinguish active structures from passive ones.


Passive structure: a static that structures organize or connect things. For example, a classification hierarchy like the Dewey Decimal System, a matrix like the chemist’s periodic table, or a network structure, or an organization chart, or a timetable for work to be done. These patterns are interesting, and some may call them systems, but they are not systems of interest in the sense here.


Active structure: an actor who can not only connect in some way, but also interact to advance the state of a system. The systems of interest both include active structures, and may be seen as active structures. Think of:


·        a solar system in which orbiting planets interact by gravity with a star and each other

·        a windmill in which parts rotate and interact to grind corn and make flour.

·        a tree in which parts interact to consume sunlight and carbon dioxide, and produce a tree trunk and oxygen.

·        a boxing match in which boxers and a referee interact to advance the state of the match, as recorded on the judges' scorecards.


Bear in mind the passive/active distinction is a matter of perspective. A sandstone pebble is an entity composed of sand or quartz grains connected by cement. It is a discrete entity - created, changed and destroyed by events. We normally see it as a passive structure.- as acted on rather than an actor. However, it can be recast as an active structure; it can be presented as dynamic system that absorbs and excretes a small amount of water.


Note also that some descriptions can be animated. Narrative and mathematical procedures are passive structures that may be executed. Human and computer actors not only read descriptions of activities to be performed, but also perform those activities.

Three kinds of emergence (evolution, hierarchy and interaction)

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


Some systems thinkers see this as the concept most definitive of systems thinking. This table (bottom to top) presents a history of the universe from the big bang to human civilisation.



Elements or actors


Human organizations

Human sociology

Humans in groups

Social psychology

Animals in groups


Animals with memories


Living organisms

Organic chemistry

Carbon-based molecules

Inorganic chemistry



Matter and energy


In relation to this table, we can identify three kinds of emergence:


·       The emergence by evolution of higher levels over time

·       The everyday emergence of higher-level phenomena from lower-level phenomena.

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


The first idea is widespread outside of systems thinking.  In On the ontology of space-time”, Gustavo Romero wrote that reality seems to be organized into.at least 6 levels. Each step up to a higher level adds some emergent properties and laws - adds complexity you may say.


Romero’s level

Elements and rules

1 ontological


basic events, which precede the emergence of physical things at the levels below

2 physical

matter and energy, which obey the laws of physics.

3 chemical

molecules, which interact according to the laws of chemistry.

4 biological

animals, which embody biological processes.

5 social

behavior, which requires interactions between people

6 technical

see note below


Level 1 "is formed by basic events and precedes the emergence of physical things at the physical level." This reflects the presumption that events come before entities. Every entity in the universe is created, changed and destroyed by events. What we perceive as persistent structures are the transient side effects of behaviors. An entity can be modified by events in ways that change its state (e.g. cold to hot), or change its type or the activities it can perform (e.g. caterpillar to butterfly).


Level 6 is oddly placed in terms of complexity. Consider a pendulum, a very simple technology, a physical system that operates at level 1. That it was designed by a person does not make it more complex than a person or a society.


It seems reasonable to say that complexity (in the orderly rather than messy sense) increases or emerges when moving from a lower level of thinking to a higher level. Chemical evolution after the big bang complexified the periodic table and the range of possible chemical reactions. Biological evolution tends to complexify the organisms of a species, and widen the range of an animal's possible behaviors. Animal behavior evolution has led to complex social interactions which depend on perceptions and memories at the psychological level, synapses at the biological level, chemicals at the chemical level, and atoms at the physical level. Human evolution has created actors who can envision and make new things (e.g. paintings), new activities (e.g. card games), new tools and software systems.


Misconceptions about emergence include that it requires manyness, surprise, complexity.

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. Designed systems are intentionally designed to produce specified results or effects. 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”.


For more on complexity science, catastrophe and chaos read this chapter.

Two kinds of actor (automaton and agent)

Automatons are actors that have no choice other than to follow given rules. If they do select from a fixed range of activities, then it is by following some deterministic rule. Or else by making a random choice (perhaps qualified by some degree of probability).

Agents are actors that can choose to exercise some degree of autonomy.


The term agent coves a spectrum of possibilities, since an agent may be free to:



At the extreme, an agent may be entirely self-serving. And act in ways that are regardless of any other actor or other actors' aims.


Two kinds of system (abstract and physical)

different observers of the same phenomena may conceptualize them into different systems”. Ackoff

"a system... is independent of the concrete substance of the elements (e.g. particles, cells, transistors, people, etc).” Principia Cybernetica Web


When people speak of a system, they may speak of an entity (the unknowable whole of a thing or situation, regardless of how observers look at it) or a pattern of activity (a regular or repeatable behavior observed or envisaged in a thing).


In cybernetics, an entity is only a physical system in so far as it realizes an abstract system - a description of regular behavior relevant to “some main interest that is already given” as Ashby put it. This table shows how an abstract system can be realized as a physical system by a physical entity.


How an abstract activity system is realized

Game of poker

Abstract system: rules, rules and variable types

The rules of the game

Physical system: activities giving values to variables

A game of poker

Physical entity: physical actors able to perform the activities

A card school


This book contrasts and relates activity system theory, which is about an abstract system and how it is realized, with social entity thinking, which is about a physical entity, and the activities (or activity systems) it realizes.

Two kinds of state (physical and conceptual)

The passage of time is revealed by changes. The universe is an ongoing process in which the state of things changes over time.


E.g. Consider a tennis match. At any moment in time, the players, the balls and the tennis court have a current material or physical state, which we don't attempt to measure. At the same time, the tennis match has the conceptual state shown on the score board. In practice, we represent the physical state of a thing as a conceptual state, as a vector containing the values of state variables.


E.g. In physics, the vector of an object might contain its spatial coordinates at a moment in time. And in economics, the vector might contain the current inflation rate and unemployment rate.

Two kinds of behavior (process and state change)

In business and software architecture, a behavior is usually a process. Any action, activity, operation or procedure that takes time to perform. Perhaps shown in some kind of flow chart. In system dynamics, a behavior is a state change trajectory, a line of behavior that shows how the value of a state variable changes over time.

Two kinds of state change (homeostatic and progressive)

Systems thinking is concerned with things whose state vector changes over time. Some things advance from one state to the next. E.g. a moon rocket, or a tennis match. Other things cyclically return to the same point, or stabilize themselves in an equilibrial state. E.g. a solar system, or a thermostat-controlled room temperature.

Four kinds of causality

Ashby discussed how a system may respond to input events or disturbances in two ways – deterministic and stochastic. When observing actors’ behavior, we can classify their responses to stimuli into four kinds.



In theory, when an event happens, we can predict


exactly which action an actor will perform in response.


how likely an actor will perform activity type A or activity type B.


the actor will choose from the range of activity types in our model.


nothing – because actors can invent activities outside any model made.


Classifying causality types this way helps us to characterise what makes social entity thinking different from activity systems thinking.


In a poker game, the range of actions is limited to those that characterize the system. The rules do not tell players whether to "call", "raise" or "fold".  Players strive to make their choices unpredictable. They also try to detect probabilities in how others choose between the actions allowed in the system.


Where the actors in a system are anthropomorphic rather than computational, we assume they can not only act in the first three ways. but also be innovative, in the fourth way. Outside of their role in playing cards, the same human actors are innovative; they invent new responses to events and conditions.


A social entity is not well called an activity system. A complex adaptive system (CAS) with human actors might be called a complex adaptive social entity (CASE).



Sociology may address all four kinds of causality above. Whether our psychology is deterministic or not at the biological level is irrelevant. At the sociological level, we must treat people of sound mind as having free will.

Two kinds of social system (actor centric and activity centric)

Two branches of social systems thinking may be distinguished. Activity system theory - about regular activities, performed by actors playing roles (e.g. a poker game). (The actors are changeable, and may act outside the system.) Social entity thinking - about actors, who perform activities (e.g. a card school with a pack of cards). (The activities are changeable.)


Other views are possible (e.g. aim-centric and state variable-centric). But the dichotomy above is the best explanation of why so much systems thinking discussion is confused or confusing.


Why say social entity rather than social system? Paradoxically, some systems thinkers promote anarchical social structures and/or irregular one-off behaviors. They propose human actors respond to events and conditions as they choose, learn from experience and respond to novel situations in innovative ways.  Which is fine.  The trouble here is, it undermines the general concept of a "system". If actors continually exercise their freedom (as autonomous agents) to innovate, then there is no regularity or repetition, and no recognizable activity system.

Two kinds of freedom (to select and to invent)

Freedom might be defined as the degree to which the actors in a system can make decisions and act as they choose. Freedom to select activities: In an activity system, actors can select from a range of regular activities. The range is limited to what the system allows. Giving actors a wider choice of actions, a higher degree of freedom, increases the system’s complexity. Freedom to invent activities: In a social entity, actors may invent their own activities, even their own aims. They may do this without any overarching change control, and make ad hoc decisions that lead them down novel paths. Ironically, every decision is a constraint in the sense that choosing one path denies another.

Two kinds of social entity (purposive and purposeful)

The purposes of a purposive social entity lie in the desire of external actors for that entity to produce particular state changes in the state of its environment. The internal actors may be seen as slaves to that end. The purposes of a purposeful social entity are found in the desire of its internal actors to produce internal and/or external state changes, as they choose to do.

Two kinds of purpose (intent and outcome)

Purpose as desired outcome: an intention – an aim we have - a reason to do things. Purpose as actual outcome (POSIWID): an outcomes of system behavior we observe, or a use we find for things


We can't have it both ways. One of the most powerful ways to influence the behavior of a social entity is through its purpose or motivation. By changing its purpose(s) you can change the activity system(s) that a social entity realizes, and the outcomes it produces.

Two kinds of organization (order and social structure)

In social entity thinking, the term organization is used to mean two different things.

The management structure of a social entity under which human actors are arranged to perform the required activities.

A pattern of behavior, a repeated pattern of inter-actor communication in a group or network of actors from which properties emerge.


Eight kinds of change

System change can be classified in three ways: continuous or discrete, state change or mutation, natural/accidental or designed/planned. Representing change as a three-dimensional phenomenon helps us to think about what it means to model change and design for it.





State change


the growth of a crystal in liquid


asleep to awake, or day to night


analogue light dimmer switch


light on to light off



maturation of child into adult


parent to child




version 1 to version 2


X? Continuous mutation may occur in nature. In design, it can be simulated by dividing changes into discrete steps frequent and small enough to appear continuous. Moreover, we can design a system that acquires new features or abilities in discrete steps, by introducing a higher entity or meta system of the kind discussed later, in the chapter on system change and evolution.

Two kinds of "anti-fragility"

The terms robustness and resilience (or some say, antifragility) are used variously with reference to how a system is designed or evolves to survive. The terms are perhaps most simply distinguished as follows. To survive in the face of a changing environment or disruptive input, a robust system handles disruptive or unwelcome events and conditions (think, homeostasis, or immunity to infection). Whereas a resilient system mutates to handle new events and conditions (think (evolution).

Two kinds of agility

Agile system mutation: Agile system development implies a designed system mutates in small ways, and frequently, from one generation to the next. Agile activity system: An agile activity system is one that can handle changes in its environment, without having to mutate when the environment changes.

Three kinds of self-organization

Discussions of self-organization can refer to at least three kinds of change; the last is the most interesting here. Goal-seeking state change: An entity is drawn to one or more "attractor" states and resists being moved from such a state - as in homeostatic biological and electro-mechanical control systems. Self-assembly or growth: This is another kind of state change. An entity grows incrementally by adding more elements or actors to its body.  E.g. the growth of a crystal in a liquid, or a plague of locusts. Self-improvement: This is a kind of state mutation. Self-improvement implies changing from "bad to good" in some way. Ashby's way of thinking about this is discussed in other chapters.