This article picks up where Social systems thinking part I left off. It adds more system theory ideas, like double loop adaptation, and thoughts on applying them to a social entity like a business.

Contents: Complex systems. von Foerster's "second order" cybernetics. Ashby's self-organizing system. Beer's VSM, Luhmann's social communication networks. Culture and identity.

Preface (repeat)

To illustrate my position, I draw triadic graphics that relate four concepts.

In short systems thinkers <create and use> abstract systems to <represent> real systems they <observe and envisage> in real world entities.

Q) Why distinguish systems from entities that manifest or perform them?

I read Ashby as doing this in “Design for a Brain” (DfB) and “Introduction to Cybernetics” (ItC). In DfB (chapters 2 and 14) and ItC (chapters 2, 3, 4 and 6) he used words inconsistently, but in my reading.

• My thinker plays the role of Ashby’s observer.
• My entity (mechanical, organic or social, and describable using countless variables) is what Ashby typically called a real machine, or real system.
• My abstract system (a set of state variables and state transition rules) is what Ashby called a system, or regular system.
• My real system (the realization of an abstract system by an entity) is what Ashby sometimes called a machine.

In some cases, we do speak of an entity and a real system as if they were one and the same.

• A clock is an entity or system whose state advances under its own internal drive.
• An organism is an entity or system that is stimulated to act in response to input events. Every cell is bound to its role in manifesting the system encoded in the entity’s DNA.

But a social entity is far, far more than any of the many systems we might observe it as manifesting. There is nothing like an overarching design or DNA to which all actors in the entity must conform. Even if there was, some human actors (aka agents) would depart from their roles and violate the rules

So, distinguishing systems from entities helps me to make sense of applying systems theory to social entities.

Q) What use is my definition of a system, and the triadic graph?

They help me draw correspondences between ideas in philosophy, systems thinking, enterprise and software architecture. For example

And they help me distinguish systems thinking from other approaches to management science and organization theory.

Complex systems

The term "complex" is widely used and abused. Complexity science seems to me a jumble of ideas - each interesting on their own - but not adding up to coherent whole.

There are scores of different complexity measures, and no agreed measure. You might start here . Or in the context of system theory, you might try this - but I don't recommend it.

Since complexity is a measure we make of a thing with respect to a given description, and there are many different ways to describe the same thing, a thing may be said to have many different complexities. Several abstract systems, of differing complexities, may be abstracted from the behavior of the same social entity.

Does complex mean unpredictable? A simple coin toss is unpredictable. A software application with a billion lines of code can be predictable, meaning that under the same conditions, it will always give same reponse to the same stimulus.

Some define complexity as a property of a system that - thanks to chaos theory - is deterministic yet unpredictable in the long term - because tiny varations in its state today can lead to massive differences in its state after some time has passed.

Some define complexity as a property of a system characterised by learning. Google tells me the brain of an adult fruit fly has tens of millions of synapses connecting 140,000 neurons. It can learn simple tasks and form memories. It can associate an odor with a rewarding sugar stimulus or a negative experience.

Suppose, we model a fruit fly's neural network, as it is today, as a deterministic system. Its response to a stimulus would be unpredictable tommorrow, because its structure evolves as experiences reorganize its synaptic connections. And by tomorrow, the real system will have departed from the system we abstracted today.

Some define complexity as a property of a system characterised by evolution or self-organization, which brings us to "second order" cybernetics.

von Foerster's "second order" cybernetics

In Heinz von Foerster’s second-order cybernetics, the focus is on the observers of a system, and their interactions with it. No doubt von Foerster was a brilliant man. However, I read he saw himself as a dilettante, dabbling in this and that. And I am not sure he added much to Ashby’s cybernetics.

HvF’s approach was more philosophical than rule-based. Arguablyt, his “second order cybernetics” is closer to Churchman or Checkland’s soft systems than to cybernetics.

HvF suggested observers construct their understanding of a system through interacting with it. They may well form an abstract system model from observations of a real system’s state and behavior. However, they can also envisage a system (construct a new abstract system) without interacting with a real system.

HvF suggested observer-system interactions change both the system and the observer. However, observing the motion of a pendulum, or the progress of a card game, does affect the state of the observer; but it does not affect the pendulum or the card game.

Fans of HvF may not realize that Ashby had already discussed how an observer can play a role not only in monitoring the state of a system, but also in controlling and organizing a system.

Ashby's self-organizing system

Ashby's cybernetics is about the control or regulation of a system modeled as a set of state variables and rules that govern state changes.

Ashby is known for two principles for the controller or regulator of a system whose desirable state is represented by a set of state variable values.

• The good regulator theorem: every regulator of a system must have access to a model of that system.
• The law of requisite variety: the model must have sufficient variety to enable the regulator to monitor the performance of the system and steer it in the right direction.

The term "self-organizing system" seems a contradiction in terms, since to reorganize a system (to changes its vairables or its rules) is to make a different system. Maturana deprecated using the term.

Ashby noted the term is used with various meanings, and went on to demonstrate that a system can be reorganized by coupling a "higher" (my word) observer to a lower, organized system.

Note that one actor can play several of thse classical cybernetic roles.

• Consumer - an observer or user who receives output from a system.
• Supplier - an actor who provide inputs that change a system's state or definition.
• Participant - an actor who performs activities within a system.

For convenience below, I use the term “observer”, but this observer can both observe and control (monitor and direct) the behavior of a target entity - meaning it can both consume output from a system, and supply inputs or changes to a system.

Single loop adaptation

The principles of cybernetics (after Weiner, Ashby, McCulloch and others) can be seen in a thermostatically-controlled heating system in which a thermostat (Y) observes and controls a heater in a room (Z). Y monitors the state of Z, and directs the heater to maintain the room temperature in a desired range.

• Y <monitors and directs> Z

Note that Y has no other role in the YZ system than to monitor and direct Z. The actor and role are in a 1-1 relationship.

Double loop adaptation (specific intelligence)

In his “Homeostat”, Ashby added a higher level observer (X) to monitor the state of YZ, and redirect it now and then.

• "The Homeostat was capable of adapting itself to the environment; it exhibited behaviors such as habituation, reinforcement and learning through its ability to maintain homeostasis in a changing environment. It was built by William Ross Ashby in 1948" Wikipedia

When X observed YZ was in distress, it reset the desired temperature.

• X <monitors and directs> [Y <monitors and directs> Z].

The Homeostat gave Ashby a model for the organization of a brain. Will Harwood tells me Ashby was interested in what a minimal controller, making random parameter changes (like biological evolution) could do. But he was aware that, given some success/failure criteria, a controller might make a more purposeful change - change a rule or even a state variable.

Again, in this “self-organizing” XYZ system, X has no other role than to monitor and direct the YZ system; its intelligence is entirely specific that. The actor and role are in a 1-1 relationship.

N loop adaptation (specific intelligence)

Following the same principles, we can design a recursive hierarchical structure, in which a YZ system sits inside an XYZ system, which sits inside a WXYZ system.

• W <monitors and directs> [X <monitors and directs> [Y <monitors and directs> Z]].

We can carry on to build an ABC…...XYZ system, containing a tower of 25 observer/controllers from A down to Y.

Scaling up to a social entity

Ashby wrote of cases where one control system regulates one target system. The coupling of systems in reality is often more complex. Ashby’s law of requisite variety does not mean or imply that:

• variety is all that a controller needs
• a controller knows all state variables of what is controlled
• a controller should maximize its own variety
• there is always one controller and one target entity.

Things get complicated when

• one controller regulates several entities, or
• several controllers cooperate or compete to regulate one entity.

Suppose, in Ashby's Homeostat, we replace the mechanical observer by a human being. On hearing people in the room say it is too hot or cold, that person resets the thermostat's desired temperature range. This human observer is much more than its observer role:

• one person may observe several unrelated systems
• one observer role may be played by several people.

When we scale up to a biological organism or a business organization, we see controllers and controlled entities are related in a many-to-many network rather than a hierarchy.

Moreover, there is a many-to-many relationship between actors and the roles they play. The observer role may be played by actors who also play roles inside the observed system. An actor can step outside their role in a system to change the rules of that system.

For example, during a card game, players may stop the game to change the rules, then continue the game.

Is the brain a controller or controlled?

In writing "Design for a Brain", Ashby hoped that cybernetics could and would be scaled up from his examples to explain the behavior of very large and complex biological and social entities, and the evolution of intelligence.

Ashby thought of the brain as an observer and controller of the body, in a 1-1 relationship. Mind-body dualism is the idea that the mind and body are distinct and separate. It has its roots in philosophy and originated in ancient times. A well-known version of the theory is attributed to the 17th century French philosopher René Descartes.

Ashby's presumed the brain-to-body control system works in a monitor-direct feedback loop, relating one brain to many dumb sensor and motors in the body. By contrast, psychologists now speak of "embodied cognition". Intelligence is seen as distributed throughout the nervous system, even the whole body.

This video reports evidence that the brain-body relationship is a two-way interaction. Split brain experiments show the two halves are different. The right brain is visual and spatial. The left brain is intellectual and verbal. To an extraordinary extent, the left brain justifies what the right brain directs the body to do, by interpreting signals and instantly rationalising what it sees the body has done.

• "The idea of downward causation may be confusing our understanding. What is going on is the match between ever-present multiple mental states and the impinging contextual forces within which it functions. Our interpreter then claims we freely made a choice." Michael Gazzaniga

This video also suggests the brain contains several semi-conscious modules that cooperate and/or compete to direct what the body does next, and a more intellectual module that takes in signals from other modules, tries to make sense of them and justify what body to has done.

Might the same be said of a business? In practice, do business directors direct? Or are they buffeted by external forces and internal events, request solutions to problems, sponsor initiatives suggested by employees, and create stories to please customers and shareholders?

Given the many-to-many relationships described above, I am not sure general intelligence is well described as an observer at the pinnacle of a tower of controller-controlled systems. It may rather be a collection of observers, which sometimes reach different conclusions.

Beer's deference to Ashby's law

When Stafford Beer wrote the “Brain of the Firm”, he saw himself as applying cybernetics principles to business management. The book title is an echo of Ashby's "Design for a Brain" and he made much of Ashby's law of requisite variety.

The designer of a cybernetic control system must identify the critical quantitive state variables of the entity to be controlled. That is challenging when thing to be regulated is a large and complex social entity, buffeted by events, in which human actors act as they choose.

Friedrich Hayek was a Nobel prize-winning economist. He met Beer in the 1970s. The quotes below are from his prize acceptance speech.

• “Fooled into believing that the measurable variables are the most critical, economists propose “solutions” that actually worsen the problem.” Friedrich Hayekc

For sure, business directors should have access to measures of business performance. Business systems should record and maintain variables that describe the state of a business and its environment, and directors should use that information to direct operations.

However, directors cannot rely on Ashby’s good regulator theorem or law of requisite variety, because a business is an immeasurably complex social entity that is far more than any system of state variables we can abstract from it. It is buffeted by external forces outsid, and by evolution in the variables that matter.

And when we scale up Ashby's cybernetic principles to large societies and economies, we hit the issue that control-target relationships are not 1-1, they are manifold, bi-directional, fragile and transient.

Beer's Viable System Model

Beer devised his Viable System Model (VSM) as design pattern for the management of business operations. The VSM stretches the idea of N loop adaptation. It arranges the management functions of a business in a recursive hierarchical structure, in which higher level directors monitor and direct the performance of lower-level directors, until you reach the bottom level where the core business operations, the essential work of the business, is done.

So far, there is a 1-1 relationship at each level of the tower:

• one observer monitors and directs one lower-level system; and conversely,
• one system is monitored by one higher-level observer.

The VSM allows a 1-to-many relationship, in which one management body monitors and directs several operational systems.

Do people find Beer’s VSM useful as a design pattern, because it looks cybernetical? Or because it captures some of Beer’s more general experience of business management?

Beer had worked as a manager for British Steel - a manufacturer that made a single product. The kind of large enterprise that feels the need for enterprise architecture has:

• diverse business operations and customer types
• a mess of operational systems, more or less coupled and synchronized, which may overlap, duplicate, cooperate, compete or conflict
• several management or governing bodies, with possibly conflicting objectives, who are little or not coordinated, and sometimes compete to direct operations.

Given the two contrasting patterns, a) a formally structured and centralized control hierarchy, and b) a many-to-many relationship between controller and controlled entities, is either ideal? Or does the optimal structure lie in between?

Imposing a hierarchy on anarchy

The VSM is a particular design pattern for business management, and if people find it useful – great. Yes, Beer was inspired by Ashby's cybernetics, but I question how well that scales up.

The general question is how far the cybernetic double-loop adaptation model scales up to a n-loop tower in a large and complex biological organism or social organization. I remain to be convinced it scales up very far.

The power of incremental gains - of bottom-up improvement by small incremental trial and error is well established. Models for that include boological evolution, Dave Brailsford’s principle of marginal gains, "cntinuous Improvement” in Six Sigma and Lean, and “sprints” in agile software development.

The notion of a one-to-many control hierarchy has its limits. I see a large and complex organization as many-to-many network of feedback loops. And success as being as much to do with motivating people and capitalizing on their good ideas, as leadership.

The universe is an ever-evolving network of things related this way and that.

• The solar system is a network of interacting bodies.
• The biosphere is a network of interacting organisms.
• An organism is a network of interacting cells.
• A brain is a network of interacting neurons.
• An economy is a network of parties playing customer and supplier roles.
• A human society is a network of interacting people.

The human instinct is to make sense of, and manage, a network by imposing a hierarchy on it.. In the 18th and 19th centuries, scientists found hierarchical structures in nature.

• In cosmology, galaxies contain solar systems, which contain planets.
• In chemistry, molecules contain atoms, which contain particles.
• In biology, there is an abundance of hierarchies; for example, human bones contain osteons, which contain lamellae, which contain fibers, which contain fibrils.

In enterprise architeture (EA), a business as a whole is not so much a system, as a social entity in which a mess of discrete activity systems can be observed. EA is about how a business evolves through the design and planning of changes (under change control) to those many systems. There are many contrasting design patterns. It is common to impose a hierarchy on an anarchical structure, the more easily to understand and manage it.

But in matters psychological and sociological, the temptation to impose hierarchies on human nature can mislead us.

In psychology, Maslow's hierarchy of needs is questionable. He originally proposed you have to achieve level N before you can achieve level N+1. In reality, your achievement at each level fluctuates over time, and you may progress or regress on several levels at the same time.

The first sociologists observed that feudal societies, bureaucratic government organizations, and factories were organized in a hierarchical fashion. But Marx and Engel's attempt to impose a two-level structure human society was naive, and proved disastrous. In reality human society is an ever-evolving, ever-shifting network of people related this way and that.

Some might relate the idea of control in cybernertic system theory to Marxist idea of power in socio-cultural systems thinking. But we are ill-advised to see human society as a collection of groups related in a hierarchy of one-way power relationships, as discussed in this other article .

One sociological philosopher who did see society as a network of communicating individuals is Luhmann, but his system is stateless, and his system theory is an outlier.

Luhmann's social communication network

When the sociologist/philosopher Luhmann discussed the roles played by message senders and receivers, he said something that is true but widely misinterpreted: "Communication is made possible, so to speak, from behind, contrary to the temporal course of the process."

Some have misread this statement to mean that for any communication event, the receiver is the sole arbiter of what a message means. This, in turn, has misled some to believe the dangerously wrong premises of "my truth".

Society depends on people communicating their ideas (right or wrong). Successful communication requires receivers to find, in messages, the meanings that sender intended. Both parties bear some responsibility for ensuring that happens, and for forgiving the other for miss-speaking or misinterpreting.

In practice, when accuracy matters, people talk back and forth until they are sure they understand each other. Or they use the controlled language of a scientific, technical, engineering or mathematical discipline, which helps message senders to be confident message receivers will understand them perfectly.

In Luhmann's theory, a human society is a network of messages about a theme, in which every message can (sooner or later) trigger a recipient to send another message on the same theme.

Why is Luhmann's theory an outlier in the field of systems thinking? Partly because his system is a collection of events rather than entities. But more because his system has no persistent state. And since it is receivers who determine the meanings of messages, observers cannot decide what theme a given message is about, nor even what themes exist.

Luhmann's system cannot be detected, observed, measured; it is a metaphysical abstraction.

Culture and identity

The concept of a "culture" is hard to pin down and exemplify. It is typically defined as the ideas and customs of a community, or the norms of a social entity.

English culture today probably differs as much from the culture of two generations ago as it does from French culture today. And within either country, there are communities with widely different, even conflicting cultures.

So, is a culture a fixed social entity with an ever-changing set of norms? Or is it a fixed set of norms, with an ever-changing set of actors?

Either way, one person may be seen as a member of several cultures, with differing degrees of commitment to each. Or to put it another way, every person has a unique identity, and lives in a unique culture of their own.

Remarls

Social organizations are very unlike biological organisms.

In Zachman’s vision, an enterprise’s architecture is a repository of artefacts that describe the enterprise as a joined up system. (Analogous to the countless engineering drawings that are maintained of a Boeing airplane and all its parts). This vision faded as people learnt how messy an enterprise’s estate of business systems is. And how impossibly large and complex a comprehensive model of all those systems would be (billions of elements).

Beer’s VSM might well be a good and useful design pattern for the management functions of an enterprise. But like Zachman’s framework for EA, it is perhaps more an inspiration or vision than a model to be documented and maintained. If the subsystems are logical functions rather than organization units, then responsibility for functions must be mapped organization units. This mapping makes for an impossibly complex model that no business manager can understand and cannot be maintained.

Further reading

If you want to read this article in the context of a book, watch this space. Related articles include:

• System theories
• A systems thinking triad
• Social systems thinking - part I
• Rules and violations
• Social systems thinking part II