Chapter 5. Systems Thinking
Chapter 6. Developing a Standard Approach
Chapter 7. Vantage Points
Belief is the end, not the beginning, of all understanding.
Johan Wolfgang von Goethe, 1749-1832
Models and Vantage Points
Complexity, as we have seen, can be part real, part perception. To clarify our perception of complexity, we need some strategy which will allow us to cut through the web of complexity and see the essence of whatever it might be that we are addressing. There might be several ways of approaching this goal, and we will address two that seem to be most effective: -
- building models to represent some aspect of a complex whole that interests us
- finding vantage points from which to view the complex whole so that we are, in effect, "outside" the situation, taking a detached, third-person view
To be useful in the context of complexity, models are most likely to represent some aspect of complex, open, interacting systems. Models may be: -
- static models, looking perhaps at groupings, structure and relationship
- dynamic models looking at changes, often over time, within and without of the whole in which we are interested
- minimalist, if they are not to become as complex as that which they seek to clarify
- Vantage points are positions from which to view a complex "something". They include:-
- third-person viewpoints, not unlike the so-called "out-of-body" experiences (OBEs) that some people recount, perhaps after an operation during which they seemed able to look down upon the operation, seeing themselves, the staff, the operating theatre, etc.
- time-compressed viewpoints, rather like time-lapse photograph, or using fast-forward on a video cassette recorder, through which it is possible to see patterns of behaviour not visible from our normal perspective
- hierarchy shifts, particularly upward shifts, which allow us to see the behaviour of a system in the context of wider issues
We will examine these ideas in more detail, since they may hold the key to cutting through complexity.
A model is an abstraction, a simplified representation of a more complex thing: -
- complex models represent many parts, many relationships and much detail. Each component is represented as an approximation only, however, which inescapably means that the whole is many products of many approximations. It is not only difficult to understand such models, but one cannot easily see the flaws which they may contain, nor the implicit assumptions upon which they are inevitably based
- simple models, on the other hand, cannot represent the detail that many consider essential. It is certainly the view of many experienced practitioners that a minimalist model cannot represent enough detail to offer substantial results
- models of complexity seek to reconcile these different viewpoints by carefully modelling the underlying essence of the problem/ issue/ situation
As an example, consider the many battle models, computer simulations generally, that are in common use around the world. Typically, they represent two opposing forces, often with one attacking and one defending. By representing the attackers’ and defenders’ weapons, organizations and tactics as faithfully as possible in the computer, it is hoped to predict the outcome of a potential battle. Such models are used to evaluate different starting points, different tactics, and to persuade purse-string holders that such-and-such a weapon systems is the best buy.
Typically, such models run from some starting condition to an end, with both sides suffering (simulated) casualties. It is in this area that models seem to deviate radically from reality. The models generally fail to represent the changing human condition. In the models: -
- soldiers do not become afraid
- each side fights on until one or other runs out of resources, weapons, etc.
- decisions are infallible, being represented generally by no more than a time delay, rarely (if ever) by mistakes
- training has little or no effect
- combatants and generals do not learn during the combat - their performance is the same throughout
- any belief that combatants might have in the justness (or otherwise) of their cause is irrelevant
We know these omitted factors to be important, so why are they ignored? For three reasons, it seems:-
- first, to include them would introduce a high degree of complexity into the models, since we would have to represent the dynamic behaviour of human beings; and
- second, because we would find it difficult to validate such models against reality, whereas
- it is believed that models without human frailties can be valid.
The notion that models of battles between humans need not represent human behaviour may, on reflection, seem utterly bizarre - but is the current paradigm.
To overcome such limitations, and in the process to cut through complexity, it is evident that a different approach to models is needed. Following topics introduce one such approach, and there may be many others. The approach followed below is formed around the concepts associated with the idea of a "system".
A system model
The concept of a system seems to be fairly simple. We all know intuitively, or from everyday experiences, what a system is. Figure 20 shows one simple representation of a system as a collection of parts which operate together to transform inflows to outflows. We are systems in this context, we convert food to energy, ideas into actions.
Definitions of "System"
We need, sometimes, to be a little more precise about this idea of "system". Here is my definition: -
A system is a set of organized, interacting parts which, when complete, exhibits properties or capabilities of the set as a whole which are not attributable exclusively to any of the parts
This definition is rather deeper than, perhaps, it may seem at first. It contains the following concepts:-
- the notion of sets, that is groups of things which share some common label, purpose or characteristics
- organization, that the set members are not simply scattered around, but that they exhibit form and structure. From this notion emerge systematic concepts
Figure 20 . A simple representation of a system comprising internal, interacting parts which, together, transform inflows into outflows. Implicit in the figure are notions of organization, structure, capacity, work, consumption and dissipation
- the idea of completeness, that the set must be complete for the overall features of the set to emerge
- emergent properties, those features of a system which are exhibited by the whole, but which are not attributable exclusively to any part (e.g. self-awareness from the human brain, flight from an aeroplane, picture from a jigsaw)
- the importance of interactions between the parts as contributing to the whole, rather than just the sum of the parts. From this notion emerges systemic concepts
- the concept of holism, that there are properties of the whole that cannot be reduced
- •the basis for hierarchy, which is perceptible when a set is complete, when that set exhibits emergent properties, and which then allows the parts and their interactions to be considered as an entity at a higher level of perception
An alternative definition, equally valid but rather more abstract, might be: -
A set of interacting entities which, with their relationships, together reduce local entropy
This definition also reveals system concepts, some of them different: -
- structure and organization, without which entropy will not be reduced
- relationships as being on par with entities in terms of importance
- localization, that any system’s influence is a local phenomenon
- metrics, in that entropy can be calculated
Upwards and Outwards - Simple Nested Systems
Starting with the simple model of a system, it is straightforward to develop a further model showing hierarchy. Figure 21 shows a 3-tier hierarchy with some system-of-interest (SOI) which we wish to consider at the lower right. Within the SOI can be seen intra-connected sub-systems at a lower level of hierarchy. On a level with the SOI are sibling systems which exist and interact with each other and the SOI in an environment. Where all the siblings, including the SOI, form a complete set, then the set can be given a new name, called Containing System in the figure. The Containing System exists at a higher level of hierarchy than the SOI, and is itself interconnected to other systems, its siblings, at this higher level.
Figure 21. A 3-tier hierarchy of systems showing a System-of-Interest interconnected with two siblings inside a Containing System. Within the System-of-Interest can be seen contained systems, or sub-systems, which are intra-connected
Simultaneous Multiple Containment
The two models above are not wrong, but they do paint a potentially over-rosy picture. It is possible for a SOI to exist within more than one Container at a time - see Figure 22.
Figure 22. Simultaneous Multiple Containment. A System of Interest can exist within several Containers at the same time
The idea that a system may exist within several Containing Systems at once becomes obvious with some examples: -
- a fighter pilot exists within a flying system, a family system, a squadron system
- a power unit exists within a radar system, a power distribution system, an interference suppression system
- a shoe-manufacturer exists within an industry, an environment, a market
This concept of Simultaneous Multiple Containment reveals potential combinatorial explosion of inter-relationships, and exemplifies one of the cardinal differences between soft and hard approaches to understanding systems and their behaviour. Hard viewpoints seek to simplify hierarchical arrangements into simple vertical arrangements. Human ancestry, in this mode of thinking, would be traced back in time through fathers (or mothers in a matriarchal society). A softer approach might recognize that the genes of the offspring derive from both parents, and will trace the source of genes along all ancestral paths. The soft approach is both more complex and more comprehensive.
Simultaneous Multiple Containment can be managed in many circumstances by identifying the objectives of each Container and showing how the SOI contributes, with its respective siblings in the appropriate Container, to those objectives. So, the fighter pilot contributes to: -
- the goals of (e.g.) the air defence system of which he is part
- the nurture and support of the family, nuclear and wider, of which he is part
- the morale and esprit-de-corps of the squadron of which he is part
Notice that there may be interaction between these Containing System objectives. A pilot may find himself behaving less boldly when he has a wife and children to think about.
Ideas of Closure and Completeness
Another approach to handling complexity is Causal Loop Modelling (CLM). CLM, as the name implies, models in loops of cause and effect. There are very few things where actions are not accompanied by reactions, by feedback of some kind. Push, and you will feel some resistance. Pull, and you will feel some inertia. Soak up ink, and the blotting paper will gorge, reducing the soak-rate. Tell someone to "get a move on" and see the resentment. Introduce a new law and see all the efforts to find a way around it.
CLMs get to the heart of complex issues. The search for direct cause and effect results in minimalist models, and transcends particular disciplines, so that a group of people with quite different backgrounds and viewpoints can come together and agree on the "way things work" via the medium of a CLM.
CLMs are easy to develop. Consider population. Population is increased by births, decreased by deaths. More than that, the rates of births and deaths also increases with the size of the population. This is represented in Figure 23.
Figure 23. Causal Loop Model of Population Dynamics. The solid-headed arrow indicates an inverse relationship, i.e. as the number/rate of deaths rises, the population falls.
Figures 24 and 25 extend the simple model of Figure 23.
Figure 24. Developing the CLM of Figure 23. Disease is added as a special cause of death, the model implying that the greater the population, the greater death due to disease. Food is also shown, as increasing births and reducing deaths.
Figure 25. Extending the previous CLMs further by adding available space. The more space, the more spread out the population, and hence the less the occurrence of stress-related and infectious diseases. The more space, the greater the number of births, simply owing to opportunity and the basic fecundity of the population. Note, there has been no definition of population type. Human? Rabbit? Crocodile?
CLMs and Completeness
CLMs are valuable to establish completeness, especially in a set of ideas. Consider the possible causes of inefficiency in a business. Scratching the head might come up with: -
- Too much waste
- Excessive rework
Following the Causal Loop Model approach might result in Figure 26.
Figure 26. A CLM for improving business efficiency
Figure 26 is best read from the bottom left, Desire for increased Efficiency: this desire enables the perception of a shortfall in efficiency. This perception must result in the establishment of some arbitrary level of efficiency, which might well include some minimization of rework and waste, to accommodate the original "head scratchings". To be achieved, targets require some strategy, followed by a plan, followed by resources to implement the plan. Once the plan is in operation, efficiency may be measured against the targets, and there will be either an improvement or there may need to be a rethink and a different strategy.
In order to complete the CLM of Figure 26, to return to Perceived Efficiency Shortfall, it was necessary to introduce ideas not on our original list. In other words, closing the loop encourages us to complete our set of ideas. We could now produce a much better list for causes of inefficiency: -
- Lack of established efficiency targets
- No strategic plan to improve efficiency
- No measurement of efficiency against targets
- No management of efficiency improvement
- ...not forgetting...
- Too much waste
- Excessive rework
In producing the list from the CLM, we have established a simple, but satisfying, approach to a complex issue which, somehow, does not seem to be quite so vague and complex any more.
Positives from Negatives
Figure 26 and the efficiency example shows one other, sometimes seemingly-miraculous, phenomenon. We humans are especially good at suggesting the possible causes of some malaise or deficiency. For example, most people find it much easier to state the likely causes of poor morale than of high or good morale, or the basis for ineffectiveness than for effectiveness. It seems to be some deep-rooted psychological feature that we find it easier to see the down-side, the negative aspects. The efficiency example, and CLMs in general, can make use of this feature, as follows: -
- Collect all the possible causes of deficiency, defectiveness, down-side, etc.
- Dropping the pejorative terms, form a CLM from the remaining statements.
The resulting CLM will present some idealized system. So, we may turn our predilection for looking at the down side to advantage. Consider truancy from school. Possible causes might include: -
- Lack of parental discipline
- Lack of school supervision
- Dull, uninteresting lessons
- Lack of lessons aimed at particular student needs
- Glamorous perception of "bunking-off", or truancy, in the eyes of some students
Such a list is relatively easy to produce and, particularly with the help of truant students and professional teachers, to extend. The list is full of pejorative terms: "lack of", "dull, uninteresting". Consider now Figure 27. The model contains the elements of the bulleted list above with pejorative terms dropped, and the various items have been formed into a set of loops, each loop containing a closed set of ideas for reducing truancy, the sets combining to create an "idealized" model of truancy management.
One loop proposes closer co-ordination between parents and the school. Schools in the UK operate in loco parentis, in the rôle of parent, once the children are in the school, but it has to be up to the parent to ensure the child arrives at school in the first instance. The loop from "school supervision" to "control of students" shows the school’s responsibility. Both of these loops are about control, but control is a negative concept. Yes, it may be possible to make students attend school, but the very act of making them attend may destroy their interest in learning. The main loop shows the school addressing the main issue - making school relevant, interesting and "the place to be" for students.
This kind of ideal model shows that partial solutions to issues are unlikely to work. So often we see successive governments attempting to resolve complex issues by addressing only one facet, by treating the symptoms rather than the cause. Not only are such efforts doomed to disappointment - they are generally counter-productive in the long run.
Figure 27 also shows the futility of trying to weight the complementary approaches to something like truancy. All the approaches are needed - it is not sensible to pick and choose.
Figure 27. An idealized model of truancy control. Filled-in arrow heads indicate an inverse relationship between the elements connected by such arrows, i.e. if the element at the arrow-tail increases, then the element at the arrow-head will fall, and vice versa.
Pairwise Comparison to Reveal Hidden Structure
Sometimes there are so many factors in some issue that we find it difficult to see the wood for the trees. One way to get a handle on such complex problems is to employ pairwise comparison. As with CLMs, this technique makes use of our psychological makeup. We find it difficult to organize, say, ten things into a prioritized list, and asking five people to produce the list together would cause all sorts of difficulties. On the other hand, we generally have much less difficulty in prioritizing between only two things. It is usually easy to say that A is better than B, or that A comes before B, or that A is more important than B. If we can take a list of many factors and compare them two at a time, it is possible to integrate all the comparisons into a network. An example will reveal all!
What are the attributes of some proposed new airliner? Suppose they are: -
- Good freight capacity
- Ease of conversion - freight to passenger
- Good passenger capacity
- Minimal maintenance cost
- All weather operation
- Good short-haul economy
- Passenger appeal
- Operating economy
- Good resale value
By asking the questions, "does Attribute A strongly contribute to Attribute B, or is it the other way around, or to they mutually re-inforce, or is there no relation?", it is possible to create a matrix of relationships (Figure 28) from which a tree of relationships may be formed.
The attribute enhancement tree at Figure 29 can be drawn directly from Figure 28 using notions of reachability and transitivity illustrated in Figure 5 shows that all of the attributes of the airliner are connected through the concept of operating economy, which enhances customer appeal provided it results in lower fares, and will contribute to the eventual resale value. (Note that the tree employs the concepts of reachability and transitivity - see Figures 4 and 5.)
Figure 28. Matrix of Contributions - printout from author’s ©CADRAT computer tool. N.B. some of the titles have been truncated by the program
Figure 29 represents a simple case, and the answer might have been reached easily by trial and error. Using the pairwise comparison technique enables large numbers of factors to be mutually-related, in situations where trial-and-error methods would be out of the question.
Figure 29. Attribute Enhancement Structure, drawn from pairwise comparisons in Figure 28
Developing a Standard Approach
To have conscientiously studied the liberal arts
Refines behaviour and does not allow it to be savage.
Ovid, 43 BC-AD 17
With all the potential for complexity, and all the variety we see about us, it would be valuable to have a standardized way of approaching complex issues and situations. The models in previous chapters are leading towards that goal by allowing us to view the world from a rather simple perspective. In many ways, much of the complexity about us can be reduced by organizing our perceptions into a hierarchy of systems, with complexity of detail being masked by viewing from a higher perspective - by standing back, and viewing from a distance.
It may not be quite that simple, of course. If we observe carefully how people approach the creation of new concepts, or tackle some complex issue, a pattern of behaviour emerges: -
- First there is a process of disordering, of increasing entropy in order to encompass a richer variety. In everyday terms, issues seem to become more complex when we scratch below the surface, and we soon find ourselves embroiled in a wealth of detail
- Second, there is a process of connecting up, or linking, the profusion of varieties to establish their relationships
- Third, once the varieties have been linked in some way, they may be grouped into sets or clusters, so reducing either real or perceived entropy, or both
This three stage process up and down the hierarchy/entropy tree occurs in many walks of life. It is as though, to reach a better state, we have to disorder the current state, throw all the cards in the air, and then rearrange them into a new, hopefully-better state.
Given simplifying systems models and a standard, three-step process, might it be possible to approach complex issues in a straightforward manner? For this to be true, it is necessary that: -
- many different situations can be seen from similar perspectives
- we find some standard systems model to accompany the 3-stage process
One such model is the Generic Reference Model (GRM), which describes any system, and which has three parts, Form, Function and Behaviour. Why the three parts?
- some systems have only form, e.g. the solar system for which it may be difficult to ascribe purpose without resort to (arguable) transcendental causes
- some systems have form and function, e.g. human heart, which is both highly organized and which clearly has the purpose of pumping blood around the tissues of the body
- some systems have form, function and behaviour, e.g. a management group which has structure/organization, which fulfils a rôle, and which undoubtedly exhibits behaviour if only because no two management groups perform identically.
So, although the GRM purports to address any system, the user of the GRM is required to select the appropriate parts of the model to suit the context. Moreover, the GRM looks at the insides of a system only. For instance, a GRM view of a human would see skeleton, organs, central nervous system, senses, response to stimulus, etc., but would not reveal external appearance.
The Generic Reference Model
Figure 30. The Generic Reference Model
The Generic Reference Model is shown at its highest level in Figure 30. It represents those parts of a system which correspond to Being or existing, Doing or functions, and Thinking or behavioural response. The GRM does not explain why a system exists, functions or behaves, but shows what features must rationally exist for properties and capabilities to arise. For example, the model does not predict particular behaviour; instead it indicates what features must be present in any system which can respond adaptively to different stimuli, i.e. those features which go towards the management of behaviour. Having said that, the model is based on extensive research into a variety of systems to find those features which are common to them all.
Generic Reference (Function) Model
Figure 31. The Generic Reference (Function) Model
Figure 31. shows one version of the Generic Reference (Function) Model (GR(F)M). The model shows the three subdivisions interacting with each other as follows: -
In principle, in the management of a mission (an undertaking to achieve some goal) can be organized under the following headings: -
- Information management: the collection, organization, storage manipulation and presentation of information
- Setting Objectives based (in part) on the Information
- Formulating Strategies and Plans to achieve the Objectives
- Executing the Plans
- Co-operating with others in the Environment in the Prosecution of the Plan
Figure 32. The Generic Reference (Function) Model showing Mission Management, Resource Management and Viability Management features. The System of Interest is shown as interacting with two external environments, an Operations Environment and a Resource Environment. Note the central nature of Viability Management, represented by the heavier arrows. Reproduced by permission of John Wiley and Sons
For any system to persist in its pursuit of the Mission, it must remain viable. Viability is founded on five pillars: -
- Synergy, co-operation and co-ordination between various parts within a system to produce some external effect
- Evolution, the (slow) adaptation of a system in line with/response to changes in its Environment
- Survival, management of threats, generally by avoiding detection, self-defence or damage tolerance, or some combination of these
- Homeostasis, the regulation within a system to provide the internal parts with a suitable Environment
- Maintenance, detection, location and replacement of non-functioning parts
In principle, the management of resources is encompassed under the following headings: -
- Acquiring Resources needed to maintain Viability of the System of Interest
- Storing Resources until needed, either explicitly or simply by virtue of slow transit
- Distributing Resources to the point where they may be employed
- Converting Resources from their raw form into some more useful form for maintaining system Viability
- Disposing of Residue, Waste, dissipation or anything not wanted in the system
These various parts may be considered acting together along the lines shown in Figure 32.
As the figure shows, the various parts of the GR(F)M cannot really be treated individually, since they are mutually interdependent. Nonetheless, it is practicable to identify individual parts of a system first, so long as they are subsequently intra-connected into a whole.
Generic Reference (Form) Model
The Generic Reference (Form) Model also arises under three headings, as follows: -
- Boundary, or the periphery of the system of interest, through which interactions with other systems must pass. The boundary may be hard or "fuzzy"
- Sub-systems, those systems contained within the boundary
- Connections that enable intra-actions between the sub-systems and interconnections with other, external systems
- Relationships, perhaps by location or through some third factor, which falls short of connection, e.g. two parts both contributing to centre of gravity
- Cohesion, influences within a system which bind the parts together. For stability, these must balance dispersive influences
- Dispersion, influences within a system which encourage the parts to separate
- Environment, that which mediates the Cohesive and Dispersive Influences, i.e. enables the effect of those influences
- Power, the summation of the various discrete sources of power within a system, giving the rate at which energy might be convertible
- Capacity, the amount of energy and/or stuff which can be stored within the system. With power gives the total energy within the system
- Redundancy, the replication of parts, often to manage failure or damage
The whole form model may be represented symbolically as in Figure 33.
Figure 33. The Generic Reference (Form) Model
As might be expected, the Generic Reference (Behaviour) Model (GRM(B)) is a little more complex than the other two parts, although nowhere near as complex as a full model of behaviour might be - remember, we are dealing here with the management of behaviour, which is altogether easier to address than behaviour itself.
Figure 34. The Generic Reference (Behaviour) Model
As with previous parts of the GRM, the Behaviour Model features may be tabulated: -
- Tacit Knowledge, low-level knowledge of simple everyday things, e.g. that things fall under the influence of gravity.
- World Models, perceptions of how the world is populated, within which representations of things behave according to Tacit Knowledge
- Cognition, recognizing and interpreting sensations, external and internal to the whole system
- Beliefs, mutually consistent pre-conceptions of how the world behaves/should behave
- Rôles, categories and pre-conceptions of rôles that should/do exist
- Values, Ethics, Morals, pre-conceived views of what is "correct’’ or "proper"
- Experience, that which has been learned through Nurture, training, situations and events. May lead to predictable behaviour - "Second Nature"
- Constraint, those practicalities or perceptions which limit category and/ or range of behaviour
- Nature, inherited behavioural features such as instincts, character, etc.
- Activation, the "drive" accorded to a selected behaviour. May be associated with Nature
- Motivation, the strength of intent. May arise from experience. Two forms: achievement; conformance
- Stimulation, the product of Motivation and Activation in support of Behaviour Selection
These features of the GR(B) M may be presented symbolically as in Figure 35.
Figure 35. The Generic Reference (Behaviour) Model
One other point about the models so far. We have seen that Behaviour Selection is essentially some contest or trade-off between Nature and Nurture. In the lifetimes of social species such as Homo Sapiens, social evolution occurs. Social evolution is the pattern of ideas, values, behaviours, that we learn from the moment we are born. Social evolution is the manners and lessons (for good or evil) which are drummed into each and every one of us until they become Second Nature. And that is the connection. Our Belief Systems are the seat of our Second Nature - beliefs, values, rôles, ethics, morals, these become our Nature, not instead of our Darwinian heritage, but as a layer on top of that heritage, not separate from it, but modifying it. If we can accept that we have a Second Nature, then we can correlate Darwinian and Social Evolution models, comparing Nature in one with Second Nature in the other
Mission and Behaviour Models - Interactions
Although the parts of the GRM have been presented separately, they are not separate and cannot be considered so with any validity. Overlooking such mutual interactions is the hallmark of reductionist thinking and leads to a gross under-estimation of complexity. It is just such overlooking of interactions which allows people to convince themselves that battle simulations can be useful when they omit human behaviour.
Interactions are very strong between Synergy Management and the elements of the Form Model, for instance. Survival often invokes redundancy to tolerate damage or defect, as in the human design where we can all survive quite happily on only one kidney. To illustrate the importance of interactions between parts of the GRM, consider the relationship between the Behaviour Model and just Mission Management (part of the Function Model) illustrated in Figure 36.
Figure 36. Interactions between the GR(B)M and Mission Management, part of the (GR(F)M.
In the figure, Mission Management is shown along the top, starting at the left with Collect Information, and finishing at the right with Co-operate with Others. A simplistic view might consider that operations management, say, was a matter of collecting all the necessary information, working out a plan and acting on it with other team members - i.e. Mission Management. The figure shows differently. Information is interpreted, if recognized at all, through a filter of World Models and Tacit Knowledge, both based on Experience and influenced by Belief; the idea of objective interpretation of information is immediately challenged. Similarly, acceptable Strategies and Plans are those confined within the boundaries of Belief and Experience. Other strategies will either not be thought of, or will be rejected out of hand as inappropriate - not because they are necessarily "wrong" but because they do not fit pre-conceptions and perceptual constraints.
Belief can be seen from the figure to be at the very heart of Behaviour Management - we shall return to Belief Systems later, but for the moment, consider the following quotation: -
Could it be that we substitute our Beliefs for rational thought in an effort to cope with the complexities of the world around us? Beliefs certainly allow us to size up situations rapidly - and perhaps wrongly.
The Annex to this Chapter contains an example of the GRM, used to support the development of a complex design concept for a manned base on the far side of the Moon. Readers may, if they choose, skip over the Annex or read it at their leisure.
Thinking in Loops
The Generic Reference Model proposed what one might find inside any system. That is fine so far as it goes, but it offers a rather static view; the interactions between any system and its environment determine the dynamic behaviour in which we are often just as interested.
Figure 37. A system, any open system, is shown interacting with others as it transforms inflows to outflows. Since systems exist in a never-ending web of interactions, some of those interactions connect outflow inevitably back to inflow, probably via some transformations.
Any system’s outflows inevitably interact with other systems in the environment to affect its own inputs - see Figure 37. Introducing a new system into an existing set disturbs all of them, existing set and newcomers alike. Subsequent dynamics may militate against or for sustaining the new system, and could be inimical to the set. In the general case, as the figure suggests, predicting the behaviour of the new set may be very complex, and yet we commonly launch new systems into existing environments without too much thought, witness the current social experiments in the UK with schools, hospitals, railways, utilities and many more. I call them experiments because, as far as one can see, they have been neither tested nor thoroughly modelled prior to introduction and, despite the protestations of politicians to the contrary, the outcome of such careless changes is really quite unknowable. Note, I do not propose that the outcome need be "bad", simply that we have no way of knowing, since the changes were driven by political and economic ideology, not as a result of some careful trial and subsequent correction of deficiencies.
One way to explore such situations, and therefore to have some prospect of anticipating the unexpected, is to use models, to place the new system into the existing set of systems, but in a computer simulation before doing so in the real world, to reduce risk.
Thinking in loops, as in the figure, encourages: -
- completeness of sets of interactions;
- identification of the "relaxation condition" of multiple interacting systems, the likely eventual state of the set
When using causal loop representations: -
- open interacting system loop models tend towards a stable state - static or dynamic - representing some minimum lattice energy condition
- the outcome of model runs may be counter-intuitive, and so...
- there is potential for predicting counter-intuitive behaviour of complex systems
Counter-intuitive Dynamic Behaviour
As an example of counter-intuitive behaviour of a system, consider the models of Figure 38. The two models are very simple, and are identical in shape and general operation; they are concerned with the supposed value for money provided by training and education respectively. On the left, training is presumed to be a short affair lasting some 3 months only, resulting in staff trained to a specific skill. That skill lasts for some 24 months, due to staff being posted to other duties, the skills becoming redundant as new facilities are introduced, etc. On the right, the same staff population is given education which, being broader and seeking to instil understanding rather than specific skills, takes longer - 12 months in the model. On the other hand, that understanding is assumed to last longer, too - for 8 years in the model. So here is the dilemma: which model results in best value for money as seen from the perspective of the organization; which costs the most; which gives the greater population of either trained or educated staff at any time?
Figure 38. Two models representing training and education respectively. In the models, training is assumed to refer to short-lived, specific skills, while education refers to longer-lived, but rather more general levels of knowledge and understanding. questions arise about the relative cost-effectiveness of the two approaches for middle-ranking staff in an organization over an extended period of time
Figure 39. The graph shows the number of educated and trained staff predicted from the closed loop models above.
Figure 39 shows the numbers of educated staff to be double that of trained staff, once initial turbulence has settled out of the model. Other results show that the cost of education is only half that of training, long-term. All other things being equal, then, education gives four times the value for money, since it results in twice the numbers for half the cost. Of course, all other things will not be equal. Any organization needs both trained and educated staff. But it is amazing how often such spurious arguments as that about the cheapness of shorter-duration training vs. that of longer duration education can, and do, arise.
Finding the Common Exchange Medium
In any complex system there are certain "fluxes" that flow through its various parts. Finding the flux is important, but can prove difficult, and is best explained by example.
What is it that runs through a car industry? Many people might say "money" and they would not be wrong, but money flows through every business, and choosing money may not result in a good differentiator for the car industry. Besides, money tends to flow in the opposite direction to "things". Instead, consider the flow from resources through manufacture, assembly, sales, usage and, finally, scrap/recycling. Keeping the flow going is the one essential, unifying theme to which all parties in the industry, be they suppliers, assemblers or scrap merchants, can subscribe.
What is it that is common to members of a political party? Is it, perhaps, a shared Belief System, an Ideology. Is seeing the world through the tenets of "the one, true Belief System" the essential, unifying theme?
What is common to the participants in an air defence system? The continuing provision of weapons to the point of delivery against an encroaching enemy is, perhaps, the single unifying theme.
These unifying fluxes are important to identify. Commercially, there tend to be two flows running in opposite directions. In the car industry, for instance, if supplies run clockwise around a system from second to first tier suppliers, to assemblers, and so on, then money runs counter-clockwise around the same system.
Thinking in Straight Lines - Occam's Razor
Sometimes, when there is insufficient information, or when the situation is too complex to even start sensibly, then some way of cutting the Gordian Knot becomes necessary. William of Occam (c.1280-1349) gave his name to one such method - Occam's Razor: entia non sunt multiplicanda praetar necessitatum - beings ought not to be multiplied, except out of necessity. There are many interpretations of William of Occam’s famous dictum. Mine is as follows. ;
If there exists a number of potential paths, routes options, etc, between which it is necessary to choose, and if there is insufficient data upon which to make a judgement, then choose the path, route, or option which contains the least steps or parts.
The reasoning is simple. For each step in any process, there is an associated degree of risk or uncertainty. The uncertainties for each step multiply together, making the path with more steps more uncertain. Given no other data about the actual uncertainties, then, the only sensible judgement is to choose the option with the least parts, steps, etc. The dictum also results in choosing the simpler in favour of the more complex. We see today many large-scale project failures and cost overruns which would have been avoided if Occam's Razor had been applied. Interestingly, the Japanese seem to have embraced Occam in their approach to Lean Production, where the cry is to streamline and simplify rather than to cope with ever-more complexity as we in the West seem intent on doing.
Annex to Chapter 6
Example of GRM in Action
The Generic Reference Model may be used in several ways: -
- As a guide to identifying the parts of a complex system
- As a check to see that the design or description of a complex system is complete
- As a basis for rapid design of a new system, especially one without precedent
Designing unprecedented systems is by far the most fun, so consider a new base, to be established on the far side of the Moon, with the dual purposes of: -
- Housing and operating a deep space radio telescope
- Supporting missions to the solar planets, both for outgoing missions and particularly to acclimatize returning missions
The design may be approached by presenting the GRM in tabular form.
Table 1. GR(F) Model Table for Lunar Base
Table 1 shows the elements of the Function Model, alongside spaces where the designer may write in the functional components of the lunar base which are to provide/create/enable the function. So, information is to be acquired and handled by the Commcen (Communications Centre) and by the Image Processing Centre, evidently for sensor images. Similarly, Objectives are to be established by CPRM (Contingency Planning and Resource Management). Continuing to the second column, Synergy is the responsibility of the LDSC (Lunar Deep Space Centre) Management. Under Resource Management, there is the Management of Conversion, covering both OJT (On the Job Training of staff) and Atmosphere Conditioning, showing that resources include all resources, human and material.
Table 2. Behaviour Model for Lunar Base
Table 2 shows some of the Behaviour Model, similarly filled in with functional components. Using threat analysis techniques provides further functional components. The following list aggregates all the functional components: -
(The functional elements above represent individual, isolated, disconnected variety - we have developed a complex, disordered view as the first stage in a three-stage process see page 37.)
Closely associated functional components can be co-located in the development of a physical design from the functional components - see Figure 40, which uses the genetic algorithm method illustrated in Figure 19.
(Figure 40 represents the second stage in the three-part process - connecting the variety to show sets and systems, so that we may move towards the third step - reduced entropy, as follows.)
The core design shown in Figure 40 can then be connected to all the necessary and associated external inflows and outflows, required and un-required, as shown in Figure 41.
(Figure 41 represents the simplified, reduced-entropy final state, the third stage in the three-stage process of creating a reduced-entropy design concept.)
This example shows only one way in which the GRM is useful. The principal alternative use would present itself when seeking to understand, and establish credibility/completeness of, a system. For instance, auditing a system design, or reviewing a system to understand the cause of failure, or seeking to improve some limited capability.
Figure 40. The Conceptual Organization of an Unprecedented Lunar Deep Space Centre
Figure 41. Connecting up the Lunar Deep Space Centre Design.
It is not best to swap horses while crossing the stream
Abraham Lincoln, 1809-1865
Archimedes, 287- 212 B.C., is reputed to have said: "Give me somewhere to stand, and I will move the Earth." (Pappus, Synagoge, ed. f. Hultsch, Berlin, 1876-8, VIII, 10, xi). Archimedes had reasoned that he might be able to weigh the Earth if he could balance it on a fulcrum, like a ball on the end of a lever. In arriving at this idea, Archimedes had already addressed the most difficult part of weighing the earth - projecting himself to a "third person perspective" from which he could see the problem. Many of his contemporaries could not see the issue so simply.
This capturing of vantage points is a special skill, which can be practised and improved, There seem to be several vantage point categories: -
- spatial, like Archimedes
- temporal, as in time-lapse photography, speeding up time to reveal underlying features not evident at normal speed
- hierarchical, as in looking across a whole economy or industry, rather than at a particular part
- analogical, by comparing understood with less well-known
Third-Person Vantage Point
Most vantage points seem to be third person views. I recently came across an example of a third person vantage point at a company that invents virtual reality arcade games. One of the games represented Formula 1 motor racing, and the software was being tested by one of its authors. He found himself, in virtual reality, approaching the first bend when a car in front reared up and descended on him. Although this was only a computer simulation, and although he had written much of the simulation himself, he nonetheless felt quite afraid. As a result, the software has been rewritten so that, in the event of imminent danger to the game player, he or she is snatched to a third-person viewpoint. In the case of motor racing, the "at-risk" player is snatched to a trackside view and sees the crash happening as though to someone else.
Temporal Vantage - Project vs. Functional Organization
Figure 42. Project versus Functional Organizations
Figure 42 shows two archetypal ways of organizing a typical manufacturing company. One approach is to consider that the organization is made up, essentially, from a number of projects, running in parallel and with little or no cross-feed. The second approach is to organize around major functions that have to be performed, such as design, development, manufacture and commissioning.
Comparing these two archetypes is rewarding. The project organization has clear goals and each project team is aimed at its respective goal. It seems to be, potentially at least, lean and efficient. The functional organization, on the other hand, quite often turns out to be more effective, because the people become expert at their functions, and can carry over learning between projects.
Figure 43. Organizational Dynamic Stability
Organizations typically switch between project and functional organizations periodically, and at first it seems hard to understand why. Looking at an organization over several cycles of functional/project change, however, reveals a rationale for the switching which may not be evident even to those in the organization.
During the functional phases, the functions learn and grow and become more expert as an organization. This results in increasing effectiveness, but also in increasing cost - which comes to the attention of accountants. Now accountants have a simplistic view of such things, and they blow the whistle because they do not understand, and cannot measure, effectiveness. All they see is increased overheads. The subsequent switch to project organization makes them happy, but unfortunately for the organization, its ability to learn is virtually destroyed by the imposition of sharp internal barriers between the projects, which allow neither ideas nor people to cross. The organization is now efficient, however. Passage of time sees two factors emerge:
- People working in parallel projects start to talk, exchange ideas and experiences, and cross links between projects become institutionalized
- The organization finds new business more difficult to acquire as it is unable to compete with organizations that have kept on learning, and people recognize the need to become more expert
After a suitable interval, the company switches back to functional organization, with all the accompanying disruption, and the process of organizational learning recommences.
Figure 43 shows what is going on. The organization is surviving by increasing its average knowledge and capability (effectiveness) while holding its average overheads in check. Switching between project and function is the stable state! Of course, there may be a better way to grow a company...
Temporal Vantage - Successive Fashions
One advantage of the temporal viewpoint is that it enables one to see fashions and fads more clearly. Consider outsourcing, the idea that a company should employ outside contractors to undertake virtually anything except its core business. Part of the approach is so-called Market Testing. We know about market testing; that is what those people who come up to you in the street with some new product are doing, yes? No. One of the ways to recognize fads and fashions is by the use of meaningless titles e.g. market testing. In this case, market testing is the process of offering a company’s internal functions to the market place so that others may bid for it. For instance, a company making shoes may decide to outsource its information systems (IS), and would present a tender document to various IS companies in the hope that a bid would undercut their internal costs for the same activity.
Now, you would go far to find a riskier strategy than this one. Not only are you letting another company handle your "crown jewels", but you are burning your boats in the process, i.e. once the step is taken, it would be very expensive and difficult to retrace. Moreover, once the IS company has your business, you become dependent on them - which can prove an uncomfortable position. Last, but not least, there is inadequate capacity in the IS sector to cope with a sudden increase in demand.
So, is outsourcing sensible, good business or fashion? Let us look at recent fashions, using our temporal viewpoint to telescope the years: -
- Mid eighties - corporate diversification is the panacea as a hedge against changing environments. Its crie de coeur might have been "flexible portfolios for survivable revenues"
- Late eighties - consolidation (or divestment, a more sensible term) became the panacea when diversification gave companies new responsibilities they did not understand and failed to manage. The consolidation crie de coeur? - "stick to core business, it’s what we understand"
- Early nineties - outsourcing or market testing, the panacea by reducing right down to core business, whatever that might be. Again, a special crie de coeur- "get rid of overhead drivers, outsiders/specialists can do things better, we are only good at our core business...by the way, what is our core business?"
Seen in context, outsourcing is just the latest in a long line of unsuccessful fashions. Not that outsourcing need be a bad thing. There are some prerequisites for successful outsourcing, however, as practised successfully around the world.
The Outsourcing Dilemma
To explore this topical issue further, consider a car manufacturer. What is the core business? The company could: -
- subcontract manufacture of engines, bodies, electrics, hydraulics, instruments, seats, everything
- outsource marketing, corporate IT, accounting, selling/franchising...
- employ subcontractors to be the actual assemblers - where is the limit?
Looking from this "piecemeal" perspective offers no answers - a "systems perspective" is called for, and may be found by adopting an analogical viewpoint.
To examine IS outsourcing let us use the project versus function model developed above. To make use of this model we need to move up in hierarchy level and look from a third person perspective on a set of interacting industries. We can then consider whether successive fashions are more a group-subconscious switch between effectiveness and efficiency, gaining knowledge, then trimming fat.
To acquire a rational view of outsourcing let us look around world for other forms of outsourcing, examine environments, successes, failures. The global car industry has outsourced for years; it is popular in Sweden and Japan, yet unpopular in the USA. The reason for the differences is not hard to find; outsourcing depends upon the existence of a robust small industry environment, less evident in US.
From this and other evidence, we may make the following (rather obvious) deduction: the ability to outsource is determined principally by the environment and its ability to sustain and support outsourcing, rather than by corporate intent. If fashion dictates that outsourcing shall happen, then the IS environment - which takes many years to evolve - may well be swamped.
Is it possible to predict the outcome? It is always dangerous to predict the future, but we do have the function/project model to guide us. In many ways, outsourcing parts of a company’s activities is analogous to switching from project to function, but at one hierarchy level up. That suggests strongly that: -
- Outsourcing corporate IS will stimulate growth of IS companies
- IS companies will
- - learn,
- - become effective,
- - grow fat,
- - become less efficient
- Corporations will regain their own IT after a suitable interval, i.e. switch back from function to project, and a switching cycle will ensue
Of course, there is more to it; for a start, we have no idea of the length of the cycle, so predicting when the cycle will repeat is difficult. On the other hand, we might consider that project/function switching is a stable state in which companies survive and grow. Perhaps outsourcing /insourcing is a dynamically-stable state. There may be a better way to grow a company...
Part 2, Getting a Handle on Complexity, has presented a variety of ways of looking at complex systems and situations. All are designed to get to the essence of a situation which can be expressed in a simple, visible, easily-understood manner. Cutting through the veil or web of complexity is never easy, but - like most things in life - does seem to become easier with practice. Following parts will make use of these ideas.
Fractal 5. 3-D view of Mandelbrot
Last updated: Sep 2002