.png)
Systems thinking in the analyst's toolset
- Igor' Arkhipov
- Sep 16, 2020
- 9 min read
Make sense of complex systems and relationships to drive better analysis outcomes
Many years ago, back in my university days, my professor in organisational analysis would always say: “Stop this nonsense about individual departments, units, and processes — and start thinking in systems”.
At that time, it didn’t really make sense to me, what does it mean to think in systems? It took a few years and some real-life projects to appreciate the advice the old man had given me… but let’s start from the beginning.
What is a system?
A “system” is a very ordinary word that we use daily to describe sets of related items, be they as huge as planets (e.g. solar system) or as tiny as molecules (e.g. the nervous system in a living being). It can cover sets of homogeneous items (e.g. measurements for a coordinate system) or very diverse elements (e.g. a computer system consisting of hardware, software, networks).
There is, of course, a formal definition for what a system is [1]:
“A system is a group of interacting or interrelated entities that form a unified whole”
However, just like with many other formal definitions, it is ultimately correct and absolutely useless without further explanation. :)
There is one thing that is important about systems. Being sets of separate things — software, hardware, people, chemicals, etc. — they interconnect in such a way that they produce their own unique pattern of behavior over time. These patterns describe the responses of the system to external stimuli.
However, it is never the external force that makes a system behave in a certain way. Any system has its own internal structure that responds to an external event, resulting in a behavioral pattern that can be observed.
I remember reading a great example of this in a book called Thinking in Systems. A primer, by Donella Meadows [2]. She used a Slinky toy as an example.
A Slinky is a precompressed helical spring toy. It can perform a number of tricks, including travelling down a flight of steps end-over-end as it stretches and re-forms itself with the aid of gravity and its own momentum, or appear to levitate for a period of time after it has been dropped.
Imagine playing with it — you move your hands, this forces the toy to change its form and flow between the hands. However, these are not the hands that made it behave in a certain way.
The hands that manipulate it trigger the behaviour patterns that are latent within the structure of the spring.
Donella calls this a central insight of systems theory — the system, to a large extent, causes its own behaviour! An outside event may unleash that behaviour, but the same outside event applied to a different system is likely to produce a different result.
What makes a system a system?
It’s important to understand what keeps a system together, for it is not just any collection of elements — there must be some kind of glue that holds them together. Let’s consider the purpose of the system as the glue.
A system that does not have a common goal for its elements eventually breaks down into independent systems with their own goals. This happens often with groups of people who may be brought together by an external force (e.g. have been invited to the same party) but have nothing in common. They tend to break out into smaller groups based on their interests.
Similar things happen in the corporate environment, when individual groups of employees form their own closed bubbles (information silos) with their own agendas. When this happens, hidden agendas come into play. It is a signal that the business does not have an inspirational vision shared by its members; and it endangers the company’s future [3].
When the system has a goal, it develops a structure that keeps the elements together and helps it achieve that goal. Think about the organisational structure in your company or the organisation of molecules in cells. This structure is something that enables the system to display a certain behaviour that brings the system closer to its goal. These are the three important aspects of any systems: a goal, a structure, and a behaviour.
Analysing systems
When a system is put together, magic happens: it starts to display characteristics that none of its elements have. These characteristics are called “emergent properties”. A person can experience love, but none of the parts of their body can love; take a person apart, and the phenomenon of love will be lost.
There are many characteristics like that when you examine systems — characteristics that cannot be called properties in their traditional meaning. They cannot be analysed or measured separately from the system as a whole. Moreover, they cannot be deduced from individual properties of the parts.
This raises an interesting question: can we use traditional analysis techniques, such as decomposition and induction, when talking about systems? The answer is “no”.
But does it mean we cannot study systems? The answer again is “no”.
It just means that when we talk about systems and system analysis, a different method will need to be applied. System analysis, unlike traditional reductionism analysis [4], embraces the fact that systems are complex and cannot be simplified.
It does not start with breaking a thing down to then synthesise the insights. Instead, it starts with taking a step back and looking at a system in the context of other systems (i.e. a super-system). If the system under analysis is considered an atomic black box in a bigger environment, what is its purpose? What is the glue that keeps it together? What are the behavioural characteristics of the system that we can observe? And what are the external triggers that cause that behaviour?
This approach is called systems analysis. It starts with synthesis (taking a step back, embracing a bigger picture) and goes into analysis later.
Once the purpose and behaviour are understood, one can dive deeper into how the system is organised. Systems modelling is a separate art on its own, but the core to it is to look for feedback loops in the system: structural elements that result in the system showing a complicated behaviour.
A feedback loop is an occurrence where the output of a system amplifies the system (positive or reinforcing feedback) or inhibits the system (negative feedback).
A simple example is bank deposits. You deposit funds to get some annual interest. Every time you receive the interest, it increases the base amount of money, increasing the interest — this is a positive feedback loop. Or an even better example (from the Donella Meadows’s book): when a kid pushes another kid, that kid pushes back causing the first kid to push harder. This is a typical example of a reinforcing feedback loop that leads to chaos!
Modelling systems
Understanding the cause-and-effect connections between measurable parts of the system is the key to modelling it. System modelling distinguishes the two key elements of the system:
a stock,
and a flow.
A stock is a measurable element of the system. This is how Jamshid Gharajedaghi defines it in his book Systems Thinking: Managing Chaos and Complexity [5]:
“State of beings, state of a variable, things that accumulate, are measured, or are quantified, for example, customer base, market potential, and cash in bank accounts. It can be constraints (carrying capacity), buffer (slack), or inventory that accumulates on store shelves, in transport trucks, and in warehouses. It can also be used to represent conveyors to define transit times and/or delays. Basically, anything that you can measure and that can change its value over time.”
A flow is an action that changes the value of a stock over time.
It “represents action over time, the beat (periodic repeats of a predefined set of operations), and the rate of change that is the action (minute, hour, day, month, or year). It is the means to change the state of the variables under consideration (adding or subtracting). It also defines activity, things in motion, earning, spending, getting angry, becoming frustrated, learning, selling, buying, hiring, and firing.”
Any time a level of stock changes, it is the result of either an inflow or an outflow.
Let’s have a look at our previous example with deposits. Here is what the diagram will look like.
This is a typical notation for operational modelling of systems. A box represents a stock, a valve represents a flow, an arrow shows the direction of the flow, a circle represents external factors that may affect the flow rate.
Positive cause-and-effect connections are marked with the letter “R” (it stands for “reinforcing) or the “+” sign. This model was built in a specialised tool for systems modelling called Stella [6], but you will see the same or similar notation implemented in any other tooling for dynamic system modelling.
Building such a model of causes and effects allows you to understand and visualise the dependencies within the system, and to predict how a particular change will affect the whole system.
What if a certain stock reaches a certain level? What if one of the flows starts to happen more or less often? What if we change the relationship between certain elements of the system? All of these can be potentially answered if you build a dynamic model that allows you to do imitation analysis of the system. Dynamic analysis helps you not only understand what is wrong, but also to learn how we got there and which actions are more likely to help.
Imagine that for our bank account modelling we also introduce monthly savings that are added to the deposit (some random amount of money we save every month within known boundaries). This will be another inflow into the deposit. We can also include the mortgage account that constantly draws money from the deposit account, and also calculates negative interest on the mortgage. So now we have two feedback loops in the model: the more money you have in the deposit, the higher monthly interest you gain; the higher the mortgage amount, the more mortgage interest you have to pay.
To introduce the concept of a negative, or Balancing (“B”) feedback loop, let’s assume our deposit account also serves as an offset account. In some banking systems, an offset account is used to reduce the base for calculating interest on mortgage. The more money you have in this account, the less mortgage interest is calculated — this is a negative feedback loop.
If we run this model, we can see how our deposit is slowly but steadily depleted by repaying the mortgage, despite the monthly savings inflow and interest.
Adding more and more flows and stocks will allow you to create a more detailed and reliable model of a real system, and mimic its reactions to different external inputs. The key is in identifying the feedback loops that make the system tick, and establishing the magnitude of those loops.
You don’t have to go to that level of detail though. Quite often it is enough to just identify the purpose and behaviour of a system, draw its structure (system analysis), then identify the cause-and-effect loops to uncover the root causes for the problems you try to solve (operational analysis), and use these insights to design a solution.
Design thinking and systems
Design ability is one of the three fundamental dimensions of human intelligence [7].
Everything we have around us has been designed. Design ability is, in fact, one of the three fundamental dimensions of human intelligence. Design, science, and art form an ‘AND’ not an ‘OR’ relationship to create the incredible human cognitive ability.
— Nigel Cross
And design thinking to a sense implies system thinking by definition. If you see design as [5]
“an ability to create feasible wholes from infeasible parts”
then going though a design process you ultimately create a new system via giving it a purpose and a structure, that in turn enables expected behaviours.
To design a feasible and viable solution, one always needs to consider the context in which the solution operates (analysing the super-system) to define the desired behaviour of the solution. The solution performance in turn will eventually be design-driven — driven by decisions made though the design process.
However, design process is not the only place where systems thinking can help a business analyst. You also apply it when dealing with complex problems that require root cause analysis. Seeing the bigger picture of how different elements of the enterprise or technology solution interact to achieve the goal, and understanding the cause-and-effect loops that underpin it, enables you see the causes of observed behaviour — which is especially crucial when the behaviour is problematic or unwanted.
This is especially important when you try to implement a change addressing the identified issues. Systems analysis will help you assess the impact of a change and anticipate potential blockers. You can also experiment by combining it with other tools, such as customer journey mapping and service blueprinting. In the latter case, you can use the customer journey to map front-of-house behaviour and a dynamic system model to map the back-of-house complexity. [8]
Just be aware that everything we think we know about the world is a model, regardless of whether it is documented or lives in our head. And just as any model, it is never perfect no matter how sophisticated it is — the real world is always more complicated than we think.
So you need to find a good working balance between what to include in the systems models you produce, and what to exclude. Remember also that system models are not the key outcome of systems thinking, it is the ability to see the forest for the trees: view the bigger picture, discover how the system operates in the context of other systems, and understand what drives it from within.
References
Thinking in Systems: A Primer, Donella Meadows, 2008 https://www.amazon.com.au/Thinking-Systems-Donella-H-Meadows/dp/1603580557
https://dougdickerson.wordpress.com/2016/02/04/how-hidden-agendas-impact-organizational-success/
Systems Thinking: Managing Chaos and Complexity, Jamshid Gharajedaghi, 2011 https://www.amazon.com/Systems-Thinking-Complexity-Designing-Architecture/dp/0123859158
Designerly Ways of Knowing, Nigel Cross, 2007 https://www.amazon.com.au/Designerly-Ways-Knowing-Nigel-Cross-ebook/dp/B000SNUQ7Q
Using Systems Thinking to Design Better Services https://medium.com/@mikelaurie/using-systems-thinking-to-design-better-services-905b62ca10b7
Comments