Complexity 101

Complexity 101

How to recognise, and understand, complexity.

What is this?

Complexity. It feels like a lot. When I ask the uninitiated what comes to mind when one hears the word ‘complexity’, I often get phrases like ‘a tangled mess’, ‘something with many components’ or even ‘something scary to address’. One even may imagine an image like this:

This is the Foresight Obesity System Map: a complexity map of the obesity ‘problem’ and all the relevant players in that informal and formal system. It’s not a bad representation of complexity in its formal definition, but it is a bit hard to swallow, and to interpret. It certainly looks complex! Paradoxically, complexity itself is a simple concept to grasp.

Complexity itself feels fairly nebulous, but once understood it can change people’s ways of looking at and responding to challenges. Managing and responding to complexity demands different approaches and different methods. To paraphrase CECAN, complexity is present in many social and natural systems and in the major challenges we need to address as international development specialists: it is pivotal in accounting for the success or failure of interventions. Making it explicit can result in better, more effective delivery, and ignoring it risks failure. Most importantly, complexity does not pose new challenges to us, but it simply intensifies the ones we are already working with.

I always introduce complexity in contrast to things that are not, and (as is beloved or tedious to my programme teams, depending on how many times they’ve seen it) I always refer to the Patricia Rodgers example [1] that involves interpreting these situations through the analogy of cakes, rocket ships and babies (see below).

In essence, we usually find ourselves in one of three situations. Well, technically four, but the fourth situation is a chaotic one: it’s like a tornado, you just run away. Let’s avoid that one shall we? These three situations follow as so.

Simple situations, like baking a cake: you have a clear end result with a process that is predictable, and linear. There are known knowns, and known unknowns. It is also repeatable: you’ll get the same result (unless you make a mistake).

Complicated situations, like building a rocket: again you have a clear end result but this time the process is less straight-forward. You will find it has has multiple causal pathways and varied TA interventions. There will be known knowns and known unknowns, and maybe an unknown unknown, but it will be surmountable. Ultimately the process is still relatively linear and the whole thing can be repeated to get the same rocket ship.

Complex situations, like raising a child: anyone who has ever raised a child knows that you may have a good idea of what you want the end result to be, but there is no guarantee that you will get there. The pathway is highly complex: you have to learn as you go, change strategy, and have multiple unknown unknowns. There will be feedback loops and unpredictable environmental factors, and a high degree of adaptation. Ultimately, much like raising a child, this cannot be repeated to get the same result. This is the domain development work usually finds itself in.

This is why, much like my MEL mentor, I like the following concept as a representation of complexity:

This is because when we talk about complexity we ultimately end up speaking about complex systems, and a flock of birds is the finest representation of that.

You see a beautiful shape moving through the sky because every individual bird (‘actor’) is doing something. While they are doing something, they are also interacting with and reacting to the birds in immediate proximity to it who are also doing something. and while they are behaving and reacting to others’ behaviours, they are each responding to the environment: to gravitational forces, magnetic fields, weather, predators, obstacles… you name it. So this combination of individual behaviours, reactive behaviours and environmental factors give rise to a flock in a certain shape that moves in a certain way.

[1] Rodgers, P., Using Programme Theory to Evaluate Complicated and Complex Aspects of Interventions, Evaluation, 2008

The ecosystem present on a forest floor exhibits complexity, just as our social and cultural practices

How do I recognise it?

Complexity should be recognisable the second you meet the conditions mentioned above under the ‘complex situation’ definition. However, there are roughly 11 features of complex systems as identified by bodies like CECAN that I’ve tried to explain in plain English below. These are helpful both in understanding exactly what you’re dealing with, as well as identifying whether you are working with a complex system or not.


The first feature of a complex system is that it adapts. This means that parts of the system – its components – are learning and evolving, and thus changing how the system responds.

An example of this is anti-microbial resistance: bacteria evolve to be resistant to antibodies.

I’ll also be using a more ‘developmenty’ example with a case study. An example of adaptation would be that as the Programme Leaders in the Civil Society Organisation (CSO) in the made up land of Teraria have learned leadership skills, the CSO sees a shift in how it behaves: all voices start to be included rather than the voice at the top.

Emergence and self-organisation:

Complex systems see emergence and self-organisation. In plain English, this means that when the parts of the system (‘components’) interact, you see higher-level properties arise from their interaction. These properties will likely be new and unexpected properties, making them hard to predict: these are called ’emergent properties’.

Social norms are emergent properties of social systems.  The self-organisation and interaction of components in our current social systems have led to an emergent property of serious inequality in terms of social and working practices for the black community. The current shifts in that organisation and interaction (mass, global protest) could lead to changes in social and working practices with regard to the black community. These would be emergent properties of that.

In our case study, the problem of lack of effective delivery of programmes is an emergent property of the Programme Leaders (components) having low skills and failing to communicate (interaction), as well as the lack of procedure and process (affecting how they self-organise).

Unexpected indirect effects:

When interacting with complex systems, you will see unexpected indirect effects. What this means is that a change in one part of the system can lead to unexpected change in what seems like a remote part of it. This happens because there are long causal chains of interaction between components in that system, such that they are able to carry an effect far down the system; much like how a spider on one side of its web can feel a fly land on the other side because of the links of silk.

An example of this was that in the UK we saw a decreased resilience to flooding. This was caused by changing agricultural practice (winter planting), new house building policy (building on flood plains), and climate change (increasing flood risks).

In our Terarian CSO we saw that an increase in leadership, team-building and communication skills has led to reduced working hours in the CSO. This is due to more efficient working and awareness of other practices.

Feedback loops:

Complex systems will have multiple feedback loops present. What this means is that a result of a given event will then influence how the given event (when repeated) causes change. This can be direct or indirect, and can suppress or accelerate change.

An example of this is that when alcohol percentage was reduced in a Swedish national beer, people increased the amount of it that they drank: a negative feedback loop from an initial event of reduction aiming to reduce alcohol intake. Instead of drinking less alcohol, they ultimately drank the same amount with additional calories to boot.

In our Terarian CSO, when hierarchical and unempathetic leadership practices decreased, members had an increase in motivation and productivity: a positive feedback loop that reinforces the new approach to leading.

Levers and hubs:

Complex systems will always have levers and hubs. Levers and hubs are effectively the idea that certain parts of your system will have the power to change everything, or block change. This is a concept we are familiar with: many stories feature a silver bullet or a holy grail that can change the tide of time. This concept acknowleges that in complex systems, some components in the system have disproportionate influence over the whole due to the structure of their connections.

For example, a key individual in an organisation may be an obstacle to change due to veto powers, or a gatekeeper due to approval powers: country leaders can slowed change by suspending funding to bodies and organisations.

Aisara is a key lever in the CSO due to her stature and the degree of respect for her. If she starts doing something differently, others will follow suit.


Complex systems are rife with non-linearity. What this means is that unlike with cake-baking or rocket-ship-building, pathways aren’t linear. A small change could cause a huge result in one situation, but the same small change may lead to no change in another. Likewise a big change may have small results in a given context, yet that same big change could have a huge result in another.

For example, the urban fox population size didn’t increase dramatically during the covid lockdown when their food source increased: people had more food waste from an increase in take-away consumption (and domestic bins are easier to access than heavier-duty restaurant bins). Instead, it stayed the same as they would still be limited by space.

The small change of a few extra CSO computers to work with with has led to a large change of projects running more and in a more effective manner as they are more quickly produced and better formatted.

Domains of stability:

Complex systems are hostage to domains of stability. Effectively, this concept states that a given system can have more than one state in which it is stable. These change as the context does. By nature, complex systems gravitate towards those stable states, and will attempt to remain in them unless something external forces them out of it. These are usually binary concepts, with the ‘in-between’ being uncomfortable and sometimes chaotic.

An example of this is that a culture exists stably either with or without electricity, but the process of installation and education on what it is and how to use it is uncomfortable and unstable.

The CSO is stable in its old state of ad hoc implementation, and in a new state of set processes and working practices. However the intermediary of changing ways of working causes instability and confusion in the system.

Tipping points:

Complex systems are often victim to this concept of a tipping point. These are thresholds such that once they are reached, the system goes through a rapid change into a different state. This state may be challenging to reverse.

Neighbourhood gentrification is an example of this as it is usually gradual at first, then suddenly tips over and rapidly changes its demographics and character.

The CSO members were slow to understand and take onboard new ways of working, but once enough had, it suddenly tips and had a ripple of rapid behaviour change.


This is a fairly intuitive concept, in that a complex system’s future depends on its history; how it got to its present state, and what that present state is currently. The order in which intervention activities are introduced affects their cumulative impact: what they add up to.

Evolution is the best example of a highly path-dependent process, as organisms cannot radically change from their predecessors. They change via mutations of existing adaptations, and the only way to understand their current state of being is to look at the history of mutations. To understand why a giraffe looks as it does, you need to look at the order of mutations that meant the species have incredibly long necks: a gradual series of changes over time. This is also why evolution often doesn’t find perfect solutions.

Similarly, an effective, well-led CSO will entirely depend on the process by which it got there. To understand the new order, you have to understand what activities happened at what point in time.


Complex systems are often Open. This means that complex systems often have links and connections into other complex systems such that a change in a separate system can end up impacting a different one due to those links. Links could be information exchange, in/outflow of money or people, etc.

A good example of this is delayed transfer of care. A delayed transfer of care occurs when a patient is ready for discharge from acute or non-acute care and is still occupying a bed. The health system will have some responsibility for this, but also a social care system. If the social care system has failed to make room for that patient to leave the hospital and enter the social system, the patient remains in the health care system. This then impacts the health case system, as longer stays in hospital impact waiting times int he rest of the health system.

In our Terarian example, the CSO doesn’t operate in isolation. To understand why it implements ineffectively you have to understand the other organisations, authorities, and individuals that engage with it and contribute to the problem, such as angry citizens placing pressure on it, their own member turnover, or the flow of people through local authorities.

Change over time:

Finally, complex systems develop and change their behaviour over time. This is partly due to openness and adaptation of its components, but also because they are systems that are usually out of balance or unstable and therefore in a continuous process of change.

Social norms are excellent examples of this as customs and cultures radically change over time and can never be said to be at an end point. Covid-19 has led to a change in our social norms where more is happening online. We also cannot say this is the final state of our social norms

The CSO has shifted from a chaotic and inefficient to one that is more process-orientate and effective. However this is an interim state that is still in a process of change. We cannot say it is the final state.

Why is this relevant?

So those are your 11 features and a brief inro into what complexity is. The question may be ‘why should I care?’. This is relevant because whether working domestically or internationally we will be interacting with complex systems, because complex problems operate like that flock of birds. Just as the flock’s shape is an emergent property of the arrangement of birds and their behaviours, a problem is the emergent property of a complex system of actors interacting and reacting.

In terms of further resources, I usually recommend that folk check out CECAN‘s work as I think they break down difficult concepts into excellent guidance. They and the Travistock Institute were behind the updates to the Magenta Book that help HMG organisations start to think about this more. I am in the process of writing a lightweight guide for my programme teams that I will upload here in due course.

Complexity 101 for development practitioners: upcoming.

Problems are emergent properties of these complex systems

Want to talk MEL? Let’s connect.

Take me back to:

%d bloggers like this: