Complexity and Systems Thinking: Approaching Massive Problems and Managing Knowledge


Thinking systemically is an attribute that can lead to considerable growth and progress toward creating change on an individual, communal, and global level. While I think everyone can and should develop their ability to think this way, I personally feel that the unique educational experiences that Engineers undergo equip them very effectively to think systemically, and furthermore facilitate massive change in the world around them. As engineers, we learn from day 1 to break every problem we encounter into bite-sized pieces; into segments, or by separating a larger problem into smaller ones that require differing expertise… components that we can analyze individually, but in the context of the entire system. Whether it is looking at the frequency response of a circuit, designing a robot, the behaviour of a beam under certain loading conditions, the response of a mass-spring-damper system, or looking at the behaviour of fluid flow, there are commonalties that can be drawn across all facets of engineering which relate back to thinking systemically. What are the different elements/impedances in this system? What is the output of the system? What are the through/across variables we’re dealing with? What can we change/manipulate? How will such a change affect the output of the system? What output do we want from the system? What is preventing this result from being achieved? What can we learn about the system? What can we measure? What data can we collect? Engineers are accustomed to approaching problems from an analytical, systemic perspective.

The true power of this methodology is leveraged when tackling massive social, environmental and geopolitical issues such as poverty, homelessness, universal education, climate change, species endangerment, loss of biodiversity, international relations, or peacekeeping, to name a few.‘Systems Thinking’ makes it more possible to address the root causes of these huge problems. This process of problem analysis aims at taking a big-picture approach to solving problems, and making a lasting, sustainable difference. This is done by utilizing methods that look at the full extent of a system, as opposed to fragments or ‘symptoms’ of it. For example, instead of giving food and aid to people in poverty, looking at what factors are inhibiting these people from being able to grow or access food or seek medical attention when needed.



Well, all this is fine and dandy… thinking systemically is a powerful way to approach big problems, and lends itself to be readily applied to an array of social and environmental problems as well as technical ones. But how does it actually work? What results come of all this fancy talk? That’s where social innovation comes in! The term Social Innovation refers to the work people do to toward addressing and solving the root causes of complex social problems such as those mentioned above. Social innovators are usually people with incredible vision, thinking and actions that bring distant ideas for innovation into reality.  A social innovation could be a change in process, a product, or a program that profoundly changes the way a system operates, and subsequently reduces the vulnerability of the people and environment adversely impacted as a result of that system’s current behaviour. Some examples of social innovation could include the Registerd Disability Savings Plan (RDSP), geared at helping elderly people living with a disability to achieve financial independence. Other examples could be Universal Healthcare, the invention of the wheelchair, and the continued advancements made in technology every day. Through these innovations, the system being disrupted will grow more resilient, and produce a more desired output. We are at a time when novel, innovative ideas are not only in high demand, but are critical in order to tackle the biggest problems facing our world today. It is the development of new products and programs, and the continued effort of people and groups creating policy and legislative amendments and improvements that will yield lasting positive change on all scales from the individual in a system, to the system in its entirety.


Systemic change… systems thinking… system this, system that… the way ‘system’ is getting thrown around here, it’s looking more and more like just another buzzword that people like to use in vague statements that really don’t mean anything. I hate buzzwords… they gloss over lack of understanding and enable people to duck out on providing detail, like development, or sustainability. ugh… what does that even mean? So let’s dive in a bit deeper – what is a ‘system’?

define system

Great – thanks, Google! This concludes my paper on Systems.

Just kidding – let’s give this definition some context. We are in a world comprised of countless systems that impact us all, both individually and collectively. We are involved in many systems right now, in fact. Can you think of any? Individually, we are being kept alive by the different biological systems our body is experiencing. Our cardiovascular/circulatory system, digestive system, nervous system, muscular system, etc. There are many systems that compose our societal infrastructure, such as the criminal justice system, or the judicial system. Right now, we’re using the internet as actors in part of a global information technology system. We consume food, and use land – we are part of environmental/ecological systems. On an individual level we are impacted by the different class & social systems that exist (power distribution in society, hierarchy at work, cliques at school), as consumers and/or suppliers, we are part of an economic system. Systems are all around us. And while some are stand-alone, others interact with one-another, creating larger systems, and as a result, we live in a space composed of an entire ecosystem of systems! WOW! So how can we categorize all of these different systems? How can we begin to understand them? Well, for starters, the vast majority of the systems we interact with and experience can be grouped into three broad categories: Simple, Complicated, and Complex. A fourth space can be called Chaotic… We’ll touch on this later. But what’s the difference between all of these categories of systems? Look no further!

Simple Systems

A simple system is a type of ordered system, meaning that an established series of steps that can be taken to overcome a problem or accomplish a task. Cause and effect exists, and is evident (think “if you do X, you get Y”). For these systems, minimal experience and expertise is required to solve such a problem, or disrupt such a system. The margin of allowable error from one step to the next is typically quite high. The confidence of the results being reliable and replicable from one iteration to the next is usually high as well.  A strong example of this would be baking a cake. Aside from knowledge of how to operate a blender, stove, or to access required ingredients, minimal expertise or experience is required, as long as the steps are followed.  It is acceptable for the steps to be followed with a fairly high margin of error; for example, if you put an extra quarter of a cup of sugar in your cake mix, you probably won’t ruin your cake.

Complicated Systems

Dealing with a complicated system can be a bit more involved. This is another type of ordered system, however cause and effect are not immediately evident. Considerable experience and expertise in a given field are typically required to devise a solution. While, like simple tasks, a list of steps must be adhered to in order to accomplish a task, the individual steps for a complicated problem are much more difficult to accomplish and must be adhered to much more strictly – the tolerance for error is much lower. For example, sending a rocket to the moon is a complicated problem, as a high volume of tasks need to be accomplished seamlessly with extreme precision. However, assuming that all tasks are accomplished correctly, the outcome can be predicted with a high degree of certainty. Additionally, every time a complicated task is completed, the ‘recipe’ can be updated or improved, and the likelihood of accomplishing the same task again becomes higher.

Complex Systems

Addressing a complex task is quite a bit different than the other two. This kind of system exists without causality. It imposes light constraints on its agents, with the agents constantly modifying the system. While simple and complicated tasks typically have a ‘recipe’ that can be followed for optimal results, rigid protocols are less applicable to complex problems. The majority of issues that EWB deals with are complex problems (poverty, for example). An effective example of a complex problem would be raising a child. In complex problems, an individual component can’t be evaluated without considering its context within the rest of the system; the relationship between each component of a complex system impacts their behaviour, and therefore, while expertise and experience can definitely play a role in attacking complex problems, its effectiveness is limited by the psychology of the individual child. Every person (child) has had different experiences, holds different perspectives, and are in different places in life and in maturity. So going back to our ‘raising a child’ example, while a parent may logically understand the reasoning for a particular decision (ie “go to bed at this time.”), the child may see it from an entirely different perspective, and therefore explaining your decision on a level that an adult would understand may not be an effective way to convey information to a child. For this reason, the ‘cut and paste’ process that works more effectively for simple and complicated problems does not apply as much to complex ones. That is to say, even if a parent successfully raises one child, it does not necessarily mean that they will be able to successfully raise a second.

Chaotic Systems

This type of system is discussed considerably less than the first three, however I think that it is equally, if not more important. In general, these systems are unwanted, and steps need to be taken to avoid creating them as the result will often be some sort of crisis. For this reason, I think they are vitally important to understand. In Chaotic systems, no cause and effect relationships can be determined. If it is entered intentionally, it is innovation. If entered unintentionally, steps need to be taken quickly to stabilize the situation.

Now we know a bit more about the different types of systems. But that on its own isn’t going to be very helpful. How can we manage these systems? How can we solve problems in these different domains?

Introducing: The Cynefin Framework!!!


David Snowden, a knowledge management expert from Cognitive Edge, developed the Cynefin Framework, which outlines the differences between chaotic, ordered, and complex systems. One of the key notions that he emphasizes is embracing complexity, and learning to manage complex systems, rather than to treat them in a different way.

The Cynefin framework is a sense-making model, which should not be confused with a categorization/decision-making model. A categorization model exists on a basis such that the framework precedes the data. That is to say, when data is dropped into the (typically) 2×2 matrix, it will enable you to make a decision accordingly. Categorization models are useful for exploitation of data, not for exploring solutions to problems.

A sense-making model, on the other hand, is comprised such that the data precedes the framework; the framework emerges from the data. The Cynefin framework is an analytical framework derived from decision theory. Its applications can be seen in knowledge management, IT design, Project Management and other areas, which recognizes the casual differences that exists between different types of systems and give people a quick and easy way to flip between them and apply an appropriate methodology for a particular system. It has the potential to propose new approaches to decision making in complex social environments.

So, what all that means is that we can use this framework to help us determine the best course of action in different situations.

In simple systems, the decision model is sense-categorize-respond.

SenseEvaluate the environment; consider the different variables an parameters relating to the situation.

Categorize: Group the elements coming into the system into relevant headings (As a simple system, these elements will likely be easily grouped into previously identified/understood groups).

RespondMake decision. As an ordered system in which the relation between cause and effect is evident, apply ‘standard operating procedures’ which in this situation would be “best practice.”

So going back to our Cake example, when making a cake, one would sense that they are in a kitchen, with utensils, a blender a stove, ingredients, etc. categorize these things into relevant groups, such as ‘wet ingredients’, ‘dry ingredients’, and ‘containters.’ and finally respond, which would be to apply best practice, and follow the cake recipe.


ASIDE: BEST vs GOOD PRACTICE (This will become relevant shortly)

-“Best Practice” refers to a methodology to approaching a situation that has been generally accepted as the ‘go-to’ as it has proven to yield the exact result required/expected whenever it is implemented. This methodology can continually evolve as new technologies/innovations become available.

-“Good Practice” refers to the methodology chosen when a range of possible approaches or solutions exists. It is the most effective solution selected after all the different parameters of the situation are considered.

For example, in a modern hospital,  with plenty of funding and the latest & greatest in technology, an emerging new technique for treating a medical condition may be considered ‘best practice.’ In contrast, a hospital in a developing nation that does not have the technology or infrastructure in place to sustain such an advanced operation may select a lower-technology treatment method as their ‘good practice.’

To simplify, ‘best practice’ is the optimal solution to any problem, but cannot always be implemented, and thus an alternative ‘good practice’ must be consulted.



In complicated systems, the decision model is Sense-Analyse-Respond.

Sense: Evaluate the parameters of the situation/environment you are in. Remember that despite being ordered, complicated systems & problems are often open-ended, and have a range of possible solutions.

Analyse: Analyse the system. What can be measured? What data can be collected? Whose expertise can we employ?

Respond: Make a decision. Based on the different factors influencing your decision (funding, accessibility, etc). Apply “good practice.”

In complex systems, the decision model is probe-sense-respond.

Probe: Do something to the system. *****Conduct Safe-fail experiments, do not do fail-safe design***



-“Safe-fail experiments” refers to an experiment conduced (ie an change made to the system, or a microcosm of the system) that can provide value information about how to proceed, but which is conducted in a safe manner that will prevent new problems from arising, or additional complexity being added to the system.

-“Fail-safe design” refers to ‘mistake-proof’ design, with the intent of preventing errors; to ensure the output of a system is the exact same every time. While this is implemented frequently in the manufacturing industry, for example this past semester I was exposed to designing Poka-Yokes, a component of the Six-Sigma methodology.

In the complex domain, fail-safe design is not very applicable, as the system is continually evolving, and a definitive cause and effect do not exist. It is strongly encouraged to employ safe-fail experimentation and restrict fail-safe design while operating in the complex domain.


Sense: Evaluate the system’s response to the change. Is the safe-fail experiment successful? Unsuccessful? Inconclusive?

Respond: Based on new information acquired about the system and its behaviour, establish next steps to be taken. In general, if a safe-fail experiment is successful, it’s amplified. If it’s unsuccessful, it’s dampened.

Chaotic Systems

An important attribute I would like to point out regarding this framework is called the simple/chaotic boundary. Looking at the framework, you can see a ‘cliff’ in the bottom centre, originating in the simple domain, and ‘falling off’ into the chaotic domain. It represents the following: If elements that were originally categorized as simple, and are hence ordered, it is believed that past success in dealing with these elements will yield an invulnerability to future failure. This is a dangerous classification, as it leaves elements vulnerable to future change. In the event that “doing X does not result in Y” the element ‘falls off’ this proverbial cliff into chaos. In this situation, there are no answers. Data is random, meaningless, impossible to make sense of as-is. Dealing with this type of problem requires devising novel solutions, and taking creative action. For this reason it is CRITICAL that you only move a very small amount of material, content, information, etc. into the simple domain, as it become very susceptible to rapid or accelerated change. Furthermore, it is a powerful attribute to become proficient in managing systems and problems in the complicated/complex domain.

In chaotic systems the decision model is Act-Sense-Respond

Act: Take creative action! Do something to the system.

Sense: After taking action, evaluate what the result is. Did anything happen? Did anything change?

Respond: Based on the new information, devise next steps. Will you impose a different novel solution, will you give your first one a tweak and try it again?

The last note I’ll make on this segment is that in general, people tend to make decisions based on their preference for action. In other words, if someone prefers dealing with simple problems, and are comfortable in that domain, they often tend to always approach problems from this perspective. This is a dangerous mindset that is not reflective of the array of systems and problems one will encounter in the world. For this reason, I feel it is critical to embrace this framework – or at the very least consider approaching problems from a different, -and potentially new -perspective.


Well, this should hopefully give you a slightly clearer (or perhaps less clear) understanding of systems, what they are, and what you can do with them. We have all encountered systems in our own lives.

My personal experience with simple and complicated problems is one that I dove into on a daily basis this past Winter semester. As a Mechanical Designer, the designs that I produced could typically be grouped into 2 categories: Assemblies, and Parts. Creating a part would be an example of a complicated problem – there is a ‘recipe’ that can be followed to produce the same part every time. However, considerable machining expertise is required to produce required parts. Additionally, in order for the same part to be reproduced, the features must be identical, otherwise the part is useless. Producing a part would typically play out to the effect of “take a block of raw material with these overall dimensions. Cut slots & keyways of this size in these locations, adhering to these dimensions and this level of accuracy. Drill and tap holes of this size in these locations with these tolerances.”

An assembly is a combination of parts. This is another example of an ordered system, and putting together an assembly could be considered a simple undertaking. In order to successfully produce the assembly, protocols and steps need to be adhered to. “this face must be mated with this face. The locating pin must connect these two parts. This sized bolt must fasten these two parts together.” In order to successfully produce this complicated system, the blueprint that facilitates the production of an individual part, as well as the actual assembly must be adhered to.However, the margin of error is often much larger during assembly than in the production of parts. In addition, there are infinite ways that a part could be created incorrectly, while a limited number of ways which could cause an assembly to be put together incorrectly.

When I think of complex systems being treated as complicated systems in my life, a few key examples come to mind. I would treat the Guelph EWB chapter as a complex system. Member retention, for example, is proving to be a huge challenge for us, and while we have successfully retained particular members, there is no guarantee that the methodologies that led to their commitment to EWB will guarantee that of others. Additionally, every person that comes to EWB is an individual, and it cannot be assumed that they all have the same interests, intentions, or thought processes. By not continually addressing these elements of the system, chapter success has been limited as a result. The overall uncertainty surrounding the success of recruiting members, and the ineffective methodologies for approaching complicated problems further leaves me with the conclusion that we are in the complex domain. In addition, the direction that the chapter takes is another complex problem. What is our role in the school? In the organization? What goals do we need to accomplish?  I have been a member for 2 years, and in that time I have largely seen it being treated as a complicated system. By this I mean that the plan-execute-reflect approach was typically used. A ‘semester plan’ would be constructed at the beginning of the year, and largely adhered to throughout the year. As a result, when unexpected events would arise, the rigid plan would fracture. While elements of the plan were successful, other elements were not, which I believe served to push particular people away, and has resulted in the current state of the Chapter – which I see as having considerable room for improvement.


As you’ve learned here, when dealing with systems (especially complex systems) there is often an element of research and experimentation involved… whether it is in attacking complex problems, part of your analysis in dealing with complicated problems, getting yourself out of a chaotic crisis, or devising a new and improved way to manage simple problems, R&D is a critical skill in approaching systems and solving problems. As has been mentioned, safe-fail experimentation is a strong tool to employ regarding this endeavour. However, and this is especially true in the complex domain, a time will come when it is necessary to scale.

Scaling – yet another buzzword perhaps?

Well let’s give it a bit more meaning then… from a mathematical perspective, scale would refer to the ratio of size of a model or other representation to the actual size of an entity.

Now this definition isn’t perfect for us, but gives us a bump in the right direction. In terms of complex problems, it’s easy to think about scaling in terms of 4 large categories: Individual, Network, Organizational, Institutional. Change in these four categories, and an idea of the scaling between the four, can be seen in the representation below:


[Fun fact: when I inserted this picture it was too small, so I had to scale it up. Ha. Ha Ha.]

So who cares? Scaling is of critical importance if you have any intention of doing anything in a larger context (as you typically do). So in framing you decisions, and deciding on a certain problem solving approach, you should always consider the scalability of the solution you are implementing.

Anyway… an effective model for scaling systems is the Adaptive Cycle Model (seen above), which has 2 key aspects, scaling out and scaling up. When change happens in a connected way across scales it becomes stronger and more effective.

Scaling Out: A replication of an innovation. Efforts aimed at making a good initiative occur more frequently, and in more places. This scaling occurs at the current level of a system.

Scaling Up: Increasing an initiative’s impact in the broader system in working toward addressing the root causes of a problem. This often involves the initiative looking or working a bit differently, in order to compensate for the additional factors that come into play upon applying an initiative to different or additional elements of a system. This scaling occurs across one or more levels of a system

Bouncing back to our chat on decision modelling in the complex domain, we established the importance of safe-fail experimentation, and amplifying successful experiments. This is a huge example of when you would implement scale. If an experiment you conduct on a small sample set/representation of your overall system is an overwhelming success, you will want to replicate it out and upward toward addressing the rest of the system…. Right?

OK – so how do you do it?


Great question!

There are a number of different ways to go about scaling up a social innovation. In my opinion, the ability to do this is the back bone of any entrepreneurial venture. SiG has identified a number of different methodologies, or proverbial pathways, that exist for achieving success, such as the Beanstalk, the Umbrella, LEGO, and Polishing Gemstones. I am going to discuss the Volcano pathway – I find it most relevant here, as it was developed by EWB. This methodology is based on embracing the energy and good intention of passionate and enthusiastic individuals, and through fostering this behaviour EWB was able to scale out. Their intent was that once a ‘tipping point’ is reached, the subsequent ‘eruption’ would lead to massive, disruptive change, across all levels of the system.


A Theory of Change defines long-term, full-scale goals and attempts to map backward the required steps to achieving these goals.

I would like to note that the usefulness of this theory is limited, because (as we know) complex systems cannot be truly mapped backward in this way.


The Theory of Change for this methodology occurred through continued internal learning and experimentation. This approach resulted in an inclusive and participatory organizational culture. It became apparent, however, that with such a diverse, inclusive, interdisciplinary team working on an array of different projects, their work became meaningless as it did not address the context of the situation. In working on isolated projects, the root causes of the problems they were working on remained ignored. For this reason, EWB chose to narrow their focus on specific objectives, as they only had finite resources available to them. In defining a strategic focus, narrowing down to 5 key projects, they were able to work toward addressing the larger complexities of the projects they were working on, and hence scale up. The risk that comes through this strategic focus is that the inclusive, participatory culture upon which the organization was based could be at risk as their ability to generate energy and enthusiasm could become more difficult with narrowed focus.

Further Learning: I found this video excellently relates the Cynefin framework to systemic change:



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s