
James Gleick, the author of the classic book “Chaos: Making a New Science” has written a terrific review of Jill Lepore’s new book “If Then: How the Simulmatics Corporation Invented the Future.”The book covers the origin of data science as applied to democracy, and comes as conversations about social media, algorithms, and electoral manipulation are in full swing due to the US election and the release of The Social Dilemma.
Gleick’s review is worth a read. He covers some basic complexity theory when working with data. He provides a good history of the discovery of how the principles of “work at fine granularity” helps to see patterns that aren’t otherwise there. He also shows how the data companies – Facebook, Google, Amazon – has mastered the principle of “data precedes the framework” that lies at the heart of good sensemaking. For me, both of these principles learned from anthro-complexity, are essential in defining my complexity practice.
Working at fine granularity means that, if you are looking for patterns, you need lots of data points before seeing what those patterns are. You cannot simply stake the temperature in one location and make a general conclusion about what the weather is. You need not only many sites, but many kinds of data, including air pressure, wind speed and direction, humidity and so on – in order to draw a weather map that can then be used to predict what MIGHT happen. The more data you have, the more models you can run, and the closer you can come to a probable prediction of the future state. The data companies are able to work at such a fine level of granularity that they can not only reliably predict the behaviour of individuals, but they can also serve information in a way that results in probable changes to behaviour. AS a result, social media is destroying democracy, as it segments and divides people for the purpose of marketing, but also dividing them into camps that are so disconnected from one another that Facebook has already been responsible for one genocide, in Myanmar.
Data preceding the framework means that you don’t start with a framework and try to fit data to that matrix, but rather, you let the data reveal patterns that can then be used to generate activity. Once you have a ton of data, and you start querying it, you will see stable patterns. If you turn these into a framework for action, you can sometimes catalyze new behaviours or actions. This is useful if you are trying to shift dynamics in a toxic culture. But in the dystopian use of this principle, Facebook for example notices the kinds of behaviours that you demonstrate and then serves you information to get you to buy things in a pattern that is similar to others who share a particular set of connections and experiences and behaviours. Cambridge Analytica used this power in many elections, including the 2016 US election and the Brexit referendum as well as elections in Trinidad and Tobago and other places to create divisions that resulted in a particular result being achieved. You can see that story in The Great Hack. Algorithms that were designed to sell products was quickly repurposed to sell ideas, and the result has been the most perilous threat to democracy since the system was invented.
Complex systems are fundamentally unpredictable but using data you can learn about probabilities. If you have a lot of data you gain an advantage over your competitors. If you have all the data you gain an advantage over your customers, turning them from the customer to the product. “If you’re not paying, you are the product” is the adage that signals that customers are now more valuable products to companies that the stuff they are trying to sell to them.
Putting these principles to use for good.
I work with complexity, and that means that I also work with these same principles in helping organizations and communities confront the complex nature of their work. Unlike Facebook though )he says polemically) I try to operate from a moral and ethical standpoint. At any rate, the data we are able to work within our complexity work is pretty fine-grained but not fine-grained enough to provide accurate pictures of what can be manipulated. We work with small pieces of narrative data, collecting them using a variety of methods and using different tools to look for patterns. Tools include NarraFirma, Sensmaker and Spryng, all of which do this work. We work with our clients and their people to look for patterns in these stories and then generate what are called “actionable insights” using methods of complex facilitation and dialogic practice. These insights give us the inspiration to try things and see what happens. When things work, we do more and when they don’t we stop and try something else.
It’s a simple approach derived from a variety of approaches and toolsets. It allows us to sift through hundreds of stories and use them to generate new ideas and actions. It is getting to the point that all my strategic work now is actually just about making sense of data, but doing it in a human way. We don’t use algorithms to generate actions. We use the natural tools of human sensemaking to do it. But instead of starting with a blank slate and a vision statement that is disconnected from reality, we start with a picture of the stories that matter and we ask ourselves, what can we start, stop, stabilize or create to take us where we want to go.
In a world that is becoming increasingly dystopian and where our human facilities are being used against us, it’s immensely satisfying to use the ancient human capacities of telling stories and listening for patterns to create action together. I think in some ways doing work this way is an essential antidote to the way the machines are beginning to determine our next moves. You can use complexity tools like this to look at things like your own patterns of social media use and try to make some small changes to see what happens. Delete the apps from your phone, visit sites incognito, actively seek out warm connections with real humans in your community and look for people that get served very different ads and YouTube videos and recommended search results. Talk to them. They are being made to be very different from you, but away from the digital world, in the slower, warmer world of actual unmediated human interaction, they are not so different.
Postscript
Over the past few years, my work has taken shape from the following bodies of work:
- Dave Snowden’s theories of anthro-complexity, which forms the basis of my understanding of complexity theory and some of the tools for addressing it, including facilitation tools and Sensemaker.
- Cynthia Kurtz’s Participatory Narrative Inquiry, which is a developmental evaluation approach that uses stories and methods of sensemaking that she partly developed with Dave and then subsequently. I use her software, NarraFirma, for most of our narrative work now.
- Glenda Eoyang’s Human System Dynamics is a set of tools and methods for working with complex adaptive systems.
- The facilitation and leadership practices from the Art of Hosting which help us to develop the personal capacity to work dialogically with complexity.
Share:

I adore Alicia Juarerro’s work. So much so that I just watched a short video and spent the last hour writing about it. Here’s what I’m learning
Alicia Juarrero is a philosopher whose thinking about causality, complexity, action, and emergence has been critical to some of the ways in which folks like Dave Snowden have thought about this field. Her book Dynamics in Action is a really important read, packed full of thinking about complex systems and constraints. It’s a hard book to get into – indeed advice I have had from others is “start in the middle” (a helpful enabling constraint) – but worth the read.
But if reading philosophy is not your idea of a fun pandemic activity and you’d like a tiny primer into her work, I strongly encourage you to watch this 27-minute video of her presenting on emergence, constraints, and closure. Watch it first and then come back to these notes, for I am going to summarize her ideas and bring them into more common applications. I’ll probably end up carving massive holes in her thinking – so feel free to correct my takes here – but here’s what got me thinking.
Juarrero presents on three main topics, emergence and constraints, context-free and context-sensitive constraints, and closure.
Emergence
Here are her main points:
- Nature uses constraints to generate emergence and sustain it. Constraints both limit and enable.
- Evolution selects for resilience, adaptability, and evolvability.
- Resilience is sustained by micro-diversity.
- Ecosystems are sustained by distributed control rather than governing control. The key is in the links.
- The emergence of novel practices – innovation – cannot be caused, but novelty can be enabled. You do this by catalyzing conditions that allow innovation to occur.
- Think of constraints as phenomena that change the likelihood of things, and the probabilities of what is going on.
SO the conclusion from this section is pretty straightforward. One cannot simply say to people “INNOVATE!” and expect emergence to happen. In order to create the conditions for novelty, one must change the interaction between the people in the system. You can do that in any number of ways, by changing a constraint. Everyone will be familiar with what happens when you are given a task with a constrained amount of time in which to complete it. The pressure of a deadline sometimes creates the conditions for novel practice. By cutting your available time in half, you will discover that a solution that requires an hour will not work, and you may discover that you can find a way to do the task in 30 minutes.
Folks are. discovering this all the time right now. Being forced to work from home is suddenly creating all kinds of novelty and innovation. Many people are discovering that the commute is simply not worth it. Some are finding that they cannot do their work from home and so must find new jobs or new ways to do what they did before. Being forced to isolate has created the conditions for emergence and innovation, and not all of it is successful. Complexity-informed governments have created temporary universal incomes to enable people to be safe to fail. This is not the time to force people to “stand on their own two feet.” If you want people to stay at home, you have to enable them to do that in order to disrupt the pandemic, otherwise, they will have no choice but to head out looking for jobs, thereby increasing the spread of COVID-19.
Context-free and context-specific constraints
This is important and dense stuff, and Juarerro gets this from Lil Gatlin who wrote about it as far back as 1971, but here are the main points:
- In a system, the probability that something will happen vs. something else happening is due to constraints.
- A system with no constraints is “smooth,” in other words there is an equal probability of anything happening.
For example, if I give you a random number sequence like 761893826544528… what do you think the next number will be? In a random system, there is an equal probability that the next number will be between 0-9.
Now If I give you this number sequence: 123456… there is a much higher probability that the next number will be a 7. Why? Because the way to make some things more likely than others is to provide constraints. In this case, the constraint is your bias that the number sequence is not random and you are entrained to expect a 7.
So then what of constraints. Juarerro says:
- All systems come with built-in probability: it’s more likely to be one thing or another. Probability is determined by two types of constraints: context-free and context-sensitive.
- A context free-constraint is like a bias, or an assumption, or a preference.
- A context-sensitive constraint is something that is conditional on a state in the context.
For example, you might say “I like walking on the beach” and that is a context-free constraint that might help you get a date. But a context-sensitive constraint like “If it is raining, I hate walking on beaches” is helpful for your date to know so they don’t invite you out for a beach walk on a rainy day, thereby ruining the chances of romance.
(“But you said you liked to walk on beaches!” is not an endearing thing to say to a waterlogged and miserable partner)
This is useful for innovation because a context-free constraint – like a shared purpose – can help give a sense of direction to work. Developing a new shared purpose will cause some things to be more likely than others. If you decide to stop farming and start building cars, you will be unlikely to be found buying seeds, discussing the weather, or thinking about crop yields. You will be more likely to be focused on supply chains, manufacturing efficiency, engineering, and roads. But in both cases, the higher level context-free constraint is the need to make money.
Context-sensitive constraints begin to give a system coherence. A context-sensitive constraint creates an interdependence or an interrelationship between to parts of a system. Hating rain makes one’s mood dependant on rain, and that can govern or enable a whole set of behaviors. If you end up with a friend who loves rain and one that hates rain, the probability of enjoying each other outdoors on a rainy day decreases radically. But it also means that two people may find that they both love being indoors playing board games while it is raining outside. Sustainable long term relationships are dependant on people finding novel ways of being together as their context-specific constraints change. This is called resilience: the ability to maintain coherence while changing.
Juarrero then talks about some useful kinds of constraints:
- Linkages and relationships: innovation requires interaction and collaboration and interdependence among what will become the components of a larger system.
- Catalysts: things which, given their presence, make other things possible. Catalysts act to break patterns or to create new ones and can sometimes become attractors in their own right.
- Feedback, especially positive (reinforcing) feedback between parts in a system which increases the likelihood of emergence.
- Rhythm, gait, cadence, sequence, order, and timing – temporal constraints – which are very helpful context-sensitive constraints that make things interconnected and interdependent in time as well as space.
In my work as a facilitator and a consultant that helps people innovate, I catalog these attractors with the ABCEI acronym, standing for Attractors, Boundaries, Connections, Exchanges, and Identities. These constraints can all be found active in systems and sets of problems. When people tell me that they are “stuck” we can usually find some of the constraints at play that are causing that state of affairs. Once we have put our finger on something, it’s a good idea to try catalyzing that constraint to see if we can break it or tighten it as need be, to create the conditions in which another course of action is more probable.
For example, today I was coaching someone to use Zoom. She had read the documentation and watched videos, but she had context-specific questions about the application. Clearly she needed more connection with someone who had more experience than she did. So I tightened that connection with her and focused the exchange of information. I started by giving her a tour and I showed her things, but when when I was going too fast she slowed me down, and ask me how she could do those things. Responding to this new constraint on our session – her desire to learn hands-on – I shifted her identity and handed her the power to host our meeting and she took a turn making breakout groups. The whole session took a funny turn when we ended up chasing each other through ten breakout rooms we had created.
By the end of the session, she had enough information to be able to schedule and host a Zoom meeting. She took on the mantle of “Zoom host” an identity that an hour previously, we didn’t even know existed.
Learning like this is emergent and one can work with constraints to discover new ways to teach, new ways to learn and play, and new things to do to address old problems. Constraints-led learning is major field of pedagogy and my friend Mark O Sullivan, a football coach with AIK in Stockholm, is one of the leading proponents of this way of learning skills and teaching the complex sport of football.
Closure
The last part of Juarrero’s talk is about closure, the essential dynamic that makes emergence possible. She says:
- Loops create novelty and innovation. When a loop closes, what emerges is cohesion and cohesiveness.
- Autocatalytic, circular causality and closed positive feedback loops generate novelty.
- Parts interact and when the loop closes, an emergent whole is created, and when that loops back it influences the parts: cultures, systems, organizations, communities, identities,. These are all cohesive and influence parts that come into the system.
Stuart Kauffman’s work on evolutionary biology and autocatalytic systems describes this process beautifully. Essentially the ancestors of all living things are small contained systems of molecules that act on one another. A interacts with B to create C and C interact with A to create B, and suddenly you have a coherent system that “creates itself.”
At the cultural level, look at the way that feedback loops and closure create communities online, for better or worse. In highly partisan contexts, “echo-chambers” are simply autocatalyzing social systems, where biases are reinforced, shared purposes are strengthened and new identities are formed and stabilized. This can create such deep attractor wells into which people fall, almost like cults. Family members can no longer relate to them, they become unable to work with people who are different than they are, especially those who are considered “the enemy.”
Closure creates identity and landscapes of mountains and valleys that Juarrero talks about toward the end of her talk. A mountain might represent an idea that is unthinkable – having dinner with your racist uncle – and a valley might be a much easier, more preferable, and more possible outcome, such as going to a rally for racial justice with your friends. The way in which constraints have closed and looped and fed back information to you in your life will determine which of these two scenarios is most probable. When you choose dinner with your uncle. everyone will express surprise. They never saw that coming. You must have climbed a mountain to make that possible.
Juarrero ends with a really important point about what happens with context-specific constraints operate in a closed system: you get identities, cultures and mindsets, which themselves become context-free constraints for new things entering the system. If you have ever had the experience of moving to a new place you know this well. On our island where I live we have a “Newcomers Guide” that talks about practical realities of becoming a Bowen Islander. It contains a helpful mix of tangible facts – like where the school is, and how to check the ferry schedule. But it also contains insider information about the emergent characteristics of Bowen Island life that have grown out of our interactions with each other and our environment over many decades. These include things such as “Someone flashing their hazard lights in the rearview mirror is not being a jerk. They are a firefighter on their way to a call” or “Don’t ask online for whom an ambulance siren was sounding…” The original guide was written in 2016, and I can already see where things need changing, although the heuristics by which one shod live here, seem robust enough for now.
Like everything associated with complexity these three simple concepts – emergence, constraints, and closure – are easy to see, difficult to unpack, and powerful in practice. Go read and listen to Alica Juarrero though, and be grateful, as I am, that someone as brilliant as her has done the heavy lifting for us.
Share:

From the time my daughter was born in 1997, my partner and I went hard on studying learning theory to understand how kids learn, what’s good for them and how to support their growth. These little beings don’t come with instruction books. It’s hard enough to learn how to feed and maintain them, let alone figure out how to help their brains and hearts grow.
We studied for a lot of years and gradually landed on the work of John Holt, an educational psychologist who, in the 1960s and 1970s, studied how children fail in the Boston school system. Motivated by that work, he later wrote a book called “How Children Learn” which was a seminal text in what became the movement of “unschooling” or “life learning.” This is, to some, a radical approach to homeschooling children.
In the early 2000s, along with a few other families on Bowen Island, Canada where we live, we created a publically-funded homelearning support community called Island Discovery Learning Community. There, our children could come together with other kids and adults, with teachers and resources, and even with curriculum and assignments, to engage in self-directed learning in the community.
Unschooling is a serious commitment and we did this with our children until they were 13 and 10 respectively, following their leads, and guiding them until they chose to go to school. At that point, we treated their choice as another step in their learning journey and at the end of every year, checked in with them about whether they wanted to keep doing that. They said yes, and have both since made their way into university – our daughter first as a jazz musician and now studying psychology and criminology, and our son going part time to explore subjects that might interest him, currently focusing on economics.
I share with you this history so you know that I have some experience in what many of you are facing right now. Kids at home, not feeling like you are qualified to teach them anything, not knowing what to do and maybe even afraid that without school they will be set up for failure in life. It’s all real.
SO to give you some hope, I want to share a few key principles and practices that work when you are homeschooling kids. Your mileage may vary.
First, relax. Even if your kid took a whole year off school, it is not going to lasting damage to them. You are not falling behind, and your kids isn’t losing an advantage by spending a tremendous amount of time away from a classroom. Things will be fine. Trust me.K
Don’t replicate “school” at home. This is a recipe for failure. Your home is not a school and probably the last thing your kid wants is a full scale conversion of their living and playing space into a school run by a nervous parent who is trying to replicate a mass education institution with no good grounding in theory or practice. Your home needs to be a home, especially now, and it needs to be a place of safety and security and love for your kids. Try to avoid doing things that place pressure on your relationship and that cause the child to become angry, resentful, or distant. If your school district is giving your child work, make sure it doesn’t take up the whole day. Remember that they need time to goof off and let off steam. So do you, probably.
Notice that you are all learning all the time. Leaning always happens best in context. Your kids will have ample opportunity to practice reading, math, epidemiology, art, music, video editing, writing, research, cooking, animal care, mutual aid and support, ideation, design, technical skills acquisition, and life skills right now. Just like they do every day. Just like you do every day. Learning doesn’t stop, especially in a context that is always challenging and offering up new experiences. What you can do is take time to notice what they are learning, collect examples of their work and build a portfolio together. Homeschooling families do this all the time because if you never go to school, this is how universities court you to attend their programs. On your body of work.
Kids learn at different speeds. For busy parents who are not intimately involved in their kids’ education, it might come as a surprise to realize that your kids all learn things at different speeds. Our son taught himself to read at 4 years old. Our daughter didn’t start reading until she was 10. They both learned to read in a couple of weeks when they were ready to. If you are getting homework from the school and it seems to be taking your kid ages to grasp a concept that is because it takes them ages to grasp a concept. They might not even be ready to grasp it. They are not broken. There is not something wrong and they are not “losing.” You might need to put aside that concept and do something else. Don’t forget there is nothing essential for them to learn right now in this moment. You could spend months trying to teach a kid something when they aren’t ready to learn and find out that a year or two later, they get it right away. Don’t force it.
Adopt this simple pedagogy: STREWING AND CONVERSATION. Seriously, these two practices took us through a decade and a half of support our children’s learning. Strewing means that you flood you environment with interesting things – books, websites, podcasts, videos, games, challenges, work, interesting people – and you watch to see what they attach to. When they show some interest in something, engage them in conversation with genuine curiosity. Ask them questions so that they can teach YOU about the topic. Don’t quiz them or judge where their attention goes. Even if they spend hours playing Fortnight, get in there with them and understand what they are doing. Ask them questions about how they make decisions, come up with a strategy and work together. I daresay that you will learn something from having them teach you about situational awareness, rapid-cycle strategic iteration, and real-time collaboration.
Love them above all else. Can I just bluntly say, that being a parent right now is fucking hard. You’re not failing if you’re feeling that. Your kids are anxious, worried, and carrying a lot as they move through this disruption in their lives. They can’t see their friends and they are possibly even beginning to hear stories of people they love who are getting sick. If they can’t focus on schoolwork, don’t force them to. These are traumatizing times. What they need right now is probably a good hug and a cry. I’m not sure that is an age-dependent need, actually. The most important thing of all is to love them and care for them right now. Make them as happy as possible right now, because that is what will help them stay resilient, and that is the most important thing.
Share:

Every year, to celebrate St’ David’s Day, Dave Snowden has shared a series of posts on the evolution of the Cynefin framework. This year he introduced the newest version. The framework changes, because as we use it, it has an evolutionary journey towards “better” and more coherent. Not every branch in its evolution has had helpful components, but I find the current iteration to be very useful because it is both simple to use, easy to introduce, and yet has quite a bit of depth.
During the pandemic, I’ve been using this version of it to help people think about what to do and this is how I propose to tour you around it as well.
First, it’s helpful to orient people to the framework. To begin with, it has five domains: the one in the middle, plus four others. It’s helpful to think of the domains as a slope, starting high in the bottom right and tapering counterclockwise around to the bottom left. The domain in the middle is the most important for me, and the most underappreciated. It is the domain of Confusion (it used to be called Disorder). The domains on the right side are “ordered” meaning that stuff there is largely knowable and predictable, and problems are solveable. These Clear or Complicated domains are, distinguished by the number of interactions going on – the more parts in the system, the more Complicated it is – and the level of expertise required to know what the answer to a problem should be.
The domains on the left side are “unordered” meaning that situations are unknowable and unpredictable. This is the world of Complexity and Chaos. These are distinguished by the way the system changes, self-organizes, and creates emergent phenomena. Complex systems exhibit emergence and self-organization, and Chaos exhibits the lack of any meaningful constraints a sense of randomness and crises.
The further you go counterclockwise, the more unordered and unstable the system is. If you go clockwise, you introduce stability and order to the system. Stability lies clockwise of where you are now and instability lies counterclockwise. It is important to note that this is true until you get to the boundary between Clear and Chaos. That is like a cliff. One falls off of the Clear domain into chaos and it is difficult – if not impossible – to recover and clamber back up to the well-ordered world with Clear answers.
Most helpful for understanding strategy and the use of the framework is understanding how constraints work. From Clear to Chaos, one can move through the framework using constraints: Clear systems have fixed constraints that can break catastrophically and can be repaired easily f you know what you are doing. Think of a water leak. If you know how to repair it, it is a simple matter to do so. If you don’t, you fall off that cliff into Chaos quite quickly, and it takes a lot of time to get back to normal.
Complicated systems allow for a little more latitude in practice and so have governing constraints, such as laws and procedures. Break them at your peril, but also discuss them to make sure they govern activity in the system well.
Complex systems are characterized by enabling constraints which give rise to all manner of creativity, emergence and self-organization, but which can also be immutable. Think of the laws of physics or principles of evolutionary biology that seem to generate a huge variety of systems and living beings. But we don’t have a creature that can breathe by oxidizing neon, because neon doesn’t oxidize.
Constraints in complexity can be quite tight and still contribute to emergence and creative action. Think of the way the rules of the haiku form don’t tell you what to write, but instead offer guidance on the number of syllables and lines to use: three lines of five, seven, and five syllables. These simple constraints give rise to tremendous creativity and inspiration as you work to create beauty within a distilled form.
In Chaos the absence of constraints means that nothing makes much sense, and all you can do is choose a place to act, apply constraints and quickly sense what comes next. This is what first responders do. They stabilize the situation and then figure out whether a technical expert is needed (to operate the jaws of life) or whether the situation needs to be studied a bit more (so we know how a pandemic actually occurs and the different ways a new virus operates in the human body).
That’s the basic orientation to the framework. There are additional features above that are helpful to note, including the green zones of liminality and the division of the Confusion domain into A and C, standing for Aporetic and Confused. Aporetic means “at a loss” and indicates an unresolved confusion, or a paradox, which is just fine. Sometimes things need to remain a little murky for a while. “Confused” refers to the state of mind where you just aren’t getting it, and you don’t understand the problem. It’s often the result of a failure to see past one’s own biases, habits, and entrained patterns of solving things.
Contextualizing your problem
One meaning of the Welsh word “Cynefin” is “places or habitats of multiple belonging.” The name of the framework references the fact that in any situation of confusion, you are likely to have all five types of problems or systems at play. So when you are working on trying to understand a situation, start by assuming you are in Confusion. As much as it is tempting to look at all situations related to COVID-19 right now as Chaos, they aren’t. In fact, the desire to do see them that way is actually a key indicator that you are in Confusion. When I am teaching this framework, I sometimes label this domain “WTF?” because that is precisely what is happening here. We don’t know what’s going on, we’re confused, and we’ve never been here before. Any data you collect about a problem should all go into the Confused domain first.
From there you can ask yourself where things belong. This is called a Cynefin contextualization and is a core Cognitive Edge method for working with Cynefin. It works like this: you literally put as many aspects of your situation on individual post-it notes as you can, put them in the middle of a table and sort data into basic categories according to these criteria:
- If the aspect is clear and obvious and things are tightly connected and there is a best practice, place it bottom right.
- If the aspect has a knowable answer or a solution, has an endpoint, but requires an expert to solve it for you, put it top right.
- If the aspect has many different possible approaches, and you can’t be sure what is going to work and no one really has an answer, put it top left.
- If the aspect is a total crisis, and you are overwhelmed by it, put it bottom left.
- If you can’t figure out which domain to put the aspect in, leave it in the middle for now. NOT EVERY POST IT NOTE NEEDS TO GO IN THE FOUR OUTER DOMAINS.
Now you have a table with five clusters of post-it notes. You can do lots of things with your data now, but for me, the next step is to have a look at the stuff on the right side. Make a boundary between the stuff you can do right now (Clear) and the stuff you need an expert to help you with (Complicated). You can cluster similar pieces of data together and suddenly you have little projects taking shape.
In the top left corner (Complex), make a distinction between things that are more tightly constrained and things that are less tightly constrained. Think of this domain as a spectrum from closed to open. For example, moving my work online is constrained by needing a laptop and some software, and a place to work and some hours in the day to minimize interruptions. Those are fairly tight constraints, even though I know that I’m not going to get it right the first time around and no expert will solve it for me. I have to make it work for my context. Figuring out how to manage a team of eight people from home is much less constrained, and even comes close to chaotic. So that gives you a sense of the variety possible as you move from the boundary between Complicated and Complex and the boundary between Complex and Chaos. And you can see now why the liminal spaces exist there too. It’s not always clear cut.
Anything else on the left side that is overwhelming is in Chaos, so leave it down at the bottom left. If it is an actual crisis, you probably should take care of it right now and then come back to your framework later!
Stuff that is still confusing stays in the middle and you might want to take a crack at sorting things into Aporetic and Confused. An example of Aporetic might be trying to figure out whether you have the virus or not without being able to get tested. Because you can’t know for sure, you have to hold that knowledge in suspension and let your actions be guided by the idea that you might have it, but you might not too. But you might. You just can’t know right now.
So now you have options:
- For Clear aspects, just do them. Don’t put them off either, because failing to do so will drop you into chaos. WASH YOUR HANDS OFTEN FOR 20 SECONDS WITH SOAP. That’s an order. Orders work well here.
- For Complicated aspects make a plan. You might be able to find someone to help you learn the technical aspects of setting up a zoom meeting. You’ll definitely find videos and technical documentation to help you do it. You can learn that skill or find someone who knows it. This is what is meant by Sense-Analyse-Respond. Do a literature search, listen to the experts, and execute.
- For Complex problems, get a sense of possibilities and then try something and watch what happens. Figuring out how to be at home with your kids is pure complexity: you can get advice from others, talk about with friends and strangers, read blog posts and tweets, but the bottom line is that you need to get to work and learn as you go, engaging in a rapid iterative cycle and see if helpful patterns emerge. As you learn things, document practices and principles that help guide you in making decisions. If rules are too tight, loosen them. If the kids need more structure, apply it. Finding those sweet spots requires adaptive action, and learning as you go. Here we talk about Probe-Sense-Respond. Don’t worry about collecting tons of information before acting: it won’t help you past a certain point. Act on a hunch first and monitor the results as you go.
- Liminal complexity means that you are choosing to enter into proximity to either Complicated or Chaos. if you apply constraints (like enforcing rules on the kids) you are moving complexity towards the ordered domains. That might work, but too much rigidity will create problems. On the other hand, if your constraints are too rigid you may find yourself unwittingly creating patterns that make it hard to flow with the changing times. And so you release the constraints until you can discover something new and helpful, and then apply constraints again to help you manage in complex times. An example might be adopting the assumption that you are a carrier of the virus and letting that assumption guide your behaviours. That helps you to make choices that will probably benefit you and the people around you. (And here are some heuristics to practice with if you have kids at home during the pandemic)
- For chaos, you are going to need to apply constraints quickly and maintain them until the situation stabilizes. That might mean self-quarantine if you are infected and sharing a house with others. It might mean relying on emergency services to impose those constraints for you.
- For confusion, think of this as the top of the fountain and as new data enters your system, add it there until it trickles into the right domain. I like to revisit things that are in this domain from time to time, because as I get to work on stuff, sometimes my confusion about other things disappears, or sometimes I find a true paradox that can never be resolved and those are delightful in themselves.
So, to conclude
In summary:
- Cynefin is a five (expanding to seven) domain framework. Whatever you are doing probably has aspects of all the domains at play at any given time.
- If you need to learn something, or discover new things, loosen constraints. If you need to stabilize a situation, tighten constraints.
- In the Ordered domains, rules, laws, and experts are helpful and should be obeyed. In the unordered domains, principles and heuristics should be adopted that are coherent with goodness, safety, and care, to guide behaviour and learn new things.
- In chaos, stabilizing the situation is most important. Act now to restrict your actions and once things are stable, make the next move.
Be careful, be aware, be connected, learn and share as you go. None of us have been here before, so offer grace and support. Try to look at what is happening and suspend your judgement. Don’t spread information unless you know it is reliable, and help each other out.
Share:

I travel around many different kinds of organizations. Many of them preach the mantra that goes something like “it’s okay to fail here. Please take risks and try new things!” Unfortunately, when I look around I can’t see much infrastructure in place that allows the work context to be safe enough to fail.
An organization needs to build learning and experimentation into its operations, especially if it is required to respond to changing conditions, improvements in services, or new ideas. And so the idea that “we want people to take risks” is promoted, often alongside an exhortation to do so prudently but really with no further direction than that.
Anyone who has worked in a large organization will know that risk-taking is perilous. There are many ways to be punished for doing something wrong, and the worst punishments are the invisible ones: shaming, exclusion, a tattered reputation, eroded trust, political maneuvering that takes you away from access to power and influence. Not to mention the material punishments of reduced budgets, demotions, poor performance reviews, and limited permission to try new things in the future.
Failure in context
Before going any further, let’s talk about what I mean by failure. Using Cynefin, we can focus on the difference between failure in complicated contexts and failure complex contexts. When we have a complicated failure in a stable and linear and predictable system, the answer is to fix it right away. Ensure you have the right experts on tap, do a good analysis of the situation and apply a solution.
In complex adaptive systems, failure is context-dependent. Here failure is an inevitable part of learning and doing new things. Because complex problems demand us to create emergent solutions, we are likely to get somewhere when we can try many different things and see what works and what doesn’t. Dave Snowden calls this “safe-to-fail” and it means taking a small bet, based on a hunch that what you are doing is coherent with the nature of the system and where you want to go, and acting to see what happens. If it fails, you stop it, and if it works, you support it.
I think I once heard Dave say something like “probes in a system should fail 8 out of 10 times or you aren’t trying to find emergent practice.” That is certainly a rubric I find helpful. This means that in developing new things, you should expect to fail 80% of the time and to do that requires that you put into place a system for supporting failure and learning.
Stuck on a cliff
Imagine you are free rock climbing – no ropes or belyaing – and there is a handhold you are reaching for that requires you to do something you’ve never done before. Your partner says “you’ll never learn to solve this problem if you don’t try something. Don’t be afraid to fail.” Far from being imbued with confidence, you are likely to be frozen with fear, seeing all the ways that things could go wrong. Better to just stick to what you know, and don’t try the move.
If however, you are in this same scenario, but you are roped up and belayed by someone you trust, you can feel safe to try the move knowing that if you fail, you will be caught and you will have a chance to try a different strategy. As you develop mastery in the move, you can use it more and more in your rock climbing life, and you may loosen the safety constraints as you develop more capability
Implications for facilitation and leadership
Safety is about creating good constraints so that your people can take risks and know they will be safe if they get it wrong. The job of leaders is to set the constraints for action in such a way that a safe space is available for work. This can take the form of limited time, money, the scope of action, or other things so that folks know what they can and cannot do. Within that space, leaders need to trust people to do their learning and create feedback loops that share the results of experiments with the bigger system. If you can have people all working separately on the same problem – working in parallel as we would say in Cognitive Edge-speak – then you increase the chances of lots more failures and also of finding lots of different ways to do things. This is called “distributed cognition” in complex facilitation and keeping people from influencing each other increases the creative possibilities within constraints.
The next level of this practice is to honestly incentivize failure. Give a reward to a person or a team that has the best report of their failure, the one that helps us all to learn more. You could easily do this in an innovation meeting by having different groups work on a problem in a fixed amount of time. Watch for the group that fails to get anywhere by the end of the time and ask them to share WHY they failed. Their experience will be a cautionary tale to the whole system.
Almost every organization I work with says that they embrace learning, tolerate failure, and want their employees to take more risks. When I ask to see how they do this, it’s rare to find organizations that have a formal process for doing so. Without that in place, employees will always respond to these kinds of platitudes with a little fear and trembling, and in general, take fewer risks if it clashes with their stated deliverables.