Dave Snowden

Liminal Cynefin, stepping over the threshold

RSS Feed

I’ve been frantic the last few months with a lot of long standing wishes and dreams that are all coming together at the same time. The sort of thing that you initially say is the sort of problem you want to have, but when reality strikes one’s view is rather different. More on those in future posts, but part of what I have been doing in idle moments on walks, planes and in the boring parts of meetings is playing with the liminal version of Cynefin. I’ve said before its one of the most important changes I’ve made, and the framework, as people know, has been through many changes.

Part of the power of Cynefin is that it is multi-layered in adoption. It can be used as a simple two by two, and often is when people get started. The five domain model is increasingly used with the need to remind people of disorder and more people are using dynamics – the second classification within the Cynefin eco-system. The essence of a good sense-making framework is that it can be used at multiple and growing levels of sophistication without loosing its original simplicity. So we are ready for the next stage, which builds on both domains and dynamics and to some extend replaces some of the domain models (the three by three matrices for each domain some will be familiar with).

The motivation for the liminal version came from a desire to create a visual representation of the reality of complexity, namely that it is not one state. I also increasingly realised that many of the methods and tools that are valuable in this space (like SCRUM) actually don’t fit within domains but are shifts between domains. Doing some of the background theoretical work on project management I also started to see the boundary between complex and complicated as a a domain in its own right. Earlier posts on this were over complicated as I started to think it through. But a few weeks ago I simply drew a single line and shaded the areas and it all came together. Still some refinement and we will hold a premium member seminar on it shortly so that might be a motivation to sign up! It will also feature in the Cynefin Retreat (only seven places left)


Liminality (from the Latin word līmen, meaning “a threshold” is the quality of ambiguity or disorientation that occurs in the middle stage of rituals when participants no longer hold their pre-ritual status but have not yet begun the transition to the status they will hold when the ritual is complete. During a ritual’s liminal stage, participants “stand at the threshold” between their previous way of structuring their identity, time, or community, and a new way, which the ritual establishes.

I pulled the above from wikipedia in the context of anthropology and while it’s not complete it gives us a good starting point. The reason I called this the liminal view of cynefin is that it introduces the concept of induced and manufactured threshold states between complexity and its two adjacent domains.

The liminal state between Complex and Complicated is the state of increasing constraints to create authentic repeatability that allows for scaling as is. Loosely translated that means that parallel experiments are starting to fuse and produce consistency over time so we can now assume a degree of predictability. In this state we continue to iterate ideas but reduce the parallelism of experimentation in the complex domain. The longer we can hold things year, the lower the risk of premature convergence on a solution (or solutions), but the more time it gets to move from exploration to exploitation. So there a a trade off here, but relatively low risk.

The liminal state between Chaotic and Complex is what some people mean when they talk about the edge of chaos; Reducing the constraints, wearing the connections until a state of randomness is reached. This should not be confused with the complacency zone in the obvious domain which results in a catastrophic collapse into chaos. This is the zone of deliberate movement to create sufficient ambiguity for true novelty to emerge. This is difficult to do in the concrete as old patterns assert themselves to quickly, hence the use of abstraction to prevent that. A body of SenseMaker® and Cognitive Edge methods relate to this and overall its a safe space. The energy gradient is to complexity not to chaos.

More on this tomorrow when I will update the constraint work and then start to cross link back. So consider this a starting post …


The image that opens this post is from the 2011 Burning Man work Liminality by aleXander hirka

  • Andre Heijstek

    When I teach Scrum, I always link it with Cynefin. For me, some parts of Scrum fit in the Complex domain: especially the realisation that you cannot predict the business value/impact of a sprint. So Scrum events like Sprint Planning and Sprint Reviews where you try to probe business impact belong to Complex.
    The execution of a sprint is more of a Complicated activity. It is like running a project with a fixed team, fixed scope, fixed time, with a group of experts. Nothing really complex here (but definitely not Obvious).
    Would you agree with this?
    Given this interpretation, I can imagine that some sprints, where we build novel features are clearly Complex, whereas other sprints where we continue to develop existing features are in the boundary between Complex and Complicated.

    • Dave Snowden

      The iteration of a single strand rather than parallel strands makes SCRUM complex becoming complicated in my view

      • Andrea Tomasini

        One way to come with it is to define a high level Goal for the Sprint, as a direction or intent. Then chose Options from the backlog (as User Stories, Spikes or whatever else) and work with pairs in parallel during the Sprint, with minimal synchronisation. This allows for a sort of multiple parallel small experiments happening, towards achieving a common Goal, which is per definition, not well defined, nor shaped, but provides a direction. Nobody has a clue how to get there, just some ideas about what could work. To me this is the nature of Scrum, it is the way I teach people to do it, because if we understand R&D, and creative activities, making upfront decisions by describing solutions to problem we don’t yet understand, is not only ineffective, but is also very bad risk management… :-) But you know I am different 😉

        • Anthony Green

          What you’ve described is the Complex to Complicated transitionary phase. Deeper into the Complex domain the high level Goal will be unknowable and emergent.

    • Anthony Green

      Complex represents more that the inability to predict business value/impact of a sprint. The emergent nature of knowledge and the co-evolutionary nature of catalytic probe and system response means that you can’t make an upfront commitment to deliver something (one of the key tenants of Scrum) because that ‘something’ does not yet have a consistent form, moreover there be may contradictory, competing evidence suggesting you do something else, to which you need to pay attention and explore.

  • http://kneaver.com/bruno BrunoWinck

    I mentioned on Twitter that it reminds me of the knot theory. A few weeks ago I was scanning my (old) post-grad courses and some drawings are similar. The semantic is totally different though. The Cynefin drawing can be seen as 4 + 1 regions. It can also be seen as a set of lines. Those lines could be seen as part of a larger drawing forming a knot (a knot because they cross or seem to cross in 2D). The first time I saw the liminal I though of moving lines, not sub regions of the main regions. There is another way to see it as a projection. You expand the lines in 3D, first by creating general cylinders on them, then by bending and shaping those surface. The projection gives a similar drawing with the major advantage of showing which moves are possible, which are not. The surface build on the lines gives the dynamic (the differential dimension). One can imagine who those surface expand further and more interesting the shape of the boundary of the disorder zone. This is purely a imaginative view, a mental game.
    Picture is just to support my explanation.

    • http://kneaver.com/bruno BrunoWinck

      Second though. What I explained before can be a bit complex. An alternative is to imagine the lines drawn on an fantasy landscape. I would see Disorder as deep hole, chaos as the lowest level (rejoin the idea of Energy level/height = potential energy) , border between complex and complicated as moat, Maybe a cliff between simple and chaos. Landscape is a simple image. Play Doh was our tool to ideate.

      Drawing on surfaces is a common technique of Knot theory as you can see on this link (Scroll down a few pages to see more examples). Changing the nature of the underlying surface was the clue for Thurston to solve more knots.

      Knots are knots in the plane, when we place them on surfaces like torus, hyperbolics we can discover new point of view and how they could be morphed into other figures. The whole idea is to search alternate spaces than the plane, even if eventually we use a plane because screens and paper are plane.


  • Anders Dinsen

    The choice of word “liminal” is interesting. Its origin is limit, which is indeed a stage, somewhere we can act and perform on. Being “on the limit” or “crossing the limit” can be a human event with significance and possible outcomes. In a leadership perspective, I want the team to get the most out of that.