Dave Snowden

The Origins of Cynefin - Part 5

RSS Feed

Screen shot 2010-07-10 at 12.55.46.pngThe Cynefin framework is frequently (and legitimately) used as a categorisation model around the four domains of simple, complicated, complex and chaotic. Working at this level it allows people to understand the difference between the four domains, the decision models associated with them and the necessarily limits of best practice. Shawn Callaghan of Anecdote produced a four minute explanation of Cynefin considered as a categorisation model which gives a good basic introduction and has proved popular. For a lot of users that level of use is more than good enough to produce results. Adding in disorder and the catastrophic boundary adds meaning when a more sophisticated approach is needed but it's not always necessary.

At its most sophisticated, and in full operational use, Cynefin starts life as a sense-making framework not a categorisation model. In a sense-making the framework emerges from the data, while in categorisation the model is pre-given. The advantage of categorisation is that it is efficient, the danger is that if the context shifts then it may result in significant category errors. Now this is best understood by describing the process by which it is constructed using the narratives of an organisations' past perspectives and possible futures. The domains, and the boundaries between the domains are defined by narrative as that is the primary sense-making mechanism by which we create common understandings; it has boundaries because without boundaries humans will not distinguish between different types of action and analysis. We are not good with gradients.

A brief aside on illegitimate approaches

Screen shot 2010-07-10 at 13.45.41.png

Screen shot 2010-07-10 at 13.42.40.png

Cynefin is even drawn as a cross from time to time, with the occasional token inclusion of a circle. Most of the time I live with this although every now and then we get absurdity. The danger of making things “fit” regadless of loss of meaning is well illustrated by the two illustrations that flank this paragraph. I didn't know whether to laugh or cry when I saw the allocation of data, information, knowledge and wisdom to the four domains. Managing in any of the domains without data would be absurd, but I suppose confining wisdom to randomness says something about the judgement of the author.

The misuse of the OODA model is equally foolish. The OODA cycle is all about making decisions, and decisions have to be made in all domains. Looking at how the OODA loop would apply in each domain is interesting, in fact I will make it a future post. Observation of what? Orientation to what? It will differ by domain. Looking back it was those two examples that finally shifted me from irritation to sympathy with the consultant concerned. The examples display a level of profound ignorance of the fact that Cynefin is a framework that allows you to understand different types of system (ontology) not different processes or things which may be common across all domains albeit with different attributes.

The desire to create neat ordered and tidy structures (in the case of this author even the curves have been regulated), to cross map any model you like the look of onto every other model is a part of the tendency to universal solutions that has been all too common in the last few decades. It's the sign of a failure to understand complexity and more importantly the aesthetics of uncertainty.

Enough; lets get back to the real thing, not shadows on the wall of a pit from a flickering candle stub.

Social construction of the Cynefin framework

Over the years a range of techniques were developed to allow the boundaries in the Cynefin framework to emerge from the data. The one which (for me) is the most authentic is illustrated below. There are simple approaches which involve some social construction but are primarily categorisation based such as the four tables method. This is used in a lot of strategy and conflict resolution work. Long term use, and its instantiation as part of a new language of strategy within an organisation requires a bit more investment of time and effort upfront; although its a lot less than most of the approaches that characterise the systems dynamics period and critically avoids reductionism.

Screen shot 2010-07-10 at 13.12.53.png

This method, known as four points involves a pre and a post process and can be summarised as follows:

  • Pre-process: generation of several hundred examples of exemplar narratives of key moments in the organisations own history, alternative histories and imagined futures. This can be done using another method Future Backwards , but it can also be achieved by brainstorming or, best of all, by a broad capture of identity micro-narratives using SenseMaker®.
  • Workshop: (virtual or physical), can be parallel process with subsequent synthesis
    1. A representative group are asked to select the four exemplar narratives that define the extreme states of the Cynefin framework (although the framework is not explained). Instead the language used is along the lines of the case where the right answer is most self-evident; that where experts or due process should be able to produce the right answer; the case where with the benefits of hindsight we would all know what to do, but not in advance; the most chaotic/random/unexpected event.These four exemplars are then placed on the extreme corners of a large workspace, ideally a vertical one with lots of natural light and space for people to move around. In a virtual environment this can be done through polling.
    2. That complete, each subsequent narrative is placed onto the work space in dynamic tension between the four corners and also with all the other narratives. This can take time and should not be rushed, people should be allowed to modify the narratives or create new ones as they occur to them.
    3. Once all the items are allocated then boundaries are created using ribbons Do not under any circumstances allow people to draw them with a pen, in fact I remove all marker pens from the room before this stage to remove temptation from the alpha-males, or the even more scary alpha-females that now hover around the glass ceiling in many a corporation. The boundaries are drawn around those items which are unambiguously in one of the four states described in step 1 above. We can now see disorder, and as illustrated, it is normally a very large domain at this stage. Our objective is to reduce disorder to allow an authentic and ontologically aware decision making process so we move onto the next stage.
    4. The items clearing in the domains are now split into two, those which can define the space and those which are extreme examples (and negative) The extreme items represent a boundary zone of the domain, back into disorder. Some readers will know that these are known as the illegitimate extremes. That complete the group now proceed to bifurcate, trifurcate or quarter the items in disorder to create the defining boundary objects between the four major domains. This over we have a framework, the boundaries of which have emerged from the data and which allows us to define domains and boundaries in language that is understood within the organisation, as it is comprised of the identity narratives of that organisation. This contrasts with other strategy models and processes which are defined in abstract ways, or using case based examples, in the main from US manufacturing industry.
  • Post-process: Now the model is in place it can be incorporated into training programmes so that it becomes part of the common discourse of the organisations: Hang on, its like these three examples and they are all complex, so we need to create safe-fail experimental probes not analyse, or Its a dead ringer for this cluster so why aren't we simply applying best practice?. Human language is intimately linked with and dependent on common narratives, and the naturalistic approach that underpins Cognitive Edge approaches reflects that. Once established the populated framework can also be used to test for cultural affinity, show different silos how their different perception of the domains is creating conflict or misunderstanding etc. etc. We now have the advantages of a categorisation model, but if the model starts to stretch we can always re-set using the emergent process, moving from exploitation to exploration then back to exploitation again.

Cynefin, especially coupled with the ability of SenseMaker® to provide continuous monitoring and feedback is a major new approach to strategy that is more dynamic than its predessors, but which legitimes those predessors within boundaries.

The first full roll out of the above method was in a government workshop held over several days in Singapore. I persuaded Cynthia to travel for the event and we worked together to refine the approach. We also used Cynthia's great invention (although I claim the origination of the name) of butterfly stamping as a pre-process. One further process we used during that Singapore workshop was to get people to create a metaphor based description of each space. That worked well but I didn't really take it forward. However the idea was planted, and it now forms a part of new work on metaphor based command languages of which more at some future date. The other major development which came from this approach was the question of sub-domains, and with that the idea of Cynefin Dynamics which I will deal with tomorrow.

Why this method is important

One of the general issues that emerged in discussions between myself and Cynthia (along with others) as part of the seeing eyes interaction (of which more in tomorrow's post) was the question of boundaries. In a very real sense the method above is a result of that fruitful debate. Boundaries are necessary for human sense-making. If we gave people a spectrum from chaotic to stable then people would settle in the place of their most comfort. If we create boundaries, then if we can create a first step which involves a choice as to which side of the boundary we are (backed up by narrative based definition which is amenable to coherence based evidence). With that done it is a lot easier to get people to accept that in a particular context they should do something they are otherwise uncomfortable with. By socially constructing the boundaries from an open space we enable people to see things in a novel and interesting way, something that imposing a two by two categorisation framework would never achieve.

Top