This week CE hosted an open webinar on the topic of Cynefin & Theory of Constraints.  This webinar featured CE practitioners and TOC experts, Steve Holt, Jabe Bloom, Greg Brougham, and Hilbert Robinson.  Steve Holt was the lead host and the others served as the panel of experts, some with more Cynefin experience while others with TOC experience and many with both!

We are sharing this webinar openly with all who would like to watch.  To watch the webinar recording we ask for your email address and we kindly request you give our mailing list a try!  You can always opt out 🙂

Here’s a link to the webinar recording.

If you enjoy this webinar be sure to sign up for the full 2-day Masterclass on Cynefin & TOC with Steve Holt and Dave Snowden July 11-12, 2019 in Chicago!  Here’s a link to the Masterclass.

In my final post in this update series on Cynefin and I’m reserving the right to change things when I think about it more.  I’ve been playing with options here on trains to and from particularly dark productions of tragic operas and the experience may have influenced by thinking.  Lets start with a variation on the normal representation of Cynefin which I’ve drawn a few times, but this is the first time I’ve published it.  Here I show the danger zone between obvious and chaos as an extension of the inauthentic aspect of disorder.  In earlier years this was sometimes called the complacent zone.  It is a type of liminality but its one you are not aware of until you fall over the cliff.  So its less a state of suspended transition, more a case of being suddenly suspended with no prospect of recover – remember the Coyote in Roadrunner if you want an illustration.

Many many moons ago, before IBM in my Data Sciences strategy days, the first time I ever got the thought leader  label I created a matrix between known and knowable, and then a second overlapping one between knowable and unknowable.   I used it a lot in C level workshops and it came over into my work in IBM and I presented in a few times in Washington before the language got toxic.  I didn’t originate it and, before you ask, I didn’t get it from Johari windows.  I learnt about that when someone referenced it after a session in the Dental Practice Board.  its not original although the extension to unknowable is I think mine.   When I was working on the HBR Cynefin article Mary and I decided to rehabilitate it and for the domains of obvious, complicated and complex I have made no changes.  However I have shifted chaotic from unknowable unknowables and added labels to the two luminal zones and the boundary with chaos.

I should say that I like this version of Cynefin but I haven’t got the drawing skills for it yet.  It may be more of a brush stroke in which entry into it is imperceptible.   Its now how I would start a discussion but one of the values of CYnefin is that you can dive deeper.   I’ve also got some work to take the domain three by threes I developed for each of the four main domains and add liminality and this language.  That will take a bit of time.  And apologies for leaving the selection box on chaotic when I did the screen shot.  I just can’t be bothered to do it all again!

So lets run through how I am using this:

• In the three main domains the descriptions are pretty self evident.  If its obvious we all know all that there is to be known and we all agree on the pathway.  If complicated then even if we don’t know, we know how to know and we know the range or nature of our ignorance.  Complex gets more interesting as we don’t know what we don’t know but its not inherently unknowable.  If we take the initiative to run parallel safe-to-experiments then we start understand what is possible.   We can use MassSense to to identity different views and also outlier perspectives as another approach to gaining insight here.  
• I decided the question of imagination was important to bring into play here.  So in the chaotic domain we again have unknowables but they are beyond imagination.  In the complex space generally someone will know, its a matter of finding them and paying attention (MassSense again).  But in case its beyond imagination which is why we get stress based novelty.  It is a domain where naïveté may an advantage as it breaks patterns and see things differently.  This is a domain of fleeting possibilities, have seen, rarely realised.
• The question of unknown knowns will be familiar to many people in large organisations.  Getting attention of senior executives is difficult.  See things differently from the interests of your immediate managers and you will never get near the real decision makers.   Also god help you if you publicly went against the norm and then get proved right, no one will ever forgive you.   Distributing situational assessment (human sensor networks, MassSense, anaymous reporting, crews are all ways to do this) make you more resilient and shift away, up through disorder into complex.
• The liminal domain between complex and chaotic is a state of suspension.  We’ve contained interactions, generated skink works or similar so the prospect of knowledge now exists, its no longer unknowable.  But it is as yet unimagined so here micro-scenarios planning, trio based investigate, social network stimulation and lots of other methods we’ve developed over the years come into play.   I think this is one of the most fertile areas to explore and includes the ideas of anticipatory triggers and mechanisms of paying attention to the unexpected.  More on this in the next few months.
• Finally we get the liminal domain between complex and complicated.  We now have an inkling, we’ve imaged what’s possible, we’ve experimented and new possibilities are emerging that we can now stabilise and exploit.

As I said I’m not 100% sure of the labels and descriptions yet – but I’ll develop it and ideas and comments are welcome

Banner picture is a view at dusk from the col between Mynydd Perfedd and Craned y Filiast looking towards Elidir Fair with Snowdon in background.   I’d planned to complete the ridge and descend to Nan Ffrancon but the walking and weather were so good I didn’t want to leave the tops so retraced my route back to Y Garn and down – a long but rewarding day and I wish I had the better camera for this picture.

I was searching for my original known-unknown-unknowable matrix for today’s post and, as you do, came across a variation of Liminal Cynefin I used at and event to talk about the whole issue of control within organisations.   I think it was one of those slides that I put together at a conference where I am getting irritated with a preceding speaker who is either peddling snake oil and/or is fatally naive.  It was probably at an Agile conference as the large red arrow has SAFe and Six Sigma written on it; more on that later.  It’s why I like to tune into a conference before speaking and I’m generally changing and adding to the slides (or notes if I am going without slides) right up the point where I arrive on stage.  Either way it was interesting and available so  consider this a supplement to the St David’s series and I’ll get back to that tomorrow.

Again I am using Cynefin here in the context of what I long ago called the principle of Bounded Applicability; with some exceptions few things are wrong, most are right within boundaries.  To put it another way they are context specific not context free.   The question of what if anything is context free I’ll leave for another day as that one is still worrying me.  The great value of Cynefin is that it provides a quick way to understand, and more critically, apply that principle.   I always emphasis that Cynefin was designed as a conflict resolution framework as much as for understanding decision making – all aspects of sense-making.

So in this use of Cynefin I was looking at three such aspects and one major danger:

How we deal with exceptions, or deviance from the norm

• In the obvious domain there is little excuse for deviance.  In the UK we should drive on the left hand side of the road, in Germany on the right.  But if a child runs onto the road in front of us, then I would make an exception.   Not allowing any variation results in a ‘jobs worth’ mentality that sooner of later ends up on the edge of chaos and a catastrophic failure.  
• For the complicated domain exceptions are more likely, experts with the right qualifications know that few rules or processes are universal and mandating behaviour based on analysis in the centre of a normal distribution missed the fat tails that are more likely in practice.  Here exceptions should allow if they are transparent, ie they can be subject to some type of review and the decision maker knows they can be held accountable.  Falling back to the rule compliance to avoid blame is all to frequent these days and that will result in a shift to the inauthentic aspect of disorder.  Both of the ordered domains can be handled by process and shouldn’t involve senior decision makers on a day to day basis.  
• In the complex domain by contrast every exception needs to be visible fast at a senior level as they may represent opportunities.  I sometimes term this finding the 17% which is a reference to inattentional blindness; we need the minorities who are seeing things differently before they are homogenised into group think.  Something that MassSense was designed to handle on a near real time basis by shifting to the liminal aspect of the chaotic domain.  
• Finally in the chaotic domain itself where we are in crisis exceptions are also important either to kill off fast or to extemporise solutions as they become visible.  Again the need for real time feedback and cognitively diverse situational assessment is key.  We will be launching something to measure cognitive diversity in the near future by the way.

Planning and control

• If you are in chaos (not the liminal aspect) them it’s all about control and little about planning.  Not only that you haven’t got time to be nice or consultative outside of a very small group (give is a good rule of thumb) so the imposition of control is likely to be draconian in nature.  The real value comes in working out what you impose and its shouldn’t be a final solution; the essence of good crisis manage is to create enough constraints that you shift the problem into the liminal area of complexity which means some patterns are starting to emerge.  At that point you can shift into safe-to-fail very short cycle experiments (this can be augmented by the use of IT) to see which future pathways are coherent enough to explore; and you should never just go with one of those.
• Obvious is about about control but it is achieved through creating and mandating process, delegation of authority and so on.  Both middle management and senior executives monitor for deviance but they have different purposes.  The former are focused on compliance the latter on checking if its time to change.   If you are getting lots of reports of deviance it may be that the rules and processes are not longer relevant to the current context and need a radical rethink; a shift diagonally into the complex domain to allow for re-discovery of what is sustainable.   The predictability here is both a benefit and a danger.
• Peer review and acceptance is a key aspect of control in the complicated domain.  Planning and control can be far less explicit because the basics of what are needed are embedded in the professional practice and training.   Ethical and other standards are key and the imposition of hard targets can create compromises here – think of the failure to report minor non-compliance in manufacturing which often arises from rigid targets.  Get that wrong and when things go wrong they will do so catastrophically.  This is a major area of concern in the modern organisation where several decades of excessive codification and measurement have destroyed professional standards and practices that have evolved over centuries to stabile sustainable and resilient decision processes.  Blind trust in experts is equally stupid by the way.
• In a complex system we can only manage the evolutionary potential of there here and now.  Planning is at best a sense of direction rather than a set of goals and targets. Rapid improvisation using and reusing what you already have (exaptation and bricolage) are key and the ability to rapidly mobilise cognitively diverse networks to support decision making vital.   I’ve been arguing for years that we need to create networks for ordinary purpose in times of stability that we can activate for extraordinary need in times of instability and rapid change.  It’s too late when it’s happened.  Rapid feedback loops, fast failure and learning, parallel experimentaion are all key.  The plans and the controls will emerge as you do that and you can then shift into the liminal domain for codification and scalability.  But you really don’t want improvisation in the ordered domains – total waste of energy.

Measurement and Targets

• KPIs abound in government and industry alike but they are always focused on outcomes and pre-determined pathways with little attention to the unintended consequences in the complex domain.  In the obvious domain they work, are valuable and necessary.  But they need to be easy to understand, few in number and not require constant reference to manuals or half remembered training.  The use of ritual boundary crossings is one way to achieve the same effect with less codification.
• In the complicated domain targeting goals is OK but needs to be generic enough for compromise and variation.  Standards that create boundaries within which variation is permitted are a good idea and this is the one domain where alignment and common purpose and goals are useful as they act as a constraint on deviant behaviour.  Too may targets and you loose respect, especially when those targets are created by bureaucrats or consultants (actually they then to be the same) who lack experience of the context and/or membership of the profession.  I had an argument recently with someone over this in terms of medical targets where he said the ones they had had come from nurses.  I asked him which nurses volunteered for the task  and after he went away to check he came back and said he got my point.  The nurses with a high reputation for care and empathy had better things to do.  Anyone in a professional group who volunteers too write standards or rules should be looked at with extreme suspicion.  Better to drag some people kicking and screaming into a mediated session where they can create something minimal.
• For the complex domain we can only really measure direction and speed of travel from the present.  This is something I have termed as vector measure and they are linked to the whole vector theory of change that I and colleagues are working on at the moment.  A sense of direction can be achieved by visionary goals, but their use has to be with care which is why I have placed them in the liminal zone with chaos.  For every story you hear about a visionary leader setting seemingly impossible targets that resulted in beneficial change there are a thousand who failed; enough people have done it that some are bound to succeed.

The big red arrow

So to the one major danger which is all to common; faced with complexity people don’t absorb it they try to eliminate it.  The fluidity of Agile is far too scary so we get SAFe, evolved practices are not explicit enough and BPR hasn’t worked all the time so we get Six Sigma and so on.  The premature codification of complex issues into neat and tidy solutions is a large consultancy disease and a fertile field for snake oil sales people and charlitans.  Most mission statements and (God help us) definitions of Mindset are little more than a collection of meaningless platitudes, the lowest common denominator of workshops based on keeping people happy rather than enabling learning.  There are better ways ….

Banner picture photo is by Aditya Siva on Unsplash you can work out the significance for yourself!

The introduction of liminality to Cynefin necessitated a slight rethinking of the main dynamics within Cynefin.  To be clear multiple dynamics are possible but over the years these have emerged as the dominant ones.   As the name implies the dynamics in Cynefin are all about movement between domains.  Such movement can be achieved by shifting the nature of the constraints – for example using heuristics (a type of enabling constraint) allows for emergent behaviour while rules (a type of governing constraint) in effect create a boundary condition that limits the actors an the context to a greater degree.  Constraint mapping, its representation in Cynefin and links to the whole known-unknown-unknowable matrix will form my final post in this series tomorrow.

Key to Cynefin is the idea that no domain – with the exception of inauthentic disorder – is either right or wrong, it just is.  I’ve seen too many people fall into the trap of saying something along the lines of my method is complex, theirs is merely complicated  and that goes along with un unthought demonisation of Taylorism that in many ways showed more respect for people that the current popular forms of systems thinking.   Domains have different qualities and we need to understand that and act accordingly.  If we want a behaviour or outcome which is not authentic to the domain then we first have to shift the context – move between domains to allow that to be possible.

Dynamics also link to ideas of resilience and sustainability.   Think back to the energy gradients of yesterdays post and you can see that keeping things in the complex domain, shifting them to complicated is hard but if you do, then and only then can you fully exploit what you have created and critical scale by aggregation or repetition.   In the complex domain scaling is only achieved by decomposition and recombination at an optimal level of granularity which means there is a degree of constant lovely, retrospectively coherence but not predictable.  

Those familiar with the old dynamic framework will see the differences with the older representation and they rest on an inflection point in the liminal area of disorder, the gamma point above.  I’m still not sure on aspects of this and its a work in progress.   But to get there lets describe the four dynamics shown above.

1. The most stable pattern (blue) is the move between complex and complicated, and if needed back again.  Parallel safe-to-fail experiments in the complex domain reveal a way forward.  Iterations in the liminal domain prove that stabilisation is possible and we shift to the complicated domain.  The reason we know this is possible is that the imposition of constraints creates repeatability – if it doesn’t: stay liminal.  What most people forget is that when you have stabilised it you need to periodically check that this is still the case.  If it isn’t then you cycle back in the complex and start the experiments again – and you do start them again you go back into complex not liminal complex-complicated.
2. If you get long term stabilisation (also blue) then, and only then can you shift it into the obvious domain.  But you make sure you do so with some flexibility in order to avoid the border with chaos.  It’s the beta inflection  point on the map and it carries a danger as there are three options.  It may be stable and you can shift, you may realise when you check (and double check) that this is not the case – variations and exceptions are increasing  in which case you return to complex.  The real danger is that you don’t realise there and then and by the time you realise a smooth transtition is not possible so ….
3. … you shift the red dynamic, also down as a shallow dive into chaos.  This is a last ditch recover at the delta inflection point. You’ve allowed the culture to develop immunity to the constraints so that they appear to work but in practice they only do so because the informal networks and practices have developed work arounds.  At this point you need some type of radical disruption although a double check at inflection point gamma may reduce the severity of this need and allow you to move directly to blue.
4. Finally with have what I called the grazing dynamic where the situation is highly fluid and any exit from disorder is not possible.  The best you can achieve are temporary stable patters in the liminal area between complex and complicated – more over slightly deeper dives into chaos are needed to create constant novelty.

Now there is more work to do here – particularly on the inflection points but for the purpose of this update that’s it!   Final post tomorrow.

Banner picture was taken between Glyder Fach and Glider Fawr shortly after the scramble up Y Gribbin

The introduction of the liminal version of Cynefin was probably the most significant change in recent years.  One of the reasons I used it was a difficulty people had in seeing that Cynefin had two classification systems: domains and dynamics.  By creating the liminal zones I managed to make dynamics a domain and that has made it easier for people to use.  It took a few iterations to get it right but when I managed it as a single stroke of the pen I was happy and there was the added bonus of finally resolving some issues around the domain of disorder.

Liminality as a concept is all about standing on a threshold (literal translation of the Latin origin namely limen) between states.  It carries with it a sense of ambiguity and uncertainty, possibly fear or trepidation before commitment.  In anthropology it is used to describe the point in a ritual where the actor stands between what they were and what they will become.

It is important to emphases this concept of a threshold.  Key to Cynefin is the idea that three ontological states or order, complexity and chaos are subject to phase shift boundaries.  Think here of latent heat; if I heat up water to 100ºC it doesn’t immediately become steam I have to continue to apply heat before this happens.  When water becomes ice it drops to 0ºC but then more heat has to be thrown out before the transition happens.  I’ll develop the idea of energy gradients more later in this post.  My use of phases shifts is a key difference with others such as Glenda Eoyang and Ralph Stacy who see the differences as a gradient.  Ralph also sees the differences as only about perception while I see the differences as being about reality, knowledge and perception (there I think Glenda and I agree).  Now I think the idea of phase shifts is more authentic to the mathematics, but even if it wasn’t then creating a phase shift boundary is essential to human sense-making.  Give us a gradient and we settle where we are most comfortable, create a boundary and we can behave differently on the other side.  It’s why ritual is so important in human society, it allows for change to be triggered in a sustainable and scalable way.  We dress differently for different events and we think and feel differently in consequence.  Crews in the emergency services have strong ritual acts to change identity in order that the interacting team can be effective and so on.

Its why I choose the image of a group of people standing on top of a cliff.   The shift between order and chaos is shown as a catastrophic fold as you walk over the cliff in mist without realising what you have done until it too late.   If you are choosing to change states and that involves climbing a gradient – making an energy commitment, the longer you can spend checking this is the right thing to do, and if it is, what is the easiest pathway, the better.  The domain shifts between the ordered and complex domains and between those of complexity and chaos are generally matters of choice and with choice come consequences.

On the right I’ve produced a version of Cynefin created bu Rob England.   His first version showed chaotic as a lower level than complex but be changed that after I pinged him.   Along with the planar view of Cynefin which introduced the catastrophic fold into the framework, this is one of several different representations of Cynefin along with the domain models (I am not linking as I intend to revise those in the near future to handle liminality).  Rob created this with a specific purpose and its adds automation to Obvious (Simple when he produced this) and I would make other changes.  But it is useful as it shows that effort is needed to shift to higher levels of order but little effort is needed for things to be complex – enabling constraints multiple interactions and fluid mess are a natural human condition after all!   I would reduce the size of the gap between complex and complicated and increase it between complicated and standard.   I might also represent liminality as a slope towards the left (but not to the right) which reduces the climb.

Now there are purists within the Cynefin fold who argue that physics states that chaotic is a lower energy state than complex and in their field they are correct.   But here we are talking about human systems and those are open not closed.  The nature of humans is that we create connections and constraints intuitively as well as deliberately and we do so very quickly.  To create a system with no effective constraints, a state of near randomness takes a lot of energy both in creation and to sustain the state.  Liminality allows us to create additional transitionary domains and I’ll be interested if Rob attempts to draw them and I may attempt it myself.   However for the purpose of this post I want to address the three liminal states within Cynefin.  If you want the visualisation look at the green areas in yesterday’s post.

1. Between complex and complicated (one way)
   In the main areas of the complex domain we understand what is possible (or plausible) through parallel safe-to-fail experiments but as patterns emerge and stabilise we enter the liminal domain.  We now have a sense of what is possible and we can radically limit the range of our experiments but we still need to carry out more testing, iterate to test understanding and so on before we can make the transition too complicated.   When we make that transition its a big commitment so we need to hold as long as possible before doing so.   I show this one as open as towards the left the gradient between complex and complicated is shallow and easy to traverse, while as it approaches disorder it is steeper.
2. Within Disorder (two way)
   The big moment of insight for me when I finalised the liminal version was the realisation that it resolved the nature of disorder.   In the green area we are in transition between domains, its a moment of legitimate uncertainty and tension in which we ontological uncertainty (not the case in the two liminal states above and below) and there are many directions we can take.  In the grey area we are adjacent to the catastrophic fold as we have no idea of what domain we are in, and we have no idea that such ignorance is an issue.   So the liminal area is transitionary, the other area is inauthentic disorder.
3. Between chaos and complicated (two way)
   This is shown as a closed state in contrast with that between complicated and complex.  The reason is hinted at above – to create chaos a state of no effective constraints is difficult and costly.  It requires boundaries to project it and it is always in danger of collapse in either direction so careful maintenance and frequent scanning are needed – this is a high energy requirement but powerful for innovation and distributed decision making using wisdom of crowds methods in products such as MassSense,

The boundary between chaotic and obvious is not liminal as there is no transition it is sudden.  It is two way, but one way happens with greater ease.  Climbing back up the cliff is normally a mistake, easier to shift into the liminal domain with complexity and then start the cycle again – but I will deal with that when I map dynamics onto the liminal version of Cynefin tomorrow.

By the way – I suspect the exploration of liminality will generate a lot more ideas and posts over the next couple of years.

Inset picture is by Rudolf Kirchner and sourced from Pixels

Banner picture is the summit of Glyder Fawr ,the first peak on a long walk of the ridge from Pen-y-Pass to Capel Curig – I broke a rib on the next peak Glider Fach when I slipped on wet boss while boulder hopping.  I selected it to convey the sense of uncertainty and transition involved in liminality.

It’s taken longer than I hoped but I now have the time to complete my series of five update posts on Cynefin. In my last post I explained the change from ‘Simple’ to ‘Obvious’ in the ordered domain. In the post I want to look at the different types of practice or knowledge use in each domain and explain why I changed ‘Emergent Practice’ to ‘Exaptive discovery” in the complex domain and added the stress condition to Novely in the Chaotic Domain. If you look at this latest version of the framework I’ve also made the boundary between Obvious and Complicated a lighter shade of grey to indicate that this boundary, unlike the others is not a phase shift; its more of a gradient.

In the ordered domains there is a linear relationship between cause and effect, there is a right way (or a definable set of right ways) by which success can be achieved, or at least failure averted. The good and best practice distinction is important here. Best practice means there is only one way, and for that to apply not only must the relationship between cause and effect be self-evident to any reasonably person, but all such people must buy into the solution. For example in the UK and Australia we drive on the left hand side of the road, in Germany and the USA its the right. It would be a very unreasonably person who sought to vary that practice and there would, as they say, be consequences. But in the complicated domain the solutions are less clear. Here things are more loosely coupled, there are more degrees of freedom. In a community of experts, if you have the expertise and it is acknowledged by your peers then a degree of variation is permitted. One size doesn’t not fit all, but the right to vary practice is hard won and to a degree bounded by said peer review and peer acceptance. In both these cases there has been little change since the Cynefin Framework first appeared in something close to its current form.

In the complex domain I used to talk about practice being emergent, that is to say it will not be fully known or knowable in advance; understood only with the benefit of hindsight there is little one can do in planning. Its more about staying alert and exploiting patterns of possibilities as they start to become visible. However over the last year or so I have started to realise there is more we can do here. Exaptation is something I have blogged on before, most notably in this series of three posts back in 2012 (like 2019 a Grand Slam Year for Wales). In that series I also talked about managed serendipity. It is not just about responding to patterns, its about creating the conditions in which radical re-purposing is easier to stimulate and easy to recognise. There is a lot more to write here and some of that is for the book, but to optimise the conditions for serendipity you have to optimise (which generally means reduce) the granularity of your knowledge objects, increase the abstraction (SenseMaker® does this by design) through which you discover new ways of using things and radically change connectivity. The key on exaptation is to repurpose things in which you are already competent, so the learning or energy needs are reduced. IBM for example repurposed their expertise in punch card management to give them first mover advantage in the emerging field of computers many decades ago. So a major reason for the change was the realisation that we can do more to manage this space, its not just responsive.

Finally I realised that novelty is university, but the difference in the Chaotic domain is that the novelty arises from stress, from a complete break with the familiar. This means that radically different approaches will be accepted while in normal times they won’t. Interesting connection to the Obvious domain here – to be effective they must be accepted by all reasonable people. The difference in the chaotic domain is that the context makes unreasonable stubbornness more difficult to sustain.

So that indicates management possibilities; more about compliance in order, more about awareness and preparedness in unorder. Tomorrow I will move onto the liminal domains, what that means for disorder and also the link to energy gradients between domains.

The picture is of Tryfan from Y Garn at Dawn – all the posts in this series will have banner pictures from the Ogwen Valley and the Glyders