One of the most frequent corrections I have to make to public uses of Cynefin is to clarify that it is a framework, not a model; in some ways, it is more irritating than pointing out that you can’t have five quadrants. The latter I find more amusing in direct proportion to the educational level of the person committing the error. Put simply, a model seeks to represent reality, with the implication that following the model in practice will provide security of some sort. In contrast, a framework, in this context, and to quote the Oxford English Dictionary, is a basic conceptual structure underlying a system, study, etc. It is a way to look at reality, in the case of Cynefin, from different perspectives and against which models, tools and methods can be tested. Models and frameworks have very different claims in respect of their validity. Cynefin provides for three primary domains (ordered, complex & chaotic) with the fourth aporetic state where it is equiprobably into which domain the issue or solution will fall. They expand to nine when you add liminality, and in full use, we also focus on shifts or movements between domains. It is designed to satisfy a simple test – can people draw it on the back of a table napkin with minimal training? From now on, I may refer to it as the Serviettentest, with acknowledgement to Hubert Magis for bringing my attention to the German compound noun, which sounds much better. English, as we say in Wales, is too good for the English as it has a beautiful habit of following other languages down dark alleys, knocking them over and going through their pockets for loose grammar and vocabulary (I adapted that from a Facebook meme). But it never acquired compound nouns, which is probably why words such as weltanschauung are part and parcel of British Philosophy. Welsh allows for combining two nouns or an adjective and a noun, but the German use is in a different league.
The purpose of a framework is to test ideas against core theory. So, in Cynefin, taking an ordered approach in a complex world is open to challenges on ontological grounds. This reduces the risk of an inauthentic approach working in one context and then being generalised. Or the danger of what passes for empiricism in many an agile coach and management consultant where “I seem to remember this working for me before, so I will do the same thing again and suggest everyone else copies my approach” is all too common. Some even create whole certification schemes based on that. As I have said on many an occasion, never attempt to scale a what until you know why. The AIMS framework I have been working on for a few years and is now in the public domain is designed to focus people not on a model of reality but on a set of principles that apply in the Complex domain of Cynefin. It reminds me that the two things I can change in any system are the Actants (actors, constraints & constructors) and their interactions. I need monitors to quickly sense emergence and let energy flow to what I want to amplify and take away from things I want to prevent. Scaffolding here (and I will blog on this shortly) is a type of actant or actant that I stabilise to give a degree of certainty within the system as a whole – a skeleton does that for the body, to give one example. Cynefin passes the Serviettentest; AIMS is in the mnemonic form to make it easily memorable. These are not abstract principles but concrete theory-based checkpoints designed to make real change faster, more sustainable and above all resilient.
Not three different events triggered this post:
Interestingly, all three make similar claims and patronise people who challenge them. Raise any issue, and they tell you it isn’t a problem for them. I had a few claims of wide adoption as proof of efficacy, but as I pointed out, that is a dangerous path to take as it also validates known pseudosciences such as the MBTI. There is a lot of retrospective coherence in the first two cases, cherry-picking aspects of past cases to support a new current model. In the third case, I have a lot more respect for those involved. They developed vast knowledge, but science has passed them by. This and what they have done is not to say that there is still not much to learn, but the model doesn’t stand up against the complexity science framework. In the context of Cybernetics, of which VSM is a part, I would also argue that Bateson’t work doesn’t really fit with what that name now means. His work is consistent with complexity science, as is that of his daughter Nora.
One of the problems with models is that they have mesmeric qualities; again, to quote the OED:
“A therapeutic doctrine or system, first popularised by Mesmer, according to which a trained practitioner can induce a hypnotic state in a patient by the exercise of a force (called by Mesmer animal magnetism); the process or practice of inducing such a state; the state so induced, or the force supposed to operate in inducing it.
Mesmer’s claims were not substantiated by a scientific commission established by Louis XVI in 1784 including Benjamin Franklin and Antoine-Laurent Lavoisier. His techniques, however, had great popular appeal and were variously developed by other practitioners in the late 18th and early 19th centuries, ultimately forming the basis of the modern practice of hypnosis.“
This links in with one of my concerns about AI; it is building models that seek not so much to represent reality as to change it, and if we succumb to the mesmeric qualities thereof, then we will lose the human capacity for novelty and abductive thinking in question. Models also create context-free solutions for a context-specific world; the granularity is mostly too coarse to allow for combination and recombination. In my social media exchange with the VSM advocates, I was asked to name a competing model that was as effective. I responded that similar claims were made by all consultants advocating a model. But no one who understands complexity science would ever claim that a single model could do what VSM claims to do, or for that matter, be fractal in nature. You need multiple frameworks, some models, many methods and a few tools that work together in different ways in different contexts. Any valid model must be more finely grained than any abovementioned types.
We also need to break the pendulum effect, as evidenced in moving from agile everything to slowness; both have validity in different contexts. I represented that with a Newton cradle as a true pendulum crafts more ambiguous patterns. But here, the bright, new, shiny thing seeks to replace what came before, but it all bounces back. The fad cycle is far more disturbing and represents a form of gaslighting of corporate executives and government administrators.
Assemblies that, in use, can trigger assemblages over time is, to my mind, the future, and it has been my focus for the last few decades. I realise the last statement is a little enigmatic, but expect more over the next few months to make it clearer. In the meantime, please don’t confuse frameworks with models, and don’t put your trust in one model: None should rule all others or in the darkness, bind them.
Cognitive Edge Ltd. & Cognitive Edge Pte. trading as The Cynefin Company and The Cynefin Centre.
© COPYRIGHT 2024
There is a little of the old and the new in the imagery I have ...