St David’s 2024: 4/5 Intervention

I established some principles for managing a complex system and a list of things to avoid many years ago.  Those have stood the test of time but need an update, especially given new frameworks such as AIMS, which cover some of the same material.  Historically, we had  (i) change the granularity, (ii) distribute the condition, and (iii) disintermediate the decision makers, and those complimented the need to avoid pattern entrainment, along with the dangers of retrospective coherence and premature convergence.  

Central to this approach is the crucial task of stimulating and monitoring the generation of emergent properties, a far more effective strategy than attempting to determine outcomes.  I shared my insights on the necessary but not sufficient conditions for emergence last year in preparation for my lecture at Hull University, where I emphasised that talking about seeing the system holistically is deeply problematic in complexity work as (i) you can’t, there are too many things going on and, (ii) thinking that you can will bias the system and means you will miss weak signals and possibilities that lie outside the compass of your imagination.

Another fundamental principle is that complex systems scale not by aggregation or imitation but by decomposition to the lowest level of coherent granularity and recombination.  This approach is not reductionist;  it’s not about breaking things down and assuming that putting them back together will yield the same result. Nor is it about believing that the properties of the parts can determine the properties of the whole. It’s about understanding the unique dynamics that emerge when elements are combined in different ways.  This is a fascinating aspect of managing complex systems, where interesting things will bubble to the surface if you monitor for emergence. 

As a reminder, the AIMS framework covers agents, interactions, monitors, and scaffolding and, as described in the second post in this series, covers the things you can and should manage within any system or system.  This post is concerned with some of the principles of intervention design in general and in relation to actions arising from Estuarine Mapping, Knowledge Mapping or more general Cynefin use.  I’m considering adding a checklist to the action forms to reinforce this.    I should also emphasise this is a developing list and may change more than any of the frameworks in this series.  

1. Obliquity, a concept with a hat-tip to the book of the same name by John Kay, is a powerful strategy in problem-solving. As anyone with teenage children already knows, the most effective way to solve issues is rarely to tackle them head-on other than when they are trivial or as a refuge of last resort.  Kay’s book contains examples of how complex and intractable problems were solved by working on a related issue.   Working obliquely also means less public commitment to solving a particular problem and more time to watch for unintended consequences.  It also gives more space for solutions to emerge rather than be manufactured or imposed. 
2. Granularity matters and this is a Goldilocks solution. Too finely grained, and things get random; too coarsely grained, and novel interactions that would give rise to new emergent properties are suppressed. At the right level of granularity, things can combine and be recombined in different ways. The best way to date (and there will be more) to achieve this is the Estuarine Mapping instruction, which keeps breaking things down until you agree.    Some aspects of object orientation in software development, including polymorphism and inheritance, are relevant here.  In terms of Assemblage Theory, changing the granularity is a deterritorialisation method that can change the affordances for distributed action within a system.
3. Sensors are critical, and they need requisite diversity and, as far as possible, provide real-time feedback so that the weak signals of emergent patterns become visible early enough so that the energy costs of amplification or disruption are low.  SenseMaker® was designed for this and has various techniques to identify where there is sufficient consensus of action. It also critically aims to find the 17% who have seen the picture of a gorilla before they talk to the 83% who didn’t.  This used to be called distributed cognition, which is another way of seeing it.
4. Abstraction enables abductive insight and reasoning.  At its simplest, this can be the use of a metaphor to get people to think about things differently.  We have also used metaphor-based games to break entrained thinking patterns, and archetypes (situational and persona) can achieve similar results.  SenseMaker® also has an application here; in exaptive innovation and design, we bring together needs and capabilities through high abstraction.  Metadata to suggest connections that might not otherwise occur to people.  It can also be used in decision-support applications to identify novel patterns.  And then, there is the range of aporetic methods developed to handle the central domain of Cynefin;  I wrote about all of this recently for those who want more details.
5. Patterns are fundamental to human sense-making. Our work on cultural mapping shows narrative patterns rather than spider diagrams and the like, which assume causality and privilege the interpreter (which may be why some consultants like them). Humans are highly sensitive to patterns in other people’s reactions, the landscape around us, etc. We have strong peripheral vision and can quickly sense that something doesn’t seem right if, and that is an important qualification, our attention is drawn to something through some form of anomaly.  Visualisation is one way to achieve this, and it’s an important aspect of SenseMaker®, which has the added advantage of the explanatory stories that sit behind the patterns.
6. Disintermediation removes interpretative layers between the decision maker and the raw or finely-grained data on which a decision is based. The more layers of interpretation there are, the more that gets filtered out, and the easier it is to manipulate the decision-maker based on what is presented to them.  Our work on cultural mapping takes people directly from a statistical pattern to the stories of the water cooler without interpretation.  The new body of work I signalled last year on distributed decision-making goes further in removing bureaucracy, which is progressing, looking at nondirected search mechanisms in nature.  Expect some major announcements over the next few weeks.
7. Fractal, decision making and initiation of action are critical.  You don’t want homogenised context-free interventions; every situation or aspect of a system has a different starting place.   A principle here is never to aggregate an aggregate but always use the same, finely-grained source data assembled at the level of people’s competence to act.  A fractal is a geometric shape in which each aspect has the same statistical characteristics as the whole.  This is pretty easy to understand in mathematics, but the word is often abused in social systems and theory.   If in doubt, remember: never aggregate an aggregate.
8. Anomalies are critical to engaging people in thinking about a problem.  If you walk down a street, then you don’t pay attention until you stumble, and then you start to actively think about the placement of your feet and the conditions.  83% of radiologists do not see the picture of the Gorilla, but if it was animated they would.  Methods to create divergent outcomes are key to the Cynefin eco-system and there are a wide range of methods to trigger what we have long known as descriptive self-awareness, not telling people or guilt-tripping them through facilitation, but taking them through parallel processes so they become aware of differences in such a way they can take action.

So that is it; note these are not a list of the characteristics of a complex system, but the sufficient but not necessary qualities necessary for intervention in complex systems.  They represent my current thinking, but I do not have a monopoly of wisdom here, and I am pretty sure I will change and refine them further over time.  Cillier’s list remains a good one so I have added it here.

Paul Cilliers

Complexity is the result of a rich interaction of simple elements that only respond to the limited information each of them are presented with. When we look at the behaviour of a complex system as a whole, our focus shifts from the individual element in the system to the complex structure of the system. The complexity emerges as a result of the patterns of interaction between the elements.

1. Complex systems consist of a large number of elements. 
2. This is necessary but not sufficient. The grains of sand on a beach do not interest us as a complex system; interaction is vital.
3. Said interaction is fairly rich, with every element in the system influences, and is influenced by quite a few other ones.
4. The interactions themselves are non-linear, which guarantees that small causes can have large results, and vice versa..
5. The interactions usually have a fairly short range. As a result, it can be enhanced, suppressed or altered.
6. There are loops in the interactions. The effect of any activity can feed back onto itself, sometimes directly, sometimes after several intervening stages. This feedback can be positive (enhancing, stimulating) or negative (detracting, inhibiting)..
7. Complex systems are usually open systems, they interact with their environment. It is often difficult to define the border of a complex system. 
8. Complex systems operate under conditions far from equilibrium. There has to be a constant flow of energy to maintain the organisation of the system and its survival.
9. Complex systems have a history. Not only do they evolve through time, but their past is co-responsible for their present behaviour.
10. Each element in the system is ignorant of the behaviour of the system as a whole, it responds only to information that is available to it locally.

The opening picture is by Vinicius “amnx” Amano, and the banner picture is cropped from an original by Tim Marshall, both on Unsplash.