Dave Snowden: Cynefin dynamics

Dave apologised for not being able to be present. Over the last 2 weeks he has delivered 15 keynotes by telephone and is getting pretty good at it although it is still difficult due to the lack of feedback (in both directions).

His goal is to give some basic tools for managing complex spaces. The fundamental metaphor can be understood by looking at the difference between 2 methods of managing a birthday party for a group of 12 years:

Method A: Assemble parents 6 months in advance to set corporate mission goals & objectives in advance. Draw up a project plan, resource allocation map, set of key milestones etc… On the day, commence with motivational video for children to tell them their role. Senior adult uses a power point presentation to show how achievement of goals will lead to rewards. After the party an After Action Review is used to evaluate performance of children & adults, leading to the creation of a “party best practice” database.

Method B: The “senior adult” present starts by drawing a line in the sand and saying “cross that you little bastards and you die.” (This is known as a ‘boundary’ in complexity theory). Then adults stimulate the party with multiple attractors (games, video, etc...) to see if beneficial patterns form around the attractor. Breathe a sigh of relief when one of them works and try to stabilize the pattern. Then head upstairs to disrupt the negative pattern forming around your 14 year old daughter and a bottle of vodka.

In Method B the focus is less on total design (input/output) control. Contrast with systems incapable of that kind of structure.

Boundary & attractor conditions allow beneficial patterns to emerge. After it’s over you know whether the party was good or bad, but it wasn’t predictable before hand.

Something else to consider is the huge difference in cost to do “method A” solutions compared with doing “method B” solutions. When you apply a “Method A” solution in a “method B” situation you never get to the AAR. You would have a riot on your hands. “It costs a lot of money to make things worse.”

This is particularly problematic in the public sector. How do we deliver increasingly high levels of service with decreasing funds? Informal social networks at hospitals manage things outside the formal processes quickly & efficiently. “Slot in” formal processes to back up things already done, inhibiting their capacity to act.

Draw a dividing line between applicability of ‘method A’ and ‘method B’. Have structured scalable approaches to manage the environment of B. Deal with a very real threat.

At the heart of the distinction: Systems of order and of unorder.

Order: Things happen the same way, again and again. Causality is based on fundamental equations that we know or we know we can research. The high investment in method A can be recouped by re-use.

Unorder: Not “order as we know it Jim” it’s something different. It is the domain of ‘retrospective coherence’. When you look backwards everything makes perfect sense but there was no way, at the time, of predicting that particular outcome. What happens is that ‘patterns of interaction’ stabilize and you can then understand how this occurred. But it could not be predicted in advance (Anti-terrorism: Emergence outcomes. Attempts to predict them can make you more not less vulnerable to threat.) Causality is apparent after the event.

Pattern of cause which become apparent, as particular assemblies of events repeat themselves, are not inherent properties of the system but coincidences which cannot be relied upon to happen again. Structured methods from business schools and management systems do not cope. Instead an approach based upon complexity theory (boundaries & attractors) is required.

So

Ordered systems => A Unordered systems => B

A simple explanation comes from their work for the European commission on the future of KM (2003) the so called ‘Lisbon agenda.’

Emergent	Mathematical Complexity	Social complexity
Design	Engineering	Systems thinking
	Rules	Heuristic

Contrast rule based with heuristic approaches.

On the one hand the Jesuit confessors’ manual which lists every sin you could expect to find along with the appropriate penance. On the other hand is the “Sermon on the mount.” Generic flexible, value based, guiding principles with necessary or sufficient ambiguity to allow them to adapt to new situations. What they gain in ambiguity they loose in control. When to slacken control?

This matrix breaks the 2x2 “consultants” convention. All spaces are important, not just top-right.

Ordered systems: The aspects of organisation that are highly structured and where things repeat. Reached a culmination in Business Process Re-engineering. Best practice. Engineered solutions. Dominated thinking for the best part of a century & still prevalent in Government. Industry fell into the trap again with CRM which is essential a social system but governed by rules which fail to acknowledge this. There is a huge amount of value in ordered systems thinking, don’t throw it out. Just understand where it is appropriate.

Peter Senge: Systems thinking/Tom Peters... A rebellion against Taylorism. Learning organisation (LO), mission/value alignment. Not a rejection of order. System can be managed based upon choice. Fundamental thinking is that the leader stands outside the system and makes choices. Recognition of the ambiguity of human systems. Shift towards heuristics expressed as value alignment. Leader makes choices and aligns the organisation to them. Individuals sacrificed to the “greater good”.

Manifested as technology vs. culture approach.

Mathematical complexity: Science of complexity challenges order: from biology and psychology. Order emerges as the result of the interaction of agents. Boyd’s algorithm: simulate flocks of birds. 3 simple rules for each agent simulates flocking behaviour of birds. Computers simulating organic principles. Rules for agents & interactions between agents.

Lots of value & benefit however some people have started to challenge this, not for it’s applicability from it’s original fields, amazes him that it requires saying that humans are not the same thing as ants or crystals!!

Human beings don’t have a single identity making it hard to work out what the agent is in a human system at any given time. In every situation the same actor may express different behaviour, possibly radically so. Additionally multiple collective identities can be held in sequence or in parallel leading to a 3rd level of complexity.

Agents in complex systems are themselves complex systems!

Confusion between simulation & prediction: KEY DISTINCTION

Human decision making isn’t made on logical choices. Defined rules, optimizing outcomes, etc... Neurophysics & cognitive science tell us that we actually make decisions based upon complex patterns. We match data to a repository of patterns in our (single or collective) brains. We respond in terms of the matched pattern. Retrospectively we make it coherent because that’s our intellectual tradition. It’s not a best fit pattern, but the first available pattern. This is sensible given our prey beginnings. Don’t scan a predator against a catalogue. “Oh my gods it’s a lion!” decision.

Pattern based decision is an essential behaviour of human beings. Anything a human being finds easy is a near impossible task for a computer and vice-versa. We’ve implied a machine like model of human decision making underpinning BPR.

3rd distinction in human systems: We can behave in an ordered way, by social convention, ritual, taboo, legal systems, etc… collectively create predictable ordered systems in our lives. C.f. birds always behave in an unordered fashion. Humans have managed the boundary between order & unordered behaviour. Intuitively understand that you behave differently in the different spaces.

Mathematically Complex: Doesn’t challenge the concept of rules (although it does the concept of order) because it uses mathematic rules.

Human systems share emergence with mathematically complex but share with systems thinking the extra unpredictability => heuristics.

Use this model to create a framework to understand:

Social complexity; Participative complexity: All human interactions are based on conversation, humans are essentially unmanageable

Contingent complexity: Humans are capable of being ordered.

Observations: These schools are based in Europe. The US is dominated by goal based culture. Europe is more diverse in its thinking and has the opportunity to dominate social complexity systems!

Basic principles of how we manage unordered space:

Cynefin System

Complex unorder Hidden Order Chaotic Unorder Visible Order

Visible Order: Controlled, centred, organisation. Agent interconnects are irrelevant.

Hidden Order: Expert community. Everything is tightly connected but the centre is still in control. High degree of consensus about world view (Danger if the world changes!), c.f. Longitude and cognitive dissonance of scientists vs. clockmaker.

Complex Unorder: Network is connected but the centre can’t see what’s going on until after.

Chaos: Turbulence.

Networks give you a simple clue to what to do. Unorder: Exploration. Order: Exploitation of what we know.

Secret: Understand how to go to unorder to explore, and then move back to order to exploit. Unhealthy to stay in just one. Problem solving happens on the left.

We use this to allow people to categorize problems. And hence understand how they should act. Left/Right. Being in the wrong domain can blind you to how to solve the problem.

When dealing with a problem Categorize Understand the underlying relationships with define the nature of the system

Understand whether you want to shift the situation E.g. move from chaotic unorder => visible order

Classic crisis management approach breaks down catastrophically. We’re unhappy. Great leader comes in and pulls all the network controls into themselves, tightening to shift the problem back into the visible ordered space, e.g. General Motors.

Oscillation between anarchy & control which is the common method of crisis management. A single point attractor solution to a crisis.

Multipoint attractor. Multiple points of stabilization. Automatically shifts the problem from the chaotic to the complex space. Select patterns that are beneficial and exploit them – kill off undesirable patterns. Multiple attractors create a more stable approach to crisis management. Execs can stay aside of the problem.

[Ed: Dave’s book may be coming out this year and will explain the Cynefin dynamics in a great deal of detail.]

Situational assessment: what is the nature of system we are confronted with? Then make appropriate response. Do we want to shift the problem? Permanently or temporarily?

Innovation Dynamics:

Hidden order => shallow dive into chaos => stabilizes into visible order

Expert community: Pattern Entrainment: Hard to acquire but creates an inherent conservatism resistant to new insights. History of science is dominant science ignoring new science. Seen as a challenge. The nature of human pattern entrainment which we just have to live with.

Expose experts to multiple possibilities. In complex order facts are coherent only retrospectively. People can then “explain away” facts contradicting their position.

To achieve a horizontal shift, you must first do a diagonal shift to shake up peoples beliefs so that they can start to see new possibilities.

Break all linkages. Stimulate network linkages to shift into complex space.

Social network simulation to simulate formation of informal networks.

Don’t give resource: Creativity and resource are neither necessary nor sufficient.

Starvation & pressure: How human beings change how they see the world, provoke effort.

REQUIRES A DIAGNONAL SHIFT!

Dangle bureaucrats over the edge of chaos to show them how bad it would be, and then move them into a complex space to stabilize the situation in a different way.

SHALLOW MOVES INTO CHAOS.

Faced with a situation. Where am I (which domain?) Where do I want to go (which domain?) What do I do?

Abide model: Opportunistic use of a crisis (published soon).

Uses simple principle of mnemonics: managing unorder meaning manipulating:

Attractors (positive, attract – single point) powerful idea, individual, legal framework, etc… agents /identities flock to these. Basin attractors. If the attractor goes it doesn’t work, hence there is instability. Multiple-point attractors are more stable – less control, more stability and adaptability. Strange attractors: (not sure why, they’re not particularly strange). Think about a U shape valley, lots of freedom to walk around the floor, but not much to go up the side. Constrained but with a lot of freedom at the bottom... U shape moves towards V shape (=>towards single)

Boundaries (negative, repel) Brittle – any rigid, when it breaks it breaks catastrophically with no recovery Elastic - middle Permeable – crossed all the time but people know they’ve crossed them The more you constrain the less adaptable you have.

Identities Dissent/Diversity Environment

In unorder you’re deeply manipulative (find any parent who isn’t!!)

If unordered look at the ABIDE elements.

Q1. What attractors are there? What are their natures? Q2. What boundaries are there?

Public sector case (same thing applies): Experiment which works brilliantly in Newcastle but fails in Birmingham regardless of demographics. Reason you’re dealing with a complex unordered system. Different attractors and boundaries (unique). Retrospective coherence leads to linear causality.

Should be saying: What are the current attractor & boundary conditions in Birmingham: how to we shift them?

No two children’s parties are the same. But we all know what is or isn’t successful. Unique combination of attractors and boundaries. Manage these not desired outcomes. This is cheaper!

Human beings have multiple identities.

In any complex intervention what are the identities in operation? Individual or coalescence? What do we want them to be? What’s their nature? Where are they? What do we want them to be? How do we change them?

Stop managing output/input and manage attractors, boundaries, and identities.

To introduce strong values: Long training under stress, people bond. Create a reliable form of identity. If you challenge that identity the system becomes more complex faster.

Approaches which treat individuals as widgets fail because people are identities in their own right and also composite identities. Disrupting one creates disruption in others that can be unpredictable.

Degree of dissent/diversity

To increase the possibility of new/novel behaviour: Increase dissent. Wildcard. Genetic diversity.

Increase the amount of diversity in a system but there may be boundaries to protect structured ordered systems.

In human systems diversity is expressed through dissent.

A balancing act.

Proximity of identities, degree to which they are close to each other.

How much pressure (agitation) is on the system?

Triggers to stimulate the emergence of complex behaviour.

Applies from the micro (child’s party) to the macro (governments)

Dominant management theory is based on the existence or assumption of order.

[Editors note: my original notes after this point got very sketchy…]

Something which only happens every 100 years or so, a completely new set of concepts, recognize when you’re in a method B situation you can produce good results quickly and cheaply.

Drive to effectiveness.

Contrast with efficiency.

Cynefin framework developed through Action Research. Dynamics is about how you shift between domains.

Newton spent most of his life as an alchemist. He discovered modern physics but couldn’t break through.

Complexity is the next major wave in organisational thinking but is 5-6 years away. An article by Tom Steward (in Business 2.0) summarizes the model in his own way.

Whilst the model may be new they have never had a problem with the concept when applied to a real problem. A good example is the situation of tax inspection [in South American if I recall, Ed.] where increasingly strict rules did not lead to a reduction in tax avoidance. Using the framework they were able to come up with a self-organising regulatory device which limited tax avoidance.

The framework is a socially constructed template. You define the boundaries & domains. Categorize according to multiple case studies from the organisation. There is a critical distinction between hidden order and complex unorder. This can be hard to determine because of retrospective coherence. 9-11 was hidden order: we didn’t join the dots.

Mathemtical complexity.

4 dots, 6 linkages => 27 pathways (patterns) 10 dots, 45 linkages => 3.4 trillion pathways

There are so many orders of magnitude difference in considering a problem of 1 pattern among 27 and 1 pattern among 3.4 trillion. It is a particular interaction of agents that leads to a pattern.

Everything is right within boundaries. Dependent upon which domain you are in even contradictory theories can apply.

Do peoples personalities fit naturally into a domain? Personally people are entrained into patterns. Most innovation occurs under the age of 17 and over the age of 50. If the context shifts we are capable of changing. Personality can also act as a pattern entrainer. “I just discovered that wisdom comes with age. I didn’t believe it before, but I do now!”

The difference between perspectives of organisation/employee. At times you align with the organisation, at other times you are encouraged not to. UN peace keepers need permeable boundaries/rules, to be guided by underlying principles.

In a riot situation officers delegate to NCO’s. Officers understand strategy; NCO’s understand tactics and execution.

IBM is the world’s biggest beuracracy. It’s informal systems allow it to survive, to make the formal systems work. (55,000 informal networks).

The problem is that people designing the formal processes do not see this. Apparent conformity validates their model/view leading them to believe it is working and put in even more processes!

Think about yourself as a bundle of identities. Designed schizophrenia!