NYC Lean Kanban Meetup - Presentation October 28, 2015 - Jocko Selberg What do we really mean when we say that a problem is "complex"? Do we simply mean to say that a given problem is extremely complicated, or are complex problems something fundamentally different? We typically assume we are operating in a deterministic, ordered system where we can identify a cause and effect relationship, when in actuality we are often operating in a non-deterministic complex system, where these relationships can not be known in advance, if at all. How can we sense which context we are operating in and how might we act under varying degrees of uncertainty. Complexity Theory is a term used to describe a field that is focused on the study of complex systems. Complexity science is not a single theory— it encompasses multiple theoretical frameworks, seeking answers to some of the fundamental questions about continuously changing, dynamic systems. Cynefin is a framework developed by Dave Snowden and Cognitive Edge which seeks to helps us "make sense of the world, such that we can act in it". By understanding the fundamental differences between directed (ordered) systems and emergent (unordered) systems, we can modify our approach to match the context of the problem we are facing. The Cynefin framework takes a science based approach to dealing with critical business issues, drawing from anthropology, neuroscience and complex adaptive systems theory to improve decision making. Complexity Theory and Cynefin have an undeserved reputation for being difficult to grasp. In this introductory talk we will break down these approaches so that we can effectively use them to help us to better act under conditions of uncertainty. About Jocko Selberg Jocko Selberg is currently a Project Manager for The Nielsen Company with over 15 years experience in the interactive industry. He is a non-sectarian agilist and does not own a TV.

1. NYC Lean Kanban Meetup
Cynefin and Complexity: A Gentle Introduction
Jocko Selberg
@JockoSelberg
Jocko21@gmail.com
NYC Cynefin Group - Meetup
Oct 28, 2015
#nyclean
* Lean Coffee to follow talk – create your topics

2. Story Time!
Mr. Narrative Fragments
&
Little Miss Context
#nyclean@JockoSelberg

3. A Very Special Message
#nyclean@JockoSelberg

4. 3 Vignettes
#nyclean@JockoSelberg
Our brains instantly construct a story based on our perception of the context.
1. We see car arrive - followed by Mr. Skinhead turning and scampering away.
2. We see Mr. Skinhead engaging Mr. Trenchcoat man.
3. We see Mr. Skinhead saving Mr. Trenchcoat man from falling bricks.
Is Mr. Skinhead a criminal or an undercover cop?
Is Mr. Trenchcoat man a banker or a drug dealer?
Who is in that car?

5. Stories in Context
#nyclean@JockoSelberg
These stories can be radically different based on your point of view (identity).
The ‘Bro Officer O’Malley The Director

6. First-fit Pattern Matching
We often find ourselves responding to events
without fully considering the context.
Once we categorize something, it is cognitively
very difficult to change your mind about it.
– we categorize!
#nyclean@JockoSelberg
For example, as a PM or an executive, you may anchor the context in assuming you need for a detailed project plan for all work
Categorizing allows us to make quick decisions based on limited information

7. Sense & Respond
How can we make sense of a situation such that we can
respond intelligently given incomplete information?
(whenever a decision is required, we are ALWAYS limited by incomplete information)
#nyclean@JockoSelberg
Given the fact that we are very prone to categorizing everything…

8. Context is Everything
There are no universal solutions, but there
are contextual solutions
There are different ways of doing things based
on the context you are operating in
#nyclean@JockoSelberg

9. Making Sense of the Problem
How can we more accurately sense the context we are
operating in so that we might better act?
#nyclean@JockoSelberg

10. Order!
#nyclean@JockoSelberg
(S)he who defines the constraint,
defines (anchors) the solution space.

11. What is Order?
or·der
/ˈôrdər/
noun
1. the arrangement or disposition of people or things in relation
to each other according to a particular sequence or pattern.
How does order happen?
Must we create order, or can we allow order to emerge?
#nyclean@JockoSelberg

12. Emergent vs Directed
Emergent Order
Relationships between
cause and effect are
unknown, unpredictable or
not perceivable
Relationships between
cause and effect are
well established or
discoverable through analysis
Order emerges out of initial starting conditions
Directed Order
Order is directed or imposed
For the purposes of Cynefin, we’ll rename
Directed Order to Ordered
and
Emergent Order to Un-ordered
#nyclean@JockoSelberg

13. Characteristics of Order
Un-ordered
Relationships between cause and effect are unknown,
unpredictable or not perceivable
Relationships between cause and effect are well
established or discoverable through analysis
Order emerges out of the initial starting conditions
Ordered
Order is directed or imposed
Higher degree of Certainty
(highly predictable)
Lower degree of Certainty
(less predictable)
CausalDispositional
Linear relationships
(output is directly proportional to the input)
Non-linear relationships
(output NOT directly proportional to the input)
Exploitation - Commoditization
(Repeatability)
Exploration - Innovation
(Novelty)
#nyclean@JockoSelberg

14. Managing Differently
Un-ordered
Relationships between cause and effect are unknown,
unpredictable or not perceivable
Relationships between cause and effect are well
established or discoverable through analysis
Things Grow
Ordered
Things are Made
Define Ideal Future State
(close the gap)
Describe the Present
(influence the direction of change)
Manage the Plan
(set rules, scope - implementation)
Manage the Constraints
(set attractors, boundaries - experimentation)
Monitor for Adherence
(governance)
Monitor for Emergence
(patterns, weak signals)
Architect the Solution
(engineer)
Amplify / Dampen Attractors*
(feedback loops)
#nyclean@JockoSelberg
* An attractor is a state or set of states toward which a system tends to evolve. System values that get close enough to the attractor remain close.

15. Triaging
Un-ordered
Relationships between cause and effect are unknown,
unpredictable or not perceivable
Relationships between cause and effect are well
established or discoverable through analysis
Are you looking for a
repeatable solution?
Are you looking for a
novel solution?
The realm of learning
Ordered
The realm of planning
Nobody in the company, the industry,
or the world has solved this type of
problem in the past.
You, your team, the company or a
specialized group has solved this type of
problem in the past.
#nyclean@JockoSelberg

16. Systems
#nyclean@JockoSelberg

17. 3 Types of Systems
• A system is any network with coherence
• An agent is anything which acts on or within the system
• In nature, there are 3 types of systems:*
• Ordered Systems: the system tightly constrains agent behavior.
• Complex Systems: the system loosely constrains agent behavior.
The system and agents (collectively) modify each others behaviors.
• Chaotic Systems: the system imposes little or no constraints on agents.
Agents act independently of each other, behavior appears random.
* Constraints-based definition
#nyclean@JockoSelberg

18. Let’s Map These Systems
Ordered
(Directed Order)
Un-ordered
(Emergent Order)
• Complex
• Chaotic
• Ordered Systems• Complicated domain
Cynefin breaks Ordered Systems into 2 domains:
Complicated & Obvious, based on peoples
perception of the relationship between cause and
effect.
• Obvious domain
• Dis-order domain
Systems
Systems
domain
domain
#nyclean@JockoSelberg

19. Chaotic
domain
Complicated
domain
Complex
domain
Obvious
domain
Un-ordered
Ordered
Cause-Effect relationship perceivable only in
retrospect. Not Repeatable (except by accident)
Cause-Effect relationship discoverable via analysis
and/or expert knowledge.
Cause-Effect relationship is not perceivable
at a systems level.
Cause-Effect relationship is obvious to all
Perceivable, Predictable & Repeatable.
Introducing the Cynefin Domains
The boundary between Obvious and Chaotic is represented as
a cliff, or a catastrophic failure arising from complacency.
#nyclean@JockoSelberg

20. Notes on Sense-making
Cynefin is a Sense-making framework, it is not a Categorization Schema
It’s all about the dynamic movement between domains
#nyclean@JockoSelberg

21. Complexity 101
#nyclean@JockoSelberg

22. Flow and Self Organization
• Autonomy – each “actor” is making
individual decisions
• Locality – behavior dependent on
decisions of immediate neighbors
• Distributed Cognition – no centralized
control
• Dynamic Interactions – via continuous
“micro-decisions”
• Simple “rules” – Match speed, avoid
collision, fly to the center of the flock
• Agents are generally ignorant of the
behavior of the system as a whole
#nyclean@JockoSelberg

23. Agents
An agent is anything which
acts on or within the system.
Examples of agents might be
molecules, ants, memes,
myths, identities, etc. *
Relationships between the
agents are more important
than the agents themselves.
Agents are autonomous and
self-organize.
* People are rarely agents. They are complex systems in and of themselves.
#nyclean@JockoSelberg

24. Agents Interact with Agents
Agents interact with
each other locally and
modify each others
behaviors via their
mutual interactions.
Proximity and
connectivity are key.
Simple “rules” act to
both constrain and
enable behavior.
#nyclean@JockoSelberg

25. System and Agents Interact
Agents and the System
also interact with each
other. Each may also
modify the behavior of
the other.
They co-evolve together.
Order emerges, it is not
imposed.
#nyclean@JockoSelberg

26. Complex Adaptive Behavior
New
Patterns
Emerge
Patterns
Modify
Behavior
(+)(-)
As agents continually interact, patterns emerge.
Positive and negative
feedback loops develop…
…further modifying the
behavior of the agents.
These patterns modify the
behavior of the agents.
...which act to amplify or
dampen existing patterns…
Rinse, Repeat, Concurrently - billions and billions of times.
#nyclean@JockoSelberg

27. Systems are Hierarchical
These systems may also
act as agents in a larger
system. They are self-
similar and hierarchical.
That is, systems both
contain sub-systems and
are contained by larger
systems. They’re fractal.
At each level of
complexity, entirely new
properties appear.
#nyclean@JockoSelberg

28. Complexity in a Nutshell
• Systems are hierarchical – cells, organs, organisms, societies…
• Systems are dynamic – systems change in patterns over time
• Agents self-organize – There is no centralized control
• Order emerges – it is not imposed
• Multiple autonomous agents – interact with each other and the system
• Simple ”rules” (constraints) – produce complex interactions
• Systems and agents co-evolve
• Co-evolution is irreversible – There is only the current state*
#nyclean@JockoSelberg
* and its history – The present is informed by the past.

29. Designing a Complex System
Can such a system be designed or engineered?
“Not bloody likely.”
- Eliza Doolittle (Pygmalion)
That is, can a complex system be driven towards an Ideal Future State?
#nyclean@JockoSelberg

30. Why is This?
Complex systems are fundamentally different
• Small changes may produce massive impacts, or may do nothing at all
• You cannot know in advance which solution will work best because of
the dynamic nature of the network of interactions
• These systems evolve, there is no Ideal Future State to “achieve”
• These systems do not lend themselves to a reductionist approach
• There is only “fitness for purpose” within the “situated present”
#nyclean@JockoSelberg
Irony Alert – The present is generally more certain than the future

31. Working in Complex Systems…
…requires a shift in how we view our work at a systems level
• Monitor for emerging patterns, suppress the urge to categorize: Probe
• Identify patterns as they emerge (in real time)
• Seek opportunities that allow the system to move in a desirable direction
• Amplify / Dampen patterns rather than trying to drive system: Nudge
• Beneficial coherence can be amplified (within attractors)
• Negative coherence can be dampened by denying energy (resources)
• Manage the constraints rather than “the plan”: Bound
• Apply context sensitive constraints – the level of constraint is defined by the domain
• Manage via constant, small interactions - within boundaries
#nyclean@JockoSelberg
The only way to manage a complex adaptive system is by interacting with it

32. My Mayonnaise is Complex
“A jumbo jet is complicated, but a mayonnaise is complex.”
- Paul Cilliers (1956-2011)
#nyclean@JockoSelberg

33. Cynefin
“It infuriates me to be wrong
when I know I'm right.”
- Molière
#nyclean@JockoSelberg
The “right” solution in the wrong context
is the wrong solution.

34. Knowns, Gnomes and Unknowns
“As we know, there are known knowns; there are things we know we know. (obvious)
We also know there are known unknowns; that is to say we know there are some things we do not
know. (complicated)
But there are also unknown unknowns - the ones we don't know we don't know. “ (complex)
Donald Rumsfeld – (a known gnome)
February 12, 2002
#nyclean@JockoSelberg

35. Let’s Make Sense of This…
…using the Cynefin Framework
#nyclean@JockoSelberg

36. Chaotic
domain
Complicated
domain
Complex
domain
Obvious
domain
Un-ordered
Ordered
The Cynefin Domains
Cause-Effect relationship perceivable only in retrospect.
Not Repeatable (dispositional, not causal) Flux
Cause-Effect relationship separated by space & time.
Requires analysis and/or expert knowledge.
Cause-Effect relationship is not perceivable
at a systems level. Turbulent.
Cause-Effect relationship is obvious to all
Perceivable, Predictable & Repeatable. Stable
Retrospectively coherent causality
(Unknown unknowns)
Incoherent causality
(Unknowable unknowns)
Knowable causality
(Known unknowns)
Known causality
(Known knowns)
“Emergent practice” “Good practice”
“Novel practice” “Best practice”

37. Chaotic
ComplicatedComplex
Obvious
Un-ordered
Ordered
Responding in Cynefin Domains
Complexity Based Approach
Create multiple, safe-to-fail interventions
(identify patterns, gain insights)
Manage in the present.
Focus on the evolutionary potential of the present
Systems Thinking Approach
Not Self-evident (test & refine hypothesis)
Engineering (reductive)
Trust the Experts
Focus on goals
Crisis Management Approach
(stability-focused intervention, impose order)
Urgent Rapid-Response
Define the constraint – while you still can
Standard Operating Procedures
Self-evident (automation, rules)
Trust the Manual
Focus on efficiency
Sense – Analyze – Respond
Act – Sense – Respond Sense – Categorize – Respond
Probe – Sense – Respond
#nyclean@JockoSelberg

38. Contextualization
#nyclean@JockoSelberg

39. Un-ordered
Ordered
Contextualization
Submitting
an invoice
Refactoring code
Building
a House
Trouble-shooting
an outage
Operating
a Start-up
Creating
'product buzz'
Finding
new markets
Shooting
Rampage
#nyclean@JockoSelberg

40. Un-ordered
Ordered
Let’s Build a House!
Building
a House
Finding a Location
Construction
Medical Emergency Making Payments
Securing a Mortgage
Building a
House on Mars
You’re Fired!
Building a
Bird House
Building Shelter
(from a storm)
#nyclean@JockoSelberg

41. Conclusion
A complexity based approach asks the following:
• In the context of what is going on at the moment, what can we change?
• Out of what we can change, what and how can we monitor for the
impact of that change?
• Out of the things we can monitor for impact, where can we amplify
success or dampen failure?
#nyclean@JockoSelberg
What is going on at the moment?
“We evolve forward into a future state, which is more sustainable and
resilient, but which we could not have anticipated.”
- Dave Snowden

42. Thank You!
NYC Lean Kanban
MeetupSpecial Thanks to:
Dave Snowden
Jocko Selberg
@JockoSelberg
Jocko21@gmail.com
NYC Cynefin Group - Meetup