(Ikke tilgængelig på dansk.)
Often we find that it is the borderline areas of concepts that really define their true nature. So it is with complexity.
Dave Snowden has once again, true to his nature, been pushing the boundaries of Cynefin. For some months we have seen his various ideas on how to cope with borders of complexity. There is the border country between Complex and Complicated as shown above with the green arrows; and there is the border country between Complex and Chaos shown with the red arrows.
As expected, Snowden has come up with a challenging word to describe this extension to Cynefin, “Liminal Cynefin”. The word “liminal” can mean a transitional stage of a process or a position at, or on both sides of, a boundary or threshold.
Second, spaces such as the threshold of a door are “liminal,” lying between otherwise defined areas without belonging to either of them.
1999, Sarah Iles Johnston, Restless Dead
For the present discussion we will use the following sketch, CHCO and COCO are the new liminal territories:
We mentioned in the previous article how an agile pattern like Scrum lives and thrives in this area between Complex and Complicated. It is therefore natural to especially expand on this area.
Snowden explains this COCO territory as one of increasing constraints on specific challenges until it ends up in the Complicated domain where solutions can be properly analyzed, designed and executed. The longer the situation can be kept in the COCO territory, the better chance there is to avoid jumping to conclusions too quickly; on the other hand, it takes longer to reap the benefits. There is therefore a balance to be found. Lean Software Development: An Agile Toolkit, explains “the last responsible moment” in this context:
Concurrent development makes it possible to delay commitment until the last responsible moment, that is, the moment at which failing to make a decision eliminates an important alternative.
Mary and Tom Poppendieck
It is also defined as follows: “The last responsible moment is the instant in which the cost of the delay of a decision surpasses the benefit of delay”.
The problem is of course that in the Complex domain it is not always known when that is. “Cost of delay” is good concept to keep in mind when prioritizing, but it is dangerous to oversimplify it by making a neat formula.
The COCO territory requires plenty of energy and constant work at the controls. It is a bit like a modern jet fighter, inherently unstable and only able to fly because there is very effective control system of feedback loops that constantly keep making small adjustments to the course.
The Chaos-Complex border country (CHCO)
This is often what people mean when they use the words “borderline-chaos”. According to Snowden, in CHCO constraints are reduced and relaxed until there is certain ambiguity and a sense of randomness, so that old thought patterns can be left, and sensemaking take place. Our brain is conservative and prefers to stick to well-trodden paths. Entering CHCO is a deliberate movement to create room for new solutions (innovation) to appear. Various tools exist to help accomplish that.
The purpose of entering CHCO is to come up with new ideas for solutions and get back to the Complex domain and start iterating through experiments to build up useful knowledge. It is like setting out to the unknown sea, casting the net and trawling for ideas.
The beauty of time boxes
There are however more lessons to be learned from understanding these liminal territories:
In most Agile disciplines work is executed in series of Timeboxes, Sprint or Cadences. This means that under certain conditions it is fair to say that iterations as entities exist in COCO territory. A Sprint is a collection (typically a couple of handfuls) of Product Backlog items (delivery narratives, stories). We always recommend Teams to estimate or assess complexity of the backlog items:
- Some will be in the Obvious domain; there is an accepted best practice to deliver that sort of thing.
- Some will be Complicated; a solution can be found through analysis or research.
- Some will be Complex; a solution needs to be found by trying out different hypotheses combining existing fragmentary knowledge in new combinations with new ideas.
- Some might even be in Chaos where no cause and effect relationship can discerned; a solution needs to be found completely by finding new ways.
Initially before anybody has looked carefully, everything is in Disorder; the state of not knowing the domain. Gradually as work progresses, clarity evolves and the backlog items can for the most part be placed in the correct domains according to the Team’s judgement. When backlog Items reach the top Sprints of the backlog there should be clarity about the nature of the item, i.e. which domain it falls in.
There are then different scenarios possible for a Sprint, following the Scrum Pattern:
During Backlog Refinement in the middle of the previous Sprint and the Sprint Planning, a set of backlog items are identified as top priority. These items have all been developed to the point where they are either Complicated or Obvious. This means that predictability exists with some variation; a complete Sprint Goal can be established. Remaining analysis and design can be done in the Sprint. Effectively the Sprint is in the Complicated domain. This is what most seem to imply is the territory of Scrum.
Alternatively, all backlog items are deep into Complex territory and parallel experiments have to be run on multiple fronts. Some Items can have experiments whose duration or effort can be estimated with some variation; others cannot, but it is possible to timebox the experiments and define how much time is spent. The concept of a timebox or a Sprint can still be legitimate and it will produce a result, but nobody can say what it is; most likely the result will be new knowledge harvested. The Sprint as a whole is in the Complex domain. This is fairly typical of the early days of an initiative.
In some situations in the Complex domain, Sprints don’t make sense. Sometimes it is not possible to decompose large Epics into smaller stories with parallel experiments. Instead, experimenting with each item seems to open up completely new avenues like driving in the mountains, coming around a corner and seeing a totally new perspective. In those situations we prefer to work with a giant relationship diagram, trying to discover relationships between items and cause and effect hypotheses, still keeping the visibility high and the feedback loops working at even higher speeds like daily or twice a day.
And then, what we find is in fact the most common scenario for a Team or a Circle to be in: A state where some backlog items are Complex, some Complicated, some Obvious and even some in Chaos. The timebox or the Sprint will contain some of each. There will be some predictability and repeatability but also a great deal of unanticipated results. This, in our opinion, is COCO territory; constantly iterating trying to produce something while at the same time experimenting with something else.
In many situations it is a desirable state to stay in as a stream of individual items are dealt with, not just one. By working on a collection of items (the Sprint), uncertainty is reduced because of the averaging going on. By being clever during Sprint Planning, the Sprints can be made somewhat similar (mixing Complex, Complicated and Obvious items in roughly the same proportions) and consequently compared legitimately. That means experience at the macro level (the Sprint) is built up, although variation will be great at the individual Item level. Estimates and variation can make sense at Sprint level although it is illegitimate at the Item level.
In our opinion, this is an often overlooked fact and benefit. Hardcore Kanban folks are so focused on the single item, that they often miss the benefits of seeing the bigger picture across a Sprint. The Sprint performs a sort of low pass filtering on the events allowing us to see trends and extract meaning.
What about the CHCO territory?
The question is how to deal with the CHCO territory; are Sprints and Teams used or is this something else all together? The fact is that most people dealing with this territory are focused on the tools and methods in creative session trying to innovate. There are many innovation games or workgroup methods available, but very little on the cognitive foundation of moving into CHCO.
In Snowden’s definition, the borderline chaos of CHCO is entered with intent – it’s not an accidental move. It has previously also been described as a shallow dive into chaos. The new liminal definition fits this concept, CHCO is just on the Chaos side of the border and yet distinctive from real Chaos. The purpose of entering CHCO is to come back to Complex with potential new solutions.
The move into CHCO can be performed by a Team or a Circle normally working iterating between the Complex and the Complicated domains. New challenges may have come to the Team or they simply may have decided that after a longer period of constant improvement (Kaizen) they need to see if there are some major discoveries to be made (Kaikaku), and they go fishing, trawling for innovation or exaptation.
The move also can be initiated as a transient secondary Circle or Team being put together, with appropriate skills and different perspective, there has to be some dissent in order create the instability. People with good “Teaming” skills are essential in order to form quickly and to work; not in the sense of aligning completely, but being able to work respectfully in a climate of instability and competing views.
We believe that it is a suboptimal solution to make specific static Teams and Circles for this purpose. Skunkworks per definition has to operate either subterranean or transiently to experience the freedom that spurs the trawling of ideas. If it is institutionalized it becomes detached from the real work, people lose contact with the core of the organization and it has a propensity to develop into cult-like brotherhoods.
The methods to go trawling are outside the scope of this article, but it normally involves extreme visualization of ideas and concepts. Often it is a good idea to have a sort of artist or illustrator involved to capture ideas and proposal for connections and solutions. A room where people dive into CHCO territory starts to look like a criminal investigation room with evidence boards etc.
The other areas in Cynefin
Chaos is a transient domain; it is not possible to stay in Chaos for long. The trick is to intervene to gain some control of the situation before the world stabilizes anyway, probably in an unfavorable way, according to Murphy’s Law.
When an organization finds itself in Chaos, it needs to act very quickly as described elsewhere. Typically a Crisis Team or Circle is formed out of very experienced people with authority. They work pretty much like described under CHCO, with the added challenge of typically being under extreme time pressure and threat of dire consequences. They always have the two options in front of them: Save or Salvage? Sometimes a situation is best left to fail and the pieces picked up afterwards.
The same situation exists at the finer granularity in many Teams’ and Circles’ work. An issue appears and people have no idea of how to solve it; they can see no cause and effect relationship. Many problems in products have this nature and require concentrated effort. Emergency problems will often need very rapid response and will thwart all plans, Sprints and commitments.
The border country between Obvious and Chaos is already known in Cynefin as “The zone of complacency”. It is not a place any reasonable person goes voluntarily. In a crisis, catastrophic failure lurks under the surface, and a sudden drop into Chaos can occur. It is therefore not a place to involve in normal operation of an organization, but more of a warning sign to avoid smugness and complacency with regards to over-regulation and over-structuring processes and trusting in that.
In the Obvious domain solutions are, well, obvious. Simple checklist and best practices are sufficient.
Complicated is the area where traditional planning methods are legitimate and well documented.
The border between Complicated and Obvious is in our understanding not a domain or territory in its own right, but rather a very soft transition. Less and less research or analysis is required as solutions move towards the Obvious. It is a very gradual shift, no steep incline or decline. Perhaps this is the reason why it is so easy to drift deep into the Obvious domain approaching the Zone of Complacency without noticing. It almost feels the same all the way and rules and regulations are mostly very comforting. So the main thing about the territory is to stay awake and not let solutions slip into Obvious inadvertently.
Cynefin is a powerful thinking hat when trying to make sense of things. The new addition of the border territories in Liminal Cynefin promises more insight into how to select an appropriate course of action. It is not just a theoretical abstraction, but a pattern that can be made actionable.