H

Our Blogs  

 

Scaling: the perversion of order

By Dave Snowden  ·  July 24, 2014  ·  Reflections

To my final post on scaling, at least for the moment as it's a big and developing subject.  There is a tendency in some complexity writing to simply abandon the whole question of management to vague and idealistic statements about self-organising and natural systems.  The simple fact is that choosing not to do something is as much an intentional act as to do something; sins of omission are at the same level as those of commission.  The issue is not management, but what is the focus of management.  To put it simply it's all about what you can manage, and what types of things give you indicators of propensities and dispositional states.  Propensities are innate features of things and connections, dispositions are the current state of an evolutionary system and its plausible vectors.   I'm using specialist language with (hopefully) precision here that references back to other posts and writing.  The sequence probable-possible-plausible for example comes from the need to manage in the tail of distributions where there are few precedents that will repeat and we need to focus on resilience.

Now the question of what can be managed, and monitored, in a CAS was the subject of our recent workshop in New York for UNDP where we had development experts and a group of scientists I brought to the table.  We ended up with this list:

  1. Boundaries and their nature: are they rigid (in which case they can easily become brittle); are they permeable which makes them more resilient; are they self maintaining or do they take a lot of energy?  Are they naturally present or do they need to be created.
  2. Constraints may of course be boundaries, but the key question is their nature.  Are they governing constraints associated with order where deviance is not tolerated?  Are the enabling constraints which allow for emergence?  A permeable boundary is a type of enabling constraints in that you have to achieve a certain status of capability to pass through it.  Heuristics to ensure diversity in the formation of self-organising teams is another example
  3. Granularity has been the subject of many comments in this series.  Too finely grained and you lack coherence, to coarsely grained and there is no change or adaptation let along exaptation.  Small, coherent objects interacting with other objects from different sectors to create new and beneficial coherences are key, so we measure that up front.
  4. Gradient has become easier to explain since Piketty's book gained popularity.  If everyone earns the same there is no change, but if the income disparity is massive there is either no change or revolution.  Some gradient, some misunderstanding and some difference of interpretation is vital.  That is one reason why the fad of alignment in systems thinking is dangerous. Too much or too little is bad.  So again its something we can measure, and against which we can assess proposed actions.
  5. Catalysts are the mechanism by which attractors are stimulated.  I can't create an attractor but I can experiment with different catalysts until I get one which is favourable.  By their very nature catalysts are often oblique in nature.  So a strategy for catalysing change against unknown outcomes is one of the potential measures of resilience.
  6. Micro-diversity links back to granularity and forward to coherence.  It is about variety of numbers, but also of type.  The higher the level of uncertainty the more (up to a point) diversity we need to operate effectively.  Again this can me measured, managed and designed whole outcomes cannot.
  7. Explanatory narratives are the way we create understanding and also the potential for advocacy, we need to test the description of the past and present against the micro-narratives of day to day life as a reality check,  The post-hoc narratives are also tested for coherence and change.  In SenseMaker® this moves on to a quantitative measure of impact without that impact having to be over defined up front.
  8. Cycle time stability is a more abstract idea, but it relates to dynamic as opposed to static stably.  Product and design life cycles, cadences etc. etc are all examples here.  While a lot of budgeting is necessary, the cycle of budget creation is also a form of attention gaining and reflective focus.
  9. Coherence is the key word in this whole new approach to evidence based policy.  Is the idea coherence enough to be a worthwhile experimental pathway, or is it incoherence?  its easier to agree that an idea is coherence that too agree its right so this is a way of enabling conflict resolution and more diversity in the early stages of definition.  My favourite illustration is that while we know much of evolutionary theory is coherent to the facts we know it is incomplete.  Creationism on the other hand is incoherence and not worthy of further exploration.  

Now this is a further development and expansion of the key phrase in the Childrens' Party Story namely that we manage the emergence of beneficial coherence within attractors within boundaries.  Now with the UNDP this basic typology/taxonomy has been converted into an experimental assessment process for new projects.  The idea there is that if we can manage, or at least better understand the starting conditions we have a better measure of the plausibility that some investment will produce a beneficial outcome.  Small investments can more easily be made into safe-to-fail experiments in a balanced portfolio before we commit significant resource.  

As in development, so in pre-scrum processes in Agile and to resolution of strategy conflicts in government and industry alike. This friday we start to move the method and principles into the software development sector and the wider use in industry is already well underway with adoption at board level in several companies and government agencies - a lot outside of our direct control or influence.  The thing about complexity is that when you get it life becomes a lot easier, but for many people the world is all to simplistic and they want tightly defined and aggregative measures that remove ambiguity.  For these people, particularly in the software community, the blue bill is the safer (sic) route while the red pill is all to scary as it involves becoming an adult.

The assessment structure outlined above is being tested at the moment and there will be an automatic test system using SenseMaker® in place for when I get back to run seminars in Brisbane and Sydney on the subject in just over a weeks time.

The principles of scaling in a complex system are actually quite simple, the practice is evolving.  The problem is the perversion of order which pervades thinking, the desire of neat tidy structures in which everything has a place, and is in its place.  Sick Stigma, SAFe, many evaluation programmes based on gamed outcomes and so on all represent that type of perversion.  I use the word perversion advisedly by the way, not only is it unnatural, but their adoption produce unnatural and twisted outcomes.