H

Our Blogs  

 

The OODA loop & Cynefin

By Dave Snowden  ·  November 10, 2012  ·  Reflections

Earlier this month I picked up a tweet from Clay Spinuzzi of Austin University asking how the Cynefin framework related to the Competing Value Sets (CSV) model.  This first saw the light of day in a 1981 paper from Quinn & Rohrbaugh entitled A competing values approach to organizational effectiveness.  My initial response was to say that one would need a different version of the CSV for each Cynefin domain.  Cynefin and CSV do different things so overlaying one on the other makes little sense although some superficial similarities might allow it with some reorientation.   

I promised to post about the CSV links later, but first I needed to get hold of the original paper and also read Quinn & Cameron's Diagnosing and Changing Organizational Culture: Based on the Competing Values Framework.  If you are going to play around with other people's models and frameworks integrity requires you to read their material.   A daughter with access to a University Library satisfied the first, and Kindle the second.  Odd really, I thought I had that book but an email home revealed that if I did, I had lent it to someone who had failed to return it.  People who do that by the way are amongst the lowest form of pond life but that is a story for another day.

In agreeing to do that I also realised I had a long standing task, something requested by several readers, to link Cynefin with Boyd's OODA loop. Both OODA and Cynefin work in the decision support arena as both Cynefin and CSV work in the cultural mapping domain (and others).  I also used the OODA loop a lot in the early days of knowledge management to illustrate how using knowledge would allow you to move around the loop faster, and thus gain advantage over the information user.  My eternal thanks by the way to James, an ex Navy Officer and part of the GENUS team at DataSciences for introducing me to both that and complexity theory three decades ago.

For those not familiar with it I've used an image from Wikipedia above. OODA stands for Observe, Orientate, Decide and Act.  The idea is that in conflict if you go through the process faster than the enemy you win.  Boyd's use is sophisticated, including a recognition that the Orientate stage is not just about analysis but includes aspects of our personal background, genetic history and cultural traditions. Its a model that has been used in both military and non-military environments and has several adherents including yours truly.

Now there are some remarkably foolish ways to integrate models.   Probably the most silly is to simply mash things together in the hope they will fit. I gather this is called smooshing by the way but why anyone would choose to embrace (sic) such a name I don't know but it may be indicative of a wider unwillingness to do basic research!  According to the Urban Dictionary the word something to do with an embarrassing type of male failure and synonyms include such delights as whiskey dick.  Either way I'm prepared to believe that it might ultimately provide something of value, probably in the same way as all those monkeys with typewriters are meant to produce the works of shakespeare. I prefer something with a little more intellectual vigour (sic).

One example of smooshing, from an old post by Tom Graves, is shown here, originally referenced in my article on the History of Cynefin. I show it as an attempt to overlay OODA onto Cynefin and illustrates my concern about the dangers of simply throwing things together in the hope that something useful may emerge.  It misses the point completely.  The essence of Cynefin, and even the most basic reading of the literature would demonstrate this, is that each domain carries with it a different mode of decision making.  So the OODA loop does not move us around the domains as a whole although it may shift us between domains as a result of the acts that arise from the process.

The real point is that each domain needs its own variation of OODA and some indication of the direction that would be taken by its successful application.  Now OODA is sometimes drawn as a circle as shown at the start of this post and I want to use that simplified form to look at its application by domain within Cynefin.  I did think about doing this as a table but that left a few blank cells so this is going to be a mixture of narrative and another picture.

Its key to remember that Boyd did not see this as a single loop, but multiple parallel loops, which given he was thinking of war is appropriate.  I'd argue that this also applies to the complex domain of cynefin, where we are dealing with unknowable relationships between cause and effect in advance of action.  We can know something of the systems dispositions, and we will have some sense of purpose (see my post of yesterday).  

So in the complex domain we need to:

  • Radically expand the range of observers and things observed
  • Accept multiple competing hypotheses (this includes holding options open in individual decision making) each of which translates into a safe-to-fail act
  • Observe the pattern of response over all the actions to determine which are successful, or more coherent to amplification
  • As and when possible increase the constraint to the point where we shift to the complicated domain

Now I have shown that in the illustration to the right.   I've taken the circular model and coloured the action transition red to make the OODA orientation clear.   I've also shown transition arrows moving from their appropriate point on the cycle.  This means:

  • The actions of the smaller parallel OODA loops create the observations at the start of a larger more formal OODA loop in the complicated domain.
  • Assuming a non-deliberate entry into Chaos, then decisive action is needed to create sufficient stability (constraints) that safe-to-fail experiments are viable, so the exit is again from action, but moves into decision in the complex domain
  • As the cycle continues in the complicated domain some decisions may allow the level of constraint to increase to the point where best practice is viable; so this enters the simple domain at the orientation point.
  • Pattern entrainment and disruptive innovation requires a specific action in the complicated domain to relax the constraints sufficiently to trigger a purposeful entry into chaotic at the orientation point of OODA

Now there are probably other transitions, and I am open to argument, modification etc. on the above.  Its a first cast at what proved to be a difficult task.  Well at least more difficult than I thought it was going to be.

I think there is more work to do here on the nature of observation and orientation, in particular looking at what filters are permitted or used and the degree to which information is sought, stumbled upon or ignored.  Boyds various forms of orientation also merit further treatment but for the moment this is it!   I've been playing with it for the best part of a year time to put it in the public domain.

The CSV model I think will be less about dynamics and more mapping orientations on that model appropriate to domains and domain transitions in Cynefin, but I have some more reading to do first and some thinking.  If anyone else has picked up on the original tweet and has some thoughts let me know.  I need to pick up on PDCA as well sometime but that is more complicated (I choose my words with care).