Dave Snowden

The mantra of resilience

RSS Feed

Screen shot 2011-03-10 at 05.39.07.png We’ve been involved in an interesting project on social resilience and I have been working tonight on the presentation for tomorrow. Aside from the analysis, I needed to get some of the theoretical stuff around null-hypothesis/abductive research together in an easy to explain format for economists and others of that ilk. At the Academy of Management last year in Montreal we made more progress on integration of SenseMaker® into Power Laws, something that will move on another stage in San Antonio later this year. Max Boisot did the summary slot at that event, putting together material from myself and others and I have been playing with the slides he used since.   

The basic starting point here is risk assessment. In the majority of cases governments and industry asses risk based on a bell curve or Gaussian distribution, shown as reddish brown in the diagram. Now this works well for things like tossing coins, but when it comes to natural events such as earthquakes and floods its less effective. If we use a power law (or Pareto distribution) then a double log scale of size against frequency produces a straight line (shown below but wait for it), which matched on the the curve above shows a fat tail, as shown on the blue curve. So an outlier event, dismissed as an outlier on one distribution turns out to have a much higher probability if we use the more naturally occurring Pareto one.

Net effect of all this is that so called Black Swan events, or better low probability high impact events have a much higher probably than we thought. I sometimes refer to this as the New Orleans Levees problem. If you assess risk using the wrong model then you create a false sense of security. The problem is that with a fat tail the cost become prohibitive, there are simply too many things we would have to take into account. Cynically I think this is the reason of lot of people are fixated on the Gaussian Universe; it’s not just that it is familiar, the alternative seems far too costly.

Now the alternative to this sort of self-dillusional behaviour is to start to learn to live with failure and to use it for strategic advantage. This is the switch from robustness to resilience which is the main theme of my strategy work (if you want to know more book on one of the seminars).

The mantra of resilience is early detection, fast recovery, rapid exploitation.

Agility properly understood is not about succeeding faster, but managing failure better.

Screen shot 2011-03-10 at 05.41.47.pngNow that has implications for how we assess evidence, do research and generally think about the world. That is what our second illustration is all about. Here we a double log scale of frequency of event against size. and we see the power law line again.

Now small event size and high frequency is no real problem. Power laws generally flatten at this end as this is the Gaussian universe. The number of data items mean that the statistics work. This is the domain were we can make statements about events that are probably and where we should be able to make confident predictions. That means the level of available certainty (its complicated) is high and hypothesis based inductive research methods (cases per business school ideology) work. Its the domain of good and best practice, of their being a right thing to do and a reasonable expectation of robust solutions.

However as we move to the right, frequency reduces and size (impact) of events increate. The numbers no longer work and we in the Pareto Universe. In the borders between the two we may not be dealing with the probably, but we are dealing the the imaginable possible. Hypothesis techniques with a judicious sprinkling of human judgement can cope. However as we move to the right this is less and less the case and we are in the domain of null hypothesis research and abductive reasoning. While SenseMaker® can handle inductive with ease it comes into its come in this domain, the full complexity of the Cynefin framework.

In a Pareto world strategies can not be based on probability assessments, instead we are asking questions about which safe-to-fail experiments have plausibility of coherence. This is the new dynamics of strategic thinking and its very exciting as we can increasingly use technology to augment (but not replace) human judgement. It is a world of praxis, theory informed practice in which we need to engage our workforce, our customers and people in the decision making process. More on that over the next few days.

Top