von Neumann and complex, dynamic, evolving, loosely coupled chaotic systems

This is an interesting perspective.  I have done a lot of risk consulting work over the years.  One phenomenon that is almost always true is that negative outcomes in a complex, dynamic, evolving, loosely-coupled chaotic system are almost always multi-causal in nature.  Root cause analysis only gets you so far because in such complex systems the roots are many and inter-tangled.  

It is comparatively easy to wring out risk from static simple systems.  You engineer it away (risk elimination) and buffer it with both risk reduction and risk mitigation.  You can do so because the root causes are few and the relationships simple and direct.   In complex, dynamic, evolving, loosely-coupled chaotic systems?  Not so much.  

In reading John von Neumann Seen by his Brother by Nicholas A. von Neuman, there is an interesting passage dealing with risk in complex systems.  Nicholas and John are discussing different aspects of the 1912 Titanic sinking.

The "Titanic" catastrophe remained in the headlines for many years, particularly after details of the USA Senate and British Board of Trade investigations became better known. I got quite excited about the argument that if only one of the many contributing factors had not occurred, then the tragedy could have been averted. John assured me that my "if only" approach is unreasonable and that I should relax. What I should say is: if another random set of circumstances had existed; and then some of them would have been worse (trade First Officer Murdoch's in-the-retrospect-erroneous order "hard a starboard, full speed astern" for a high wind or storm) , and none of the lifeboats could have been lowered!

The "Titanic" also involved discussion of a broader moral issue: responsibility for training persons in command positions in how to behave in situations of stress or panic. Indirectly, this also has CNS context. However, at that time our discussion was limited to such questions as who would have

been responsible for training First Officer Murdoch in high speed maneuvering; who would have been responsible for equipping the ship with the additional lifeboats for which the davits were designed; responsibility for "Californian"'s Captain Lord's failure to order his radio operator on the air at first sighting of distress signals (accompanied by detailed analysis of the question whether the "Californian" could have arrived on the spot in time to be of help); was J. P. Morgan in any way responsible in his capacity as sponsor, organizer, and financier of International Mercantile Marine, owners of both the "Titanic"'s White Star Line and the "Californian"'s Leyland Line; etc. , etc.

CNS refers to Central Nervous System, one of the many fields in which John von Neumann was a pioneer.

Richard von Neuman is focusing on the traditional cause-and-effect simple system perspective captured in the adage of For Want of A Nail and reflected in Shakespeare's Richard III, "A Horse! A Horse! My Kingdom for a Horse!"  It has deep roots in western thinking and in American traditional wisdom.  

In 1758, Benjamin Franklin included the following poem in The Way to Wealth, observing that "A little neglect may breed great mischief"

For want of a nail the shoe was lost,

for want of a shoe the horse was lost;

and for want of a horse the rider was lost;

being overtaken and slain by the enemy,

all for want of care about a horse-shoe nail.

This describes a simple, tightly coupled, static, cause-and-effect system.  There is no ambiguity or probability.  No nail = lost shoe.  Lost shoe = lost horse. etc.  A lost nail is 100% cause of lost battle and lost kingdom.  

Introduce loosely coupled non-linear, dynamic and evolving chaotic systems and you end up with probabilities.  An example with made up percentages:

No nail leads to a 40% probability of a lost shoe.  

A lost shoe is a 20% probability of a lost horse.  

A lost horse means a 70% probability of a lost rider.  

In a world of loosely coupled non-linear, dynamic and evolving chaotic systems, the probability of this happening is 5.6% (40% X 20% X 70%) rather than 100%.

John von Neumann's insight is one I have not seen before but gets to the heart of the issue.  Complex, dynamic, evolving, loosely-coupled chaotic systems are hard to shape, much less steer, because there are so many moving parts in configurations which cannot be anticipated.  The deterministic approach of root cause analysis only goes so far in a complex, dynamic, evolving, loosely-coupled chaotic system.  There is too much to take into account.  

Von Neumann's insight is that in such systems which are not simple, tightly coupled, static, or governed by strict cause-and-effect rules, what you really have are constantly shifting portfolios of risk scenarios.  In fact, an infinite number of such scenarios.  

You have to think of portfolios of scenarios rather than simple linear cause-and-effect models.  Applying cause-and-effect thinking to loosely coupled non-linear, dynamic and evolving chaotic systems is a category error.  You have to think about relative probabilities of portfolios of scenarios.  Which is hard work.