Current design methods need a radical rethink. We must shift from a fail-safe design to safe-fail experiments. In my keynote at the KMWorld & Intranets Conference in September (available as a podcast at cognitive-edge.com), I elaborated on that idea and introduced the science of complex adaptive systems as a key piece of underlying theory to allow us to implement and, as importantly, to scale the change.
When I talk about a system, I am not just talking about the technology aspects, but all of the human and environmental aspects. When I talk about an agent, I mean anything that acts. An agent can be an individual, but it can also be a collective identity, an idea, a myth or—more technically—an object. Most management theory is familiar with the following two systems:
- Ordered systems, in which there are repeating relationships between cause and effect that can be discovered by empirical observation, analysis and other investigatory techniques. Once those relationships are discovered, we can use our understanding of them to predict the future behavior of the system and to manipulate it toward a desired end state. Critically in an ordered system, the nature of the system constrains agent behavior to enable that predictability.
- Chaotic systems, in which the agents are unconstrained and present in large numbers. We can gain insight into the operation of such systems by the application of statistics, probability distributions and more. The number and the independence of the agents allow large number mathematics to come into play. In recent times, we have seen some popularization of that with varying degrees of success and intellectual integrity in phrases such as crowd sourcing and the wisdom of crowds.
During the 20th century, natural science (mainly chemistry and biology) became aware of a third type of system called complex adaptive systems. Here the system lightly, but not fully, constrains agent behavior, and in turn the agents through their interactions constantly modify the nature of the system. The technical word for it is co-evolution. In essence, each agent in a co-evolutionary relationship exerts selective pressures on the other, within an environment that itself creates pressures, thereby affecting each other’s and the system’s evolution. The result is a system that operates in far from equilibrium conditions with some important characteristics for strategy:
- High susceptibility to small changes in starting conditions can magnify consequences quickly and in unexpected ways.
- There is a constant danger of observers using retrospective coherence (more colloquially hindsight) to assume linear causality that is not present. Hindsight does not necessarily lead to foresight.
- Complex systems constantly adapt to local interactions. In a human system, the influence of family, work colleagues, etc., has a disproportionate effect.
- Change in complex systems can be sudden and catastrophic in nature with very little prior notice.
- Complex systems are non-aggregative in nature and, therefore, non-reductionist. The whole is never the sum of its parts; it may be more and it may be less.
- With co-evolution comes the associated phenomena of irreversibility. In a complex system, we can only move forward from the present; we cannot reset and start again.
An understanding of complex adaptive systems theory gives a scientific base to what we know as common sense. In our day-to-day interactions with friends and children, we manage for the emergence of beneficial coherence. We respond to weak signals and amplify or dampen our response based on a vague idea of the overall objective we want to achieve.
To put it simply, we don’t organize a party for a bunch of children through learning objectives, mission statements, milestone targets and after action reviews. Instead, we create some simple, enforceable boundaries and take a safe-fail experimental approach to directing play. We need to learn to live in a state of requisite ambiguity in our strategic lives at work, as we do without issue or conscious thought at home and with friends.