Why Complex Adaptive Systems Can't Be Managed

Our usual way of thinking about management is controlling predictable variables for repeated planable outcomes. Management works with simple and complicated problems where we engineer solutions from proven formulas. One of the hallmark characteristics of simple and complicated contexts is that all of the variables are learning disabled. No variable has capacity to learn. Everything can only do what it does unless some outside agent or force alters function by altering form. Building a subway car is a simple context. Building a subway system underneath a large urban city is a complicated context.

A complex context is one where the system involves variables that have learning capability. It is a complex effort to get people who never used subways to use them. Learning makes behavior unpredictable and therefore the consequences of actions are not predictable. In complex contexts change cannot be managed, it can only be enabled. Our capacity to learn becomes more important and relevant than our capacity to conform to formulas. 

In complex contexts, when we experiment with something on small scales, we can work to spread any success gleaned rather than try to scale. Complex contexts literally have minds of their own because of their learning capabilities. We cannot engineer compliance in second application contexts any more than we can engineer compliance in second children based on the successes of the first.