"Every society honors its live conformists and its dead troublemakers."Mignon McLaughlin
"An important scientific innovation rarely makes its way by gradually winning over and converting its opponents... What does happen is that its opponents gradually die out and that the growing generation is familiarized with the idea from the beginning."Max Planck, The Philosophy of Physics, 1936
In the world of yesterday the clockwork toys ruled. Our metaphors were mechanistic, our understanding was fixed in the static world of the repetitive illusion. The planets revolved around the sun, people lived and died, rain fell and evaporated, all progressed in a deterministic, predictable, controllable pattern. And then the sky fell in.
A mushroom cloud of change swept the Earth, mountains collapsed, rivers dried up, the cosy world of horse and sail exploded into a multi-fragmented cascade of invention. Industry had arrived, and along with it a new metaphor based on evolution, on change, on progression. The old was discarded, the new embraced, yet here too the pattern remained deterministic, controllable, linear. Effect followed cause, profit followed investment, achievement followed plan. But a new era dawns.
This is a vision of tomorrow, a world out of control, yet promising a level of understanding and achievement far beyond anything yet seen. This is the era of the Complex Adaptive System, a 21st Century metaphor relating innovation, cybernetics and complexity, within a new organic cosmos filled with living machines and emergent organization.
Late in the 20th Century we began to realise that linear prediction, far from being the ubiquitous panacea of success imagined in the past, was only a simplified viewpoint applicable to relatively few systems. Most of the world about us is nonlinear, and this has vast implications on our views of control. In nonlinear systems we find that knowing the exact equation for their behaviour is of little help in predicting the outcome. Sensitivity to initial conditions, in systems containing chaotic interactions, precludes accuracy in long term predictions. The implications for plans are obvious, we cannot set out plans that will come true as expected, except in very specific cases. Real world systems, especially those involving people, are generally too nonlinear to qualify. Management by imposed control, in other words top-down rule in the way practiced for two and a half millennia, is ineffective or at best inefficient.
To overcome this problem we can only do one thing, and that is to move the control from the past to the present. Instead of planning an action in advance we specify constraints and then allow the local conditions at the time to determine how the task will be done. This is the essence of distributed systems, each task responds to the local environment in real time, the interactions between the system and environment allow an emergent solution to arise. Power is localised, not concentrated and this allows fast responses to unforeseeable events, a flexibility that removes the rework costs inherent in more static plans. This is a open, parallel mode of operation, where multiple options can be tried simultaneously, compared to the closed, serial mode of conventional management where decisions move linearly up and down fixed command chains.
But given such open systems how can we ensure that the required result occurs ? Here we need to look at how this is arranged in machines, and this is called cybernetics, the science of control by feedback. In this field the traditional ideas of cause and effect are revised and the effect feeds back upon the cause, completing the loop. Two results are possible, called negative and positive feedback. In the first an output change opposes the original cause, returning the system to the desired state, this negative feedback leads to a self-maintaining stability (which can be seen in crime and then punishment). Positive feedback, on the other hand, is a reinforcement process, aiding growth and development (seen in the linked growth of DVDs and DVD Players - the more of one, the more likely sales of the other will occur). These interactions are common in our world and make most systems a network of many interacting feedback loops, both opposing and supporting actions.
For a system to change we need something new to happen, something to alter the status quo. Such novelty may come from random events, from unexpected repercussions of actions or from deliberate changes of rules. Each change perturbs the system in some way. If the perturbation leads to a new state, one not seen before, we have an innovation. This new system can then be compared to the old and if it is an improvement then we say the system has evolved. To do so as a result of an external change (an environmental disturbance) is what we call adaptation, and a system that can do this we call adaptive. All organisms are adaptive in this way, and so are organisations, societies and cells. Unlike the mechanical view, where systems have fixed functions that either work or do not, adaptive systems have flexible functions that adjust to the context of their environment.
To show flexibility we must have a number of possible modes of operation, and to generate such diversity we must have many parts, in other words we must be complex. Diversity comes from the combinatorial explosion possible when parts can interact in different ways. But parts alone are not enough, there must be freedom of interaction (in contrast to conventional machines or companies where lack of such freedom is the objective). Yet this freedom cannot be total, or else we will have chaos, a system unable to persist long enough to complete any function. The balance between freedom and constraint is a dynamic one, those systems able to maintain function yet change quickly will prove the most effective. These can be shown to exist at the edge of chaos, a position where structure exists at multiple levels in space and time, corresponding to our definition of complexity.
Complex Adaptive Systems (CAS) are the merger of the previous three concepts. They have many autonomous parts, they are able to respond to external changes and form self-maintaining systems with internal feedback paths. The essence of CAS is that they self-organize, to optimize function. An over-constrained system will benefit from more freedom, so a choice (random or otherwise) that allows that will prove more successful and be retained, conversely an over-free system will benefit from changes that add stability. Such systems are well placed to explore new niches, to search their fitness landscape, changing their composition to fit the changing patterns they encounter. This adaptation internalises environmental information, the system generates a model of the world outside, a distributed set of rules corresponding to the interesting or valuable aspects of their context. Machines based on these concepts will act like living cells, companies will transform from static cold-blooded dinosaurs into dynamic warm-blooded organisms, societies will evolve into diverse ecosystems, infused with freedom and creation.
We must notice however that we have several simultaneous levels here, we have parts within parts (e.g. cells, organisms, ecosystems). Each of these levels consists of a number of CAS, but we must also take into account the interactions between them. These come in three flavours, involving the agents at any one level:
These are interactions within the system, for example within a company the different people collectively arrange their functions. This is self-organization, the move to an internal attractor.
Here the agents interact with each other across the system boundaries, for example a supplier interacting with a customer. This is coevolution, changing fitness landscapes.
Agents also interact within vertical hierarchies, e.g. we have bosses and staff, different control levels. This is emergence, new levels of description, global functions.
We thus see how these three dimensions of interaction can form a cube, where each node links to three others, representing the three main aspects of complexity thinking.
But life is not quite so simple ! Each level itself has these three modes, so we have an overall fractal structure of levels within levels of interactions. We can see this in our Environment, our Society, our Business world, and the psychology of Self. A complete description of the CAS world would show that all these systems inter-relate - the objective, isolated systems of the past were a myth all along. Every change, no matter how small, has the potential to change the entire multi-level whole. A perturbation follows all the available paths of influence to affect internal, horizontal and vertical levels simultaneously, a new thought affects us, the people we talk to and ultimately the whole of society.
Most children are fascinated by magic, the unexpected, the inexplicable. Their delight at being surprised is plain to see. In the recognition of Complex Adaptive Systems we can share that delight, each of us is empowered to change the world in ways not previously considered possible. For in a CAS all agents are equal, it is true democracy, what emerges can be as much the result of changes to our own personal CAS as the results of action by the biggest company or state. Ideas can change the world, so lower your own barriers to allow new thoughts and watch the webs expand magically to show your new possibilities, the new opportunities at every turn, the power of people thinking !