More on complex systems
Friday, June 27th, 2008Brian Hayes just forwarded me a great article from Nova (published by the Australian Academy of Science) about the science of complex systems. It’s written in plain English (always a delightful surprise) and covers some critical properties:
Emergence
Emergence is the formation of complex but regular patterns from the interaction of the many simple parts of a system. The emergent collective behaviour of a system cannot be predicted merely by understanding its individual elements, or from understanding the interactions between these elements, but it can in principle be predicted by seeing how all the elements work together. It is this element of regularity in the emergent behaviour that distinguishes complex systems from complicated and chaotic systems.
You may recall this concept from a previous post about the difference between complex and complicated. A complex system is one in which the very interaction amongst its constituent parts creates characteristics (the ‘emergent properties’ referred to above) that wouldn’t otherwise exist.
This concept is key to anyone looking to take long-term, strategic views involving human behavior, which means anyone driving the vision of a company, an organization or a government. It’s why macroeconomics and microeconomics are distinct disciplines, because the way we behave in our one-on-one purchase decisions changes radically once we factor in societal interactions.
Self-organization
Self organisation is closely related to emergence and refers to the ability of the system to organise itself. The emergent features of the system appear spontaneously. There is no one in control of the system.
…Connect up a few computers, as happened when the internet was born, and you have a simple system that’s predictable and controllable. Connect up several million computers around the planet and you have a complex system behaving in ways no one could have imagined – and nobody is in control.
Self-organization is why social media panels don’t like questions about ‘viral’: nobody wants to admit that they can’t control how the system will react to a given bit of content. Self-organization is why it is more effective to get your customers talking to each other than to keep them separate. Self-organization is why something like Wikipedia can exist at all, let alone be highly effective.
Local interaction
The parts of a system tend to only interact with a small subset of the whole system, known as a local neighbourhood. For example, during our everyday lives we interact with friends, neighbours and work colleagues, as well as people on the street, but never everyone in the world. If a flu outbreak reaches a country via an infected tourist arriving at the airport, it’s unlikely that you will be infected by direct contact with that person. If the disease spreads, and you become infected, it will probably be via people you are regularly in contact with.
What really bakes my noodle about local interaction is that it is infinite and recursive. We tend to think of six degrees of separation as a flat model, but if I connect to the person on my right and on my left, and each of them do the same, where does our network begin and where does it end?













