Complex Systems

The word "complex" appears in the SiD sustainability definition for a very specific reason: to underline that the systems we are talking about are non-linear, infinitely intricate entities, almost like biological organisms, and absolutely not predictable, finite, mechanistic machines. This distinction matters enormously, and getting it wrong is responsible for most sustainability policy failures.

We differentiate between "complex" and "non-complex" systems. These terms are roughly analogous with "non-linear" and "linear," or "complex" and "complicated."

Non-complex systems have components and relations that can be fully indexed and understood. They are finite in composition and can, in some cases, be fully modeled using physics, mathematics, or other scientific tools. They are called non-complex rather than "simple" because they can still be very complicated. The electrical system of a house is non-complex. You can model it and predict its behavior. But it can be hard to figure out. Working with non-complex systems has been the dominant mode of systems analysis and innovation for over a century.

Complex systems are an entirely different ballgame.

A complex system consists of a number of objects and relations so numerous that we cannot keep track of all of them. Complex systems exhibit behavior that cannot be predicted using normal mechanical analysis or computer simulation.

The weather is the classic example. Nobody can accurately predict the weather more than a week into the future. There are too many factors, and the events that generate future patterns may not yet have happened. As we try to predict further ahead, the complexities multiply so fast that any attempt grinds to a halt.

Some complex systems may appear non-complex, or have been treated as such. This is dangerous. Ever had a bad weather forecast ruin your plans? The same thing happens to economies, when well-intended policy measures do not account for complexity. One can blame poor performance on "unexpected" events, but in complex systems, only the expected is unexpected. An economic policy that does not account for complex dynamics is simply not resilient, which means it was bad policy.

Thinking you can predict a complex system is a thinking error. Just as our brains are more than a sum of molecules, complex systems are more than the sum of their parts. They behave less like machines and more like creatures displaying "emergent" behavior. Acting on one aspect in isolation will always produce side effects. Expecting complex systems to respond like machines is a one-way road to disaster.

In the 1970s, scientists in Arizona attempted to build a comprehensive computer model of an ecological system. They collected increasingly detailed data, expecting the model to converge with reality. The more data they put in, the more the model diverged from reality. Complex systems cannot be analyzed from a bottom-up reductionist perspective in order to be predictive. The attempt itself reveals the limits of the approach.

Even though complex systems cannot be predicted, they can be studied, learned from, and their behaviors analyzed and intervened upon. We can make them more resilient. That is the work ahead.

12 Rules of Complex Systems

To help understand what a complex system is and is not, here are 12 general rules. These are practical principles, not abstract theory. Each has direct implications for sustainability practice.

1. Numerous in components. All components influence each other. They exhibit non-linear behavior emergent from interactions, beyond each component's mechanical behavior.

1. Can be understood but not predicted. Any action upon them may have unpredictable side effects. Do not make decisions based on prediction. Instead, prepare for resilience, adaptability, flexibility.

1. Grow like organisms, perish like organisms. No complex system exists for eternity. Understand and accept the natural cycle. Aim for self-reproductivity and longevity rather than eternality.

1. Require increasing resources per added unit of complexity. There are always limits to growth. Systems respond differently at different scales, but may exhibit similar patterns across them.

1. Change in revolution-like jumps as well as slow evolutionary progression. Both together. Triggers can be anything. Details matter as much as large-scale variables.

1. Do not necessarily behave the same way given the same conditions. Historical behavior is not always an indication of future behavior.

1. Always dynamic, never sit still, never entirely in balance. Even if they seem to be.

1. Not aware or alive per se, but may exhibit survival or seemingly cognitive behavior. It helps to mentally construct a complex system as a biological entity with a character.

1. Require incubation periods for changes to be registered. Be patient. Measure in the full spectrum lest you miss a rebound effect or changed state.

1. Best understood by human brains, which are themselves organic complex systems. Immersing yourself in a complex system and fully interacting with it is the best way to learn its behavior. Get out from behind your desk and connect.

1. Interact beyond their chosen system boundary. Account for this at all times. Maximize beneficial externalizations and minimize dependency on them.

1. Always offer hidden dynamic processes with both beneficial and destructive effects. Finding these patterns boosts capacity for change and prevents harmful externalities.