# Notes on complexity, drawing on Samir Rihani

These notes draw on Samir Rihani's *Complex Systems Theory and Development Practice: Understanding Non-linear Realities* (2002), adapted for organisations. They are working notes, not verbatim quotations, and all credit for the ideas is his.

## Check what kind of system you are dealing with

Some parts of an organisation are perfectly predictable and linear, and detailed action can be taken with confidence there. Other parts are complex, where people adapt to each other and to what you do. The trick is to work out which kind of system you are dealing with before you decide what to do. Command and control, detailed forecasts and rigid plans suit linear systems. In a complex one you cannot select sensible actions to reach a desired result, because the outcome comes from vast numbers of interactions and cannot be traced back to specific causes.

## How a complex system behaves

Complex adaptive systems have active internal elements that give them enough local variety to survive as they adapt to circumstances nobody foresaw. The elements interact locally, by simple rules, and that interaction gives the system the energy to hold a stable pattern, in place of rigid order or chaos. Conditions change, and the result is many minor changes and a few large ones. It is not possible to predict in advance which will be which.

Evolution does not lead to an optimal end state. "Good enough" is the sum total of what nature hopes to achieve. Progress is an open-ended run of small, effective improvements. The quick skip to paradise promised by ideologues, politicians and consultants is not on nature's agenda.

Staying in the zone between order and chaos takes constant effort, and it is easy to get wrong in either direction. Too little effort lets the system slide into chaos. Too much kills it, and unvarying order spreads everywhere. Living systems have spent millions of years learning to stay in that narrow layer.

## Calm can end suddenly

Evolving systems stay much as they are for long periods, then change fast and radically. During the calm, change is still going on locally, the system moving through many states within one pattern. Ripples come and go without leaving a trace, and then a relatively minor event shunts the whole system into a new pattern. A system that looks stable can transform for no clear reason, and there is no knowing in advance.

History shapes what is possible. Frozen accidents, chance events from the past, exert a decisive influence on what happens next. Against them sits relentless daily adaptation to changing circumstances. The contest between the two is typical of complex adaptive systems.

Gateway events go through a system like a whirlwind. Covid, the industrial revolution and the internet are examples. They open up new opportunities for some and bring disaster for others. A system needs enough variety and flexibility to survive the shock before it can profit from what the event opens up.

## Why prediction fails

Complex adaptive systems show constant adaptation, accumulating irregularities and a chronic state of unpredictable flux. Forecasting is taken as an article of faith nearly everywhere, and nowhere more than in business. Nonlinear systems are not necessarily unpredictable in theory, but in practice you would need infinite precision and computing power to trace causes and effects through millions of local interactions. To know what such a system is doing, and why, it is often necessary to watch while it does it.

Methods built on a linear paradigm give only an illusion of knowledge and control. Emergent properties are a bottomless source of surprise. The only real defence is to make sure the system is robust and pragmatic enough to survive the knocks.

The expert's need to predict the unpredictable does harm. People demand rational answers and quick fixes, and when those words are turned into action the results can be damaging. The future is uncertain whether or not anyone says so.

## What to do instead

Command and control methods may succeed for a while when applied with enough force, but they are not sustainable. Complex adaptive systems respond better to a light touch. That means constantly watching the overall pattern of performance and making judicious, small, incremental adjustments, enabling interactions to go well and not trying to dictate results. Management of a complex system is a repeated, slow and uncertain process.

Progress is made through modest local improvements that accumulate over long periods. Rigid plans and policies are inappropriate. An organisation's best strategy is to stay flexible and pragmatic, and to survive, learn and adapt over and over as its circumstances change. Nothing happens without interactions. People have to be free and able to interact, or to put it in more traditional language, to participate, for anything to happen.

The best policy is for an organisation to optimise its performance within its own opportunities and constraints, and not to be diverted by how others are doing.

## Why recipes fail

Universal prescriptions to be adopted by one and all discourage variety and innovation, and variety is essential to survival. Sameness can do positive harm. It flattens the fitness landscape, the picture of where each element does well or badly in a given set of conditions, so there are no high spots for some elements to occupy. If the environment alters radically, the average elements, perfectly at home in the flatlands, may not survive the shock.

Variety has to be local to mean anything. It is pointless to compete on someone else's fitness landscape. There is no one common landscape on which every organisation is best equipped to occupy the summits, whether the model is top-down, agile, lean or self-management. Local variety is one of the few defences available, and it is not possible to introduce variety on demand.

It is improbable that a single process of a few specific steps could meet the diverse needs of every organisation. Who leads the change, how, and for what purpose may matter as much as time and money. Leaders' eagerness to engineer an organisation to their satisfaction is understandable, but it cannot be done. The linear temptation is hard to resist because it offers the illusion of control.

Transplanting an intricate system that evolved to suit another organisation causes difficulty. As with every complex adaptive system, an ultimate model does not exist. Change is continuous and takes a long time to unfold. Learning from others is useful, but imitation is no substitute for exploring your own constraints and possibilities.

Too much connectivity can bring a complexity catastrophe. As interdependence grows, systems impose conflicting requirements on one another, resolved only by frequent compromise. They shed variety and fit just one set of conditions, and progress grinds to a halt through repeated local failures.
