Coping cool with complexity

Things will only repeat by accident – not by design.

Hindsight does not mean foresight.

Problems and solutions are co-evolving as you go along.

Constraints get difficult to handle.

You are never quite sure how people and things will behave.

Things are dispositional – not causal.

This is complexity. But what can I do (you might ask)? Is it possible to intervene in such a domain?

The answer is yes, but let’s try to get a little more specific than riding the flow,managing the patterns or absorbing complexity. In the process we may also become aware that complexity requires more management discipline rather than less.

So if you want to STOP treating the complex as though it is ordered – and avoid perpetual disappointment – then keep on reading.

Key rules for dealing with high levels of complexity

You don’t try to interpret it and you don’t try to analyze it, because you don’t have repeatability. Instead you understand complexity by interacting with it.

More precisely this means that you run multiple parallel safe-to-fail experiments very quickly, rather than doing one thing many times. Because with high levels of uncertainty you can’t afford the time to do one thing many times until you get it right.

Actually you should never run less than three safe-to-fail experiments, and ideally five to seven. And remember: in parallel – not one after the other. The cycle time should be fast: about two to three months.

But let’s take a closer look at the key rules of engagement when dealing with high levels of complexity:

1. Your experiments should be coherent. You don’t do something just because someone thinks it’s a good idea. There has to be some rationale to it.
   1. One way of testing for coherence is by employing techniques such as “Ritual Descent” or “Red Teams” where you use opposing teams trying to criticize or destroy the presented ideas
2. Your experiments must be safe-to-fail. Which basically means that if it doesn’t work you can recover. Also it means that if you cannot answer the four next questions, it is not safe-to-fail:
   1. What are the early signs of failure?
   2. What are the early signs of success?
   3. What are your actual dampening strategies when the experiment fails?
   4. What are your actual amplification strategies if the experiment succeeds?
3. Your experiments need to be finely grained, tangible and understandable, so that they are possible to measure – not abstract

Overall your portfolio of experiments should contain:

• Some experiments that are oblique in nature – trying to solve another problem related to your primary problem often is more effective when dealing with complexity
• Naïve approaches – using somebody with a completely different knowledge base and perspective, e.g. putting a biology professor in an engineering department and see what happens
• A few high risk/high return options in order to broaden the portfolio
• Contradictions – they are good in a complex field, because they provide great success/failure feedback

As you can see managing complexity is a disciplined – and never ending – process. In fact you are managing the evolutionary potential of the present. Some say that this takes 10.000 hours (or 5 years) to master, so you might as well get started.

(The article is inspired by Professor Dave Snowden).