On the ACT-KM list serve someone made a comment to the effect that we should not give up trying to predict a complex system. I just posted a reply which I repeat here. It needs more work but I offer it for comment:
Words can mean many things in the context. However I think its worth a quick refresher on this
1 – complex systems can not be predicted, they are non-causal (taking cause in its normal Newtonian sense) in nature they evolve and the same thing will not happen again twice, we can predict aspects of the system and different aspects of time but never the outcome of the whole system
2 – the concept of a non-causal system is a very difficult one to grasp as the west abandoned the idea at the time of the Enlightenment (Vico and others were prophetic in arguing against this). KM people are more likely to get this than most as they deal with such systems all the time. However anyone who mentions the word “best practice” without strong qualification has just fallen from grace
3 – a complex system can be simulated – which increases understanding but simulation should not (although it is often) confused with prediction
4 – We can understand starting conditions as a complex system evolves and we e can influence their evolution if we focus on barriers and attractors (1st and 2nd order constraints) but not if we look at the end point (so attempting to predict makes things worse not better)
5 – Humans are not ants, or crystal formations. The study of human complex systems is a new trans-disciplinary field and we can learn from but should not be constrained by lessons from biology, chemistry, climate change etc.
6 – Humans tend to premature convergence (seeing a pattern too quickly before it is stable) and also to retrospective coherence (implying past causality where there was none). Both of these tendencies are pervasive and dangerous
7 – Scenario Planning as a method is a sure fire way of falling into all the traps listed above for managers and post-modernism is a refuge for the confused academic who does not get point 1 above.