When the world is complex, how should we think and act?

“...it is not as simple or easy as we would prefer; neither is the world we are trying to manage...”

Warning: more thoughts on complexity. To some extent covering the same ground here (apologies), but more from a philosophy of science perspective, which has been at the same time interesting and thoroughly boring, unless you find the differences between genetically modified potatoes and genetically modified corn to be a significant one.

Complexity books: good covers. Credit: Amazon

Sandra Mitchell, in ‘Unsimple Truths,’ argues for a new approach to science which takes into account the world’s “tractable, understandable, evolved, and dynamic complexity”. Because if the world is complex, then our approach to representing and explaining it should be too. Her expanded, complexity-informed epistemology of science is therefore 1) pluralistic, accepting multiple explanations and models at a variety of levels of analysis, thus allowing for emergent properties when ‘the whole is more than the sum of the parts’; 2) pragmatic, recognising that whenever we represent nature it’s always for some purpose and so can be done in different, equally-valid ways and 3) dynamic, in that our knowledge of the causal structures of the world evolves as that world evolves. The latter of which is a novel thought to me, at least in that form. So the “greedy reductionist strategy” epitomised by the Newtonian physics and universal, exceptionless laws we love to hate, isn’t always wrong, but is when you try to explain everything using it.

Mitchell conveniently pulls out some implications for public policy: she, like Ramalingam and Beinhocker, takes aim at ‘predict-and-act’ models of policymaking which rely, essentially, on predicting the future, or at least assigning vaguely accurate probabilities to different outcomes. The problem is that “uncertainty about the probabilities of outcomes is pervasive, multiplicative, and often non-linear in complex systems” [such as climate change, mental health disorders, the macroeconomy etc.]. In such cases, we need to replace the ‘predict’ with models of “multiple alternative futures” and the ‘act’ part with adaptive management.

The first part of that involves mapping out multiple scenarios of the future, even if the likelihood of one or another is unknown, and comparing policies by how robust they are to the uncertainties in each scenario. This method captures the information we have about the future better than any single estimate. The second part essentially means picking an approach, then monitoring the results of that approach in the short term and modifying it based on the findings. “[A] dynamic, iterative, feedback-rich strategy for decision making that matches the dynamic, feedback-dependent reality of complex systems”.

If not, you end up with the climate change situation as is currently: the inability to come to an agreement on a single quantitative assessment of the probability of various outcomes (2 degrees warming? 10 degrees warming?) undermines any scientific contribution to public policy decisions. The inevitable uncertainty of the issue challenges the credibility of any scientific claim, leaving us to rely on ignorance and intuition: if you’re optimistic you can content yourself with the knowledge that technology will save the day; if you’re pessimistic you can content yourself with the fatalistic observation that we are already screwed.

So more flexibility in our approach to science, more scenarios, more computer modelling, more feedback, more tinkering, more informed actions, more successful policies. Only thus can we manage our “dynamically changing, complicated, complex, and chaotic but understandable universe”.