A guide for planning and strategy development in the face of complexity

By Richard Hummelbrunner and Harry Jones
ODI Background Note, March 2013. Available as pdf

“Many argue that governments, non-governmental organisations and international agencies should spend less time planning in advance and more time adapting programmes to changing scenarios and learning by doing. In the complex context of development, how can policy makers, managers and practitioners best plan in the face of complexity? Does complexity make planning an irrelevant exercise?

This Background Note is a guide, explaining how planning and strategy development can be carried out despite complexity. While it is true that complex situations require a greater focus on learning and adaptation, this does not render plan­ning irrelevant. In fact, there are ways in which the processes and products of planning can respect the realities of the situation and set up interven­tions (policies, programmes and projects) to give them the best chance of success.

The guide builds on academic, policy and programmatic literature related to themes around systems and complexity  and draws on the authors’ experience of advising development agencies and governments in both developed and developing countries.

The note covers three points:

  1. How to recognise a complex situation what challenges it will pose
  2. Principles for planning in the face of complexity
  3. Examples of planning approaches that address complexity”

Rick Davies comment: Over two hundred years ago William Blake exclaimed in verse “Pray God us keep From Single vision & Newton’s sleep”  If he was to read the current literature on complexity, planning and evaluation he might be tempted to repeat his advice, again and again, until it seeped through. Why do I think this? I searched this ODI paper for three magic words: diversity, difference and variation. Their existence in real life is the raw fuel for evolutionary processes, one that has enabled living organisms to survive amidst radically changeable environments over aeons of time on earth. And lo and behold, most of these organisms dont seem to have much in the way of planning units or strategy formulation processes. Evolution is a remarkably effective but non-teleological (i.e. goal driven) process of innovation and adaptation.

While I did not find the words diversity and variation in the ODI text, I was pleased for find one brief reference to the use of evolutionary processes, as follows:

Another option is an ‘evolutionary’ approach, whereby a plan is not seen as single ‘big bet’ but rather as a portfolio of experiments, by setting an over-riding goal and then pursuing a diverse sets of plans simultaneously, each of which has the potential to evolve. We could also adopt a ‘breadth first’ approach with ‘trial and error’ as the central aim of the initial stage of implementation, to encourage parallel testing of a variety of small-scale interventions”

One means of ensuring sufficient diversity in experiments is to decentralise resources and the control over those resources. This can happen in projects which have explicit empowerment objectives and also in other kinds of projects that are large in scale and working in a diversity of environments, where central controls can be loosened, either by accident or intention. In my experience there are already plenty of natural experiments with experimentation underway, the problem is the failure to capitalise on them. One reason being the continued fixation with a single vision, that is, an over-arching Theory of Change, embeded in a LogFrame and/or other planning formats, which end up dominating evaluators’ attention and use of time. This includes my own evaluation practice, mea culpa, notably with four projects in Indonesia between 2005 and 2010.

The alternative is to develop testable models that incorporate mulliple causal pathways. In the past I have emphasised the potential of network models of change, where changes can be affected via multiple influence pathways within complex networks of relationships between different actors. The challenge with this approach is to develop adequate descriptions of those networks and the pathways within them. More recently I have been argueing for the use of a simpler representational device, known as Decision Tree models, which can be constructed, and triangulated, using a variety of means (QCA, data mining algorithms, participatory and ethnographic techniques). The characteristics of a portfolio of diverse activities can be summarised in the form of Decision Tree models, which can then be tested for their  degree of fit with observed differences in outcomes of those activities. The structure of Decision Tree models enables them to represent multiple configurations of different causal conditions, identified before and/or after their implementation. More information on their design and use is provided in this paper “Where there is no single Theory of Change: The uses of Decision Tree models” While I have shared this paper with various writers on evaluation and complexity, none seem to have seen its relevance to complexity issues, possibly because in many writings on complexity, the whole issue of diversity gets much less attention than the issue of unpredictablity. I say this with some hesitation, since Ben Ramalingam’s forthcoming book on complexity does have a whole section on the perils of “Best-practicitis” i.e single vision views of development.

Incidentally, for an interesting but demanding read on the many relationships between diversity and complexity I recommend Scott Page’s “Diversity and Complexity” (2011)

 

%d bloggers like this: