Melissa Leach: Dynamic Sustainabilities: Taking complexity and uncertainty seriously in environment and development

Editor's Notes

• #5: In our work we have drawn on and drawn together – and go beyond - a related set of ideas and approaches that draw in different ways on complexity science, and offer specific helpful concepts and modes of analysis for different fields of development. The last is particularly significant, and we would argue vital in a development context, drawing attention to the importance of both reflective traditions in analysis, and constructivist/reflexive ones drawn from other social science traditions.
• #6: The idea of pathways is useful in thinking about development, change, directions
• #7: But central to the pathways approach is the addition of a reflexive dimension, drawing on methodological constructivism in the social sciences. That is to recognise that there are multiple ways of understanding and representing a system; and that all analysis of a system involves framing. Framing involves not just choices about which elements to highlight, and how to bound the system, at what scale, but also subjective and value judgements. Such framings are produced by particular actors – whether different local people, scientific, policy or business actors, and co-constituted with their particular institutional, political and life settings.
• #8: Some applications of complexity science tools seek to apply these in a deterministic way; if we just understand the range of tipping points, we can plan for them But of course a fundamental implication of complexity and dynamics is that there are limits to knowability, and to control/planning. Taking complexity seriously means dealing with incomplete knowledge in situations where uncertainty and surprise are inevitable, and tailoring actions and strategies to situations where dynamics of change and their drivers are not always tractable to control. Will now go on to look at each of these and their implications in turn.
• #9: But what do we mean by incomplete knowledge? The top left hand quadrant defines risk in the strict sense of the term. This refers to a situation where there is confidence that probabilities can be calculated across a range of known outcomes. Under the strict definition of uncertainty (lower left quadrant), we can be confident in our characterisation of the different possible outcomes, but the available empirical information or analytical models do not present a definitive basis for assigning probabilities the rigorous approach is therefore to acknowledge the open nature of a variety of possible interpretations.   Under the condition of ambiguity (upper right quadrant), Disagreements may exist, for instance, over the selection, partitioning, bounding, measurement, prioritisation or interpretation of outcomes. For instance in decisions over the right questions to pose in regulation: ‘is this safe?’, ‘sustainable?’, ‘sustainable enough?’, ‘acceptable?’ or ‘the most sustainable option?’. Similar ambiguities emerge when we are forced to compare ‘apples and oranges’. These might be qualitatively different forms of damage; impacts on different people (e.g. workers or the public; children or adults); or consequences over different time-frames (e.g. present or future generations) – in effect, questions over ‘contradictory certainties’ (Thompson and Warburton, 1985.   Finally, there is the condition of ignorance (lower right quadrant) where ‘we don’t know what we don’t know’ (Wynne, 1992; 2002), we face the ever-present prospect of ‘surprise’ . where the parameters are not just contestable, but are acknowledged to be at least partly unknown, implying straightforward ignorance over the possibilities themselves.
• #10:   Let us pick up the epidemics example and look at how different dimensions of incomplete knowledge are invoked in different narratives about avian flu – in the diagram, we could envisage different narratives and sub-narratives ‘clustering’ in different corners Much of the debate has been dominated by quantitative probabilistic models of risk. In 2005, for example, two models were presented in Nature (Ferguson et al, 20050 and Science (Longini et al, 2005) which together had a huge influence in framing the response as one that needed to be focused on containment at the source of the outbreak. But of course a wide range of uncertainties exist – from the big uncertainty (will a pandemic happen at all, and if so when?) to more specific uncertainties (about the impacts of veterinary control measures, about vaccination and drug efficacy, about behaviour change in situations of crisis and so on). Thus, for example, the interplay between viral ecology and genetics, transmission mechanisms (e.g. the role of wild birds or poultry, back-yard chickens or large factory units) and impacts (e.g. the consequences in immuno-compromised individuals and populations) are highly complex and contingent. There are also ambiguities: how do we define an ‘outbreak’, and its impacts and distributional effects? Outcomes defined in terms of potential impacts on human mortality globally - up to 150 million deaths in a major global pandemic, according to some estimates – contrast with those defined in relation to particular groups at risk – for example, women or children handling poultry. Surprise renders it intrinsically difficult to substantiate possible examples of ignorance ex ante . Yet possible surprises may plausibly be anticipated around the emergence of radically new strains of the virus, unexpected transmission mechanisms or unanticipated health outcomes, including those arising in complex interactions with other health/social conditions. Of course, there is always the broader possibility of the emergence of entirely novel pathogens; indeed over 70% of new infectious diseases affecting humans that have emerged over the last 30 years have emerged unexpectedly from non-human animal populations (Jones et al, 2008; Woolhouse and Gaunt, 2007).
• #11: Frequently, and around many issues, one can observe a process of ‘closing down’ towards this top-left-hand corner. The implications of system dynamics come to be treated in terms of risk, neglecting or underplaying these other dimensions of incomplete knowledge. Put another way, it is often risk-based narratives and the pathways they justify that come to dominate, over and above narratives that appreciate uncertainty, ambiguity, ignorance and their implications. Thus in the case of avian flu, we often see narratives which cast the problem in terms of risk predominating; these are the ones which are articulated by the major international agencies, and which drive outbreak-focused policy responses. Other dimensions of incomplete knowledge – despite their relevance to the problem – receive far less attention   How does this happen? A number of political, procedural, and knowledge-related pressures appear to be at play., varying according to the issue and context, but with some general tendencies . Thus the institutional remits of organisations may encourage a move from ignorance to uncertainty: ruling out surprise. the institutional remits and organisational mandates of international agencies such as the WHO and FAO are simply not geared up to dealing with ignorance and surprise; the very existence and status of the agencies is interlocked with the idea that outbreaks and their effects can be known about and thus rendered amenable to management The use of particular indicators and definitions, and the use of particular legal frameworks, may assist this. Further techniques and strategies of governance in turn may lead uncertainty to be re-defined and treated as risk. These range from bureaucratic and planning procedures that rely on, and thus reinforce, an image of a calculable, manageable world, to particular techniques of modelling, reasoning and categorisation that render the world legible and calculable in risk-based terms. Moving up the right hand side of the diagram, further techniques and strategies of governance may be used to ‘tame’ ignorance and surprise into a more manageable range of possibilities, even if these end up as ‘apples and pears’, non-congruent possibilities. Practices such as drawing on (transdisciplinary) expertise, and setting agendas, and defining organisational mandates may be significant here. However, further pressures may act to narrow a range of possible outcomes (ambiguity) further, creating a set which can be clearly defined and dealt with as risk. Political closure, strategies of ordering (and exclusion), and processes of subjectification whereby people – whether supposed policy beneficiaries, or workers within an organistaion or agency, or others) come to internalise this possible range of outcomes as the appropriate set for consideration, are all relevant here. In the avian flu case, political closure was encouraged by political-economic interests in garnering support for a massive global response, and by aggregative forms of analysi. Claims about particular sorts of vulnerability, perhaps associated with the livelihoods or social positions of particular local groups, tend to disappear as mere ‘twitter’ amidst the dramatic figures about aggregate risk that garnered public, political and media attention. Thus, through a cluster of interlocking political, institutional and knowledge-power processes, the problem of avian flu tends to be treated in terms of risk, at the expense of uncertainty, ambiguity and ignorance.  
• #13: Moving from knowledge to action, here we are talking about directing and shaping pathways to sustainability Butt recognising that sustainability is not just one thing ... Different goals Different dynamic properties
• #14: Actions aiming to promote sustainability involve assumptions about the nature, or ‘temporalities’, of changes – are these seen as short-term shocks or long-term stresses? In complex dynamic systems there may be multiple, interacting shocks and stresses in different system elements operating simultaneously. And the styles of actions that are envisaged. Is the aim to control the causes or drivers of change, or to respond to them? In complex, dynamic systems, it is likely that many changes and their drivers will be intractable or uncontrollable, rendering more conventional control-oriented management inappropriate. Instead, need responsive, adaptive management, aimed either at resilience (to shocks) or robustness (to shifts). Thus we might ask, within any given narrative: are intervention strategies aimed at exercising control in order to resist shocks (stability )? Or is there an acknowledgement that there may be limits to control, and thus that interventions should resist shocks in a more responsive fashion ( resilience )? In other circumstances, the system may be subject to important stresses, driving long run-shifts. In this case, interventions might attempt to control the potential changes – aiming at durabilit y. Alternatively, embracing both the limits to control and an openness to enduring shifts would suggest strategies aimed at robustness. These are important practical distinctions that are often elided or ignored in existing analysis for policy-making on sustainability.