How to Plan when you don’t know what is going to happen? Redesigning aid for complex systems

May 14, 2013

     By Duncan Green     

They’re funny things, speaker tours. On the face of it, you go from venue to venue, churning out the same presentation – more wonk-n-roll than rock-n-roll. But you are also testing your arguments, adding slides where there are holes, deleting ones that don’t work. Before long the talk has morphed into something very different.

So where did I end up after my most recent attempt to promote FP2P in the US and Canada? The basic talk is still ‘What’s Hot and What’s Not in Development’ – the title I’ve used in UK, India, South Africa etc. But the content has evolved. In particular, the question of complex systems provoked by far the most discussion.

Complex system US Afghan mindmapI started off with the infamous US military mindmap of Afghanistan. Although ridiculed at the time, the map looks like a genuine and nuanced effort to understand the country and is probably fairly typical of the complexity of power and relationships in any given country. The point is that such a system is complex, not complicated. Complicated means if you study it hard, you can predict what happens when you intervene. In contrast a complex system has so many feedback loops and uncertainties that you can never know how it will react to a stimulus (say $100m in aid, or an invasion….).

The crucial point is that most political, social and economic systems look like the map. Yet the aid business insists on pursuing a linear model of change, either explicitly, or implicitly because a ‘good’ funding application has a clear set of activities, outputs, outcomes and a MEL system that can attribute any change to the project’s activities – a highly linear approach. Other organizations – say forest fire managers, or the military, seem more able to cope with complexity, although I found out from a woman in one seminar who had served in Afghanistan that the power map was actually drawn up by a consultant, who was promptly sacked after showing the slide to General Petraeus, so maybe the soldiers aren’t so comfortable with complexity after all.

In denying complexity is obliged either to seek islands of linearity in a complex system (vaccines, bed nets), which may not always be the most useful or effective places to engage, or to lie – writing up project reports to turn the experience of ‘making it up as you go along’ that epitomises working in complex systems into the magical world of linear project implementation, ‘roll out’, ‘best practice’ and all the rest. That not only wastes a lot of staff time and energy, it also reduces the ability to learn about how to work best in complex systems.

So how should the aid system change? Overall, we need to think though ‘How to plan when you don’t know what is going to happen’ (my best effort at explaining complexity without resorting to jargon). Here are my bullet points, and brief explanations:

Fast feedback: if you don’t know what is going to happen, you have to detect changes in real time, but also have the institutions to respond to thatcomplexity road sign 2information (as was not the case recently in the Sahel).

Focus on problems, not solutions: Drawing on Matt Andrews’ work, the role of outsiders is to identify and amplify problems, but leave the search for solutions to local institutions. (At the World Bank, Shanta Devarajan pointed out the contradiction between this approach and NGOs’ preference for big, simple solutions – end land grabs, no to user fees. Ouch.)

Rules of thumb, not best practice toolkits: I am told that the US marines do not go into combat brandishing Oxfam toolkits and online resources on best practice. They operate on rules of thumb – take the high ground, stay in communications and keep moving. They improvise the rest. Aid workers on the ground operate far more like this than our project reports admit. If we were honest about it, we could have a better discussion on how to improve those rules of thumb.

Some possible approaches that spring to mind (and I would love to hear examples of others)

Work on the ‘enabling environment’ rather than specific projects: things like norms, rights or access to information

Evolutionary/Venture Capitalist approach: run multiple experiments and then zero in on what seems to be working best. Example, the Chukua Hatua project in Tanzania

Convening and Brokering: Get dissimilar local players together to find solutions – the outsiders’ job is to support that search, not do it themselves. Example, the TAJWSS water project in Tajikistan

But any attempt to move in this direction raises some fundamental challenges to the current structures of the aid industry:

Results for grown ups: The current approach to measuring results favours linearity. But rejecting results altogether is the wrong approach – both evidencebecause even those who recognize the central role of complex systems still want to know if they’re doing any good, and because the results people control the cash. No results, no funding. We need to get much better at ‘counting what counts’, and reclaim the idea of ‘rigour’ for qualitative and other methods better suited to complex systems.

Who to employ? Risk-taking, entrepreneurial, maverick searcher types have a hard time in an aid business dominated by bureaucratic procedures and risk aversion. Moreover, working in complex systems requires deep local knowledge of formal and informal power maps, something expats on a two or three year posting are unlikely to acquire. How do we turn the tables to attract and retain searchers, and value locally embedded knowledge?

Short Term v Long Term: Funding and project cycles are short term, change in complex systems is often long term. How can we bridge the gap, for example by combining good, plausible stories about the short term, with more rigorous impact assessment in the long term (how often do we go back and study the effects of an intervention 10 or 20 years after the funding has ended?)

How to keep/build political support given that working in complex systems means acknowledging a lack of control over what takes place and limits to attribution (no you can’t ‘badge’ the Arab Spring as created by Oxfam, USAID or anyone else, sorry). It also means greater tolerance of failure – a venture capitalist approach means accepting 9 failed start-ups for every 1 big success, but imagine what aid critics would do with a 90% failure rate. And how do we communicate and sell this approach to the public after systematically dumbing down the aid and development story for decades? (From buy a goat and save the world, to a post-goat narrative….)

Ben Ramalingam has been thinking about this for years, and writing about it on his Aid on the Edge of Chaos blog. His book of the same name is due out later this year, so let’s hope it can settle a lot of these issues (and doubtless raise many more).

May 14, 2013
Duncan Green