Book Review: The Road to Somewhere, by David Goodhart

September 14, 2017

Can the UK become a Human Economy?

September 14, 2017

Complexity v Simplicity: the challenge for Campaigners and Reformers

September 14, 2017
empty image
empty image

Had a few thought-provoking conversations on this last week. I increasingly see most problems (social, political,

How do we talk about this?

How do we talk about this?

economic) as complex, i.e. arising from multiple causes in interconnected systems, often highly dependent on the specific context and history of any given place/population. My campaigner friends generally hate such talk, because their gut feeling is that it makes taking action to change the world much more difficult. We often end up arguing about Make Poverty History, which from a campaigners’ point of view can be seen as a great success in terms of public mobilization and rich government commitments to aid and debt relief, but which I hated because of its implicit (and wrong) message that poverty could be ended simply by the actions of outsiders on aid, debt and trade.

So when (and what) is it OK to simplify, and what are the costs and benefits of doing so?

When I arrived at Oxfam, I was given its recipe for the perfect campaign: PSV  – a clear problem, solution and a villain (heroes are optional). For example, the campaign for Access to HIV medicines in the early 2000s had problem = patent rules restricting access; solution = clarify international law to allow governments to override the rules; villain = Big Pharma helpfully taking the South African government to court to stop it over-riding patent rules during the early years of the HIV pandemic. Although battles continue, the campaign helped bring about major changes in policy and practice on access to medicines.

complexity and coffeeBut when you think in terms of systems, all three PSV elements need to be heavily qualified. Problems are usually multiple, interconnected and not all obvious. Solutions vary according to time and place. Villains are seldom monolithically villainous and may also turn out to be potential problem solvers/heroes. And you are likely to be at least partially wrong in your identification of all 3 and to have to change your views as you learn more. Ouch.

What to do? There is good and bad simplification. Good simplification doesn’t overclaim: A clear problem statement can still acknowledge complexity but say ‘this needs to be fixed’. A proffered solution should emphasize that this is not a magic bullet but explain why this solution, among others, is worth a try. On close inspection a villain is likely to be itself a complex system – look inside a company or a government and you are likely to find some allies that can work with you to find a solution.

More broadly, working for change in complex systems requires a change in approach at lots of levels, starting with tyranny is the absence of complexityleadership. I’m currently reading a fascinating new book (review to follow) on China’s development success, which identifies the genius of the Chinese leadership since 1978 as partly about accepting the limitations that complexity places on leadership. It argues that after 1978 they moved from Mao’s disastrous attempts at Command and Control to a fuzzier (‘directed improvisation’) attempt to influence the decisions and behaviours of China’s 50 million (!) public servants. Leaders set the boundaries of the broad environment, but then it’s up to local officials to improvise and find new solutions (often to the surprise of the party bosses in Beijing).

Thinking about complexity has also brought me round to an argument I used to hate. If problems/solutions are complex and context-specific, outsiders might be better advised to argue about process – who needs to be involved and supported (eg civil society, women) in finding the solutions, rather than claiming to have the solutions themselves.

When I looked at how aid agencies design their theories of change for working in messy places like fragile and conflict affected states, I found an interesting bifurcation between, on the one hand, trying to engage with complexity at multiple different sites and levels (a bit like acupuncture) and arguing that this is all too difficult for outsiders. They should instead stick to some broad ‘enabling environment’ issues, usually around access to information, and leave the rest to local actors.

The Chinese Leadership also have some lessons to offer in terms of how to talk to the public/activists about all this. Don’t gleefully say ‘hey, it’s all really complex and so you have to be really smart like me’ (an unfortunate trait of a lot of systems thinkers I have run into). Do communicate through analogy (as in the memorable Chinese version of adaptive management: ‘crossing the river by feeling the stones’).

The inside of my head

The inside of my head

I’ve found analogies by far the best way to talk about complexity and systems thinking, as you can plug into people’s lived experience of complexity, and highlight the absurdity of linear approaches: Would you design an 18 year logframe for your newborn baby, setting out your parental activities, outputs and outcomes in advance? Hope not. Ditto cycling/driving across town – a plan of intended velocity and direction for the entire trip would lead to an early grave; instead it’s all about fast feedback and adaptation – learning by cycling. You get the picture (and so does everyone else).

Conclusion? It is possible (and preferable) to keep complexity in mind, while tapping into the virtues of simplicity. Aims/direction of travel/principles need to be simple and memorable; problems to be fixed and process should be kept simple. But beware simplistic solutions and caricature villains – that’s when simplicity can make changes harder to achieve.

Thoughts?

11 comments

  1. Delighted you’re enjoying “How China Escaped the Poverty Trap”. It’s absolutely fantastic, and rightly applauded by many in the World Bank & other major international institutions. Given complexity, we don’t know what will work in different times and contexts, so need to learn by doing. Most importantly, low- and middle-income governments learn by doing.

    But..

    Some donors might be be reluctant to fund this approach – concerned to let governments cross the river in their own (perhaps slightly unusual) ways. In this podcast, Matt Andrews makes two key points: (1) learning by doing is the only way to solve complex problems with unknown solutions; (2) monitoring & accountability can still be integral to these processes. https://soundcloud.com/user-845572280/matt-andrews-podcast. Draws on his open access book co-authored with Lant Pritchett and Michael Woolcock, which you reviewed in April.

    But I think there’s a further challenge. Yuen Yuen’s book doesn’t just highlight improvisation, it also highlights top-down incentives for improvisation to achieve particular goals, i.e. investment and growth. The CCP prioritised those ends. But what happens when important issues aren’t prioritised by governments? For instance, as Naomi Hossain discusses in “Aid Lab”, over 60% of Bangladeshi politicians are also garment manufacturers. They’re unlikely to perceive low wages as a ‘problem’, so may not address this via PDIA. They may not want to cross this river.

    I think the international development community can play a key role here, supporting a more enabling environment, with better incentives for pro-poor PDIA – such as if European buyers reform their practices that encourage a ‘race to the bottom’.

    1. Thanks Alice, am half way through Yuen Yuen’s book, and loving it. Also loved Naomi Hossain’s v different take on development success in Bangladesh. Aim is to finish the China book, review it, then think about the different approaches of the two authors. Sounds like we’d better confer!

      1. Naomi & Yuen Yuen’s books are my two favourite books from the past few years. Both brilliant.

        I think Naomi’s main claim is that the 1974 crisis galvanised elite commitment to poverty reduction, education & gender equality. I think there are parallels with the failure of the Great Leap Forward and the Cultural Revolution. Widespread famines made elites get real: try out new policies to radically reduce poverty. But, as Naomi notes these are exceptional cases. Elsewhere, in the Horn of Africa for instance, famines have not strengthened elite commitment to poverty reduction. Naomi suggests this can be due to horizontal inequalities, the paucity of shared identities and empathy.

        So, I think the real challenge is to think about how do you foster this kind of pro-poor commitment and iterative adaption more broadly?

        I’m hosting Naomi at KCL in November, do you want to come and debate the wider implications?

  2. Duncan you might want to also factor in to your reading ‘Butterfly Politics’ by Catherine Mackinnon http://www.hup.harvard.edu/catalog.php?isbn=9780674416604 it usefully links some of the complexity and politics deabate to gender issues. Including the notion that actually getting things approximately right early in processes- as opposed to just trying stuff and adaptively iterating – is really important given sensitivity on initial conditions. I am not sure I agree with her interpretation of Chaos theory but her analysis of practice through a complexity lens is interesting. Perhaps this is usefully combined with Yuen Yuen Ang’s stuff which also suggests that innovation doesn’t just magically happen.
    Chris

  3. Hi Duncan. That you have used my slide about policy processes being complex at the top of your article has quite made my day. In my talks I used to say “chaotic” rather than complex, but soon discovered that word was extremely unpopular, especially with policy makers. So now I usually say complex, multifactorial and non-linear, which seems to be much more palatable. Another slide I use a lot is Dave Snowden’s Cynefin Framework (https://en.wikipedia.org/wiki/Cynefin_framework) which describes five decision-making contexts or “domains”—simple, complicated, complex, chaotic, and disorder – and how to approach deciding what to do in them. In simple domains where cause and effect are clearly understood – just follow the recipe. In complicated domains, where cause and effect are well known, but might require multiple actions, just follow the blueprint. In complex domains where there is no well defined link between cause and effect (though it may be discernible afterwards), find out as much as you can about the situation, do something that you think might work, monitor it closely and adjust as necessary. And in chaotic domains just try anything that seems sensible. If it works, do more of it, if not do something different. A bit like life really, except mine is more like disorder – I can’t ever quite figure out what is going on.

  4. There is a real problem of the word “complexity” have different meanings that is starting to get in the way of understanding.

    The example in the illustration about options for coffee is NOT complex, in the usual definition but complicated. When you order a particular option from the long list you expect to get exactly what you ordered. Linear cause and effect apply if the coffee supplier is in any way competent. It is much more useful to regard that as complicated. Complex adaptive systems are complex in large part because they adapt. If what you got when you ordred a latte from the list depended on what all the previous customers had ordered, the weather, who was serving today and what clothes you were wearing – with no reliable way to predict the interaction – then that would be complex.

    Confusing complex and complicated (or whatever words you want to use) is a significant problem and a major obstacle to making any sort of progress.

    The most widely accepted consistent view certainly appears to be the Cynefin work of Dave Snowden and others http://www.cognitive-edge.com. This has the real advantage of being based on the rigorous (but very difficult) information economy theory of Max Boisot. Most other contributions – many of which are certainly valuable – are based on observation of a limited number of examples and personal interpretation. Trying to extend what is observed in one place to a more general approach is absolutely doomed to fail it is complex adaptive systems – it is nonsense to say a system is not linear cause and effect, but I am going to assume it works in exactly the same way as another system somewhere else where there is also no predictable link between cause and effect. The core definition of any complex adaptive system is that you CANNOT use linear cause and effect and what happens elsewhere CANNOT be simply copied. Cynefin work suggests that two of the key properties of complexity are -1- cause and effect can only be seen with hindsight (e.g. financial crashes) and -2- repeating the same action does not always produce the same results (e.g. elections last year and this year, e.g. buying shares in a company last year and this year).

    Until we can start to agree on what complexity means and start to take care about what the word complexity means, it is going to be really hard to advance. An example is a recent posting on one of the major development list-servers that said “complexity is surprisingly predictable”. aaaargh ! If it is predictable is stops being complex and becomes – like your coffee – complicated. Or else we need another word for non-predictable complexity.

    Useful discussion of Theories of Change has largely ground to a halt because of the gulf between linear and non-linear (or causal and non-causal) usage. Until there is some broad agreement on the meaning of complex as non-predictable and not just complicated predictable then better understanding is probably going to be blocked.

    1. ..and a brief reply to myself as I didn’t notice the title “complexity vs simplicity”. Complexity – as most commonly used to mean “complex adpative system” – is not the opposite of simplicity. Complicated is the opposite of simple. Complex is the opposite of “predictable” or “causal” or “repeatable” or “linear”.

      Russell

    2. Good point Russell, I just thought it was funny, but you’re absolutely right, there’s nothing complex about it. Apologies

    3. Agree with your point on Theories of Change, Russell.

      Also found my way to your site. ‘Reporting on the difference that an intervention made to the lives of people, not how busy everyone was during the project’ – amen. Looking forward to you populating that!

  5. I was struck by a recent comment by Richard Butterworth, who is leading the struggle on these issues in DFID: that addressing complex development challenges doesn’t necessarily require complexity in programme management. Strictly, that should probably be read as doesn’t require *complicated* interventions; but I think it’s an important and not entirely obvious observation. Some current governance programmes supported by external donors have far too complicated delivery systems, which takes away the agility needed to address complex challenges by being adaptive etc.

  6. Complexity comes naturally to humans. Simplicity we recognise as desirable, but difficult (for us) to achieve. Good article. I get annoyed by the KISS disciples, who don’t understand that it is not stupid to act as humans do. Rather, it is admirable when humans achieve worthwhile simplifications. It is hard for them to do that, and their achievements should be appreciated more. :)

Leave a comment

Translate »