Programming in Chaos. Why I think we’ve been getting it wrong.

I’ve been bouncing some ideas around with Irene Guijt on how aid agencies can/should work in what we call ‘fragile and conflict-affected settings’ (FCAS). This matters because FCAS are where a lot of the aid business (both donors and INGOs) will end up, as more stable countries grow their way out of aid dependence (and a good thing too).

Let’s start with a simple framework, Cynefin, which divides up situations into four broad types.

  • Simple/clear: (known knowns, in Rumsfeld terminology). The situation is stable, and the relationship between cause and effect is clear: if you do X (walk across a motorway), expect Y (sudden death). Response? Find the appropriate rule/toolkit and apply
  • Complicated: (known unknowns). The relationship between cause and effect requires analysis or expertise; there are a range of right answers. Example, sending a spaceship to the moon and getting it back again. Response? Combine expertise from different sources. Artificial intelligence copes well here: Deep Blue plays chess as if it were a complicated problem, looking at every possible sequence of moves.
  • Complex: (unknown unknowns). Cause and effect can only be deduced in retrospect, and there are no right answers. Examples: battlefields, markets, ecosystems and corporate cultures. Response? Trial and error with quick feedback loops for fast adaptation.
  • Chaotic: (yikes!) Response? Staunch the bleeding/fight the fire/stem the panic. Try and get back to islands of complexity.

The aid business (along with academia, businesses, think tanks, government departments etc) mainly sees the world in terms of the right-hand side of Cynefin – problems and tasks that are either simple or complicated, like building a bridge, but soluble. That’s the basis for projects, logframes and all the rest of the guidelines that the aid business takes for granted.

Critics of this ‘business as usual’ approach recognize that aid actually often operates in countries and systems on the left-hand side of Cynefin. Cue lots of discussions and experimentation on ‘adaptive management’, ‘thinking and working politically’ etc etc, all faithfully reflected on this blog.

But I’m worried that the new conversation has blurred the distinction between the ‘complex’ and ‘chaotic’ quadrants, with some concerning consequences. When I was researching ‘adaptive management’ case studies for the Action for Empowerment and Accountability research programme, we wanted to look at how DFID-funded governance programmes operate in FCAS. But in the end we had to settle for countries or sub-regions that weren’t actually that fragile or conflict -Myanmar, Nigeria and Tanzania. These had functioning governments, CSOs, laws etc, at least before Magufuli in Tanzania and Myanmar-before-the-coup. Carrying out research in more chaotic settings were simply too inaccessible and/or dangerous.

The problem is that the lessons we and others drew on how to operate in complex settings, which I think are valid and really interesting, don’t necessarily hold for truly chaotic settings, for example those where no-one knows who’s in charge, and everything changes from one day to the next. So what kind of aid programming might work there? Some initial thoughts, but please add your own.

Firstly, there is humanitarian aid. This is the default, and very necessary, response in situations where lives are disrupted and survival put at risk. Interestingly this is a simple response to sometimes chaotic contexts. It’s like fire fighters – very rule-based responses developed for ‘chaos’.

Beyond fire-fighting, no country or situation is completely chaotic – there will be islands of social and political solidity that donors can identify and support. But these grains may well be different from the normal partners (the state, formal civil society organizations) preferred in the other quadrants. Here are three candidates for promising islands worth exploring, but I’d love to hear about others:

Individuals and Leaders: even in the maelstrom, the same people will crop up, organizing their fellows, finding ways to improve things. In the DRC, Oxfam supported Civilian Protection Committees that organized local people to negotiate the day to day hassles of living in the middle of a war zone, like reducing the amounts demanded by the soldiers (both government and rebels) to get past the roadblocks between them and their fields. When forced to flee because of armed attacks, it was those same committee members who started organizing people in the refugee camps.

Which points towards activities that may seem pretty unfashionable/individualistic – scholarships, leadership programmes, stipends for activists – as being compatible with chaotic contexts in a way that more institutional programmes are not.

Diaspora: Many chaotic settings have large, active and influential Diaspora communities that retain deep connections back home. I’m not quite sure whether they need (or even want) support from the aid sector, or what form that support could take, but as a permanent and stable feature in a chaotic situation, they are definitely worth consulting.

Faith Organizations: People’s personal faith and their involvement with faith organizations is if anything even more important in chaotic settings. The Catholic Church was once described to me as the DRC’s only truly national institution. Faith organizations incubate activists and future leaders (both good and bad). Yet many big donors and even some INGOs are resolutely secular, and see any involvement with faith groups as at best instrumental (one Oxfam staffer once told me ‘sure we work with faith organizations – they distribute our leaflets’).

If donors are serious about working in genuinely chaotic settings, they will need to adapt what they have learned in complex settings to more chaotic ones. One aspect will be getting better at spotting islands of political or social stability that emerge from the swamp and finding ways to support them. Chucking lots of money at them may well be counter-productive.

Any other suggestions? And here’s more from the Cynefin folk on managing chaos and complexity in times of crisis.

Subscribe to our Newsletter

You can unsubscribe at any time by clicking the link in the footer of our emails. For information about our privacy practices, please see our .

We use MailChimp as our marketing platform. By subscribing, you acknowledge that your information will be transferred to MailChimp for processing. Learn more about MailChimp's privacy practices here.


16 Responses to “Programming in Chaos. Why I think we’ve been getting it wrong.”
  1. Mark Adams

    Hi Duncan, this is very interesting – both the analysis and the conclusions. Your focus on people and their institutions feels intuitively correct to me, and reflects my own experience of working in some of these situations, and the pressing need we have to address issues of power and “localisation”. Having worked for Oxfam and a faith-based NGO I have a better understanding of the differences and commonalities – and tensions – between “secular” and “faith-based” actors – though I believe a binary distinction between the two has limited utility. Your caution in assuming that diaspora communities want our support or involvement also seems valid to me. Some honest dialogue could be valuable in examining that question.

  2. Hi Duncan
    Perhaps it is time to retire the Cynefin framework? Why? Because it just does not have a very good fit with the real world. Two examples come to mind.

    The first is the day-to-day weather system which we are all familiar with and would probably describe as being some sort of mixture of complex and chaotic. The thing is, the ability to predict the weather in the near future has been improving decade after decade. Taking a longer-term perspective, the ability of climate change modelling to fit observed behaviours of the climate has also been improving. So where does this example fit in the 2 x 2 framework?

    My second example is prompted by my current reading of the Checklist Manifesto by Atul Gawande. In this very readable book, he describes a number of fascinating accounts of the complicated, complex and chaotic situation that can be found in intensive care wards of hospitals, where the capacities of multiple different specialists are brought to bear on the partially known and often compound medical problems of people who have had the misfortune of ending up there. The stories of people’s survival in these circumstances is nothing short of amazing, when viewed within longer-term perspective of the history of human medicine. But there are also tragedies and this to some extent is what his book is concerned with helping us to avoid in future. Often those tragedies arose not from lack of knowledge, but from forgetting or neglect of existing knowledge. He gives numerous examples of where simple devices i.e., checklists of things that need to be done in particular circumstances, which can make a huge difference. He also explores similarly complex situations in other spheres of human life – piloting aircraft and building multi-storey office buildings where checklists have been invaluable as a means of avoiding disasters. So, I ask again, where do these examples fit within the 2 x 2 framework?

    Then there is your own example. “Firstly, there is humanitarian aid. This is the default, and very necessary, response in situations where lives are disrupted, and survival put at risk. Interestingly this is a simple response to sometimes chaotic contexts. It’s like fire fighters – very rule-based responses developed for ‘chaos’.” This sounds similar to the situations I’ve just described above, except that you have categorised them as chaotic. But are firefighting situations really as chaotic as you think? If they were it is hard to imagine that any rule-based system would work.

    I do agree that “no country or situation is completely chaotic” (or complex, complicated, or simple). In just about every circumstance where we can find ourselves there will be different combinations of these characteristics. From the examples you cited of the different groups that can be worked with in the DRC and elsewhere it seems that some relatively simple interventions (contra Cynefin) can still find a useful place in that very challenging environment –”activities that may seem pretty unfashionable/individualistic – scholarships, leadership programmes, stipends for activists”. Similarly simple to those interventions described in the checklist manifesto. They may not be sufficient, but they are possible and can make a difference.

    • Duncan Green

      Thanks Rick, excellent food for thought. Isn’t the point that, like most 2x2s, reality does not fit in any one quadrant, but thinking about them helps stretch our analysis to include things we might otherwise have missed? Not ready to jettison Cynefin quite yet!

      • Well… the Cynefin framework is not really a 2×2, or even an static clasifying model: its last iteration considers 5 to 7 different spaces, and its aim is rather enabling dynamic transitioning among them, rather than reflecting on how “things” are.
        Because certainly, no situation or context is expected to fit into any of the quadrants, or stay there for long. Whatever the situation you are in, there will be aspects/issues that will fall in the clear, complicated, complex, chaos, or even the “confusion” sections (meaning that you do not even know where they fit), that you need to handle in different ways.
        (See for a quick tour on Cynefin’s last iteration).
        But in fact, anything you place in the diagram is not even “the thing” (out there). To a big extent things are not complex or complicated “as such”, but only in relation to you as an actor, and particularly in relation to your level of knowledge, understanding, infrastructure, technology… your capacities, generally, which allow you to consider and handle things in different ways. You mentioned “Chess”; it certainly used to be a complex enterprise, but now savvy enough actors can considered it “clear”: you just need to feed a powerful enough AI model with information on the basic game constraints, let it play against itself for a couple of days and… bang!!, then proceed to defeat any human master that appears.
        There were also mentions the fire of a building: is it chaotic, complicated, or rather simple? Well, it mostly depends if you are the fire department, or just a person trapped in the building (and again, not the same for one in the top floor and one in the ground floor).

        This “relational” character is very important, as it is linked not only to our capacities… but also to our “IN-capacities”, which in the case of organizations frequently result from our own “internal complexity/constraints” as an actor. This “internal complexity” may be more relevant than the “external complexity” we face, and render us unable to respond to the challenges coming from the settings where we operate.
        This, I am afraid is so frequently the case of the aid industry and the programmes we run, which bring together such an incredible assortment of wise and competent people, but force us to act collectively in utterly stupid ways.
        I almost feel that we keep talking so much about how challenging and complex the contexts we work in are, just to avoid facing the irrational internal complexity and constraints that define us.
        A silly example from last week: a programme designed to be at the forefront, the cutting-edge of adaptiveness thinking and practice, where nonetheless it is not posible to move expenses planned for March into the next financial year. God forbid those sinful and frightening things from happening in this very decent industry!! 🙂

    • Dan

      I think part of the issue with the fire-fighting example is merely that it’s not a great example of chaos (sorry Duncan!). Putting out a fire is pretty much always going to be “simple” (in Cynefin terms, that’s not to say it will be easy), or at most it will be “complicated” – e.g. if fleets of airtankers are involved for bushfires etc. But all certainly on the right-hand “ordered” side of things.

      Where there’s an analogy to be drawn for the “chaotic” quadrant is maybe if you were trying to work on a problem (possibly a complicated or complex one) and someone kept setting fire to your desk. It won’t directly help solve the original problem, but your step #1 will be to put the fire(s) out, find out who’s doing it and remove that person from the premises – putting you back in the complicated/complex worlds.

      I’m entirely agnostic about whether we keep the Cynefin framework, certainly I agree it doesn’t fit perfectly with the real world (what models do?), but it seems a whole lot better than the actually-existing model of most of the sector which (as Duncan says) only looks at the right-hand side. If there’s a shift towards people viewing “complexity”/”complex systems” not as synonyms for “complicated” (which is how they’re used in, oh, 100% of all conversations I’ve ever had in INGOs) but as something requiring a fundamentally different type of response, then that’s a positive.

      Whether or not the “chaos” quadrant helps with that is, for me, a slightly tangential issue. I’m not entirely sure I share Duncan’s concern about the blurring of chaos and complexity, especially since adapting to complexity is so little a part of anyone’s actual practice beyond lip service. So, if someone were to cut out that part of the framework and have a simple/complicated/complex triptych then I’d view that as already better than the standard approach. What I personally would then consider to be missing is the sphere where you don’t know which of the three you’re in (a very common scenario). This might be Cynefin heresy, but that’s sort of what I’ve always seen the “chaotic” sphere as representing – so the response is you have to act blindly just to find out where you are, and how to slice your problem up into complex/complicated parcels.

      More directly in response the blog: Duncan, I think it’s great, but I’d be cautious of this list of “islands” (which make sense) leading to a list of possible response archetypes (your question of “what kind of aid programming might work there?”), which would bump you back over to the right-hand side. What seems crucial is not to miss the answer you’ve already provided above: “Trial and error with quick feedback loops for fast adaptation”. I’d argue it almost doesn’t matter what *types* of programmes you do (so long as “do no harm” and all that), the absolutely critical non-negotiable is what feedback loops you’re building in and what potential you have for adaptation etc. Doesn’t matter if your programme is working with all three of your identified “islands”: if it requires a static logframe, get out of the country. And take your donor with you.

  3. John Borton

    Many thanks Duncan, much appreciated! For me though, your heading ‘Individuals and Leaders’ doesn’t quite capture the ‘groupness’ aspect – spontaneously formed, self-organised groups – ‘grassroots groups’ if you will. Yes, leaders emerge and help the process of group formation, but I would suggest that the ‘leaders’ may be outwardly ‘ordinary’ people who may not have been aware of their leadership qualities until the crisis creates a situation of need that prompts the personal passion /compassion/ motivation to organise and (perhaps) to lead. My perspective comes from observing, studying and limited participation in the ‘activist volunteer’ response to the (ongoing) European refugee ‘crisis’ – a phenomenon well captured in Sue Clayton’s recent book ‘The New Internationalists’ (Goldsmiths Press). Two points that I would highlight from that perspective/experience are: a) that intense humanitarian needs can occur within ‘developed’ and otherwise well-organised contexts as a result of hostile policies towards refugees and migrants (so not necessarily in FCAS contexts), and b) that INGOs and UN agencies providing humanitarian assistance have very real challenges in engaging effectively with such ‘grassroots groups’ as a result of their rigid and restrictive funding and accountability models and procedures. I would suggest that one of the (many) challenges facing humanitarian agencies is their general ineffectiveness in supporting and working with such locally formed groups of activist citizens/volunteers. I hope this perspective is helpful. All the best.

  4. Singe Day

    In haste and FWIW, thinking of humanitarian experiences from Oxfam’s work in Darfur, I would highlight the following experiences.

    • To improve limited understanding of the context, engage widely with a range of people and institutions – local staff, partners, specialist national and international Universities, government agencies and faith based organisations to try to map out complexity and local realities. Each group will have their biases.
    • Local organisations are probably already well versed in working in these complex environments. I doubt they refer to their work as Adaptive Management but learn from their experiences.
    • Look to the coping mechanisms people (such as displaced and host communities) are already applying and identify ways to support them.
    • Focus on “do no harm” – for example do not go around constructing water points (boreholes) that may inadvertently legitimise illegal land occupations.
    • Focus on “making a difference” – by ensuring interventions are relevant and effective. Building water supplies to high standards for example, as wider support systems will be absent.
    • Be able to respond rapidly as opportunities arise and security allows.
    • Be more open to critically questioning interventions – including scrutiny from independent national and international experts.
    • If possible, engage with donors so programmes can be more adaptive and responsive – hence less time spent feeding the “internal machine” with information.

  5. Steve Wiggins

    Many thanks, Duncan and to all commentators.
    Just to note that the Cynefin frame is not so different from that of Jimmy Thompson, written up in the 1960s. Thompson, James D., 1967, Organizations in action, McGraw Hill, New York
    He came up with a 2 x 2 frame that I’ve found really insightful when I came across it in the 1980s. Here’s some detail:
    Now, to Rick, how can it be that in chaos that simple rules work? Long ago, organisation specialists realised that authoritarian leadership works when (a) you have complete power — no surprise there, but also (b) when you have no power! Example? Secretary of local golf club. Bullying and tyranny often work when you have no real authority at all.
    Management studies discovered and documented an awful lot many decades ago: few people seem to know this material.

  6. Masood Ul Mulk

    How about an organisational approach rather than a project approach to these problems. When dealing with choatic areas here the government has been willing to fund projects but has to be persuaded to fund organisations that are responsive, flexible, adaptive, accountable and people centered. Investing in the former leads to a dead end while investing in the latter enables an organisation to resoond to unforseen events and build programmes as they go. In this process the organisation builds credibility in these areas , and leverage that to do more

  7. I would look at the latest version of Cynefin with its emphasis on liminality and then on the Aporetic aspect of the central domain of confusion (formerly disorder). Blog post updates at Cognitive Edge on 1st March 2019 and 2020 explain this. This is also covered in the new joint publication (with the EU Commission) on managing in complexity (and chaos). which also emphasises what is called the ‘aporetic turn’ with due homage to Derrida, and the creation of paradox for those who want a different language. From memory, Rick was happy with the three-domain version of Cynefin (order, complex, chaotic and the centre ‘Triple point’ of confusion where movement into the three main domains was equi-probable. The split between Clear and Complicated in Order is optional but generally helpful. It is substantially different from Thompson’s Typology of Decision Making and I may write a blog post on that.

    Aporia Five posts starting with
    Liminal Cynefin – links in the first paragraph of the above link

    One of the things I was going to contact Irene about (and am open to wider collaboration) is creating a specific version of the EU Field Guide for the development sector.

Leave a Reply

Your email address will not be published. Required fields are marked *