Overcoming Premature Evaluation

November 16, 2016

Some highlights from the first 30 book launches for How Change Happens

November 16, 2016

Fragility v Conflict – can you help with a new 2×2 please?

November 16, 2016
empty image
empty image

Struggling towards the finishing line on my paper on empowerment and accountability (E&A) in fragile and conflict-unity-is-strength-cartoon
affected settings (FCAS) – thanks to everyone who commented on the first draft, by the way). It’s nearly there but I need your help with one particular section.

I want to argue that lumping ‘fragile’ and ‘conflict’ together in one category is very unhelpful. In reality, many violent conflicts coexist with high levels of state capacity, while low levels of state capacity (either national or subnational) often do not generate or coexist with high levels of violence.

By conflating fragility and violence, the FCAS category muddles attempts to distinguish between E&A in different settings. That is aggravated by the use of the word ‘conflict’ to cover both contestation (omnipresent in any polity) and violent confrontation. In practice, those working on empowerment and accountability often appear inclined to interpret ‘conflict’ in its less violent manifestations, airbrushing out the horrors of war and violence. But that gives insufficient weight to issues of violence, trust and fear that profoundly change the risks and logic of working on E&A in dangerous places.

So what if we disaggregate the two in the inevitable 2×2 – does it help us think through the kinds of approach best suited to different contexts? I would really appreciate suggestions for the kinds of donor/NGO interventions that are most suited to the different quadrants, and for the countries/regions that exemplify the different categories. I’ve plonked a few in there to give you an idea, but am very unsure where things belong.

It probably needs at least one more axis – whether the state is willing or unwilling to promote empowerment and accountability (see top right quadrant), but do please have a go with these two dimensions and see what you can come up with.


I sent this to my always wise colleague Jo Rowlands and got this late night response:

‘My problem with your 2×2 , (as it usually is with them – I know they’re useful but they’re also limited as you yourself indicate in your comment about wanting a 3rd one!) is that you have to choose just one of the possible axes of fragility to juxtapose with the violence axis. Maybe what we need is an iterative 2×2 matrix, keeping the violence axis constant and cycling the other one through a range of fragility axes, in order to consider our situations through a range of lenses…otherwise, in our efforts to render something simple enough to think about, we simplify beyond usefulness.’

That sounds about right. I’m becoming a big fan of 2x2s (for example this one on aid intervention v context) because they help you look at hitherto ignored combinations and quadrants. But if they start looking like a typology, where you have to try and cram all known situations into one of four quadrants, then forget it. They should always remain an aid to thought, not a substitute for reality – a trampoline, not a straitjacket. A starting point, not the final word.

In that spirit, over to you.



  1. While 2 X 2 matrices have the benefit of simplicity, am increasingly finding them overly simple and restrictive. Experienced this on a scenario planning exercise a couple of weeks ago, where high and low against the different axes felt too limiting, and a middle ground on each seemed a useful addition to the analysis. In your example, having Sudan and Myanmar (or Syria) at the same level of conflict, and Tanzania and DRC at the same level of state capacity, feels wrong – and not very helpful for the purpose of this tool, in identifying relevant Empowerment & Accountability strategies/tactics for that context.

    Would a Rubix cube work be a better option, covering higher, medium and lower levels for each of the three axes (you could then bring in your additional variable of “open-ness to non-state actors” that you’ve started in the top right hand corner)? You wouldn’t have to populate each of the 27 boxes of this 3 X 3 cube, but look at which ones fit the different contexts you need to analyse for the work you are doing (presumably you aren’t looking at every potential FCAS?) – and so help identify where different Empowerment & Accountability strategies/tactics would be appropriate for those contexts.

  2. Very interesting Duncan and Jo. It sounds like we need to bring in a monitoring and evaluation expert to develop this multidimensional iterative matrix and design metrics for it to better understand the relationship between these variables. We are still at the early stages in building these conceptual frameworks that are much needed to be smarter about what matters and what needs to be measured. We are currently recruiting in Oxfam GB for a new Global M&E Advisor to lead thinking in this area of work. For more information, see here: http://policy-practice.oxfam.org.uk/blog/2016/11/real-geek-wanted-mel-specialist-on-fragile-and-conflict-affected-contexts and here: https://jobs.oxfam.org.uk/vacancy/global-planning-monitoring-evaluation-and-learning-pmel-advisor-int2968/4973/description/

  3. I personally would avoid a 2×2 like this for all the reasons that Jay sets out, but also because it risks masking more than it illuminates. Suda Perera wrote recently, for example, about how we think of countries like the DRC having low capacity when they can often ‘turn on’ much higher capacity – at least in some areas – when it suits (eg, shutting down the country’s communications systems or scrupulously sticking to bureaucratic procedures). I’m also not convinced about usability, if that’s the purpose. Sudan and Myanmar are completely different contexts, and it’s hard to see how one prescription per quandrant is going to work. Would you discourage social accountability in more difficult contexts? Simple 2 x 2s encourage simplistic decision-making…not very systems-y!
    I do like the attempt to parse out conflict from violent conflict, but that may be overstating the issue. Separating fragility from conflict, that makes more sense. But most conflict work is concerned with violence conflict. I personally think the more interesting question is how to separate out the rare occasions when violent conflict may be the only way to move towards a more inclusive political settlement.

  4. Do you remember the work I did on fragile states and complexity for Jo a few years ago. It looked at this issue dynamically – and tried to consider that fragility is about the potential that things will collapse – which is more likely if there is a coming together of social/political/economic and environmental issues – so much wider than conflict. And it tried to consider what issues are deeply embedded in a culture (if not very visible), what issues are emanating from wider geo-political pressures, what issues are on the horizon. Not sure it is easy to capture the dynamical and systemic aspects of this in a 2 by 2.

  5. Hi D, Just ran this post (and comments) past tonight’s CSO Forum in Yangon (but didn’t waste too much of their time on it). They all said “Derrhh”. I suggest you are making it all much too pointy-headed, and are ensuring that the Trumpets increase their global vote in 2020. Do it simple! Lumping conflict and fragility together is a dead-end, and the dead-end is not because the two little boxes won’t squash into the same one little box, but because they are not boxes in the first place. Onwards and upwards. Matt

  6. Scrap the generic typology.

    My understanding of is that local problem-solving builds ownership, understanding & sustains activism.

    Facilitate this process, surely? Rather than charting generic universal guidance that might not motivate, inform or engage local actors?

  7. I am often a little bit sceptical on the “low state capacity” assessment. In Zambia, we would no doubt qualify in that bracket, according to most aid agency measures. But we have the capacity to close down the independent media, fix elections, and direct the agencies of state to vigourously pursue some goals while ignoring others. Capacity may be low in places, but its high when it matters. The juxtaposition of the very political domestic nature of “capacity”, and the very neutral external assessment of supposedly the same thing, is troublesome.

    Similarly, what constitutes a tick-box for violence? Violence can be very selective, and people very rapidly change their behaviour in response to threats of violence. So, from Zambia again, there may not be widespread violence, and we are well known for being “peaceful”. But it would be very wrong to overlook the extent of intimidation that secures this “peace”. Its not the same as overt violence, but its also not what you might understand by “low violence” either.

  8. Hi Duncan–a colleague forwarded your question and suggested that I comment. Fascinating discussion. I agree with the several folks who have suggested that this cannot be reduced to a nice, simple 2X2. It IS an issue of complexity as Jean Boulton suggests. Although there are somewhat classic “archetypes” for conflict and violence, each situation warrants a careful local analysis that takes account of complexity. We could have a long discourse about the important variables to pay attention to in developing a systems understanding of any particular context. Our research (at CDA, Reflecting on Peace Project) has identified the following as consistently (if differently) important in all situations of violent conflict: security/sense of security; acknowledgement of key conflict drivers and commitment to address them; good-enough governance (which would include your high/low capacity, among other things); equitable economic structures; and social cohesion. AND, all of those are deeply affected by cross-cutting issues of marginalization/exclusion, sense of grievance, and political culture regarding the gaining and holding of power. Those factors, examined in each situation can be shown to interact dynamically in ways that perpetuate “fragility” and violence. Gaining a systems understanding provides the basis for identifying how to intervene to shift the systems dynamics and move towards greater justice, peace and security.

  9. I think you’re right to separate violence from fragility, but wrong to equate fragility so heavily on the state (I know you go into this in detail in the paper).
    Fragility can be rooted in society as well as (and not only in relation to) the state, and concepts of empowerment and accountability become too limited if they are only conceived of as in relation to the state.

    We rooted Fragile States: the role of media and communication in a quote from Collier (P.6) – “the fundamental mistake of our approach to state building has been to forget that well functioning states are built not just on shared interests but on shared identity”. (

    It argued that most fragile states (and you could argue some closer to home) are fractured states – “where the existence of different politics, religions or ethnicities makes relationships between communities especially difficult, and where the building of shared identity can be especially challenging”.

    I think it’s difficult to provide an analysis of empowerment and accountability without looking at these issues of media, communication and identity formation. A lot of what you are focused on, (and a lot of what we BBC Media Action do through our media programming in fragile states), is designed to amplify “the voices of the marginalised”.

    But we have to acknowledge that some media and communication trends are explicitly leading to or are focused on empowering specific identities in society (many of them historically marginalised) and some of this takes damaging forms leading to an increasing enmity of or suspicion of the “other” in society and a shift away from shared identity..

    I don’t in the 21st century information age think that shared identity should or can be shaped by the state (old fashioned nation building through monopolistic control of state broadcasting for example), but nor are there good platforms through which citizens can do this through public dialogue and argument across the fracture points in society (the current talk about filter bubbles and echo chambers suggest the reverse).

    In fact, I think there’s an increasingly compelling case to be made that how information and communication is accessed and deployed and controlled and manipulated is a primary driver in determining how fragile or resilient societies are, and that it is the fragility of society (e.g. lack of social cohesion, common bonds of mutual trust etc) that may increasingly be shaping the fragility of the state and making the achievement of political settlements more difficult and less sustainable. All that is intrinsically tied up with issues of empowerment and disempowerment.

    I’m going to refrain from tying this to current political developments in the West but in a nutshell, if you’re going to have a 2×2 I’d suggest having a High Societal Fragility < Low Societal Fragility axis, perhaps instead of your High/Low violence one. And I’d suggest the paper focuses more on these issues of information, communication, media and identity.

  10. Thanks for opening the discussion Duncan, I wanted to throw in some thoughts from Pakistan. I’ve been living here for three years implementing an empowerment and accountability project (which is being considered for the research programme you’re involved in) your question is one that we come across and discuss quite regularly in the team.

    I think the 2×2 matrix can be relevant for some places (Tanzania as you mention is a good example) but I think Pakistan is an example of where the 2×2 matrix of fragility and conflict would a) fall short and b) give you an unhelpful framework through which to view the country and thus design projects.

    On the face of it, one would put Pakistan in either the bottom right or top right quadrant (depending on whether one considers Pakistan to be a state with high capacity – which is another debate entirely! I would argue for the top right quadrant but this debate should also be informed by Heather’s point above about choosing to turn up capacity which I thoroughly agree with), by doing so, one would then approach work in Pakistan considering where it is in the matrix and this could be a mistake in the case of Pakistan because since the 18th amendment which devolved much of the State to the province level, one has to view a) state capacity on a province by province basis and b) the violence in Pakistan is very diverse (terrorism, violent, organized and disorganized crime) and is not evenly spread across the country, the provinces and urban/rural areas.

    For example, the province of Punjab has high state capacity but chooses to exercise it selectively (focus on infrastructure and power projects instead of public services), the violence in Punjab is low in terms of terrorism (terrorist offshoots of TTP exist but they mostly focus on making trouble for India rather than Punjab, there is also counter-terrorism violence by the army) but high disorganized violence in rural areas and high organized violence in urban areas. If we contrast this with KP province (Imran Khan’s province), it has much lower state capacity than Punjab (a coalition government and far fewer funds from the federal level) but directs more state capacity towards public services and devolving power and funds to the local level. On the violence side, KP still has incredibly high terrorism issues and is currently in the process of absorbing the agencies collectively known as FATA which have been a hotbed for TTP, it has much less violent crime however.
    Then one can flip to Sindh and Baluchistan where Sindh has much lower violence (although still terrible incidences of GBV) in the rural areas, extreme levels of organised, violent crime in Karachi and extremely low state capacity throughout the province. Poor Baluchistan sits firmly in the bottom most and farthest right corner of your quadrant with little to no relief in sight.
    And just to throw some more contradictions into the mix, the autonomous administration of Gilgit Baltistan has low violence and low state capacity.

    Of course, I completely appreciate that you never intended your 2×2 matrix to speak for the nuances of every situations others above have mentioned, but I think the Pakistan example above is a fair challenge to the matrix as it stands. I wonder if a matrix would which would be more helpful for a state like Pakistan (and I’d guestimate would also apply to countries such as Nigeria, Burma, Kenya and Morocco) would be one which looked at violence vs state delivery rather than state capacity. This would also speak to recent literature on political settlements and economic growth and political settlements and violent conflict. Of course I appreciate that state delivery cannot always be a proxy for fragility in certain contexts and so I agree with your colleague Jo that perhaps one needs different proxies for fragility against a constant axis of violence.
    It might also be helpful to flip the question on its head and have 2×2 matrices with a constant axis of violence with the other axis determined by what you want to know/do. If you are designing a CVE project or a conflict resilience project, you’ll want to look at fragility in a different way from if you are designing empowerment and accountability project or a supply side service strengthening project – that could help to determine how you class a country in terms of its fragility I think.

    1. Thanks everyone, a fairly consistent message here! Silly to conflate fragility and violent conflict, but pretty much just as silly to replace them with a 2×2. Will make sure the paper reflects this thinking. Particularly struck by
      – the fact that ‘capacity’ involve agency – parts of the state deciding to do something or not, as well as a passive ability to do so.
      – states are often simultaneously strong in some areas and weak in others
      – look at any state, and regional variation is huge (Pakistan example very graphic)
      Crowdsourcing rocks!

Leave a comment

Translate »