CEWARN Reconsidered

See update here.

I’ve been particularly critical of CEWARN over the past few years but as a colleague of mine recently noted—and rightly so—the last time I directly worked on CEWARN was towards the end of 2005. In other words, it’s been a good 4 years. In addition, my full time work at CEWARN was limited to about 16 months of work, with about a third of the time spent on-site.

So I’d like to invite my CEWARN colleagues to elaborate on some updates they kindly shared with me via email last Friday. For example, CEWARN has established local peace committees (in addition to national early warning units) which have taken the lead in initiating early warning and response activities. It would be great if CEWARN could describe in detail the actions taken by these committees over the years so that other conflict early warning initiatives in Sri Lanka and Colombia can learn from CEWARN. For example, just how successful have these initiatives been? And what failures have they encountered?

One criticism I have had of CEWARN is the limited number of “success stories” attributed to the initiative which has been operational for 6 years now. However, according to my colleagues at CEWARN, the cases I cite are only a few amongst many others and only highlight responses to Alerts issued by different actors including Field Monitors, local peace committees, and other stakeholders in the Mechanism. In other words, CEWARN has not shared information on other types of responses including cross-border communal peace dialogues, livestock recoveries, etc.

This is really good to know, and I would really encourage CEWARN to share the full extent of these success stories since it is really important to demonstrate the impact of the mechanism. To be sure, if CEWARN has not already done this, I would recommend that they compile a full detailed list of all success stories related to CEWARN’s work. Surely, this would be a prized document vis-a-vis donor relations and beyond. In other words, such a document should figure very prominently on the CEWARN website. I for one would be ready to dedicate an entire blog post series on CEWARN to highlight the initiative’s many important successes over the past 4 years.

My colleagues at CEWARN have also brought to my attention the development of an ICT for Peace Project and a Rapid Response Fund Facility for local, national and cross-border response. These new initiatives are very timely and worthwhile. It would be great if CEWARN could upload more information on these initiatives to their website. For example, how many times has the fund facility been used and what has the impact been? And what failures has the team encountered?

In sum, I’m really looking forward to getting myself more up to speed on the incredible progress that my colleagues at CEWARN have made since I left in 2005. Indeed, they very kindly pledged to help me learn more about the mechanism’s district and local peace committees by offering to provide me with the names and contacts of their early response units and country coordinators.

I eagerly await this information from CEWARN so I can revise and correct my criticisms of this conflict early warning system. As soon as CEWARN follows through with their pledge and I have completed my updated research on CEWARN, I will be sure to write a full blog post on everything I have learned. Of course, if I don’t hear back from the team, then I’m not sure how I can correct my criticisms and share their success stories and lessons learned. In any case, the record will show that I genuinely tried.

Chapter 9: FCE’s Early Warning System and Applicability to Other Countries

The 9th and final chapter of the FCE book on Third Generation Early Warning was co-authored by Kumar Rupesinghe and Tadakazu Kanno. This is an important chapter that seeks to draw on the lessons learned from the Sri Lanka experience to outline how a similar approach might be taken in other countries.

As the authors note, the Third Generation approach is particularly applicable at containing inter-communal violence. It is also very refreshing to read that the authors include a section on the weaknesses of FCE’s EW/ER system. It would be great to see other initiatives do the same.

Rupesinghe and Kanno write that “if there is no will for peace, the FCE-type Early Warning/Early Response cannot work effectively” and that the “cessation of violence is subject to the will of [paramilitary groups].” This is why I have been advocating for a tactical approach to conflict early warning and response; one that leverages the tactics of strategic nonviolent action and digital activism.

The authors also include a helpful section on “Criteria for the Application of FCE EW/ER System.” This section includes pointers on necessary conditions (e.g., inter-communal conflict) and subordinate conditions (e.g., causes of conflict are grievances). Another very helpful section of the chapter outlines how the FCE approach could be applied in specific countries such as Pakistan and Kenya.

In conclusion, Rupesinghe and Kanno write that FCE’s Early Warning/Early Response system “will contribute to saving a number of precious lives in conflict areas.” This is the last sentence of the entire book and a very important one. To be sure, the saving of lives should be the ultimate indicator of success and it is important that we apply rigorous monitoring and evaluation frameworks to assess whether we have any impact on this important indicator.

Chapter 8: Impact of FCE’s Human Security Program

The 8th chapter of the FCE book was co-authored by Joseph Bock, Patricia Lawrence and Timmo Gaasbeek. The chapter summarizes findings from six in-depth case studies carried out to assess the impact of the FCE’s Human Security Program in the East Province. I won’t comment on the individual case studies but will review the authors’ overall findings.

The authors write that “FCE’s Human Security Program has prevented violence.” I think what’s needed beyond the qualitative case studies is some hard numbers. I could equally write “CEWARN has prevented violence.” This would be a true statement—CEWARN has intervened in a dozen or so cases to prevent or mitigate violence. But one has to ask for a percentage figure, i.e., what percentage of all violence did the FCE program actually prevent?

In the case of CEWARN, there has been well over 3,000 incidents of violence documented by the “early warning” network. This would mean that CEWARN’s “batting average” is 0.004%. So yes, CEWARN has prevented violence but is the early warning and response system successful?

I find it refreshing that the authors are so up front about the difficulty of assessing FCE’s singular impact. “Because many different actors seek to resolve problems and support peace in the Eastern Province, there are few cases in which FCE was the only actor involved. Because of this, FCE’s claims at effectiveness will by default always be contested.”

The authors also note that the reports coming in from the field are “not always read because people get so much information that they do not have time to read everything. Generally, people glance through the daily reports a few times a week, mainly to confirm reports that they have already heard.”

The FCE introduced the use of SMS, which is probably the first example of a third-generation early warning system employs text messaging for the dissemination of alerts. The authors note that this mode of communication is “relevant for people working and traveling in the districts, because it helps them avoid dangerous areas.” In terms of staff outside the districts, the authors realized that “being bombarded real-time with security information if one cannot really do anything with it might cause a lot of stress [...].” Finally, two FCE staff members noted that the incident reporting via SMS was generally “correct and useful, but in very rare cases the situation is misinterpreted.”

The data analysis at FCE headquarters was found to “support early response, but it’s exact impact is difficult to measure.That said, FCE field officers “commented repeatedly about how the categorization scheme of FAST [which is actually VRA's] and the training they received from Swisspeace [...] made them think in new ways.” However, the authors note that as a macro-system, FAST was inapplicable to the micro-purpose of FCE’s initiative.

The use of a computer-assisted micro-system also has the effect of promoting forward thinking. In the authors’ own words: “To the extent that an [early warning system] requires that each location has a list of community leaders to be contacted in the event of high tensions, it fosters forward thinking.” I couldn’t agree more and also see an important parallel with crisis mapping. To the extent that crisis mapping platforms like Ushahidi require georeferenced information, it fosters forward thinking on where one might intervene.

Chapter 7: Ethnic Violence in Kattankuby & Eravoor

Heshani Ranasinghe authored the 7th chapter of the FCE book. This chapter goes into quite some detail regarding the Sri Lankan context, which I don’t have the knowledge to comment on. This is an important chapter since Heshani takes two case studies to evaluate how well the FCE’s early warning system worked. I really appreciated the author’s transparency when they note that the system had both successes and failures. Very refreshing indeed.

Chapter 6: Dynamics of Social Identities for Conflict Prevention

The sixth chapter in the FCE book was authored by Priyan Seneviathan. I’m no expert on the Sri Lanka context vis-a-vis social identities and relationships for conflict prevention. So this “review” will figure a few excerpts from the chapter that I found interesting.

Currently there are 95 such Co-Existence Committees established across 7 districts of the country comprising of 5,821 members.

The uniqueness of the co-existence committees is that they comprise of people from different trades and different religions as well as different ethnic groups.

Boege in his analysis on conflict transformation argues that for a successful conflict transformation process to take place, there should be mechanisms in place that are effective in the localities of conflict affected societies that could effectively address relational issues at the local level. He argues that solutions that are bought ‘at the top’ will not be sustainable unless they are synchronized properly with solutions that are coming from the bottom.

In a world that always tries to look at the big picture of a conflict, important aspects of communal relationships that are formed through social identities tend to lose its influence when planning for a post conflict reconstruction phase.

I wish that more had been shared on the main challenges that FCE faced in helping to form the Co-Existence Committees. It would be helpful to have a list of hard lessons learned.

Interview with National Intelligence Officer for Warning

From McKinsey Quarterly:

Kenneth Knight describes his job as helping the president of the United States and his administration “avoid surprise.” As the national intelligence officer for warning, Knight oversees a small team of analysts who serve as an institutionalized safeguard against risk—monitoring and challenging the analyses and assumptions of the broader intelligence community.

In this interview, he discusses evaluating threats, overcoming cognitive biases, and constructing scenarios—challenges familiar to most private-sector strategists. McKinsey’s Drew Erdmann and Lenny Mendonca spoke with Knight in Washington, DC, in June 2009.

The public video is also available here on the McKinsey website.

Chapter 5: Theoretical Justifications for FCE’s Early Warning and Early Response System

This chapter by Dinidu Endaragalle is definitely my favorite chapter of the entire book. I highly recommend a close read to anyone interested in the field of conflict early warning.

Endaragalle draws on two contemporary theories identified by Professor Joseph Bock Bock to justify early response interventions by the FCE’s Early Response system. The first relates to timing and nature of interventions while the second addresses the capacity to intervene.

Timing and Nature

Donald Horowitz (2001) has provided insights about the timing and nature of intervening to prevent ethnic violence. Through an extensive comparative study, he identified how collective ethnic violence displays a marked pattern. Where there has been a hostile relationship between ethnic groups and usually, but not always, after there has been a ‘precipitating event,’ there is a period of time (a ‘lull’) when consensus building for violence occurs.

The attacking group develops intense emotions. Its members build a consensus around a moral argument justifying violence (which is commonly linked to religious beliefs). They assess the threat posed by another group (often exaggerating that threat). And they downplay the anticipated risks of participating in violence.

This focus on the tactical level is one that definitely resonates with me. Understanding patterns of conflict in large part requires an understanding of what military strategies and tactics are employed. See, for example, Jen Ziemke’s excellent research on spatial patterns of civilian targeting during the Angolan civil war.

Capacity to Intervene

The second theory that can be applied to the science of early response originates from Ashutosh Varshney’s work. Varshney has “provided empirical evidence that formal inter-ethnic associations, especially in urban areas, constitute an effective capacity to intervene to prevent violence. […] he wound that violence was much less likely when associations were formalized […] and inter-ethnic in their membership.”

Taken together, Endaragalle argues that “the essence of Horowitz’s theory of intensity and timing of a ‘lull’ and Varshney’s theory of capacity to intervene provide important insights to the science of early warning and early response.” Endaragalle then applies this framework to two real case studies from Sri Lanka.

Developing a theoretical framework and then being able to apply it and draw preliminary conclusions is the mark of serious applied academic research. For example, Endaragalle relates Varshney’s formal association theory to the FCE’s Co-Existent Communities (CECs) as these are formal interethnic associations with different religions represented. By applying the theoretical framework to actual case studies, Endaragalle finds that “after a precipitating event […], the early response interveners should concentrate on diminishing the degree of consensus building for violence by conflicting parties.”

I found this insight interesting as it relates to the idea of patterns and anti-patterns that I have described here in the context of conflict and crisis mapping.

Insider Partial Mediators

Endaragalle also adds a third component to the two-pronged theoretical framework originally proposed by Bock. This one focuses on the concept of “insider partial mediation” developed by Paul Wehr and John Paul Lederach (1996). These authors suggest that the “insider partial mediators often prove beneficial in reaching a successful settlement.” Wehr and Lederach define this type of mediator as “an entity (an individual or institution) that is already involved in the conflict and, at least to some extent, is aligned with one side or other.”

In general, this means that these mediators “are people of high stature and as a result, they have credibility with stakeholders on all sides of the conflict.” Endaragalle combines this concept with the theories developed by Horowitz and Varshney. Insider partial mediators already have network ties with local communities and may thus have the contacts necessary to identify upcoming lulls in inter-ethnic conflict. “The mobilization of ‘insider partials’ can also increase the capacity to intervene in conflicts rapidly in most instances.”

There are a few points in this chapter that deserve to be written in more detail. Endaragalle refers to the FCE’s GIS maps: “they zoom into the databases and the GIS maps and single out the events data relating to the precipitating events,” but does not expand further. How are the events singled out? Do the maps facilitate this singling out in terms of spatial patterns?

Like other chapter authors, I would press Endaragalle to be more specific when writing that FCE’s early warning system “intervened in a recorded number of 174 cases of conflict.” Namely, 174 out of a total of how many incidents? And out of this 174, how many can clearly be shown to have been successful? I don’t doubt the probable success of the system, but having specifics would make the 174 figure a stronger case.

In the same paragraph Endaragalle writes that the next task for the FCE’s early response system is to “estimate the time period of the actual time period [i.e., the lull] for the actual onset of the violence.” I agree that timing is important, of course, and would be curious to know why Endaragalle does not draw on William Zartman’s concept of ripeness. I would also liked to have seen a discussion on William Ury’s Third Side approach vis-a-vis capacity to intervene.

In any case, the plan to analyze lulls is a very good one. But I’m biased given that I’m doing similar research with colleagues at the Santa Fe Institute and the Technical University in Zurich. There is evidence from other scientific research that the “wait times” between certain events often follow a specific statistical distribution, often a poisson distribution.

Chapter 4: FCE’s Early Warning and Early Response System

Written by Priyan Senevirathna, this chapter of the book is a good overview of the conflict early warning methodology employed by the FCE. While labeled as a “unique” system, we should keep in mind that several other third generation conflict early warning and response systems exist, such as in Timor-Leste and Kyrgyzstan.

Of note, however, are the 30 field monitors attached to FCE field offices and the use of SMS. The latter is “primarily targeted at activating those who are in [sic] the ground for immediate [preventive] action […].” This is an important component of the FCE’s system.

Senevirathna suggests that, “by activating an immediate ground level conflict preventive agent, the FCE Early Warning system can first contain violence that may erupt in a given location and take immediate action for spreading further into different parts of the region which are still unaffected.”

I liked this description because it reminded me of the spread of contagious diseases and the need to quarantine or contain this spread. And we know from the field of preventive health that one once of prevention is worth a pound of cure. What strikes me as missing from the FCE approach is a strong focus on preparedness. There are no vaccines for conflict, but perhaps the equivalent is preparedness. While FCE staff members “plan out the early response strategy for the respective issues at hand,” this is still reactive rather than preparative.

Related to this is Senevirathna’s definition, or stated purpose of early warning; the function of an early warning system is to send “the right information at the right time to the right people to take timely action for prevention of conflicts.”

I like this approach because it is simple and a good contrast to more traditional definitions which would focus on sending the right information to the right officials, thus sidelining civil society altogether. I also like the definition because of the link to timely action for prevention. What Senevirathna’s definition lacks however, is any reference to preparedness—which for me is a centerpiece of third generation early warning systems.

I was pleased to see Senevirathna tackle the issue of early warning for whom and for what. The author refers to Chapter 3 in which Kanno writes that “closeness to conflict areas enables one to understand the situation better and intervene rapidly and appropriately.” This important shift in discourse from mainstream approaches to conflict early warning.

In the past, the mind-set was that we could not draw on local field monitors because they would be too biased, which would skew the information collection process. Another thought was that local field monitors would be “too close” to the conflict, and like frogs in slowly heating water would not notice the increase in temperature. The chapters by Senevirathna and Kanno suggest a different and refreshing approach; one in which local communities are finally given the trust and respect that they deserve.

I was also pleased to read that a number of Embassies are involved in the FCE initiative—something I have not seen done in other third generation approaches. Representatives from the British High Commission, the Delegation of the European Commission and several others meet on a monthly basis with the FCE and other multi-lateral organizations as well as international NGOs. These meetings are mean to serve as an “advocacy forum somewhat similar to the approach taken by the Crisis Group.”

Senevirathna writes that the FCE’s “GIS tool has been very effective in generating conflict early warnings within the FCE system” but the author produces absolutely no evidence to back up this claim. My main question is whether the FCE actually leverages geospatial analysis for pattern recognition of conflict.

In closing, I completely agree with Senevirathna when he writes that “the success of every conflict early warning [and] early response system depends on how well it is customized to fit into the environment within which it will operate.”

Chapter 3: An Overview of Early Warning and Its Three Generations

UPDATE: I realize that I wrote up this review too  quickly while traveling between airports and did not emphasize how much I agree with Kanno. Instead, I mistakenly only focused on the sections that I disagreed with. While I realize that is the point of a critique, it is not a balanced review of Kanno’s chapter. So I highly recommend anyone interested in this field to read the chapter as there is a lot to learn from what Kanno has written.

Tadakazu Kanno authored the third chapter in the FCE book, which has some overlap with Chapter 2 by David Nyheim. While I do critique several sections of the chapter, criticizing is always easy and this chapter remains an important contribution to the discourse. Kanno provides a basis on which to continue discusssing third generation initiatives.

Kanno notes that the field of early warning and response has traditionally “been the purview of inter-governmental organizations such as the UN, EU [etc.].” Moreover, “early response in particular has been confined to the sphere of the UN and regional organizations.” He points to the post-election violence in Kenya as yet more evidence on the ineffectiveness of today’s sophisticated top-down systems.

Like David Nyheim, however, I obviously disagree with Kanno’s notion that third generation systems include those mechanisms implemented by regional organizations such as IGAD’s CEWARN. Again, I hope that this misunderstanding has to do with a lack of knowledge on CEWARN rather than confusion over what third generation systems entail.

Kanno writes that third-generation systems are “early warning and response systems of citizens, by citizens and for citizens.” This is not what CEWARN is since the initiative is first and foremost an inter-governmental mechanism. And yet, one page later, Kanno argues that there are two different kinds of actors in third-generation systems: local NGOs and non-Western regional governments. This is in my opinion a conceptual mistake and also a misunderstanding of what third generation systems are—citizen based.

Continuing with the confusion, Kanno argues that the “micro size” of third generation system “is the most important [point] to understand.” He writes that unlike previous generations, “actors of third generation EW/ER are smaller than those of the first and second” which means the main actors of third generation systems are “local NGOs and regional governments.” Am I the only one who sees this as contradictory? Local NGOs are civil society groups and hence citizen based. Regional organizations simply aren’t.

Kanno further argues that the information flow in third generation systems is “horizontal only from field officers to the Head Office within the conflict area.” CEWARN’s headquarters are in Addis, not in the Karamoja Cluster. And the information flow is anything but horizontal as I have argued in detail in this paper (PDF) back in 2007. See CEWARN’s organizational chart in network form below (taken from my 2007 paper). As the graphic denotes, CEWARN is hardly an example of horizontal information flow.

I was also disappointed by the definition Kanno users for early warning and response: Collection of data; Analysis of data; Formulation of best/worst scenarios and response options; and Communication to decision makers.” I had really hoped the discourse and literature on conflict early warning had progressed more than that. The above definition implies that a successful early warning system is one in which response options are communicated to decision makers, end of story.

Surely we can do better than this. Why not define a successful early warning system as one that is able to trigger an operational response? I don’t even set the bar higher by requiring that successful systems should trigger an operational response by decision makers *and local communities that either mitigates or prevention violent conflict; simply operation response. Surely we can adopt this definition.

In the section of first generation systems, Kanno writes that information collected by this generation of systems initially came from newspapers but that modern systems have “shifted to the much denser and more even coverage of on-line newswire sources such as Reuters.” This may have been true 5 years ago. We are today way beyond the use of simple key word searches of sources like Reuters, etc.

Today’s most sophisticated first generation systems employ advanced natural language processing algorithms to comb thousands of different sources in dozens of different languages. Some systems like Crimson Hexagon* go beyond standard simple event data and use sentiment analysis. Others are parsing SMS, Twitter, Flickr, etc., as well as new media content and user-generated content such as blogs. It is important that the conflict early warning community get up to speed on technology.

Kanno suggests that the shift away from first- and second-generation early warning systems can be explained by “generational trends in analytical methodologies.” In other words, Kanno argues that the development of analytical methods for conflict early warning explains the emergence of first- and second-generation systems. I would personally phrase this another way: the emergence of first- and second-generation systems is the result of Western Ivory Tower academics being hired as consultants to develop operational conflict early warning systems in developing countries.

I agree with Kanno that third generation early warning systems “are not the best systems to prevent macro-level political conflicts.” That said all politics is local, right? Local politicians are often responsible for fueling violent conflict that can quickly escalate. Still, I agree that current third generation systems do not “possess coercive power to stop impending and ongoing violence.” That is why I introduce the concept of fourth-generation systems, which specifically focus on preparedness and civil resistance.

To conclude, it is always easier to criticize but Kanno has produced an important chapter that enables us to continue the conversations and progress forward. The heart of my criticism is simply with respect to CEWARN but I realize that little has been written on the topic which makes it harder for colleagues who have not worked directly on the system to have a solid understanding of the project. Also, while I was critical of some sections, please note that I fully agreed with the vast majority of what Kanno wrote.

Chapter 2: Three Generations in Early Warning – Challenges and Future Directions

This chapter of the FCE book written by my colleague David Nyheim starts to wrestle with some of the challenges of conflict early warning systems across three generations. This is an important chapter that seeks to go beyond some of the commentary that already exists on the topic. At the same time, however, I found several sections in the chapter rather puzzling.

I was perhaps most troubled by the categorization of CEWARN as a third generation system. This is simply wrong and the categorization is unfortunate as it means one of two things: either my colleague is not familiar with CEWARN, or worse, there is confusion over what constitutes a third generation system altogether. I worked directly on the operationalization of CEWARN. This explains why it is crystal clear to me that this regional, inter-governmental early warning system is at best a second-generation system and in no-way a micro-level system.

The reason I feel somewhat strongly about “semantics” here is because CEWARN has only prevented 4 or 5 pastoral conflicts out of several thousand conflicts since 2003. This is hardly a success by any measure and clearly shows that simply having a “rapid response mechanism” does not make an early warning system successful. Calling CEWARN a third generation system is thus particularly unhelpful to those of us presenting third generation systems to donors as distinct from the failures of CEWARN.

Why is CEWARN not a third generation system? Let me count the ways. Writing that CEWARN has a micro-level focus is misleading. True, CEWARN employs field monitors and the system does not monitor entire countries but rather on specific sub-national areas. However, lets take the example of CEWARN’s three field monitors who cover Kenya’s Turkana District in the Karamoja Cluster. While the discourse labels them as “field” monitors, which suggests a “local” presence, these three monitors are tasked with monitoring events and processes taking place on a weekly basis across more than 70,000 square kilometers—equivalent to the landmass of Lebanon, East Timor and Burundi, put together.

David cites “real time monitoring” as one of the principal points of comparison between first-, second- and third-generation systems. The latter systems “access information on a real time basis” and “verify information quickly.” CEWARN’s information collection process is hardly real-time. When I left in 2005, we were still struggling to reduce the time lag between event and reporting from 3-4 weeks to 2 weeks. This is hardly real time. Even David himself notes (later in the chapter) that governmental and intergovernmental organizations with larger mechanisms were unclear—or did not share—about “how quickly or not responses were.”

Another important point of comparison between early warning systems across generations is “local ownership.” David writes that “with strong field-roots, third generation systems are sometimes able to ensure greater local ownership of warnings and responses than other systems.” This description hardly applies to CEWARN. The vast majority of pastoral communities in the cross-border areas have never heard of CEWARN because the system is hierarchical and centralized.

David also refers to “several success stories” reported by several conflict early warning initiatives like the FCE* and CEWARN. While commendable, these success stories in no way imply that the systems themselves are successful. Just because I happen to serve one ace in a tennis match does not mean I’m a successful tennis player. The question, rather, is this: how many aces do I serve on average per set and how does this figure compare to professional tennis players?

I do agree with David vis-à-vis the following: “It is important to stress that different generational systems meet different demands, institutional needs, and mandates—which means that all serve important current needs.” In addition, I also agree that, “there remains a significant accountability deficit for inaction or poor action in responding to violent conflict and state failure.” And David is right in arguing that, “more work can be done to empower institutional responders,” and “strengthen response delivery mechanisms.”

But lets be very clear that CEWARN (or other governmental/inter-governmental initiatives) are not third generations systems; they do not have local ownership.

In closing, I would suggest that more focus be placed on organizational theory when discussing the differences between first-, second- and third-generation conflict early warning systems. One should understand the implications of using “generations” as a metaphor to describe different early warning systems. For example, the term “generations” implies “evolution”, “learning,” “adaptability” etc., which is precisely why political scientists refer to different generations of warfare.

An important difference between various generations of warfare is the organizational element. Al Qaeda is far less centralized killing organization than any other in history. It is their network structure that enables them to adapt and respond quickly. If anything, we need to learn how such terrorist cells operate in order to apply the same efficiency and speed to the prevention of violent conflict. See this previous post for more details.