Category Archives: Uncategorized


Greetings! All content on this blog along with future blog posts on Early Warning are now available on my main blog iRevolution! Please adjust any bookmarks you may have accordingly and be sure to subscribe to iRevolution by email for information on the latest blog posts.



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.

Applying Earthquake Physics to Conflict Analysis

I’ve long found the analogies between earthquakes and conflicts intriguing. We often talk of geopolitical fault lines, mounting tensions and social stress. “If this sounds at all like the processes at work in the Earth’s crust, where stresses build up slowly to be released in sudden earthquakes … it may be no coincidence” (Buchanan 2001).

To be sure, violent conflict is “often like an earthquake: it’s caused by the slow accumulation of deep and largely unseen pressures beneath the surface of our day-to-day affairs. At some point these pressures release their accumulated energy with catastrophic effect, creating shock waves that pulverize our habitual and often rigid ways of doing things…” (Homer-Dixon 2006).

But are fore shocks and aftershocks in social systems really as discernible as well? Like earthquakes, both inter-state and internal wars actually occur with the same statistical pattern (see my previous blog post on this). Since earthquakes and conflicts are complex systems, they also exhibit emergent features associated with critical states. In sum, “the science of earthquakes […] can help us understand sharp and sudden changes in types of complex systems that aren’t geological–including societies…” (Homer-Dixon 2006).

The Model

To this end, I collaborated with Professor Didier Sornette and Dr. Ryan Woodard from the Swiss Federal Institute of Technology (ETHZ) to assess whether a mathematical technique developed for earthquake prediction might shed light on conflict dynamics. I presented this study along with our findings at the American Political Science Association (APSA) convention last year (PDF).

This geophysics technique, “superposed epoch analysis,” is used to identify statistical signatures before and after earthquakes. In other words, this technique allows us to discern whether any patterns are discernible in the data during foreshocks and aftershocks.

Earthquake physicists work from global spatial time series data of seismic events to develop models for earthquake prediction. We used a global time series dataset of conflict events generated from newswires over a 15-year period. The graph below explains the “superposed epoch analysis” technique as applied to conflict data.


The curve above represents a time series of conflict events (frequency) over a particular period of time. We select arbitrary threshold, such as “threshold A” denoted by the dotted line. Every peak that crosses this threshold is then “copied” and “pasted” into a new graph. That is, the peak, together with the data points 25 days prior to and following the peak is selected.

The peaks in the new graph are then superimposed and aligned such that the peaks overlap precisely. With “threshold A”, two events cross the threshold, five for “threshold B”. We then vary the thresholds to look for consistent behavior and examine the statistical behavior of the 25 days before and after the “extreme” conflict event.


For this study, we performed the computational technique described above on the conflict data for the US, UK, Afghanistan, Columbia and Iraq.

Picture 4Picture 5Picture 6

The foreshock and aftershock behaviors in Iraq and Afghanistan appear to be similar. Is this because the conflicts in both countries were the result of external intervention, i.e., invasion by US forces (exogenous shock)?

In the case of Colombia, an internal low intensity and protracted conflict, the statistical behavior of foreshocks and aftershocks are visibly different from those of Iraq and Afghanistan. Do the different statistical behaviors point to specific signature associated with exogenous and endogenous causes of extreme events? Does one set of behavior contrast with another one in the same way that old wars and new wars differ?

Future Research

Are certain extreme events endogenous or exogenous in nature? Can endogenous or exogenous signatures be identified? In other words, are extreme events just part of the fat tail of a power law due to self-organized criticality (endogeneity)? Or is catastrophism in action, extreme events require extreme causes outside the system (exogeneity)?

Another possibility still is that extreme events are the product of both endogenous and exogenous effects. How would this dynamic unfold? To answer these questions, we need to go beyond political science.

The distinction between responses to endogenous and exogenous processes is a fundamental property of physics and is quantified as the fluctuation-dissipation theorem in statistical mechanics. This theory has been successfully applied to social systems (such as books sales) as a way to help understand different classes of causes and effects.

Our goal is to use the same techniques to investigate the questions: Do conflict among actors in social systems display measurable endogenous and exogenous behavior?  If so, can a quantitative signature of precursory (endogenous) behavior be used to help recognize and then reduce growing conflict? The next phase of this research will be to apply the above techniques to the conflict dataset already used to examine the statistical behavior of foreshocks and aftershocks.

Genocide Areas of Concern

The Genocide Intervention Network (GI-NET) recently released it’s end of 2008 summary on Areas of Concern. The report (PDF) describes the analysis performed in early 2008 and updated throughout the year.

Areas of Concern

Darfur, eastern Chad, Central African Republic, eastern DR Congo, Somalia, Iraq, eastern Burma, and Sri Lanka.  Areas that may be added to the Areas of Concern list next year include North Korea, Ethiopia’s Ogaden region, and Afghanistan.

On Radar

The  “radar” list contains areas we are looking at closely but that do not currently constitute Areas of Concern. These include: Cote d’Ivoire, Ethiopia (Ogaden), Afghanistan, Pakistan (especially Waziristan and Baluchistan),  North Korea, Kashmir, Tibet, Indonesia, Chechnya, Ingushetia, Dagestan, Abkhazia, South Ossetia, Kordofan and South Sudan,  Zimbabwe, Nigeria, Israel-Palestine, Burundi, Kenya, Uygher areas of China (Xinjiang), Colombia, Niger/Mali, Orissa (India) and Northern Uganda.

Abeyance List

Areas of Concern denote countries and territories that are currently experiencing massive violence. Areas where violence subsides for six months or more are placed on “abeyance” if it is appropriate to remain engaged/watchful of these situations until a re-emergence of conflict is less likely. Currently no areas are on abeyance, as we are just finishing the first cycle of the analysis. In our next cycle, areas likely to be placed on abeyance include Central African Republic, and possibly eastern Chad.

Detecting Rumors with Web-based Text Mining System

Robert Kirkpatrick at InSTEDD pointed me to a very interesting public health project out of Japan called BioCaster, an ontology-based text mining system that uses linguistic signals on the Web for the early detection and tracking of infectious disease out-breaks.


“The system continuously analyzes documents reported from over 1,700 RSS feeds, classifies them for topical relevance and plots them onto a Google map using geocoded information. The background knowledge for bridging the gap between layman’s terms and formal coding systems is contained in the freely available BioCaster ontology which includes information in eight languages focused on the epidemiological role of pathogens as well as geographical locations with their latitudes/longitudes. The system consists of four main stages: topic classification, named entity recognition (NER), disease/location detection and event recognition. Higher order event analysis is used to detect more precisely specified warning signals that can then be notified to registered users via email alerts. Evaluation of the system for topic recognition and entity identification is conducted on a gold standard corpus of annotated news articles.”

BioCaster has specific advantages over related initiatives like GPHIN, MedISys, Argus, ProMedMail, EpiSpider and HealthMap. I’ve blogged about these initiatives here and here but BioCaster combines the following functionalities within a single system

  1. Text mining techniques such as entity recognition which aim to generalize to previously unseen terms and expressions;
  2. Text-level recognition of severity indicators such as international travel or the contamination of blood products;
  3. Ontology-based inferencing to fill in the gaps, e.g. between a mentioned pathogen and the unmentioned disease that caused it or between symptoms and diseases;
  4. Direct knowledge of term equivalence within and across languages.

The system has been operational since 2006 and offers “an intuitive mapping interface [see above] for the general reader as well as an openly available ontology for community re-use. Future work will focus on extending coverage to new languages and public health threats. A paper on BioCaster is available here.

I’m very interested in this system and would really like to apply the methodology to early detection and tracking of conflict rurmors. See this post for more on early warning and natural language parsing.

A Conversation on Early Warning with Howard Adelman

Professor Howard Adelman kindly shared some interesting insights (via email) in response to (1) Michael Lund’s new chapter on conflict prevention, and (2), my reaction to it (see previous blog entry). In his response, Professor Adelman also drew on a number of my other blog entries, which I greatly appreciate—starting a conversation on early warning was exactly what I was hoping to do with this blog.

What follows are some reactions to Professor Adelman’s email. I’ve chosen to “reply by blog” as opposed to email in order open the conversation to others who might wish to contribute. I want to keep this blog entry at readable length (i.e., under five minutes) and will therefore be biased in selecting the issues I respond to. Professor Adelman is certainly invited to share additional thoughts via the comments section.

Reading the entries on the blog and Michael’s chapter suggested to me that there is  confusion over the relationship between conflict prevention and early warning. […] Early warning not only includes the gathering of data but the analysis of that data to develop strategic options for response but does not include the responses themselves which come under conflict prevention.

Whether early response should be filed under conflict prevention or some other term is perhaps more a question for academics. I do realize the importance of having clear definitions and sharp conceptual frameworks. However, I’m more preoccupied with early response actually happening at all, regardless of which toolbox it belongs to.

Patrick observes that CEWARN’s methodology, like the majority of intergovernmental systems gets rather technical, institutional and bureaucratic very quickly, it is unclear whether he is pointing out to a structural flaw or a propensity because the system has strong governmental links. Though he is correct that, “It is easy to forget the human element of early warning when faced with fancy language such as baselines, trends analysis, structural indicators,” it should be noted that the few early successes of the system did not come from the highly developed technical side but from the very personal reporting side of those individuals gathering information before it was subjected to systematic extrapolations. Nevertheless, the systematic framework allowed the observer to ask the right questions and look for the data that revealed an impending crisis.

In my view, the structural flaw of CEWARN is the system’s strong governmental links. This is why the few early successes of the system did not come from the highly developed technical (or data-driven) side but from the personal reporting side of those individuals gathering information before it was subject to systematic extrapolations and institutional inertia.

I find it particularly telling that CEWARN’s first success story occurred in July 2003, barely a month after the system went operational, which is when I first joined the CEWARN team. None of the institutional or highly technical procedures were in place at the time so when a CEWARN field in monitor called a country coordinator to alert him that an armed group was mobilizing to raid another group’s cattle, the communication of this information to CEWARN was all done ad hoc, right through to the early response. I am skeptical that institutionalizing effective early response is possible. In fact, I see it as an oxymoron. To find out why, please see my ISA paper on new strategies for early response (PDF).

CEWARN and other such systems are intended to involve communities at the grass roots level to sideline the source of violence and initiate processes that will keep them sidelined. Further, the Ushahidi approach involving peer-to-peer, networked communication tools was not that different than the networking design and open information system at the base of the CEWARN system.

I disagree. CEWARN and Ushahidi are hardly similar or comparable, either in design or in operation. CEWARN is not an open information system by any measure—the project’s incident and situation reports are not open to the public. The online CEWARN Reporter is password protected, only the CEWARN team and select government officials have access. In fact, CEWARN’s design is an excellent example of anti-crowdsourcing. CEWARN’s network design remains far more centralized than Ushahidi’s can ever be; not least because the source code of Ushahidi will be made available freely to anyone who wants it. If there is one similarity between the two systems, it has to do with the fact that both projects need to focus far more on operational and tactical early response.

Patrick’s argument is akin to saying that when we see certain kinds of spots on the skin we know the child has measles, so why do we need greater in-depth analysis for detecting patterns of spread or for detecting the disease even before the spots appear on the skin.

Close. Why do we need greater in-depth analysis when this analysis will be sent to a hospital a thousand miles away for further analysis and not result in any response by public health professionals who have no incentive to respond? Why not train the parents directly to deal with the measles instead?

The CEWARN and WANEP systems were deliberately designed to be frugal operations rooted in community-based gather of information and data with the analysis located in the state and the region of the conflict.

Why are we not designing systems rooted in community-based early responses? Why ask communities to code data that is ultimately of limited use to them?

Alex de Waal’s depiction of the documentation provided in Sudan that allowed villagers to evacuate is but one example of one end of the early warning spectrum but does not obviate the need for more developed systems. However, Patrick’s message needs to be heeded: the latter should not be developed at the expense of community-based systems such as GI-NET in Burma using a civilian radio network to enable civilians to receive and send warning information and distress calls.

I would add that more sophisticated systems need to demonstrate cases of operational response (particularly since these systems tend to be expensive to fund). Note that I’m not even raising the bar to successful cases of operational prevention. Just responses, that’s all.

CEWARN reported more than 3,000 conflict events in the first three years of operation but has only responded to a dozen at most. That’s a “success” rate of 0.4%. On the other hand, the system can be assessed using other measures. For example, the project has successfully documented extensive evidence human rights abuses, which has forced governments to acknowledge that a problem exists and to start taking responsibility for that problem.

Recall when the CEWARN team reported its first year of data to government officials in Addis Ababa (you and I were both there, Professor Adelman). The government representatives were so taken aback by the extent of the violence taking place in cross-border regions that they refused to release the country reports (in direct violation of the CEWARN protocol which had been ratified). They eventually did release the reports six months later and by doing so have acknowledged there was a problem, which is a critical first step.

Conflict Early Warning in Central America

I just gave a keynote speech in Guatemala as part of a week-long conference on developing capacity for a regional conflict early warning and response system for Central America. The conference is supported by the European Center for Conflict Prevention (ECCP) and the Global Partnership for the Prevention of Armed Conflict (GPPAC). The gathering brought together civil society groups from across Central America and South America.

My presentation focused on human early warning systems. But I began the talk with a brief overview of CEWARN‘s methodology, which like the majority of intergovernmental systems gets rather technical, institutional and bureaucratic very quickly. It is easy to forget the human element of early warning when faced with fancy language such as baselines, trends analysis, structural indicators, etc. Indeed, it’s easy to forget that today’s sophisticated early warning systems are relatively new mechanical inventions, which begs the question, how did people manage before these systems were available?

The answer is that they managed, they had to. I took up the example of Guatemala and El Salvador during the 1980s. Populations caught in between military operations and rebel activities found ways to survive. Tens of thousand lived undercover, moving only at night, building extensive underground tunnels, growing hidden gardens and carrying out regular drills to practice rapid evacuations. They would set off firecrackers to warn neighboring villages about incoming military fighter jets. These survival stories are not unique to Central America, hundreds of similar stories can be found across Africa and Asia.

Local communities across the world do not have recourse to sophisticated early warning systems, but they survive, by monitoring their own (often less tangible) indicators and by prioritizing preparedness. Surely, the human being is one of nature’s most phenomenal early warning systems, tried and tested by evolution over millions of years. Why forgo this remarkable system completely for more technical, mechanical systems and bureaucratic structures that are not “naturally” designed for early warning and response?

The disaster management community has already recognized the importance of people-centered early warning (as opposed to system-centered, or data-centered). The purpose of people-centered early warning approaches is to empower communities at risk to get out of harm’s way. This empowerment is achieved through preparedness and contingency planning. We often hear about disaster preparedness and disaster mitigation, risk reduction, etc. Why don’t we hear about preparedness in the context of conflict early warning and prevention?

I presented two case studies to outline examples of people-centered conflict early warning/response projects. The first is a new initiative out of Timor-Leste, which specifically focuses on conflict preparedness  at the community level. The project, which I worked on in February, seeks to outline detailed, local contingency plans for early response at the community level. Conflict resolution and mediation skills are integral to carrying out these responses when conflict does escalate. Training in conflict mediation is therefore critical, and even more valuable when linked to specific contingency response measures. The second project I presented is Ushahidi in Kenya and emphasized the novelty of taking a crowdsourcing approach to crisis information by drawing on new information communication technologies.