THE PREVENTION OF ARMED CONFLICT IN ESTONIA 1993
1.3 Early warning
With these general and epistemological considerations in mind, let us now turn to consider the contemporary effort to establish an early warning system for violent political conflict, along the lines of Boulding's proposed 'social data stations' which he saw as analagous to networks of weather stations in the identification of 'social temperature and pressure' and the prediction of 'cold or warm fronts'. This is widely seen as essential for monitoring particular areas of potential conflict, and seeking ways to act early enough to nip a potential conflict in the bud where this is feasible and appropriate. There are two tasks involved here: first, identification of the type of conflicts and location of the conflicts that could become violent;
second, monitoring and assessing their progress with a view to assessing how close to violence they are.
One line of approach, which addresses Suganami’s second question, aims to establish the circumstances under which wars are likely to take place. We can take Ted Gurr’s work as an example of this approach.
Using data from his Minorities at Risk project, he identifies three factors that affect the proneness of a communal group to rebel: collective incentives, capacity for joint action, and external opportunities (see Box 30, page X). Each concept is represented by indicators constructed from data coded for the project, and justified by correlations with the magnitude of ethnic rebellions in previous years. The resulting table makes it possible to rank the minorities according to their risk-proneness (Gurr 1998a). The assumption is that the more risk-prone are those with high scores on both incentives for rebellion and capacity/opportunity. The table shows, for example, that the Kosovo Albanians have high incentives to rebel but a lack of capacity and opportunity; the East Timorese on the other hand have both incentives and capacity.
This is a political science version of the methods used in econometric forecasting. Like them, it may yield results in the short-term, though the technique obviously blurs the case-specific and context-specific information which area experts would use. If it turns out that this approach yields acceptably good forecasts, it may be possible to offer conflict prevention agencies useful information about where to
concentrate their efforts. Variations in Gurr’s indices could also be used as indicators of effectiveness of conflict prevention policies.
A similar approach, using a different starting-point, is taken by the Dutch conflict monitoring organization, PIOOM. Their studies assess risk of armed conflict using indicators of human rights violations and poor governance. As described in chapter 1, section 2, they use a five-phase model to classify countries on a scale ranging from a peaceful stable situation, through political tension, violent political conflict, low intensity conflict and high intensity conflict, and thirteen indicators of conflict escalation (Schmid 1997:74). For forecasting purposes, their work is trend- based, in that the countries with political tension or violent political conflict now are expected to be sources of armed conflicts in the future.
Barbara Harff examined a number of contemporary conflicts, including some in countries that have experienced political violence and ‘controls’ in countries with similar ethnic situations that did not experience violence (Davies, Harff and Speca 1997). Her study used the concept of ‘accelerators’ and
‘decelerators’: accelerators are events that escalate the conflict, decelerators events that dampen it, although the study under discussion only reports accelerators. Based on a coding of events reported in Reuters World Service, she plots the number of accelerator events per month before war for each of the ethnic conflicts, with a comparison for the control over a similar period. In each case of conflict that led to a war, there was an intensification of the number of accelerator events in the three months preceding the war. The implication is that similar coding schemes might offer an early warning of conflict, by reporting on the intensity of events. The basic assumption is that trend extrapolation can be used to measure the intensity of political conflict.
An ambitious version of this approach is the Global Event-Data System (GEDS) project which aims to provide near-real-time automated coding and monitoring of on-line news services, yielding a quantitative trace of the level of tension in ongoing conflicts.xxxvi
‘Enduring rivalries’, that is, protracted disputes between pairs of states or peoples, have accounted for half the wars between 1816 and 1992. These may be expected to be sources of further disputes. It is not difficult to point to regions - such as West Africa, the Great Lakes region of Africa, the Caucasus, the India-Pakistan border, parts of Indonesia, where future violent conflicts can be expected. It is less easy, however, to anticipate wholly new conflicts, still less new types of conflict.
Turning from quantitative to qualitative conflict monitoring, a mass of information is available on particular societies and situations. It includes the reports of humanitarian agencies (now linked together on the ReliefWeb site on the Internet), e-mail early warning networks of conflict monitors (for example, in the former Soviet Union), analyses by the media and by the academic community, and of course the diplomatic and intelligence activities of states. Efforts are underway to improve and systematise these qualitative sources of information and to make them available to those who could undertake a response. Qualitative monitoring offers vastly more content-rich and contextual information than quantitative statistical analysis, but presents the problems of noise and information overload. Given the current state of the art, qualitative monitoring is likely to be most useful for gaining early warning of conflict in particular cases: the expertise of the area scholar and the local observor, steeped in situational knowledge, is difficult to beat. In some cases, observers clearly realised that violent conflicts were coming well before they occurred: for example, in former Yugoslavia and Rwanda. In others they were taken by surprise. Even when observers have issued
‘early warnings’, it is by no means certain that they will be heard, or that there will be a response.
Governments and international organizations may be distracted by other crises (as in the case of Yugoslavia), or unwilling to change existing policies (as in the case of Rwanda). Given the unpredictability of human decision-making, no system of forecasting is likely to give certain results. Nevertheless, there is already sufficient knowledge of situations where there is proneness to war to justify perseverence in international efforts to provide data which might enable early and timely preventive response.