this example shows, even an established NGO such as ACORD faced challenges and a degree of vulnerability. The necessary authorisations were not always sufficient to gain access to individuals and places, or to address certain issues. Fluctuating security made travel outside the immediate vicinity of Gulu town unpredictable, and the context as a whole created high levels of suspicion, not just of me as an obvious foreigner, but also of colleagues whose personal political affiliations were not known but could put respondents at risk. Unlike an idealized Clausewitzian situation in which good and bad guys are clearly identifiable, here there was the official position on the ‘bad guys’ (the LRA), but no agreement on the ‘good guys’ – or if there even were any. Some people did not even see the LRA as primary protagonists, but rather as mere proxy fighters in a war between the Governments of Uganda and Sudan (see Chapter 4).
Possible respondents had many interpretations of our role as researchers, including that we were spies for foreign governments, or might feed our findings to the Ugandan government, or alternatively were rebel sympathisers ourselves. It was often extremely difficult to judge where a given respondent sat on the political spectrum, and how information flows worked. It was also the case that both respondents and co-researchers were likely to have occupied several different roles in their lives (e.g. civilian, soldier, rebel), each with different political overtones. And it was not uncommon for a person to have different close family members in several different political camps. They might have a brother in the bush, a sister in local government, an uncle in central government and cousins in London, Toronto or New York.
These high levels of ambiguity about identities, affiliations and political sympathies, created serious security/confidentiality concerns for researchers and respondents alike. This was particularly the case when researching the LRA, for even to ask a question about them in Gulu was liable to raise suspicion (or fear). Not only was documentation scarce, but to attempt to talk with the LRA could be viewed by the authorities as tantamount to collaboration – as The Monitor newspaper noted, ‘When Presidential candidate Paul Ssemogerere told voters in 1996 that he would talk to Kony into abandoning the rebellion, he was branded a rebel himself’1. Even those who had been authorised to make contact with the LRA were at times arrested by the UPDF, as happened when three priests, carrying a letter from the Kitgum Resident District Commissioner, tried to meet with LRA Commander Toopaco on 28 August 2002.2 Individuals offering to make contact with the LRA on my behalf often turned out to be linked with government security services. I could have interviewed returned child abductees in centres established for their rehabilitation, such as GUSCO, but I was reluctant to follow this well-trodden path. Just as Heike Behrend had observed that well before she wrote her own analysis of the Holy Spirit Movement, it ‘had already been created by the mass media’ (Behrend, 1999: 2), in the case of the LRA it was created not only by mass media but also by the UN and Non-Governmental Organisations using the testimonies of traumatised children who had escaped captivity and been processed through reception centres such as GUSCO and World Vision. While the testimonies seemed in many respects genuine, they provided a partial picture, as adult voices were largely absent. There were though few other obvious options. At the same time, the LRA were said to have eyes and ears everywhere – at any public meeting someone would comment to the effect that ‘whatever we say here will be relayed to the LRA’ – but nobody would go so far as to point out particular individuals. As a result one could never be sure who was who and had to regard everyone as a potential informer to Government or LRA – or to both.
The sense of being under surveillance was at times acute; it was not uncommon to meet a respondent for a meal or drink and then find somebody unknown sitting exceptionally close by, despite the availability of numerous empty tables further away. On one occasion in Kitgum, I spotted such a ‘restaurant observer’ at ten o'clock the following morning sitting 100 metres away from the ACORD office on a termite mound. From there he could observe our comings and goings while reading a book. Such scrutiny from security services meant that some respondents could also be at risk from being seen to talk with us, and raised serious concerns about the confidentiality of sources and security of data collected. Returning to my hotel room in Kitgum after a day's fieldwork to find that my laptop had been tampered with brought this home to me in a very direct fashion. Furthermore, for much of my fieldwork there were only a handful of telephones in Gulu town; any call from the post-office was bound to be listened to by the queue of other people waiting, and any fax received was likely to be perused by several people before reaching its intended recipient. This also led to a form of self-imposed censorship in discussing what was happening.
Conceptual Challenges
I felt I had to go well beyond simply documenting the impact of war and humanitarian interventions on people at a local level. I did not wish to simply repeat data collection that had already been done; some issues, such as LRA abductions and other atrocities already seemed relatively well documented – indeed they seemed amongst the few elements of the situation which did not require much research.
Furthermore, I felt that such documentation had a limited impact on the situation. From a humanitarian perspective, given the multiple indicators of suffering which already existed when I first went to Gulu in January 1998, northern Uganda should have been termed a complex emergency, yet it was not recognised as such, and the humanitarian imperative did not seem to be operating. Eventually – in November 2003 – it would come to be described by Jan Egeland, the UN's Under Secretary General for Humanitarian Affairs, as one of the worst humanitarian situations in the world3 but in 1998, despite being already more than a decade old, the situation in northern Uganda had attracted relatively little international attention. From day one, therefore, I was confronted by questions about whether the label had any objective meaning (i.e. was linked to specific ‘objective’ indicators) or was purely politically contingent.
Horizontal Segmentation or Vertical Linkages?
I was also challenged by the way the larger COPE research project was effectively segmented horizontally: Officially I was supposed to focus at the local level, while others (in line with their academic disciplines – political science and international relations respectively) prioritised state and international dimensions of CPEs in their fieldwork. Although in principle each member was intended to also consider the questions identified by those working at the other two levels, in practice there was little space in this project structure for considering the linkages between levels. I found this limited and limiting, but reflected in the majority of NGO reports and policy documents; while the premise of internal wars had effectively penetrated this grey literature, the more subtle linkages between internal and external, such as are found in Kaldor and Duffield's discussions on new wars, generally had not. This suggested to me that a purely ethnographic approach to understanding the dynamics of war and its continuation would be irrelevant, and that even while conducting fieldwork at the ‘local’ level it was necessary to examine the connections between different parts of the conflict dynamic and different parts of the globe. To do this also indicated the need to go beyond the confines of traditional academic disciplines; to get away from the economist's preoccupation with economic rationality to look at more politically and psychologically complex models of behaviour and motivation; to discard international relation's fixation with two-party models of wars and their resolution and replace it with multi-actor models; and to take the anthropologist's concern with local level motivations and ideologies and integrate it with more systemic perspectives. Against the backdrop of these needs I felt that development studies, as an academic setting which embraces multi- or cross-disciplinary perspectives, was an appropriate ‘home’.
I was encouraged in the pursuit of connections across levels (and disciplines) by Colson and Kottak's writings on linkages (1996). They observe that ‘Contemporary anthropologists can no longer even hope to do ethnography among people isolated from world markets or unaffected by centers of political and economic power’. Furthermore, ‘…No matter what the subject or the research locale, we need to consider documents describing the interdependencies between local systems and larger economic and political networks’ (Colson and Kottak, 1996:
