Surveillance and the African State

Surveillance as an academic topic is invariably bound up with Western governance. Much less is known about the experiences of former colonies and postcolonial states, especially in Africa, where the lack of historical analysis is most apparent. Exceptions here include Young and Turner’s (Reference Young and Turner1985) work on Zaire’s security forces and Schatzberg’s (Reference Schatzberg1988) work on Zaire’s Centre National de Documentation (CND)—the successor to the Belgian colonial Sûreté—and the environment of intimidation and fear it created through its local networks of informers. Similarly, Mamdani (Reference Mamdani1983) has explored the operation of the State Research Bureau (SRB) in Idi Amin’s Uganda and its effective deployment of informers and intimidation. Several more recent studies turn away from the organs of the state to investigate local and everyday manifestations of surveillance at a particular historical juncture, from passbooks in apartheid South Africa to checkpoints in contemporary Eritrea (Breckenridge Reference Breckenridge2005a, Reference Breckenridge2005b; Bozzini Reference Bozzini2011; see also Purdeková Reference Purdeková2011a; McGregor Reference McGregor2013). Despite being few in number, the available studies do nonetheless offer a glimpse of the rich field of inquiry that remains largely untapped.

But the gap to be closed is conceptual, too. Unlike the works that precede it, this study seeks to understand what drives the layering, intensification, redeployment, and repurposing of “mundane” surveillance technologies over the longue durée. A number of key questions emerge: How should we theorize surveillance in Africa if we want to carry out an analysis across epochs, and one that does not begin with colonialism or treat surveillance as an “import” from the West? What drives the evolution of state surveillance over time? What explains continuities in social monitoring across a century of turbulent change and despite experiences of major atrocity abetted by it? Finally, how do we make sure that by foregrounding systems of “capture” we do not overlook forms of escape?

The Rwandan state is unique in the context of Africa not only in terms of its control of the periphery, its vertical reach, and the density of its structures (Purdeková Reference Purdeková2015, Reference Purdeková2011a; Ingelaere Reference Ingelaere2014), but also the essentially indirect workings of its political power. Power works through “subtle forms of intimidation” (Jessee Reference Jessee2013) and the state’s embedding of itself in the everyday. This article aims to contribute to a better understanding of this indirect and subtle governance modality as well as its development and transformation across time. Historical work on the Rwandan state has traced the gradual territorial extension of state control and elaboration of complex administrative and extractive systems, placing much emphasis on the transformation and rigidification of ethnic identity and the resulting entrenchment of inequalities, political exclusion, and conflict over time (Vansina Reference Vansina2004; desForges 2011; Newbury 1988; Jefremovas Reference Jefremovas2002; Thomson Reference Thomson2013). While ethnicity remains an important frame for understanding Rwanda’s political history, this article shows that the story of surveillance might well represent a key continuity that cuts across Hutu- and Tutsi-dominated epochs.

Further conceptual work can also help us identify the limited effectiveness of surveillance in many contexts. The topic of surveillance often invokes the notion of “capture,” as in being caught up in the modes of both state care (i.e., its benevolent intentions to foster the well-being of its citizens) and coercion. Yet existing empirical work (Kelly Reference Kelly2006; Jeganathan Reference Jeganathan, Das and Poole2004; Bozzini Reference Bozzini2011) qualifies the effectiveness of surveillance in reaching, sorting, and identifying people. Surveillance here is not understood simply as “forms of watching,” but also as a “sphere of vision” in which seeing is complemented by being seen and in which the experience of being seen—whether real or suspected—affects the performance of those under state watch. “Sphere of vision” thus differs subtly but importantly from Foucault’s “sphere of visibility” (1975). Even as Foucault spoke of surveillance’s decentering through individual embodiment of the gaze and the resulting self-directed censure, his approach has remained fundamentally one-sided, offering little scope for resistance or co-authorship (only allowing for the latter inasmuch as one imbibes the gaze). Yet the defining feature of Rwanda’s everyday surveillance is not simply visibility as the scope and ability to oversee, but rather the interface between ways of seeing and being seen.

This framework better captures the fundamentally “interactive” nature of the technologies under study here. Interactive surveillance is one that fundamentally depends not only on the “presencing” of agents of the state (or state-aligned or state-associated actors), but also on the “presenting” of tokens by those being surveilled, in the form of cards and permits, information for censuses, taxation, displays of loyalty or a friendly roster, or attendance at meetings. In all of these interactions, there is scope for play and subversion. Yet the frame of interaction also poses some fundamental questions: Does interaction imply a balance of powers, outmaneuvering on the part of citizens and state “impotence” (Mbembe Reference Mbembe1992)? Or does the state still hold the upper hand? This will be studied through an analysis of the exactions, duties, and emotive states that the Rwandan state is able (or unable) to extract with the help of its surveillance technologies.

The above-outlined conception of surveillance also complicates a simple division and opposition between visibility as capture and invisibility as escape. Seeing does not easily translate into knowing. On the one hand, visibility itself can produce misreading and be manipulated and subverted. On the other hand, invisibility is not the exclusive domain of exit and escape. As will be shown, it is often combined with surveillance itself in the exercise of state power. States can fudge, obfuscate, and disappear, and they can reproduce themselves politically or economically through such manufacture of the unseen (see, e.g., Newbury Reference Newbury1984 on Zaire) or the haunting of the missing.

The framework of “mundane sights” captures the operation of state power in Rwanda particularly well. But by the same token it diverges from debates over the “control society” and studies of modern surveillance focused on high-end electronic technologies, databases, biometrics, and CCTVs, or assemblages of vision operating across social arenas and geographical borders (Lyon Reference Lyon2007; Haggerty & Ericson Reference Haggerty and Ericson2000). The focus here lies squarely on “interactive” technologies of surveillance embedded in the flow of everyday life. Similarly, the focus is not on special sites of surveillance such as the prison or paradigmatic technologies such as the “panopticon” (Foucault Reference Foucault1975) or “banopticon” (Bigo Reference Bigo and Lyon2005). Quite the opposite—what is of interest is the “non-special,” the mundane and the everyday. Hence the chosen focus on local administrative “overseeing,” ID cards and identification, and state-mandated public activities as ceremonies of being seen and posing for view.

Importantly, though the analysis that follows is anchored in specific technologies, the idea of “mundane sights” is meant to capture a broader state of being. From a lived perspective, surveillance is a more generalized state and an overall effect of state presencing, and it is not reducible to or perceived as a “set” of technologies or specific “sites.” Today, Rwanda as “surveillance state” is less a geographic entity or a physical composite than a state of being, a community of affect.

The focus on everyday and “interactive” technologies also suggests a particular contribution to debates on African bureaucracy. Rather than looking at political economies of African bureaucracies as a whole (see, e.g., Berman 1984; Goldsmith 1999; Mamdani Reference Mamdani1996; Heussler 1963; Bierschenk & Olivier de Sardan Reference Bierschenk and Olivier de Sardan2014), I am interested in investigating “daily governance” (Blundo & Meur Reference Blundo and Meur2009) and “street level bureaucracy” (Lipsky 1980, quoted in Olivier de Sardan Reference Olivier de Sardan, Blundo and Meur2009:56). The objective is not to unearth genealogies of state institutions in Rwanda, such as the security services. The analytical agenda is closer to Olivier de Sardan’s call for the study of “actual daily functioning” (2009:39) of “states at work” (Bierschenk & Olivier de Sardan Reference Bierschenk and Olivier de Sardan2014). The project nonetheless diverges from this agenda inasmuch as it does not offer a close-up study of bureaucrats at work, and in fact, bureaucracy as a term invokes official posts and structures; the term much more appropriate to the study of surveillance in Rwanda is apparatus. The article also focuses on the “in-between,” the space where contact takes place and vision gets fuzzy. As such, the main aim here is to outline the local apparatus of surveillance and demonstrate, at least in part, its interactive dimension—what is actually captured and produced (politically speaking) through the sphere of vision.

Surveillance technology can be put to different uses by the state; hence change in surveillance is conceptualized here through the lens of the changing nature, tasks, and roles of the Rwandan Leta (a Kinyarwanda term derived from the French l’etat). ^{Footnote 4} Continuity in surveillance, on the other hand, can be found in the ambiguity that lies at its core. State surveillance can be narrated as either beneficial (i.e., as reach devoted to the care of the citizenry) or harmful (i.e., as repression or oppression, as overreach). From the perspective of care, visibility is essential; that is why the medical journal The Lancet can speak of a “scandal of invisibility” in which an “absence of reliable data . . . renders most of the world’s poor as unseen, uncountable, and hence uncounted” (Setel et al. 2007:1569). Invisibility here is read as a form of structural violence. At the same time, states, especially authoritarian states, often both care for their citizens and target them. They can deploy visibility and monitoring as repressive strategies. The article will show how the ambiguity of surveillance has allowed successive Rwandan governments to reappropriate this state asset with ever-increasing vigor and despite its imbrications in a past of exploitation and violence.

Overseeing: Administrative Reach and Density

The most striking story of social monitoring in Rwanda begins well before colonial rule. The story is one of surveillance as “overseeing” of local life through administrative presence. The administrative structures of the Nyiginya kingdom (ca. 1650–61) were consolidated gradually, increasing both in geographical scope, local reach, and complexity. ^{Footnote 5} They primarily served the purpose of extraction of corvée labor, the collection of ikoro (taxes), and the cooptation of notables into the system through distribution of spheres of control. Historically, both seeing and being seen have been key principles in the exercise and maintenance of central power. Then and now, presencing rather than simply overseeing has formed an important “state effect” (Mitchell Reference Mitchell and Steinmetz1999) in the central state’s feedback loop of control.

Importantly, even at the height of precolonial expansion, the central court and its delegates were hardly “all-seeing” and in many areas visibility was dispersed among small islands of control. ^{Footnote 6} While “Rwandan royal traditions portray an ancient royal kingdom,” recent historiography focuses on the dynamics of patchy authority, gradual expansion, varied resistances, and the “shifting political field of constant negotiation and competing loyalties” that characterized the Rwandan court at the time of European contact (des Forges Reference Des Forges2011:xxiii; see also Vansina Reference Vansina2004).

As des Forges observes (Reference Des Forges2011:101), the presence of the state was uneven and authority “varied from region to region, and sometimes hill to hill.” The kingdom was historically strongest in the central areas of today’s Rwanda, expanded outward through a set of military expeditions, and consolidated control over areas in the north only with the help of the European powers in the first decades of the twentieth century (Vansina Reference Vansina2004; Lemarchand Reference Lemarchand1966; Des Forges Reference Des Forges2011). Full occupation was achieved only in 1931 (Reyntjens Reference Reyntjens1987).

Despite the dynamic and uneven nature of state expansion, administrative innovation and imposition were key to the successful consolidation of power. The style of centralization was based on direct court appointments of local representatives and overlapping authority structures at the local level. The analysis here is then principally concerned with the reach, density, and style of administration and how these evolved over time. Such an analysis will allow us, in turn, to assess whether (and if so, how) postgenocide Rwanda is unique in terms of administrative surveillance of local spaces.

In the early nineteenth century expansion of the kingdom brought with it the development of new administrative norms. A series of three overlapping authority structures were established in each district under the kingdom’s control: the batware (army leaders with powers over conscription and taxation), banyabutaka (responsible for land grants), and banyamukenke (responsible for pastureland). All of these individuals were delegates of the court, suppressing and replacing prior authorities (lineage heads or chiefs). The overlap assured a “more complete court authority,” and “with each delegated authority overseeing the actions of his colleagues” it also prevented any particular delegate from accumulating excessive power (des Forges Reference Des Forges2011:7). Below this intermediary layer, power was delegated to the hill chiefs, who “in turn appoint[ed] a group of petty functionaries called ibirongozi (from Swahili: supervisors, overseers) to act as intermediaries” (Lemarchand Reference Lemarchand1966).

The reach of the state was thus intricate. It is estimated that before the arrival of the Europeans in Rwanda there were about eighty district authorities and a total of about two thousand to three thousand hill authorities in this rather small territory (whose population was no greater than 1–2 million in 1900). While the German colonizers left the precolonial administration largely intact (Rumyia Reference Rumyia1992), the Belgian administration reduced the number of local authorities between 1926 and 1932 in an attempt to simplify administration. Provinces and hills were regrouped, and the “trinity of chiefs” was replaced with a single one. In 1932 there were 1,043 subchiefs with 343 taxpayers on average (Reyntjens Reference Reyntjens1987). In 1948 Rwanda was divided into fifty chiefdoms and 630 subchiefdoms (sous-chefferies) (Codere Reference Codere1973). Biographies of chiefs (abashefu) and hill chiefs (abasoushefu) collected at the cusp of independence by Helen Codere show that the population overseen by each subchief was nonetheless still small. Sous-chef Ruhaniriza, a sous-chef at Ngarume between 1935 and 1940, reminisced that in 1935 he had overseen 550 taxpayers in Nyarure, “but that was more than at Ngarume where I only had 350 able bodied men [hommes adultes valides]” (Codere Reference Codere1973:20).

In daily life, it was these lowest authorities—the subchiefs and especially the ibirongozi or, in popular parlance, umumotsi—who had most presence (Mulinda Reference Mulinda2010). The umumotsi’s “popularity” was clearly expressed in the name itself, which derives from the verb kumoka—to bark. Umumotsi was a figure associated with orders and obligations—he called for obligatory meetings, or iperu (a contortion of French appel–summons), collected prestations and taxes, and called able-bodied men to perform compulsory work (Mulinda Reference Mulinda2010). As Ruhamiriza described his tasks, “I collected taxes. I caused various crops to be cultivated. I assigned corvée work. I traced out roads. I did reforesting” (Codere Reference Codere1973:58). While Sous-chef Gasigwa spoke to Codere of his “enormous popularity,” Sous-chief Mihana was clear that “my subjects simply obeyed me out of fear. They worked hard but were malcontent” (Codere Reference Codere1973:79). The biographies of sous-chiefs are filled with stories of local political intrigue and “malice,” appointments and reappointments, easier and more difficult constituencies, but they also speak clearly of a privileged class and an intricate, centralized political organization with a clear presence in Rwandans’ daily lives.

After independence, the rhetoric of the First Republic spoke of an “absolute break” with the past of the “feudal-colonial system” (Desrosiers Reference Desrosiers2014:210). But despite administrative reform and different political functionaries (with Hutu now replacing most Tutsi), the reach of the state and the styles of its appropriation remained largely unchanged, and the administrative structure and labor requisitioning were kept intact. Based on a decree from December 25, 1959, subchiefdoms became communes (municipalities), and chiefs were replaced by burgomasters (mayors). Yet “upon assuming power, many burgomasters interpreted their role in pre-revolutionary terms” (Reyntjens Reference Reyntjens1987:90) and thus were complicit in partially restoring the very order they first hoped to destroy (Lemarchand Reference Lemarchand1966).

The state reached the pinnacle of intricacy under the Second Republic of Juvenal Habyarimana. The National Republican Movement for Democracy (MRND), a single party created in 1975, fused completely with state structures (Guichaoua Reference Guichaoua1989). The state was integrated vertically and organized hierarchically into prefectures, communes, sectors, and cells. The leadership made attempts to decentralize politics and services to the commune level, “ensuring closer ties between the state and citizens,” though reach was far from perfect (Desrosiers Reference Desrosiers2014:204).

Taking inspiration from socialist Tanzania, President Habyarimana instituted the nyumbakumi system in Rwanda—an informal system of unpaid authorities responsible for ten houses. In Tanzania the system was first instituted in 1964 and the nyumba kumi kumi, in essence the lowest arm of the TANU party (and later of the CCM, Chama Cha Mapinduzi) served multiple purposes, the most important of which was checking and reporting on security and movement, collecting party dues, and mobilizing the population for development tasks. It was these local balozi (lit. ambassadors) who were responsible for compliance with and execution of the ujamaa system. They were meant to be “the eyes of the nation” (according to Second Vice-President Rashidi Mfaume Kawawa, quoted in Levine Reference Levine, Cliffe and Saul1972:330). In practice, their loyalties were often split and oversight over them imperfect (Cross Reference Cross2013).

In Rwanda the function of the nyumbakumi was very similar—they were the lowest arm of the MRND, reported on security and movement, and mobilized the population, including supervising umuganda community work. The nyumbakumi also had the power to fine people (HRW 1999). The new system had the intended effects of greater responsibilization, visibility and compliance of the population, and a greater presence of the state in daily life. The nyumbakumi’s intimate knowledge of the neighborhood, their tasks, and their ability to draw lists contributed to Rwandans becoming “eminently findable” (Scott 2006:215).

The nyumbakumi system poses interesting puzzles. Clearly, neither the system nor its surveillance capacity is unique to Rwanda. The system itself has a long genealogy, reaching to communist countries outside of the continent (Maoist China, Cuba, Cambodia) and has in turn inspired other countries in Africa including Burundi, Uganda, and most recently, Kenya. ^{Footnote 7} Nonetheless, the nyumbakumi has always been treated as a symbol of the Rwandan state’s penetrating reach and the use of this reach in the genocide (Mironko Reference Mironko and Cook2004; desForges 1999). So is Rwanda unique in terms of reach, and if so, how?

This is where political geography and the long history of local state presence come into play, a legacy that cannot be reduced to the nyumbakumi. Rather, it harks back to the umumotsi or the imposed Tutsi sous-chefs, a past of imposition, vertical integration, and intricacy that countries such as Tanzania or Uganda do not share with Rwanda. When it comes to the genocide, state presence was an indirect contributing factor through the selective impositions and harsh extractions that it imposed. The nature of the state also affected the form and execution of mobilization during the genocide. The intricate hierarchical state structure in a small geographical area helped unroll the project, again a factor not replicated in the neighboring countries (with the exception of Burundi).

The second puzzle concerns the continuity of local reach and the nyumbakumi institution after the genocide. The same institution that aided mobilization for the killing project became the key to mobilizing local population for justice after the genocide (Nagy Reference Nagy and Srira2013; Thomson Reference Thomson2013). The nyumbakumi have also facilitated research into both of these topics (McDoom Reference McDoom2011). Perhaps rather than a puzzle, what this apparent paradox represents is the ambiguity at the heart of quotidian surveillance and state penetration more broadly. It is not inevitable, in other words, that calls for increased “security” are always motivated by repressive goals. Ordinary citizens can be caught up in the state’s modes of both coercion and care, and it is this doubling of motives that helps reproduce surveillance over time. The powerful ambivalence surrounding matters of state presence was evident, for example, in Kenyan public debates about the introduction of the nyumbakumi system of local intelligence and control in the wake of the Westgate Mall attacks of September 2013. An analysis of blogging activity in 2014 showed that while some people posted pictures of Big Brother and spoke of spies and an unwelcome intrusion of the state into private life, others excused the costs as necessary tradeoffs to assure greater security. ^{Footnote 8} In addition to ambiguity, continuities in surveillance are also tied to its ability to ease governance and to the government’s infrequent ability to oppose the “inertia of the state” (Stepputat and Hansen Reference Stepputat and Hansen2001:29).

In the case of Rwanda, care and coercion have been tightly intertwined. State presence across epochs has been coercive and overbearing, and continually excused through projections of “benevolent leadership” (Desrosiers & Thomson Reference Desrosiers and Thomson2011:437) and references to service provision and security. Both the developmental drive and securitization have nonetheless been augmented after the genocide, leading to increasing state presence and surveillance of daily life. This is so despite the abolition of the nyumbakumi in 2006 and the reduction of umuganda to a once-a-month obligation. ^{Footnote 9} The new lowest official administrative level today—the village, or umudugudu—is governed by an umukuru, who, with a committee of four, is responsible for fifty to two hundred houses. The next level up (the “cell”), consisting of about five hundred to one thousand houses, is overseen by the responsable, with a seven-member committee and five Local Defence Forces (LDF) personnel. The number of cells has remained virtually unchanged from before the genocide, despite two administrative restructurings in 2001 and 2006. And while the 2006 reforms decreased the number of higher-level units—provinces and districts—they significantly increased the number of the more localized sector offices, from 145 to 415.

Despite the above-detailed restructuring, state presence has intensified since the genocide, in line with the new national goals of wholesale social transformation (Ingelaere Reference Ingelaere2014; Purdeková Reference Purdeková2011a, Reference Purdeková2015). “Reconstruction” in Rwanda has reached far beyond the physical realm as the government asserts itself in attempts to securitize, reeducate, and sensitize the population in the name of both its developmental aspirations and its political struggle—to capture their hearts and minds and assure political consent. Politics and security have become interlocked, both calling for intensified surveillance. In the name of uncovering “genocide ideology,” “divisionism,” and “terrorists” and preventing a future slide into violence, the government, in order to suppress real or perceived opposition, has tightened surveillance of local milieus.

Surveillance in today’s Rwanda is thus a tool of political control and repression to a greater extent than previously. The memory of the genocide is used to invoke and legitimize the need for greater securitization of everyday life. In fact, public security has become an “all-encompassing indemnifier” (Nyst Reference Nyst2012), excusing further empowerment of police, army, and intelligence agents in their information-gathering roles. Security was invoked, for example, in arguments in favor of the 2013 amendment to the 2008 Law Relating to the Interception of Communications, which further increased intrusion into private lives and the ability of state officials “to listen and read private communications, both online and offline” (Nyst Reference Nyst2012). Today, all communication providers are required to implement state-acquired technologies such as keyword scanning. It is generally assumed that both e-mail and phone communications are tracked, and people certainly self-edit as if this were the case.

In the postgenocide era, then, state platforms and the responsibilities and obligations associated with the state have proliferated, not only in areas of security (amarondo patrols, plainclothes police, and ad-hoc information gathering), but also in connection with development initiatives (“special” umuganda, ubudehe schemes, imihigo contracts, sensitization sessions), politics (imisanzu, or contributions), and justice (attendance at gacaca hearings, acting as an inyangamugayo judge). Importantly, these new roles cannot be reduced to the official manning of administrative posts. The true extent of the state may be hard to gauge precisely because it overspills its official structures into multiple functions carried out on its behalf, some established (such as the activities of abakangurambaga figures or amarondo patrols), others ad-hoc (such as when someone is tasked with “keeping an eye” and reporting on the whereabouts of a person).

But state reach in Rwanda is more diffuse yet, and cannot be fully captured through the sum total of formal and informal institutions and responsibilities. The more generalized sense of “being monitored” is tied not only to the intricate structure of state presence but also to forms and perceptions related to information gathering. As an informant told Susan Thomson, “there are a lot of people watching you, checking on your actions and the people you are with.” Rwandans, she says, “all know of state surveillance. . . . Dense networks of spies are known to exist throughout Rwanda (and abroad) and the Department of Military Intelligence is rumored to pay for valid information’ (2013:123,124). The word spy, however, might not capture the diversity of informer types and the dynamics of often ad-hoc informants or people in a variety of functions asked to “keep tabs on” or to “figure out” people (in this sense “sorting” and categorizing political character, seeking answers to the key question of “What does the person think of the government, is it a friend or foe?”).

Researchers, too, attract curiosity and careful observation. I have been asked by the National Unity and Reconciliation Commission (NURC) to produce a list of interviewees and my schedule of interviews, and have been told that if I did not produce a particular report reflecting on my stay, the government might think “there is something confidential [I am hiding].” Like everyone else, I had to rely on my wits and resources when interacting with surveillance, engaging but also subverting it at the same time. My and others’ lived experiences of surveillance helped me understand that studying structures, institutions, and even systems is insufficient since surveillance works indirectly, inspiring anxiety, carefulness, and an atmosphere of “quiet insecurity” (Grant 2015). This is governance through social unease—an insecure government creating an insecure and uncertain society, even if paradoxically through its techniques of security and “ascertaining.”

In sum, while the political history of Rwanda over the past century has been undoubtedly tumultuous, the intricate state administration stands out as a constant on the political landscape, repeatedly reappropriated as a valuable asset by the reigning authorities of the land. While at the top political change might seem profound, on the hills state presence and its exactions form an important continuity in the lives of common Rwandans.

During the precolonial and colonial times, the main tasks of the state were extractive and (later) “developmental,” though the two can hardly be dissociated as construction of roads or drying up of marshes itself depended on “labor prestations.” After independence, state presence extended to political “mobilization” of the population and to the “checking” of loyalties and leanings—whether it was checking of Parmehutu card ownership under the First Republic, or kubohoza political “liberation” descents on households during the multiparty era of the Second Republic (1990–94), or the checking of membership in the umuryango (family, meaning the RPF party) today.

The pinnacle of conflation between the state and the political project was undoubtedly the genocide, in which local state structures were directly implicated. But yet again, the genocide should be read as a continuity rather than discontinuity. The call to participate was couched in the same terms as the exactions and public tasks of the state in previous decades—as communal labor, “special” umuganda, and public work (gukora, akazi, or akazi gakomeye—a big job) (Hintjens Reference Hintjens1999; Article 19 1996).

The continuities in everyday state presence pre- and postgenocide are also striking (see Desrosiers & Thomson Reference Desrosiers and Thomson2011). This is especially so considering the stark change in the dominant class, its negative rhetorical attitude to the previous regime, and its emphasis on discontinuity, rupture, and the “rebirth” of Rwanda. Yet the continuities appear time and again, symbolized not only by the nyumbakumi (which was abolished only very recently), the umuganda community works, the amarondo (informal night patrols), the animateurs (renamed abakangurambaga), and more broadly by the continued conflation of party and state. Continuities are also evident in the state’s developmental orientation, and as later sections will show, by the ID card and other forms of ever more sophisticated civil registration and various informal but state-initiated activities and forms of state appropriation of time, labor, and loyalties. The continued and increasing state reach and presence—and hence visibility—can be explained by the ambiguity that surrounds these manifestations and that derives from the ability of the state and its “mundane sights” to present a face of both care and coercion.

Identification: Counting, Sorting, Tracing

While forms of oral identification (and hence social “tracing” and “placing”) existed in precolonial Rwanda (see, e.g., Nyirubugara Reference Nyirubugara2013), it was undoubtedly the colonial government that devised and imprinted upon its colonies a distinct bureaucratic form of identification whereby identification became standardized and identity became legible and “actionable” for the centralized state. Considering the prominence of the ID card in accounts of the implementation of the Rwandan genocide, it is striking how little is known about the purposes and colonial implementation of the census, the ID card, or the passport.

The urge to sort, count, and trace the Rwandan population came as a result of a wider colonial policy during Belgian rule (1923–62). In 1933 the official colonial bulletin (Bulletin Officiel du Congo-Belge) published a decree setting out the legal basis for identity cards in the Belgian Congo and “neighbouring colonies” (referring to Ruanda-Urundi). It stated that all “indigenous” subjects should register and would receive an identity card upon doing so (certificat d’identité or eenzelvigheisbewijs). The decree did not include specifications as to the design of the card or the process of registration, with these details remaining the prerogative of the governor of the two colonies (van Brakel & van Kerckhoven Reference Van Brakel, Van Kerckhoven and Boersma2014).

In Rwanda the ID card, or indangamuntu, became best known for creating a fixed record of ethnic identity. Indeed, the very first item under the photo was ubwoko (lit. meaning “type” but referring to ethnicity) and the card offered four options—Hutu, Tutsi, Twa, and Naturalisé. Although this registration of ethnic identity seems particularly significant in retrospect, it was not the primary purpose of the card at the time. Instead, “registering ethnicity was merely one component of a broader program to increase the regulation of Belgian subjects” (Longman Reference Longman, Caplan and Torpey2001:353).

In the twilight of colonialism in the 1950s, Rwanda and Burundi had a “more or less well-functioning civil registration system” (Uvin Reference Uvin, Kertzer and Arel2002:152), originally overseen by the Catholic Church. The identity cards were introduced together with other social monitoring mechanisms—a passport (passport de mutation, to regulate movement outside and inside the colonies) and a census (Reyntjens 1985). Despite the apparent scientific objectivity of a “counting” enterprise, its politicized nature is easy to make out. The census was in fact “linked to dynamics of power and resistance in the region,” since colonial counting was tied to taxation, and repeated evidence of population growth was tied to legitimization of the colonial enterprise, a token of its “benefits” (Uvin Reference Uvin, Kertzer and Arel2002:148). The first full national door-to-door survey took place in 1978, when “for 24 hours, no citizen was allowed to leave home and throughout the country, a whole army of teachers and bureaucrats, accompanied by military personnel, went house to house collecting data on 36 variables” (Uvin Reference Uvin, Kertzer and Arel2002:153).

But despite the meticulousness of this technocratic effort, unsavory facts were deliberately allowed to remain invisible. The census created power both through visibility and invisibility. In Burundi and Rwanda major violent episodes and exodus of populations did not register on the census. In the case of Burundi, this was the 1972 genocide, estimated to have killed between one hundred thousand and one hundred and fifty thousand Hutu. In the case of Rwanda, it was the secret 1962–63 Tutsi purges, which precipitated the flight of between 40 and 70 percent of the Tutsi population, most of whom did not return until decades later. Censuses thus perhaps best show the fuzzy line between visibility and invisibility in the exercise of power, dissolving what is seen and what is hidden, shown and concealed into a “zone of indistinction” (Agamben Reference Agamben1998).

What is significant is not only the institution of the card by the Belgians, but also the subsequent decision by the postcolonial authorities to retain it. The ID cards had gained a more explicit political character at the end of colonial rule when tensions in the country ran high. In the “Hutu Manifesto” of March 24, 1957, the authorities expressed clearly that “we are opposed vigorously, at least for the moment, to the suppression in the official or private identity papers of the mention of ‘muhutu’, ‘mututsi’, ‘mutwa.’ The suppression could create a risk of preventing the statistical law from establishing the reality of facts” (i.e., the numerical dominance of the Hutu; emphasis added) (quoted in Fussell Reference Fussell, Watner and McElroy2004:64). The first president of Rwanda, Gregoire Kayibanda, kept the card, and so did his successor, Juvenal Habyarimana, whose death precipitated the genocide. ^{Footnote 10}

In 1990, another time of political turmoil, the ID card again featured in political rhetoric, and this time its future was in question. At the turn of the decade the government of Juvenal Habyarimana came under a mix of pressures. In October 1990 the exiled Tutsi-based RPF invaded Rwanda on a platform of ethnic unity. On November 13, 1990, under foreign pressure, Habyarimana announced a new multiparty system, along with “his intention, which he never acted upon, to abolish the ethnic identity cards” (Fussell Reference Fussell, Watner and McElroy2004:65). If read in its historical context, this perplexing promise is easily understood as political rhetoric and a political maneuver intended to signal an adjustment to the political platform of the invading RPF and to offer a liberal concession to the international community.

Despite the continuity of its use and its effective deployment in genocide, the Rwandan ID never presented a system of total capture. Soon after the ID was introduced by the colonial administration, the limits of registering identity had became apparent. The ingangamuntu—or in popular parlance, les pièces (documents)—became susceptible to what Breckenridge, writing in the context of South Africa, has called the “age-old faking and forging practices” (2005a:98). As in the case of the census, visibility is never total, and one can in fact “hide” behind visibility. Here, however, the hiding was enacted not by those conducting the sorting and tracing exercise, but by those subjected to it. The trends in reclassification followed the changing political fortunes of different ethnic groups.

During the colonial period, a trade in ID cards developed, and Hutu who wanted to gain better opportunities and could afford the purchase reclassified themselves as Tutsi, “generally through illicit means” (Longman Reference Longman, Caplan and Torpey2001:353). Following the muyaga, the violent winds of the 1959 “Social Revolution” and the institution of the First Republic under Hutu majority rule, Tutsi fearing persecution or discrimination attempted to purchase new Hutu identity cards. Successful reclassification was not automatic, however, and hinged on the knowledge embedded in the community. A successful assumption of a new identity was more likely for those who relocated elsewhere in Rwanda and for those moving to urban areas with greater possibility of anonymity.

Because of their prominent role in the genocide, the ID card together with the checkpoint became the emblems of a state and its bureaucracy turning against its own citizens, with visibility and order enabling sorting for selective annihilation. Indeed, those who produced cards reading “Tutsi” were usually executed immediately. At times, cards of victims were collected for accounting purposes, highlighting the bureaucratic nature of the task (see Fussell Reference Fussell, Watner and McElroy2004). Captain Oldephonse Nizeyimana, for example, “regularly received cards from his men as they reported on the progress of the killings . . . [, and] in the captain’s absence, his wife received the cards” (Fussell Reference Fussell, Watner and McElroy2004:15).

Yet even during the genocide, people did not cease to invent ways to subvert the visibility imposed by the card. One woman testified to having used methyl alcohol to erase the marks in the ID papers of her friend and successfully reclassify her as Hutu (Pottier Reference Pottier and Broch-Due2004). In other instances, authorities themselves allegedly used their power to issue false Hutu IDs and laissez-passers, as well as blank documents, and to make false entries in the Registre des Résidents to selectively save a small number of Tutsi, even as they aided and abetted the broader project of genocide. ^{Footnote 11}

Due to historical manipulation and forgery, the counterfeit ID was “not so secret after all” (Nardone Reference Nardone2010)—IDs were not considered reliable and were not fully trusted as indexes of difference. The label “Hutu” was not a salvation—its worth was eroded and it became harder to hide within the “visible.” The preoccupation with wiping out a whole category of people meant that ibiymanyi (hybrids) and abaguze ubwoko (ethnic cheaters) were also targeted (Eltringham Reference Eltringham2004). The latter term was in use already in 1973, but categorical purity became a true preoccupation at the extremist Kangura newspaper at the beginning of the 1990s. In the November 1990 edition, an author asked rhetorically: “A person that adopts an ethnic identity, which is not that of their birth and who carries the supporting documents, is this not a species with two heads (espèce à deux têtes)?” (quoted in Chretien Reference Chretien1995:102). The allusion was to a burrowing snake known as ikirumirahabiri, an image used to characterize ambiguous persons, and in a political context meaning a “double agent” (see Chretien Reference Chretien1995). ^{Footnote 12}

As a result, alternative ways of identification were deployed by the killers, most prominently “body maps” (Malkki Reference Malkki1995) or rumored narratives of “real origin”—histories of migration and subsequent reclassification. ^{Footnote 13} Small children did not carry ID cards, and when wandering alone they were better able than adults to hide and pass through roadblocks. However, “survival often depended entirely on the decisions made by the individual militia on the roadblock. ‘I passed a roadblock and the men looked at the lines on the palms of my hands, they decided that the lines did not show that I was a Tutsi and so I was allowed to pass’” (Bleach Reference Bleach2009:69). More generally, those officially classified as Hutu but who looked Tutsi were targeted.

In 1996 the postgenocide government, responding to the symbolism of the “deadly” ethnic ID card, instituted a new de-ethnicized indangamuntu, which soon came to be read as a symbol of the nation-building effort. But the underlying issue—registration and its facilitation of tracing, sorting, and targeting—was never problematized in itself. In fact, the system of identification became much stronger than it had been previously. In 2009 new digital ID cards were released, and by 2014 the brand new National Identification Agency (NIDA) (established in 2011) had issued “smart ID cards,” which collate a wide array of information in one document using biometric information, to 80 percent of the population. ^{Footnote 14} The government is now also considering issuing identification for children (World Bank 2014).

Not only was ethnicity treated as the key issue at stake, but its lack was seen as truly emblematic, symbolically potent—the state might not have been changed structurally but rather recaptured for the purpose of benevolent “care.” “De-ethnicization” became a powerful legitimator upholding the state’s ability to reach and trace the population for biopolitical ends in the name of fostering well-being and national development. Nonetheless, this hardly spelled an end to “ethnic tracing” as age-old and new alternatives continue to be deployed, including “body maps,” family names, and personal information on CVs. ^{Footnote 15} Similarly, the new ID card did not signal an end to official social sorting and implicit ethnicization (Burnet Reference Burnet2012).

Importantly, though, ethnicity postgenocide is not the key category to be “sorted” out by those in power. Today it is political sorting that is most important, and to this end cards and registers are complemented by other forms of information. The indangamuntu is complemented by other cards, such as the umuganda card, the mutuelle de santé (health insurance) card, or the Umurenge Sacco (savings collective) cards. All of these serve as mechanisms ensuring citizen compliance (participation in state-mandated activities) and facilitating government control (since these documents are a prerequisite to obtaining other permits and favors from the local authorities) (Purdeková Reference Purdeková2015). The most potent of these documents are the imihigo contracts. Introduced in 2006, imihigo is an annual pledge of having accomplished specific development goals that is signed at all administrative levels, down to the level of the household (see Purdeková Reference Purdeková2011a; Chemouni Reference Chemouni2014). In theory, the imihigo objectives stem from local priorities, but in practice they are guided from the top. The district imihigo are elaborate and comprise around forty indicators, thus “leaving few activities for which planning is not reviewed by the centre. In addition, ministries regularly keep an eye on districts through regular ‘descents’ of their agents deployed locally” (Chemouni Reference Chemouni2014:249). Of course, imihigo is not immune to subversion, even among those “overseeing” its implementation at the local level. At the same time as officials lament that imihigo “‘is killing us, no one can escape it’, [they] may resort to data falsification to reach their objectives” (Chemouni Reference Chemouni2014:250).

Importantly, the Rwandan state has made significant strides in integrating these diverse “pieces” of identification and monitoring. As opposed to the national ID, which is required for all Rwandans aged sixteen and above, NIDA’s “smart ID” is optional. The ID integrates seven identification features, including a tax identification number and details of driver’s license, passport, family dependents/members, social security (RSSB), and health insurance. According to Rwanda’s Minister of Youth and ICT, this new technology rapidly “decreases the time they [the government] use[s] to access the citizen’s full identification.” ^{Footnote 16}

To conclude, the Rwandan ID card was a product of the colonial period and was undoubtedly seen at the time as a modern and universalized system of state control, with all the ambiguities that the technology implies. The ID card is thus a good example of a piece of a wider surveillance “assemblage” (Haggerty & Ericson Reference Haggerty and Ericson2000)—a collection of pieces that can be assembled and appropriated for an array of purposes. It is one technology among others that can be configured as part of a broader, transitory constellation of control—of paperwork, data, population registers, and census results, of tracing, sorting, and checking systems. It is both tangible as a collage of objects, sites, and moments of interaction, and more intangible as a general sense, an affective state.

The analysis of IDs and identification again highlights the role of ambiguity in driving continuity and reapplication. Since the Rwandan colonial period, the card too presents an interesting continuity across a turbulent political discontinuity, and is deployed in Rwanda to this day. Undoubtedly, though, the thickness of social monitoring has increased postgenocide, with old (or renewed) forms being accompanied by new forms such as imihigo contracts (though these too are rhetorically anchored in the “traditional” past; see Purdeková Reference Purdeková2011a; Chemouni Reference Chemouni2014). Crucially, the analysis challenges any easy distinction between visibility as a space of state power and invisibility as a space of subversion, showing that it is indeed possible for the state and its citizens alike to “hide” within the visible.

The Public Show: Activities, Ceremonies, and Commensality

While postcolonial Rwanda reproduced the legacy of “close public scrutiny of all spheres of life” through a “network of controls” (Hintjens Reference Hintjens1999:245), the Second Republic of Juvenal Habyarimana saw a particular intensification. Rwanda became a developmental state (Verwimp 2013) that “stirred the hillsides” as “‘projects’ of all kinds . . . spread across the countryside” (de Lame 2004:295) and the state brought people further into the public sphere—and thus the state’s sphere of vision—through multiple new state activities, including umuganda public works and public feasts. This style and repertoire have been maintained and elaborated to this day.

Whereas the previous sections explored the ambiguity of surveillance, the resulting ambivalence felt toward it, and the confluence of visibility and invisibility in the exercise of power, this section focuses on display and disguise as epitomized by interaction in public activities such as umuganda Saturdays or communal feasting occasions. These ostentatious and obligatory public displays of “togetherness” are perhaps the best demonstration of the existence of mutual deception between citizens and the state, and their mutual “disarming” through pretense in public spaces.

The approach in this section further complicates the notion of surveillance as “seeing” by investigating the public sphere as a sphere of vision where both sides arrange and frame themselves in response to the other, resulting in mutual deception and, potentially, a “crisis of transparency” (de Lame 2004). From this perspective, it is both the state representatives and the state’s subjects that are “watching each other.” Several implications flow from this: While seeing is always also reading (as in “decoding”), greater seeing does not automatically translate into more accurate reading. In addition, since what matters is the “presentation of self” (Goffman 1959)—not only that one is seen, but what one is seen as—framing can be manipulated not only from above, but also from below. From above, the dramatic (dis)play is one of a caring state; from below it is one of a loyal subject. In this way, “revealing and concealing,” as de Lame observes (2004:305), present themselves as “two ways of playing on the same keys, of escaping any definition of one’s position vis-à-vis society.” Indeed, when it comes to the public sphere, display and disguise again enter into a state of indistinction.

Under Habyarimana’s leadership the developmental drive led to the institution of multiple new public activities, all of which increased the presence of the state in people’s lives. On February 2, 1974, Habyarimana asked every Rwandan to perform “voluntary” community work for half a day every Saturday. Taking inspiration from Tanzania’s ujaama and Zaire’s authenticité and salongo (public works instituted in 1973), Habyarimana anchored the practice rhetorically in Rwanda’s own precolonial past of communalism. Umuganda was the traditional stamp on what was otherwise known as “Collective Works for Development.”

In addition to umuganda, the Habyarimana regime initiated numerous public feasts and exaltations of the regime, weekly entertainment sessions (séances d’animation) of song, dance, and sensitization, public celebrations, and stylized gift exchange ceremonies (amaturo). For the abaturage (people), the séances took place after the umuganda public works. State employees had get-togethers every Wednesday afternoon “to practice chants and skits in celebration of the Rwandan state, its overthrow of the Tutsi monarchy, and its rejection of the ubuhake cattle contract signifying Hutu servitude to Tutsi, and most of all to honor the country’s President” (Taylor Reference Taylor, Kapferer and Bertelsen2009:175). These employee groups were also organized as cellules (and thus integrated as separate political parcels) and were sometimes called groupes de choc. They would perform publicly on national holidays, competing against each other in their official exaltation of the regime. The voice of the president was also broadcast for five minutes every day. What resulted was a dense interaction between the people and the state. For example, according to one report about the activities of a commune in 1985,

De Lame (2004) argues that the ostentation characteristic of the Second Republic and visible in activities such as umuganda served the purpose of show— “they were highly visible in organization, but not in output” (2004:289). The ceremoniousness was meant to have symbolic and metonymic power, both reflecting political authority and enacting that very authority by means of the representation itself. Similarly, Ranck argues that the key aspect of public interaction under the Second Republic was performative; it was all about “the spectacle of the state” (Reference Ranck, Simon, Rosenberg and Eppert2000:193). Verwimp (Reference Verwimp2000, 2013), by contrast, contextualizes umuganda squarely as exaction and direct tax, an institution that made an “enormous amount of unpaid labor available to the state” (2000:27). Indeed, despite the inefficiencies and evasions that were part of the system, the material results of umuganda were visible throughout Rwanda and included the145 identical commune offices and hundreds of kilometers of anti-erosion ditches (Verwimp 2013). One can conclude, therefore, that both show and real extraction have defined umuganda and other public activities to this day. The aspect of show is perhaps more important today than before, with umuganda and other public activities bringing people “together” presented as tools for community healing and nation-building.

Yet again, the continuities across the landmark of genocide are striking—both in terms of the developmental disposition of the state and the style of “public togetherness” and ceremoniality promoted. The term animation has been dropped and in its stead there are the ever-present references to “sensitization.” Umuganda, which was discontinued with the 1994 genocide, was reintroduced in 1998 along with the post-umuganda meetings and ubusabane community festivals. At the same time, the exigencies of reconstruction, development, and reconciliation have called for the introduction of additional platforms of public communality, such as the ubusabane community “feasts,” commemorative activities, and special public holidays such as Heroes Day. To community works have also been added community justice responsibilities in gacaca courts (which have only recently finished their work), policing and security responsibilities, multiple unity and reconciliation activities organized under the aegis of the National Unity and Reconciliation Commission (NURC), and ceremonious football matches, among others.

Among these new platforms of communality perhaps the most important is the vast “civic education” exercise. Civic education commenced first through selective targeting under the ingando program (see Purdeková Reference Purdeková2015; Thomson 2011; Melvin Reference Melvin2013) in which hundreds of Rwandans at a time spent weeks or months in camps in different parts of the country receiving military training as well as lessons in history, politics, “traditional” Rwandan culture, and the policies of the government. More recently, civic education was “decentralized” and extended to the general population under the program of Itorero ry’Igihugu (The National Academy; see Sundberg Reference Sundberg2016). Participation in civic education camps is mandatory and all participants receive a certificate of attendance upon graduation, which serves as a way to check and identify compliance and hence potential dissent. Government-sponsored university students must produce the intore/ingando certificate to gain entry to university and university students must present their ingando certificate in order to graduate.

Aspects of mutual play are clearly present in all of these activities. “Being seen” can at once acclaim and undermine the official. People’s perfunctory attendance and token compliance can score credit or help avoid cost (sometimes literally in the form of fines). Public activities thus often turn into surface-level acts that speak simultaneously of acquiescence and hollowing out of activities from within (see Purdeková Reference Purdeková2015). “The most important thing is to be there [at umuganda],” one of my informants in Kigali told me (Purdeková Reference Purdeková2015:123). When the most important aspect of a public activity is “being seen,” effort diminishes and often becomes perfunctory. But subversion can also happen by appropriating and playing with words, as, for example, in the ironic claim of one Rwanda citizen that citizens have participated in public works by “cleaning their living room” (Purdeková Reference Purdeková2015:124). And while one’s public voice is policed and rehearsed, personal power is often reclaimed through silence (see Burnet Reference Burnet2012; Thomson Reference Thomson2013; Rettig Reference Rettig2008). Purposeful silence is both a strategy against being “read” and a direct way to undermine surveillance, and it is an indirect yet powerful message of distance or disagreement even in the face of an intimate encirclement by the state.

But it is not only “being seen” that is a space of play and maneuver, but equally “being seen as.” Both are plays of appearance, but while the first relates more to “making an appearance,” the latter is more about “managing an appearance.” The two nonetheless are closely related. The purpose of political surveillance often is to ascertain loyalties and leanings. Being “seen as” thus refers to maneuvering in order to frame oneself in profitable ways or to avoid being framed in costly ways, essentially of appearing as a friend, or at least not as an enemy. The latter dynamic is what McGregor (Reference McGregor2013) described in the context of today’s Zimbabwe as “being watched and the risks of being seen as a traitor.”

In Rwanda, the salience of political framing and the stakes involved increased with the civil war (1990–94) and later the genocide, but continue to be salient to this day. The terms of “inspection” might have changed, but the intense use of surveillance to ascertain a political profile has not. During the civil war and the genocide, a key term of condemnation was ibyitso—referring to an accomplice of the invading (and predominantly Tutsi) RPF and expanded to equate all Tutsi with the enemy (inyangarwanda) (Straus Reference Straus2006; Purdeková Reference Purdeková2009). Today, the talk is of divisionists, people with genocide ideology, and more recently terrorists—but there is continuity in the search and identification of internal and external enemies. The resultant fomenting of uncertainty is reflected in public appeals to Rwandans to be vigilant— “banyarwanda basabwa kuba maso” (literally “to be eyes”). It is this “inspection mode” that contributes to further securitization and increased surveillance in postgenocide Rwanda.