Situating science in Africa: The dynamics of computing research in Nairobi and Kampala

Igniting the invisible

Omwenga uses the metaphor of ‘igniting the invisible’ to signify that he is developing techniques capable of visualizing Nairobi’s technology communities in uncharted ways. We adopt his metaphor to describe our own methodology, which sheds light on the computer science communities of East Africa in a different way. Our research design involved developing a robust methodology informed by long-term ethnographic research in computer science laboratories and universities of Kampala and Nairobi. The science that we observed is substantially networked into local and transnational flows of knowledge, capital, technology and expertise. As social researchers, we are part of that networking and internationalization, and thus we become connected to our respondents, who are devising algorithms, designing software platforms, experimenting with bits of hardware, training and developing new researchers, and so on. Another way of describing our methodology, following Mol and Law (2004), is to say that we, through our own knowledge practices, along with those of our respondents, are enacting computer science in East Africa.

We draw on a substantial body of qualitative data based on in-depth interviews and video ethnography (Shrum et al., 2010). Over the period 2011–2016, we conducted over 75 interviews (most recorded on video) with active computer scientists, graduate students, policy makers, administrators and tech entrepreneurs currently working in the university and private sectors in Nairobi and Kampala. We also filmed laboratory tours, public events, meetings, seminars and field sites where computing knowledge was put into use. We tried to avoid using our own conceptions of what computer science is or is not to shape our sample. Rather, we found respondents who had institutional and organizational connections to the disciplines of computer science, informatics, information systems and computer and software engineering, and then designed our research to examine how respondents conceptualize and participate in computer science.

Labor power and scientific imaginaries

While Omwenga and Sansa-Otim do not represent the stereotypical pattern of ‘brain drain’ from Africa, they are not outliers. Many respondents returned or migrated to East Africa to complete their studies and work, despite having received their training and admission offers from institutions in the Global North. Respondents repeatedly voiced the vocational draw of pursuing a scientific career in Africa. The continent is imagined as a place where there are problems to be solved that satisfy both intellectual curiosities and moral conscience, and being based locally is essential to pursuing both. The motivations for Ugandan and Kenyan researchers vary, resulting in differing perspectives on computer science in Africa: in Nairobi computing research is enacted in ways that are commercially and entrepreneurially oriented, while in Kampala computer science is more academic. However, the mobility of labor is significant in both cities because it can change the narrative about a continent in deficit.

Through this vocational appeal – conjoining moral conscience and intellectual curiosity – African societies are valorized as ‘living laboratories’ for computer science (see Tilley, 2011). Respondents became animated by what they conceive as chaotic and complex social, environmental, biological and economic problems that seem to lend themselves to computer science configurations and solutions. In its own unique way, Africa is conceived as fertile ground for experiments in computer science. The researchers interviewed for this project all pointed to the necessity of being embedded within local science and technology sectors from which they could not only discursively frame problems but also build computer science solutions. Their localities enabled respondents to better know the sector and realize its capabilities, but more importantly working from within allowed them to experiment with different forms of abstraction in problem solving.

Tonny Omwansa, Lecturer at the University of Nairobi, completed his master’s degree at Wichita State University in the USA. He had an offer to stay and complete his PhD there. Like many others, Omwansa chose to return to Kenya to continue his graduate studies in his home country. When asked if he had any regrets about not taking the option to stay in the US, he responded simply:

Not at all. Absolutely not. The reason is, I think the opportunities here are so many. And you have to be a part of the system in order to tap into that.

Omwansa is director of the Computing for Development Lab at the University of Nairobi. He has an established research career working on problems he finds interesting ranging from developing algorithms for control of broadcast frequency, to new techniques for data management. His main focus is on the lack of financial infrastructure and lack of credit in Africa. Taking advantage of his knowledge of Nairobi’s bustling commercial and donor operations (which often span outward from Nairobi across East Africa), he has turned his research on mobile money into consulting for mobile companies, foreign universities and donors in Nairobi.

In contrast, Ernest Mwebaze has carved out a more academic career heading up Makerere’s Artificial Intelligence Laboratory. Mwebaze laughed when he remembered the shock of peers in Groningen, where he studied for his PhD, when he talked about returning to Uganda to pursue a scientific career. Mwebaze established his career with a substantial grant from the Bill and Melinda Gates Foundation to apply computational techniques to diagnosing crop disease in the cassava plant based on images captured using farmers’ mobile phones. ⁵ Through this project, called Mcrops, Mwebaze can advance the field of algorithmic analysis and machine learning at Makerere, while also addressing a global development challenge, framed by the Gates Foundation as food security.

There is a scent of technofetishism in respondents’ statements, arising from an irrepressible faith in the technology emerging from computer science applications to solve seemingly intractable problems, such as crop disease eradication and inequalities of global financing. While respondents imagine African societies as ripe for innovative and impactful computer science that can circumvent centralized governance and large infrastructural systems, the very reality of moving their labor power is socially transformative. Omwansa and Mwebaze are Africans returning to advance their careers, but expatriates who migrate to East Africa also contribute to the construction of similar scientific imaginaries – though they have different racial and class positionalities and motivations for moving their labor power to the region. Jay Taneja is an Indian-American ex-patriot who was working at IBM Research Africa in Nairobi. Taneja, a recent graduate from University of California, Berkeley, declined an offer for a postdoctoral position at Lawrence Livermore National Laboratories to take up a research position in Nairobi. In reflecting on what drew him to work in Kenya he argues:

Lawrence Livermore National Lab, it’s the energy lab, right. It’s a great place to do awesome environmental energy technology stuff. It didn’t seem like it would be as unique of an opportunity as here. The risk with coming to East Africa to do this research is that you could fall flat on your face. Everything could fail here and it would just be time lost in your career. So it’s a bit risky to do right at the beginning [of your career]. … So you kind of have to trust in yourself that you can establish a good research trajectory here.

Taneja is conscious that his move to Kenya is risky for his future career, because it does not represent the typical progression of an internationally mobile scientist. Using Ferguson’s (2006) metaphor, Africa is perceived to lurk in the shadows of global scientific communities and, as Taneja laments, it is not seen as the destination for doing good science. We interpret Teneja’s story about turning down Berkeley in terms of him reaffirming his own intellectual abilities and accomplishments against the common assumption that Africa lacks intellectual strengths. Yet the lure of Nairobi lies in the strong commercial computing sector and the opportunity for him to exercise his chops modelling modular energy platforms that can reach rural communities, leapfrogging the thinly distributed centralized infrastructure. Taneja appears less aware that by moving his labor to East Africa he is redressing any historical inequalities between North and South. While his motivation explicitly resonates with a vision of Africa as ‘living laboratory’, it is the mobility of his labor power that is potentially socially transformative (Marx, 1906). His labor power extends to tutelage, institutional building and attracting research funding, bringing credibility and contributing to the local economy.

Similarly, John Quinn, a Scottish born computer scientist educated at the University of Edinburgh, highlights the contingent nature of African computer science.

I happen to come here, [at] the right place at the right time, there is this period of huge technological and social change and the conditions are right now for computing which weren’t even five to ten years ago. It was much more of a struggle than it is now… we are at this elbow of the curve … whereby things are starting to take off that weren’t before … there is the phone, the hardware, there is data connectivity that there wasn’t before, there are people [who] understand how to use technology in a way that they didn’t earlier, and there are a lot of data sources online that we can access, which weren’t around before.

Quinn points to the accelerated technological and social changes occurring in African cities. He came to East Africa earlier than Taneja, and places his career at ‘the elbow of the curve’’ of historical change towards more technologically fluent societies. Compared with Taneja, Quinn’s motivations are more academic: He sees the sociotechnical changes as providing new data sources for interesting research questions. Quinn has helped to establish the Artificial Intelligence laboratory at Makerere and has steered teaching towards machine learning. He moved into the UN Pulse Lab, one of only three data analytic labs of its kind in the world, and quickly established himself as an expert in African data science while continuing to teach at Makerere. His profile at the UN Pulse Lab allowed him to consolidate discourses on the promises of data science to transform African societies. For example, continuing his research at Makerere, he carried out a project on algorithmic analysis of local radio content to understand how development goals are discussed in various regions and languages. In addition to carrying through projects, Quinn also created a labor supply chain between Makerere AI lab and the UN Pulse Lab.

As part of the scientific community, respondents are also part of the professional middle classes in each city and their lifestyle choices, mobility and patterns of consumption reflect this. It is from within this class status that respondents are afforded a view of African societies as experimental sites for an embedded computer science. This imagination of a ‘living laboratory’ conjoins a form of social responsibility with intellectual freedom to form a moral economy that motivates their movement of labor. Yet, by mobilizing their labor against more typical international flows of academic and scientific capital, they contribute to redressing global inequities. While respondents extol the virtues of this vocational appeal of computer science in Africa, they are less cognizant of the very concrete impacts of them moving their labor power to East Africa; in part, this movement has been made possible by the privatization of higher education.

Universities in the new world order

Through changes in admissions policies, tuition fees and financial administration, universities in East Africa are meeting the demand for education in marketable subjects, such as computer science (Harsh et al., 2018). Makerere University and the University of Nairobi are constantly transforming themselves in response to neoliberal policies linking education more directly with economic growth, ⁶ which serve as universal blueprints for institution building and knowledge production. In East Africa, implementing these policies has opened up institutional space to promote new academic jobs and career trajectories in computer science, and launch labs and incubators of ‘innovation’. These changes trade on the promise of computer science and its global relevance, generating economies of expectation that disguise the tensions inherent in everyday institutional life (cf. Tsing, 2011). Our concern here is to document institutional change to show how it has afforded a vision of computer science to emerge within the universities of East Africa that is consistent with global policy. Importantly, as we show in later sections, enabling this vision simultaneously creates interstitial spaces in which diverse forms of labor and research take shape.

When Quinn chose to relocate to Uganda, he joined a growing university department, spearheaded by a charismatic academic figure at a significant moment of institutional change. Shortly after completing his PhD in 2007 at the University of Edinburgh, Quinn made enquiries about pursuing computer science in Africa. He made contact with Zachary, one of the authors (at the time a journalist and consultant for the Gates Foundation), who recommended that he look at Uganda, and in particular Makerere University in Kampala, which had established a computer science department and was graduating an increasing number of PhD computer scientists. Within days of making enquiries, Quinn had a flight and an academic position at Makerere. Under the mentorship of Professor Venansius Baryamureeba (known simply as Barya) he began contributing to the growth of its computer science faculties. Recruiting Quinn was part of a larger institutional vision pursued by Barya. Even before he became Dean, he laid out his ideas for transforming Makerere:

If I was Dean of Science, I would retrain and redeploy, retrain people so they get a PhD in computer science, open up and become multi-disciplinary, and design courses demanded by the market (Barya quoted in Mamdani, 2007: 76).

Barya was able to take advantage of the financial restructuring happening across higher education in Uganda to channel student fees and donor funds directly to his department. This enabled him to nurture a critical mass of expertise and market a proliferation of short courses (certifications on Cisco and Microsoft technologies, for example) and degree programs in computing with practical applications in the local labor market. Completing his postgraduate education in Norway, Barya cultivated relationships with Europe-based computer scientists. With the assistance of the development agencies in these countries, especially the Netherlands, these contacts evolved into capacity-building collaborations, most effectively the creation of a joint PhD program. Barya mentored approximately 30 students who travelled to Europe for part of the year to complete doctoral training at the participating European university (this includes Sansa-Otim and Mwebaze). Almost all students returned to Uganda and to newly created lectureship positions at Makerere. Barya proved to be a commanding, charismatic figure who channeled the creativity of those returning and relocating to East Africa into an institutional vision. In particular, Barya was keen for them to use their new skills and the tacit knowledge they had acquired from working with established scholars to tutor a new generation of computer scientists. The PhD program is now embedded in the computer science departments at Makerere and the degree is awarded locally.

The ambitions of computer science at Makerere were conceived in the broader context of the neoliberal restructuring of higher education (and public services more generally) in Uganda – creating revenue for teaching units by admitting students who paid higher fees. Critics of neoliberalism across the globe argue that such changes lower educational standards, demoralize academic professionalism and deinstitutionalize universities (through the insidious creep of privatization) (Holden, 2015). The case of computer science in Uganda, however, shows how structural changes do not have a blanket effect. Actors situate themselves within the tensions created by constant reforms, which in this case, resulted in the creation of a class of globally-trained, computing research professionals at Makerere.

The Kenyan universities we studied have experienced similar changes in the privatization of higher education, but with its focus on nurturing centers of innovation and entrepreneurship, the Kenyan context has a distinctly different flavor than does Uganda. Kenyan universities are more visibly influenced by prominent universities in the US such as MIT and Stanford University. These universities continue to provide the blueprint for the entrepreneurial university looking to competitively position itself in what Mirowski (2011) calls a ‘marketplace for ideas’ and in the commodification of educational provision. This is manifest in the more dynamic structures built around public-private partnerships, and include nested laboratories, collaborative research centers, and multi-disciplinary spaces for the ‘incubation’ and ‘acceleration’ of innovative products. In our field research, we were able to visually capture new laboratory spaces in Nairobi – such as the @iLab at Strathmore University, C4D Lab (Computing for Development) at the University of Nairobi and IBM Research Africa based at the Catholic University of Eastern Africa – as well as report on the rhetoric used by politicians, business leaders and university managers in re-imagining the role and contributions of universities in the burgeoning knowledge economies of East Africa.

Gieryn (2008) lends insight into how laboratories reflect not only the ‘requirements of inquiry’ but also the political, economic and social contexts in which they emerge. This argument can be extended to understanding the new computer science laboratories appearing across Nairobi and how they aspire to the Silicon Valley model of open, collaborative and fun workspaces that engender a distinct, yet familiar, model of innovation born of Mode 2 (Nowotny et al., 2001) policymaking of the US and Europe (Mirowski, 2011). We had the opportunity to record the launch of @iLab Africa, Strathmore University’s research and innovation lab, in June 2014. The attendance of high ranking politicians, influential CEOs from national companies like Safaricom, regional directors of Samsung, Google and IBM, make clear that computing and information technology are lauded as national projects that aim to link Kenya to the global economy. At the launch of @iLab, the rhetoric promoted innovation hubs as panaceas to the problem of unemployment. In improving the job market, higher education institutes become the explicit target of structural change.

But today we look at what is happening. The ecosystem has completely changed. We have innovation hubs all over. We have people who are in the universities, and actually because there are no jobs they have to create the jobs and that’s why actually to me what is happening today is a clear manifestation that we are actually moving to the next level. (Kenyan Cabinet Secretary for ICT speaking at the iLab launch, 19th June 2014)

The dreams of computing and information technology are made explicit in the rhetoric furnishing the launch events of various labs and centers. Institutional blueprints for generating entrepreneurialism, enterprise and innovation circulate with the promise of also solving social and economic problems – often couched in the language of meeting the UN Sustainable Development Goals. Implementation of these blueprints is always envisioned as a seamless process that originates independent of the context in which they are applied. And yet, generations of postcolonial critiques have argued that the local context was better off before or outside the encroachment of universal ideals, models and blueprints. Mamdani’s (2007) critique approaches neoliberalization as a pervasive external force imposing on the autonomy and intellectual freedoms of universities. Following Tsing (2011), we move past polemics about the before and after of neoliberalization, the inside and outside of global capitalism, to provide an analysis that questions the enactment of both global and local discourse in context. The global is produced through the contextualized exchanges, interactions and interpretations of recruitment strategies, financial restructuring and lab launches, which generate friction: productive tensions through which discourses normalize and yet, multiple modes of work and computing research take shape.

While the blueprint for the neoliberal university conjures a professional identity in line with its values of entrepreneurialism and privatization, the actual day-to-day experience of working inside higher education is much less stable and affords a heterogeneous working life that at once copes with precarity, but also facilitates research and innovation. There is an implicit assumption that projects form an orderly organization with distinct and measurable instrumental value, yet in practice, computer science generates a granular scale and space that makes it messy rather than orderly.

Identities and working life

The changes in the institutional landscape promoted by technocrats, and the corresponding critique of neoliberalization, take a rather standardized view of the university that belies the complexities of sustaining an academic career. In this section, we illustrate how working life among computer scientists is differentiated and not structured by simple trajectories. While economic changes have undoubtedly structured careers, respondents talked about how they engaged in many different kinds of work, ranging from teaching and research at universities (sometimes holding down positions at two or more universities), contract research and consulting (in some cases respondents headed their own consultancy firm or nongovernmental organization [NGO]), farming and land ownership (respondents reported that they own small plots of land on the edges of the city or back in their villages), entrepreneurial and business activities, working for the NGO sector and doing community development work, and policy development and implementation.

Respondents perform a ‘donor dance’, positioning their various professional identities in order to access and channel funds into doing research, ⁷ while at the same time transferring knowledge, expertise and technology through their teaching programs. By adopting shifting professional roles, respondents are not solely invested in getting work done; they also sustain their personal lives and that of others by making a living through multiple income sources. Respondents are reflexive about the performative dimensions of their varied working lives. While they are critical of the implicitly technocratic drives of international donors, they acknowledge that doing the ‘donor dance’ lends them some autonomy to make the material changes necessary to earn a living and build up an active knowledge base. In his studies of African knowledge communities, Shrum (2000, 2005) notes how particular professional identities are performed in relation to the distribution of research funding in an African context. He develops the concept of re-agency to explain how actors negotiate their identity so that resources such as funding and equipment can be transferred to them (Harsh et al., 2010; Shrum, 2005). Re-agency is ‘a contingent reaction’ between donors and recipients (Shrum, 2005: 273). Harsh et al. (2010) argue that ‘actors cannot directly control interactions or resource transfers, but can instigate, influence, and “redirect” them’ (p. 175). Given the lack of public funding, respondents acknowledge the negotiated nature and contingency inherent in accessing the resources to do research. ⁸ This generates competition amongst researchers to establish themselves as the regional or national conduit for donor funding.

Florence Tushabe has several high-profile roles at Uganda Technology and Management University (UTAMU) and Makerere University. Tushabe talks passionately about teaching a new generation of computer scientists. She also tells us about her current research, funded by the Clinton Health Access Initiative (CHAI), aimed at lowering child and infant mortality through an assessment of malnutrition. At first, we assumed that her research was funded as an academic collaborator on a global health project. Tushabe corrects us, saying that she is also the Director of an NGO called Free Hearts Uganda and that the funding was channeled through that organization. In her capacity as Director, Tushabe is able to negotiate access to funds from donor organizations, such as CHAI, and her expertise in computer science and interest in development lends her credibility and reliability. Tushabe realizes that her professional identities are to some extent performed. She slips between different roles in order to attract and obtain the funding needed to support many different activities, which include doing the research, teaching, solving health problems, in addition to raising a family and being active in her church community (cf. Shrum, 2005). And while Tushabe aims to alleviate suffering amongst Uganda’s poorest citizens, she also talked in a rather sanguine manner about herself seeking medical treatment from private providers. Her professional status means that she is affluent enough to bypass public health provision. While she may seek to help poorer, less fortunate members of society, she would not voluntarily accept the services to which they have access. The ‘donor dance’ thus contributes to a class division in countries such as Uganda whereby a growing, educated middle class, with increasing access to international funds, redirects their personal income to private providers in healthcare and education. Tushabe’s choice to use private providers both silently acknowledges the chronic underfunding of the Ugandan health system and condones donors’ preference for programmatic rather than systematic approaches to health (Biehl and Petryna, 2013; Crane, 2013).

Despite her accomplishments, Tushabe is stuck with a common paradox that characterizes historic inequalities between Africa and the rest of the world and forms part of the postcolonial condition for Africans (Chakrabarty, 2009; Ferguson, 2006). She echoes those critiques of global health and development by expressing disappointment with the technocratic policies that drive the donors and projects she partners with, criticizing their desire to find magic bullets and techno-fixes. Her concerns touch on the unequal nature of collaborative partnerships and the short-term nature of research projects, the external agenda-setting and the expectations of donors that she perform deference towards them (Biehl and Petryna, 2013; Crane, 2013; Michael and Rosengarten, 2013). In sum, Tushabe speaks to the dominant critique of technocratic approaches that they reinforce the status and limitations of African scientists such as herself. Collaborative research projects often prioritize the intellectual property rights and promotion of scientists and expertise from the Global North, leaving African scientists, such as Tushabe, in what on Malkki (1992) refers to as a ‘sedentarist metaphysics’ fixing them in place. As an example, Tushabe refers to her students interacting with an international research center based in Kampala:

but then the students who go there [to the international research center], they feel as though they are just being exploited. When you go there with a good idea, you present it. Before you know it, you are hearing that they [research center staff] are going to Washington, they are talking about your innovation but they don’t actually give you the support.

Tushabe continues to articulate a call to reclaim science for Africa and Africans. She seeks to carve out her autonomy as a researcher, despite her criticisms of research funding in Africa. The performative dimension of Tushabe’s ‘donor dance’, of her slipping into different professional roles and identities, is crucial to her own agenda to establish home-grown talent that can quite possibly escape the paradox she finds herself in. Through the training of her students, Tushabe is attempting to re-direct (cf. Shrum, 2005) the research momentum of global health, for example supported by CHAI, to create a research trajectory in Kampala. Her personal life and identity, however, chafes with her desire to redirect this momentum, because she still resists subjecting herself to the precarity of using public services.

This skepticism of the donor community is shared by Tim Waema, a Professor at the University of Nairobi, though he sometimes refuses to perform the ‘donor dance’. From previous experience, Waema came to recognize that the research agenda is skewed in favor of supporting the interests of donor organizations and the policy priorities of the countries they hail from (Krause, 2014). Waema refuses to align his professional identity with the expectations of certain donor organizations, even though he is aware it means him not receiving funding.

Very, very rarely can you form your own agenda. Very, very rarely can you form your own agenda. In fact, it is for that reason that I refuse to participate in the EU funding, Horizon 2020, and FP7 before it, because largely we are solving European problems … But me I said no. I did not come to Kenya to solve European problems. I came here to solve our own local problems.

Both Tushabe and Waema invoke ‘Africa speak’ (Ferguson, 2006: 2) to articulate a sense of agency claiming computer science and its benefits for Africans. This is not an unusual response; it has been recorded in other scholarly texts as an expression of self-determination and identity in response to the dominant development discourses that continually disenfranchise African-born innovations and condition the distribution of goods and resources across the continent (see Crane, 2013; Ferguson, 2006). Both respondents have ambitions to determine their own research agendas, and yet they mobilize their identities in markedly different ways. Where Tushabe is reformist by redirecting the project’s momentum into training, Waema is rebellious by totally resisting collaboration with some donor organizations.

While Tushabe and Waema acknowledge the inequalities of research partnerships, respondents are not ‘resisting domination’. Donors are not simply conditioning or disciplining their time. Nor are these researchers implementing policy and programs in a straightforwardly instrumental manner, as technocrats might envision. The re-agency approach helps us see how respondents negotiate their identities to acquire some contingent agency and ‘room to maneuver’ (Hilhorst, 2003). Here we have shown that respondents used funding to not only deliver on the project on which they are contracted to work, but also redirect resources and expertise to animate new technological developments in the direction of teaching and further research, particularly amongst a generation of graduate computer scientists.

There is another aspect, perhaps more obvious, to maintaining multiple professional roles: Respondents are able to make a living in the face of relatively low wages and precarious labor markets. The flexibility of working life is thus also connected to a lack of faith in the strength and sustainability of public institutions, such as the university. A lecturer at UTAMU is explicit about poor pay, when she says:

The staff at Makerere were paid very little money. You can’t believe that a Senior Lecturer, who has a PhD, and has published after her PhD, salary is less than $1,000 a month … So in order to make ends meet … you still have to do something else. You either are engaged in business, you are doing consultancies, or you are doing something; you have to do something on the side – farming, others have other kinds of businesses; so almost everyone else is engaged in some additional income-generating activity.

Whereas low salaries are manifestations of structural constraints, institutions are flexible enough to allow for staff to have many other jobs, even if this involves management ‘looking the other way’. In their publicity, institutions promote broader discourses of entrepreneurialism, economic impact and innovation, but the practicalities of everyday working life appear much more fragmented. Indeed, researchers rarely echoed or even acknowledged the mission statements of institutions, and it certainly did not shape their professional identity and workload.

The flexibility of working life supports multiple professional identities that are not articulated in and through a single institutional vision. Instead, respondents reflect on how they embody a range of identities to pragmatically manage the negotiated territory and precarity of working life in Kenya and Uganda. It is worth remembering that in Uganda, many respondents were born during the political regimes of Idi Amin and Obote II and the devastation of those regimes remains etched into the memories of their family members. While we did not collect data to support claims of an outright lack of faith in public institutions, we did sense precarity in public institutions ability to support livelihoods and career progression (Jamal, 1991). ⁹ Flexible working lives might be a method of managing the traumatic histories of the country by preparing for the uncertainty of working in the public sector in addition to negotiating the political economy of donor funding in order to build up an effective, computer science research community.

The politics of scale

Computer science in East Africa is characterized by two kinds of projects. The first are large-scale, donor-funded projects driven by humanitarian concerns to provide technological fixes to problems of need and deficit. As we illustrate below, these projects enact a modernizing discourse that broadens the horizon of expectations, by imagining better worlds, while closing down the scale of experience in the present time. The second kind of project is smaller in scale, more diffuse and adds a granular texture to computer science in these cities; it is produced through the redirection of momentum and resources generated by showcase projects. Our fieldwork led us first to the more visible and accessible large-scale projects. Yet, through repeated conversations with respondents, and in spending time in the computer science labs of Nairobi and Kampala, we began to see the proliferation of small-scale projects that were interfacing with the larger ones – borrowing bits and pieces of technology, and connecting with clusters of other projects – but taking novel formations that develop in almost fractal formations (on African fractals see Eglash, 1999).

We began to understand a little more about what Omwenga meant when he referred to Nairobi’s tech sector as hidden from the view of international metric systems. We witnessed an almost cell-like proliferation of computer science projects developed by students armed only with a laptop and some storage, intent on solving everyday problems they or their family and friends experienced. These small-scale projects were not tied to donor organizations, and thus they were not as responsive to the discursive framings of need set up by big grants. Indeed, they exhibited the kinds of fluidity and networked activity that reflects one of the main claims of STS that technology and society are emergent and co-constitute each other. To illustrate the difference in scale between these two types of projects and to produce a critical reading, we turn to the example of a showcase project, the Digital Matatus project, and to a series of vignettes about working in computer science at a smaller scale.

One of the collaborating Principal Investigators, Peter Wagacha, professor and coordinator of the PhD program in computer science at the University of Nairobi, introduced us to the Digital Matatus project. ¹⁰ Designed in partnership with US universities and funded by the Rockefeller Foundation, the goal of the project was to create a digital map of Nairobi’s informal, private, paratransit matatu bus system. Wagacha oversaw data collection that took two years to complete – by master’s students at the University of Nairobi boarding matatus and using the GPS on their phones to track routes. Wagacha then applied his expertise in machine learning (previously applied to other areas such as natural language processing) to design algorithms to help create the map itself.

The Digital Matatus map was completed in 2015 and the graphical depiction of the data was designed to emulate the London Underground map by representing an enhanced, spatial visualization of Nairobi. The map brings a version of the city into view that is conversant with universal models of how people navigate cities around the world. In this regard, it aims to be scalable and transferrable; a map of Nairobi could potentially mirror that of Manila or Lagos – other countries that are targeted for hosting paratransit informal public transport.

The Digital Matatus map is warranted by an assumption that Nairobi is chaotic and over-congested; problems that lead to pollution, ill-health, inefficient working patterns, protracted travel times and so on. Data science was explicitly applied to make the matatu buses legible, which, according to the project leaders, will lead to greater efficiency and better usage (Klopp, 2014). The map is intended to ease people’s navigation of the city and thus aims to improve the livelihoods, health and wellbeing of urban residents. Achieving these ends, the creators of the map envision its uptake among policy makers and regulators of city-wide transportation. With an emphasis on creating legibility (Scott, 1998), the Digital Matatus Map exemplifies Porter’s (1992) definition of a quantification ideal that silences and externalizes the social conditions of its own production. However, to function as a quantification ideal, the map has been in some way representative and useful to Nairobi civic governance. We challenge the representativeness and use-value of the map, and develop a different interpretation arguing that it is aimed more at advancing digital data markets.

Wagacha describes the amount of labor and resource that went into collecting the data, and the problems inherent in generating a consistent and continuous supply of data. The data had to be formatted to Google’s General Transit Feed Specification (GTFS), ¹¹ which requires a regular feed of data. The matatu system, especially because of its competitive and constantly changing nature, would require updating at least twice annually to be in some way representative. Thus, the labor of local students and faculty to collect the data that undergirds the map’s representativeness is not commensurate with the requirements of globally standardized systems. Neither the local government nor the University of Nairobi have the funding to support the kind of labor and resources required to keep pace with GTFS. As Wagacha explains,

We had a challenge … that Google is looking for people who can provide the data in a continuously updated (manner), because it’s not very useful if it’s not updated. You understand the matatu system here … there’s no way they can start collecting that data and sharing it right now. So between now and the time that we are able to perhaps collect the data, then we need to figure out how this process can be supported for maybe the next one or two years. We need at least two cycles every year … just to refresh the data.

This resource problem renders the representational value of the map obsolete. This does not necessarily matter to Nairobi residents, because they have been using the matatus for decades without the aid of any maps (Mutongi, 2017). However, long standing critiques of quantification argue that numbers are not purely representative, they configure the object they aim to measure (Porter, 1992, 1996). The matatu map is thus not innocent, it transforms the city-scape into a series of data-sensors, through which governance can be envisioned as a future goal; the map calls forth centralized transportation policy, regularity and rhythm in transit provision, and understandings of road worthiness and land use that are based on data science.

According to critical geographers such as Harvey (2001) and Lefebvre and Nicholson-Smith (1991), geographical scale, such as the city-scape, becomes fixed and deterministic through capitalist expansion. In the Digital Matatus project and beyond, we found that respondents involved in showcase projects funded by donor and philanthropic organizations described their research using discourses of scale that resonated with that of their benefactors. The continent, the region (‘East Africa’), the city (‘Kampala’, ‘Nairobi’) are configured as fixed, determinable geographical units in which problems happen and from which the projects and programs can be scaled up, standardized and universalized. While representations of scale pervade donor discourse across all sectors, the difference with computer science is that scale can be modelled, quantified and programmed into the technology itself – into software, algorithms and apps, where it becomes numerical. Space, such as the city, is extracted in order to form the basic analytical unit through which knowledge of say, transportation (in the case of the digital matatus project) is generated and then transferred quickly and cheaply to other contexts. The politics of scale become ever more remote from the practices of doing computer science in these showcase projects. Through layers of extraction and modelling, place, history, politics are de-contextualized and externalized in the service of producing a clean, transferable representational artefact (Bowker and Star, 1999; Star, 1999).

Flagship programs are indelibly infused with grand designs for governance. Similar to other flagship programs we encountered, the Digital Matatus map has yet to improve socio-economic conditions and, to the best of our knowledge, it has yet to impact on transport policy, though UN Habitat is currently using the Digital Matatus data to plan routes for proposed Bus Rapid Transit in Nairobi. However, the Matatus project has had tangible effects in developing careers, training graduates and doing the actual science at the University of Nairobi. Working on the project has increased Wagacha’s visibility to international researchers, who have sought him out to work on new collaborative projects and his graduate students have benefited from a more enriched learning experience.

In addition to the flagship projects, we encountered a granular layer of computer science activity. Essentially, to generate computer science all you need is a laptop, an internet connection and some cloud storage in order to begin working on a project. The lightness and cheapness of the field has stimulated new synergies in research. This is an open field of possibilities inflected by, but not locked into, the political economy of showcase projects.

The small-scale use of computer science is promoted by Hugh Cameron, a visiting professor at Makerere University. Cameron hails from Canada and has spent over twenty years working in the private sector conducting telecommunications research. He foresees the future of computer science in Kampala to take the form of small-scale entrepreneurial projects that are interconnected in multiple ways and work towards problems-solving at discrete, highly contextualized and localized levels. This scaled-down version of computer science takes a more polymorphous form. While smaller projects need not conform to the dictates of large donor programs, they share a sameness influenced by similar frames of problematization and a desire to generate scalable technologies that are capable of making money. However, in other ways, students of computer science do not need to converse in the same language of the donors, despite working with the same technologies. While scale and space are hardwired into computer science, the smaller projects demonstrate that these concepts are also emergent through language and discourse. Thus, to use Verran’s (2001) dictum, scale and space are recursive concepts that share sameness and difference, patterned both discursively and materially across projects large and small.

It tended to be the younger respondents, often graduate students (master’s or PhD students), in our dataset who were putting Cameron’s vision into practice. Through their Safaricom Academy, Strathmore University’s @iLab graduates an average of 40 master’s students every year. The University of Nairobi’s School of Computing and Informatics graduated 111 master’s students in 2016 (The University of Nairobi, 2017). In the same year, Makerere’s College of Computing and Information Sciences graduated 63 master’s students (Tuhereze, 2016). As part of their degrees, master’s students in both countries are required to conduct an applied research project.

Sarah Kigongo (a pseudonym), a master’s student at Makerere working with Cameron, is building on research developed by Ernest Mwebaze, who, as mentioned above, has created a mobile phone application that analyzes photographs to assess the extent of viral plant disease. Kigongo’s plan is to use the mobile software developed by Mwebaze’s team to analyze wet tissue. Her idea is to build a device that can be attached to a mobile phone and make simple diagnosis of infectious disease. Kigongo’s motivation is to save time by eliminating hospital visits and cut down wait times. She complains that people travel great distances to hospitals and health centers, and wait for hours to receive even the simplest blood test, which has the effect of causing other problems, such as missing work and finding childcare. While she understands the difficulty involved in creating such a device, Kigongo’s laptop is adorned with a sticker reading ‘get shit done’, which she gleefully points to as her mantra. While plucky in some regards, the sticker does acknowledge more serious forms of tutelage and apprenticeship because in order to ‘get shit done’, Kigongo has to learn the science underpinning Mwebaze’s application and in that process, she will also acquire the tacit knowledge that comes from doing computer science. She is learning to think like a computer scientist, and part of this involves the diagnosis of social problems. Kikongo, however, need only frame problems in terms of familiar challenges, rather than the ‘global’ ones that operate within large, determinable scales.

Another example is John Njue, a master’s student at Jomo Kenyatta University of Agriculture and Technology. Njue expresses an entrepreneurial spirit characteristic of Nairobi, with its labyrinth hubs, labs, incubators and accelerators. He says that he wants computer science to be commercially-relevant, making improvements to existing products.

Although the general research is important in the extension of knowledge which is ok, but I would wish we can have a way of trying to encourage people to look at existing products, looking at their weaknesses, doing research on those products publish the results, let the products absorb these people. Let them take this product to the next level.

Njue’s master’s research focuses on algorithms that can track, without the use of cameras, where a user is looking on a webpage; this could improve targeted advertising and make the web more accessible. In addition to being a master’s student, Njue often provides IT support to the university. He saw a clear unmet need in his university department: Globally, software is moving to web applications and so-called cloud-based computing, however in Kenya many organizations cannot afford their own web servers. He created a solution as a side-project: a desktop application that mimics a web server, which he sees as a novel and publishable idea, and one that has local commercial viability.

Computer science is taking shape in Nairobi and Kampala in the shadow of showcase projects. As Brian Omwenga stated, if we ‘shine a light’ on Nairobi, an active tech community is illuminated that is vital to the city’s economic development. Similarly, in Kampala, a community of academic computer scientists are revealed. Across the region, high-profile projects are having a trickle-down effect, inspiring leagues of graduate students to spearhead their own personal projects. The larger projects are more visible because they generate media and are subject to forms of accountability and evaluation that make them appear singular, logical and self-contained with direct, measurable impacts. In consequence, they often conceal or obscure other research activity in and around computer science, helping make Africa appear science-less – a perspective that actually legitimates further investment in big projects. However, the intermingling of large showcase projects and granular small-scale ones constitutes the fabric and expansion of computer science in African societies.

Conclusion

The growth of computer science is a discernible phenomenon occurring in East Africa. Our aim was to understand how computer science is made and what it means, how the everyday institutional and material realities in each city and its universities are co-constituted with the processes of knowledge creation. Our analysis draws on recent work in the history, geography and anthropology of globalization and science that adopts more relational approaches to understanding the interplay between different analytical understandings and material contexts. In particular, Tsing’s (2011) metaphor of ‘friction’ helps us to situate computing knowledge in Nairobi and Kampala and shed light on the international circulation of labor, the negotiable terrain of professional life, the shifting structures of higher education institutions, and the sliding scales of computer science research.

Across these four areas, the dynamism and movement that we use to describe the technological communities of East Africa also provide analytical tools that aid our understanding of how knowledge is co-constituted with social, institutional and economic worlds. When examining the labor force, there is a global circulation between East Africa, the US, Europe and Asia. The strong draw of doing computer science in Africa stems from a moral economy, afforded by privilege, that satisfies respondents’ aspirations to do good and pursue their intellectual desires. In Nairobi, traditions of entrepreneurship and commercialism create new opportunities to do and apply computer science, whereas in Kampala, Makerere University has deepened its academic credentials. This has led to atypical accumulation of labor power in Africa that potentially helps redress long-standing inequalities of wealth and development. The opportunity to pursue a career in computer science research is necessitated by the neoliberalization of higher education, which, rather than having a blanket effect, emerges through fluid and fluctuating institutions that are continually interpreting and responding to economization. This sliding institutional landscape is reflected in the shifting and multiple identities respondents adopt in order to manage their professional and personal lives on many fronts. Respondents are confronted with the scarcity of funding, demanding teaching and administrative roles, and the pressure to sustain their own livelihoods and that of their extended families and communities. They redirect the prestige and resources that come from collaborative donor funded projects towards training graduate students and stimulating a local knowledge base. There is a precarity to sustaining professional life that vitalizes granular computer science research. Flagship projects may have provided us with our ethnographic entry points, but they led us to an understanding that computer science research proliferates in almost fractal formations, implying that patterns repeat and return when seen from different perspectives.

There is considerable emphasis in STS on epistemes and the entailed political visions – and in health and humanitarianism, the biopolitical regimes – they articulate through, for example, technology transfer, fixes and interventions (Harding, 2011; Jasanoff and Kim, 2013; Redfield, 2016). Technologies are argued to carry designs for life, or at least designs for the management of life (Jasanoff, 2005; Winner, 1980). While STS scholars have argued convincingly that politics are written into things like algorithmic code, software applications and hardware, their approach has tended to lose sight of traditional sociological dimensions. By situating our analysis in the computer science communities of Nairobi and Kampala, the data we collected led us in some ways back to these social structures and relationships. In examining the dynamics of labor, institutions, identities and scale, we not only directly questioned the power and presence of those grand designs for life (especially in large, donor-funded projects), but we also recorded so many other narratives that contribute to the burgeoning literatures on digital cultures, data science, ICT and computing in African societies. There is renewed energy for understanding the impacts of computing and digital cultures, and for approaching these as significant of new ways to read the continent and its place in contemporary histories of science and technology. This may represent what Mavhunga (2014, 2017) and Comaroff and Comaroff (2012) have repeatedly argued is a turn to Africa. What this revitalized body of literature, to which we contribute, provides is a methodology invested in moving us away from deficit and despair towards hope and optimism in studying science and technology in Africa.