Higher Education in Sub-Saharan Africa
Computing and Online Knowledge
Are computers necessary for a university education? It is possible to teach most subjects well and thoroughly without expensive technology. But computers offer a solution to a problem identified by all the students surveyed: outdated textbooks and course materials. They are also vital for much of modern science, engineering, and business. Leaving students without programming skills may leave them underprepared for graduate school or unemployable in industry.
-- read this survey from Ethiopia for the trauma of trying to write a research paper without textbooks, research journals, instruction, or an internet connection.
Programming is essential for at least graduate-level science, engineering, and business: all use computer-based data analysis, modeling and numerical simulation. Computer use for data analysis can also, if taught well, help focus science instruction more on problem solving than on memorization of received knowledge. The students surveyed here, mostly participants in an international postgraduate program for science and math graduates, universally report that its chief benefit was practice in scientific computing. They also cite lack of programming instruction as one of the chief faults of their undergraduate educations.
-- read this survey from the comparatively wealthy South Africa for the conditions for learning programming at understaffed and underequipped universities
Hardware. The most common response that students gave to the question "what can international donors do for African universities?" was "provide computers." Although lack of instructors in programming may eventually become an issue, at present the first need of African universities is hardware. One student surveyed described programming classes where all programs were written on paper; there were no computers available for running them. Ahmadu Bello University in Nigeria, with 30,000 undergraduates, had exactly 19 computers and 1 printer in its libraries in 2006 (see this study of Nigerian universities).
At maximum the laptops could also be used for scientific computing. They use the same Linux operating system as do scientific workstations and can run the standard open-source software used for scientific programming and publishing (e.g. Python, TeX). The laptops have relatively small data storage and memory (128 MB), but are sufficient for most programming courses and can be upgraded for more substantial data analysis.
The students surveyed all mentioned out-dated and overly theoretical curricula as a failing of their universities. This is a symptom of low spending (lack of money to buy new textbooks or install laboratories) and of isolation (professors have little access to science outside their own graduate training). Both can be remedied given internet connections.
Lecture notes and curricula. There is a growing movement in the U.S. and elsewhere to make university-level educational materials freely available to all. A leader in this movement is MIT with its Open Courseware project, in which professors place their lecture notes, problem sets, exams, and even videos of lectures online for all to use, reaching 1.5 million users each month (see also this summary and articles from MIT and Information Week). All of MIT's courses will be on the web by the end of the 2007. Other universities participating in open courseware efforts include Tufts, Johns Hopkins, U.C. Irvine, Univ. of Notre Dame, Utah State (see here), and Yale (see article). The Open Educational Resources Commons also gathers university course material and videos of lectures and demonstrations. The movement is not confined to the U.S.: Britain's Open University, a fee-charging distance-learning university, is now making some courses freely available (see article).
Textbooks. The movement to make educational materials freely available extends to textbooks as well. Numerous authors have placed their work online for others to use. Wiki Books includes texts on many subjects. Individual highly-regarded textbooks available online include the science text Motion Mountain, with 30,000 downloads a year, and others in computer science , physics, and many other subjects. Project Gutenberg (2 million downloads a month), Google Books, and other digital library projects are placing out-of-copyright volumes online for anyone to read.
Economics of science publishing. Internet access and open courseware can be a great asset for African universities, but cannot alone fully connect African universities to the international scientific community. African universities would still lack access to the journals in which essentially all of academic research is published.
How can access be opened? One possibility is to persuade publishers to make journals available at low or no cost to developing-world universities. The UN has begun an archive for agricultural research (AGORA), and the WHO sponsors a similar archive for health studies (HINARI). (See here for e-resources at the University of Dar es Salaam in Tanzania; note their use of AGORA). Some individual publishers have also taken independent steps to open access (see article about the Royal Society for Chemistry). But a broader and simpler strategy is to pass legislation mandating that all federally funded research results must be placed in publicly accessible archives. The Federal Research Public Access Act of 2006 (Cornyn & Lieberman), introduced in the 109th Congress, would have required this (see also this summary). This legislation would force a reshaping of the publishing world that would have widespread support from libraries, universities in both the U.S. and developing world, and scientists themselves. Most people agree that the current system must change, but individual scientists feel constrained to publish in prestigious existing journals and publishers have no incentive to release copyright on their articles. U.S. legislation can break this logjam. Open research access is supported by many groups, including the American Library Association and the Alliance for Taxpayer Access.
The proposals above for strengthening African universities with online knowledge depend on being able to actually access that knowledge. That would not have been possible a decade ago: the American Association for the Advancement of Science (AAAS) found in 1999 that students could not download research articles from a number of African universities because data transfer rates were so slow that connections were dropped (see report). In 2007, using online resources in sub-Saharan Africa is a possibility. Still, serious challenges remain.
High costs. The most remarkable aspect of the rising demand for internet connections in sub-Saharan Africa is that it is occurring despite the the highest connection costs in the world. The African consumer (or university) pays 50-500 times more than an American for an equivalent connection (e.g. $3,000/month instead of $30/month for a 1 Mbps connection; prices vary by country). African universities therefore cannot afford the bandwidth they need to make efficient use of online resources. The average African university, with tens of thousands of students and faculty, has the same aggregate bandwidth as a single household connection in the U.S. (see article or this study by the Partnership for Higher Education in Africa). The principal reason for high costs is the lack of optical fiber infrastructure (discussed below).
Optical fiber links. Optical fiber infrastructure is the cheapest and most efficient way of transferring data. Only 14 of 49 sub-Saharan countries have any fiber connection to each other or to the rest of the world (NEPAD, 2004). The remainder must use expensive satellite or radio connections. A single cable runs along the West coast of Africa. Few overland networks penetrate the interior, and East Africa is completely isolated.
Lack of guaranteed open access is a concern because a single cable operated as a private monopoly will not generally reduce costs for the consumer. Although West Africa is now linked to Europe by optical fiber (the SAT-3 cable), communications costs there are generally no lower than via satellite. In the absence of strong regulatory agencies, the consortium of investors who funded the cable have kept bandwidth costs prohibitively high. The SAT-3 pricing strategy has been termed "high cost, low volume": bandwidth is sold for $4500-$12,000 per Mbps/month and the cable is underutilized. (Prices differ by country; the $4500/month figure was obtained only after a 2-year court fight in Ghana). Even the low Ghana price is 20 times higher than could be offered by a non-profit cable (less than $250 per Mbps/month, estimate by Eric Osiakwan of the African Internet Service Providers Association). The original SAT-3 operating licenses expired in June 2007 and are being renegotiated, possibly opening an opportunity to drop communications costs for W. Africa. (See article and commentary ). The Association of African Universities has pleaded for, if nothing else, special rates for universities (see article).
Note that predatory pricing by the SAT-3 monopoly consortium is not a purely African problem. Although many of the SAT-3 consortium members are African telecoms, the three largest investors were non-African: in order of stake, TCI (at that time a subsidiary of AT&T) + AT&T itself (U.S.A.), France Telecom (France), and VSL (India/Singapore). The U.S. stake in the cable may now have passed to Comcast. (SAT-3 ownership information is carefully guarded).
The E. African cable proposals, also consortium operated, have raised similar concerns about monopolistic pricing (see commentary here, and here, analysis by the APC, and the APC report mentioned above). EASSy ownership is dominated by South African telecoms (which have given South Africa some of the highest telecoms costs in Africa). Other consortium members include AT&T (USA), Verizon/ex MCI (USA), France Telecom, BT (UK), Saudi Telecom, VSNL/Teleglobe (India), and Etisalat (United Arab Emirates), and 28 other public and private Africa-based telecoms. If EASSy or another single East African cable operates on the SAT-3 model, E. African consumers will see no price benefits and E. African universities will remain isolated. Either multiple competing cables or regulated or non-profit operation are necessary to lower costs for African internet users.
Updates on both SAT-3 and E. African cables are posted by Fibre for Africa. The subject is also frequently covered in the online newsletter Balancing Act which reports on telecoms and internet in Africa.
Reverse Subsidies. An additional factor raising connectivity costs for Africa is a perverse pricing structure in which African users subsidize all data transfers to and from the continent. (Imagine that you paid telephone charges on both incoming and outgoing phone calls, while your neighbor paid for neither). These charges alone nearly double the cost of African internet usage, costing Africa between $250-500 million/yr. They also further entrench the stalemate: the African market remains small, investment in infrastructure is discouraged, and the companies that provide international bandwidth (IBPs) can continue to insist on predatory pricing arrangements.
Why does international connection cost matter so much to local users within Africa? Because the lack of optical fiber lines and local data aggregation points means that even traffic within Africa is typically routed through Europe. An email sent from the Central African Republic to nearby Kenya, for example, is routed through London via two satellite connections. One means of remedying the pricing inequity is an international trade agreement. Another is to build up local land-based networks and regional aggregation points to reduce the need of European connections in the first place. Aggregation also pushes the international providers to come to Africa instead, after which they would bear connection costs equally.
Current actions. Many governments in Africa are now planning and building infrastructure to reduce the need for intercontinental traffic. East African countries have been most aggressive in pursuit of this goal, with Rwanda as the strongest leader. Rwanda, Kenya, and Uganda are all constructing inland optical fiber backbones (Uganda with funding from the Chinese government). Tanzania is at least in the planning stage (see report "Optical Fibre for Education and Research Networks in Eastern and Southern Africa"). Connectivity Africa, sponsored by Canada's International Development Research Center, has also done preliminary work on aggregation in six countries (South Africa, Mozambique, Kenya, Nigeria, Uganda and Tanzania). A coalition of African Internet Service Providers has proposed a broad outline for regional aggregation and cost reductions (the "Halfway Proposition"). And the United Nations and the Rwandan government are sponsoring a telecommunications conference in Kigali in October 2007 to bring together African governments, businesses, and global telecommunications firms to discuss infrastructure and regulatory changes to boost connectivity ("Connect Africa"). Note however that all of these efforts still depend on construction of an East African submarine cable. It is difficult to orchestrate any solution to Africa's connectivity problems if the continent remains digitally isolated.
Additional factors for universities in particular. One additional factor in connectivity costs that may affect African universities in particular is lack of capital to buy equipment that would result in eventual savings. A AAAS study of connectivity in African universities in 1999 found, for example, that Makerere University in Uganda paid steep monthly phone bills for a dialup connection because they could not afford a one-time cost of $18,000 for a radio link. These kind of situations may or may not exist in 2007.
Helping universities in Africa. Several organizations work with African universities to help use existing bandwidth more effectively. The U.K. International Network for the Availability of Scientific Publications works with African university libraries. The eGranary Digital Library project places widely used free academic materials on local servers in African universities so that they can be accessed with only a local connection.
Many organizations are working to lower bandwidth costs for universities and increase capacity. Several countries in Africa have formed National Research and Education Networks (NRENs) seeking high-speed and affordable connections for universities, and have organized as the UbuntuNet Alliance. (Participating nations are Kenya, Malawi, Mozambique, Rwanda and South Africa). The target goal is university connections equivalent to those of the developed world: 1 Gbps or more.
The Partnership for Higher Education in Africa (including the Ford, MacArthur, and Rockefeller Foundations) has helped a consortium of 13 African universities to lower connectivity costs. The Partnership has
donated over $5 million to make satellite bandwidth available to the consortium at $2330 per Mbps/month instead of $7300 (see also here). This effort will be welcomed by students and researchers. It is not a long-term solution, however, and some local connectivity activists argue that giving subsidies to foreign satellite firms actually hinders long-term solutions. Even with the Partnership's support, universities would pay over $1800/month for the same household-scale connection that Verizon advertises in the U.S. with an introductory rate of $9.99/month.
Purchasing the bandwidth of a normal U.S. university (1 Gbps, the UbuntuNet target) would cost an unaffordable $2.3 million/month.
The digital divide will ultimately be bridged only with optical fiber, and those investments are large in scale.
Last update: October 2, 2007