Version classiqueVersion mobile
OpenEdition Books

Arguing about justice

Axel Gosseries
Philippe Vanderborght

A mobile water project: mobile-for-development meets human-centered design

Sangick Jeon, Eran Bendavid, Joshua Cohen, Katherine Hoffmann et Terry Winograd


Dans cet article, nous décrivons un projet de "téléphone mobile pour le développement" baptise M-Maji – "Eau M(obile)" en Kiswahili. M-Maji, qui en est encore a ses balbutiements, a pour objectif d'améliorer l'accès a l'eau potable – et donc d'améliorer le bien-être et la libérté elle – dans les bidonvilles des pays en voie de développement, grâce a l'utilisation des technologies mobiles désormais largement accessibles même dans les communautés les plus pauvres. Les projets de développement fondes sur la téléphonie mobile sont souvent ineffectifs, en raison de leur forte focalisation sur les aspects technologiques, au détriment des besoins, préférences et capacités des utilisateurs. Pour éviter ce problème, nous avons développé M-Maji en le concevant comme "centre sur l'humain" – soit une approche ancrée dans l'engagement ethnographique avec les utilisateurs finaux. Si M-Maji fonctionne, il permettra de "capaciter" (empower) les communautés désavantagées, grâce a une meilleure information quant a la disponibilité de l'eau, de son prix, et de sa qualité.

Texte intégral

1In this short essay, we sketch a project in process. The project is focused on access to clean water in Kibera—the largest of Nairobi’s intensely poor, densely populated, humanly vibrant informal settlements. Like many of Philippe van Parijs’s projects, from universal basic income to a bike lane in Brussels, ours is guided by a normative concern. In particular, our aim is to increase real freedom by ensuring that people who are now economically marginalized and lacking access to basic goods have easier, less-costly access to clean water—surely essential to making any use of our formal freedoms. Moreover—and again like much that Philippe van Parijs has done in his admirable career—our project sits at the intersection between theory and practice. We are developing and testing a practical idea that is informed by normative and positive theory. But, as we write (in August 2011), we are entirely unsure whether it will work. Precisely for this reason, we think the project belongs in this volume, with its animating spirit of new exploration.


2Kibera, located a few miles southwest of central Nairobi, is one of Africa’s largest informal urban settlements. Population estimates vary widely, but roughly 200,000-250,000 people now live on 630 acres, an area two-thirds the size of New York’s Central Park. Most Kiberans live on under $1 a day, and dwellings generally have one room, no electricity, and no toilet. Moreover, residents in Kibera, as in many of the world’s informal settlements, typically have no direct access to running water. So most Kiberans (roughly 75-85%) acquire water by making daily purchases at kiosks from private vendors, who (legally or illegally) tap into water supplied by the Nairobi water company. Because buyers typically purchase water from nearby vendors, the distances they walk to purchase water are not great—though considerably greater than any reader of this volume travels. In addition, while affordable for most families, the standard price for a 20-liter jerry can is eight times the lowest tariff for domestic connections and four times the average tariff in Kenya (World Bank 2005).

3However, in Kibera, as in many informal settlements, water shortages occur frequently—on average, once a week. And when shortages hit, buyers need to find new water vendors. The shortages drive prices up dramatically: from 100-600%, depending on the village in Kibera (Jeon 2011). And because the geographic scope of the shortages—which often result from breaks in pipes—is so uncertain, search costs are high: buyers search, on average, for one-and-a-half to two hours, carrying a 20-liter jerry can (Jeon 2011).

4Moreover, on good and bad days, water quality is suspect. Buyers wonder—so we have learned from our conversations in Kibera—whether they are getting a jug of water or a bucket of cholera vectors. But while Kiberans often lack easy, low-cost access to water and face pervasive uncertainty about water quality, mobile phone penetration is very high: in most of Kibera’s 13 villages, nearly every household has at least one mobile phone (Jeon 2011). And growing competition in Kenya’s mobile market—once fully dominated by Safaricom—has very substantially reduced the costs of both voice and SMS communication.

5Our project, called M-Maji—Kiswahili for ”M[obile] water”—aims to improve the opportunities and welfare of Kiberans by exploiting virtually ubiquitous, low-cost mobile access to improve limited, high-cost water access. Mixing moral-practical purpose with field-experimental social science, M-Maji aims to enable buyers to use their mobile phones to acquire information about water location, price, and quality. It builds on and tests the main claim in the growing social-scientific literature on ”mobile for development”: that mobile technology is effective at reducing search costs (Jensen 2007, Aker & Mbiti 2010). Such reduction is, in turn, typically good for creating more efficient markets (Baye, Morgen & Scholten 2007). More contextually, it builds on local enthusiasm about mobile phones, which is driven in part by the remarkable growth of mobile banking in Kenya. MPesa, Safaricom’s mobile money-transfer system, has fueled an extraordinary expansion in access to financial services since its creation in 2007, with adult access growing from 5% prior to M-Pesa to 70% this year (Jack & Suri 2011). In 2010, 15 million Kenyans used the system and pushed 20 per cent of national GDP through it.

6If M-Maji works, water vendors will use their mobile phones to advertise water availability, with information about location, price, and quality; water buyers will use their phones to query the M-Maji database to find the closest, cheapest, cleanest water. M-Maji will thus reduce information asymmetries that now impose large burdens on purchasers, enhance water market efficiency, and improve consumer welfare.

7M-Maji was designed by an interdisciplinary team of Stanford University graduate students (Anuraag Chigurupati, Katherine Hoffmann & Sangick Jeon) in Designing Liberation Technologies, a course taught in Stanfords Hasso Plattner Institute of Design ( by Joshua Cohen and Terry Winograd. The guiding philosophy of the is human-centered design—an approach to problem-solving that is applied to designing products, processes, and experiences, and is anchored in ethnographic engagement with users, pursued in interdisciplinary teams, and focused on innovative solutions.1 The Cohen-Winograd course, which is coordinated with the University of Nairobis School of Computing and Informatics, grows out of current enthusiasm for mobile solutions to economic and social problems in developing countries. Some of that enthusiasm seems well-founded (Aker & Mbiti, 2010). But mobile (and other informational technology) solutions are often poorly designed because they reflect very little understanding of the lives of users (Toyama 2010). The idea of designing liberation technologies is to deploy human-centered design to develop projects with greater chances of success because they emerge from a real and sustained engagement with end users.

System Design

8M-Maji will work in three steps (see Figure 1 for a system overview):

Fig. 1: System overview

  • 2 USSD (Unstructured Supplementary Service Data) is a technology available on basic GSM phones. In Ke (...)

9Step 1: Each day, water vendors notify M-Maji via USSD2 that they have water to sell, where they are selling it, and the price they are selling it for. They also have the option to advertise the last date of water purification and the results of any recent quality testing. These advertisements are collected and stored in the M-Maji database in real-time.

10Step 2: Water buyers initiate a USSD session with M-Maji, and obtain a location-relevant listing of local water vendors who have sent notifications of price, location, and quality, as well as their vendor rating (see Step 3 for vendor ratings). Within the same USSD session, users are given the option of re-sorting the listings by price, quality, vendor ratings, or another geographic location.

11Step 3: If a vendor misreports availability, price, or quality, buyers can file a complaint with M-Maji via USSD. The M-Maji team will then review, investigate, and evaluate these complaints, and assign vendor ratings to alert future buyers of negative histories (the details of this part of the system are still in development).


12M-Maji aims to improve access to clean water by providing low-cost information about water availability, price, and quality. By pooling water information from multiple sources, it provides information beyond what other search methods (e.g., gossip and word-of-mouth) deliver. It is also economically sensitive, relying on basic GSM (Global System for Mobile) phones, which are already broadly accessible in informal settlements. Moreover, it operates free of cost for individual users (USSD costs are subsidized). Information accuracy is encouraged by the vendor rating system and by the M-Maji support team on the ground, which will monitor data quality with regular surveys and random evaluations (for example, through drop-in testing of water quality).

13M-Maji is a novel approach to water supply problems in informal settlements. Alternative approaches aim to increase water supply or quality. M-Maji tackles the information side of the problem. People struggle daily to locate water and evaluate its quality, and may pay much higher prices than they would with better information. By providing better water information to consumers, we seek not only to reduce the individual burden of finding water, but also to equalize water prices across villages of Kibera, and make clean water affordable to larger segments of the population.

14M-Maji’s promise is supported by a number of studies on water in developing countries. Randomized field experiments, for instance, have shown that consumers who receive information regarding their water quality are more likely to seek out a clean source and purify their water (e.g., Jalan and Somanathan 2008). The demand for water is also estimated to be highly elastic at low prices but inelastic at high prices, indicating that clean water uptake can be increased through a price reduction mechanism (Kremer et al. 2009). And, more generally, mobile-for-development research has shown the promise of mobile technology in improving the efficiency of certain product markets by reducing information costs (Jensen 2007, Aker & Mbiti 2010). Whether this same logic extends from markets for perishable goods (fish) and goods of relatively uniform quality (grain) to water markets is one of the issues we hope to address.


15As a first step in developing M-Maji, we are conducting a randomized impact evaluation in Kibera to test our system’s potential for improving clean water access in urban slums. Randomized impact evaluations are the gold standard of impact evaluations (see Duflo et al. 2006), and they proceed by randomly assigning members of a population to a ”treatment” or ”control” group—groups that will or will not receive a particular intervention, respectively. After intervention, outcomes are compared across treatment and control groups to obtain statistically convincing estimates of a program’s impact.

16We are assessing M-Maji’s impact on water prices, search costs (in terms of time and distance), and clean water uptake (under-5 diarrheal infection rates and household water quality).

17To conduct this program evaluation, M-Maji has partnered with Umande Trust (, a non-profit community organization (funded in part by the Bill and Melinda Gates Foundation) that works on water and sanitation issues in under-served communities throughout Kenya. In collaboration with Umande, we have been collecting baseline data to obtain pre-intervention benchmarks on clean water consumption and supply. After completing baseline data collection, we will move the project into a pilot phase during which we introduce M-Maji to users in a random selection of villages in Kibera (our treatment group). We will then compare these users to people in villages that did not receive M-Maji services (our control group) to estimate our impact on the three water outcomes of interest: prices, search costs, and clean water uptake. Umande Trust will lead the pilot implementation: they will market M-Maji to communities, collect and spotcheck accuracy of field data, and obtain user feedback.

18The program evaluation serves both practical and scientific aims. Practically, running the pilot in this way will generate user feedback in the short run, which will be important for designing M-Maji to be maximally effective for users. In the long run, this evaluation will permit an estimation of cost-effectiveness, which will inform how we proceed with this project (including scale up and system modifications). If successful, the project will generate real human benefits.

19Scientifically, we hope to gain a better understanding of the scope and limits of information-based approaches to water problems in developing countries. The project’s focus on data collection will also contribute to our understanding of access to basic public services in informal settlements, a setting for which data is scarce and unreliable. It will help to test the idea that mobile applications improve market efficiency by reducing search costs. And it may suggest lessons for the innovative efforts to use human-centered design to develop sensible mobile-for-development ideas that Cohen and Winograd explore in their course.

Future Directions

20M-Maji represents, in our view, a second-best solution to water problems in informal settlements in developing countries. The best solution would not be to improve a poorly functioning private water market, where much of the supply is in any case of dubious legality; rather, it would be a formal and efficient water and sanitation system provided by local and national governments. Like the U.N., and like Umande Trust, we believe these services should be provided as a matter of basic rights. But for many reasons—including political corruption, insecure land tenure in informal settlements, and extreme poverty—many communities continue to be deprived of them. In collaboration with Umande Trust, we are developing M-Maji as a temporary solution for such communities.

21For the communities that M-Maji does serve, we expect that it will prove to be a cost-effective solution to problems of water quality and access. Because the project uniquely focuses on the information side of the water problem, we also expect it to supplement other water projects, which address issues of supply and quality. If results from the program evaluation are favorable, the next steps include scaling up, both in Nairobi and in other informal urban settlements in Sub-Saharan Africa. In fact, M-Maji may help in addressing water problems in rural and suburban settings as well. Indeed, it should help in any location that has multiple sources of water and pervasive information asymmetries that impose significant burdens on the time and money—and thus the real freedom—of poor consumers.

22Finally, if M-Maji works on its principal domains of application, future extensions might incorporate a method for tracking cholera outbreaks and reporting leakages—the latter being an avenue by which we might indirectly increase water supply (since up to 40% of piped water in Kibera is lost to leakages, according to Umande Trust 2007). But, first things being first, we begin this process of practical and socio-scientific experimentation by examining whether our project delivers on its promise—to improve the real freedom of people in Kibera.



AKER, J. C. & MBITI I. M. (2010), 'Mobile Phones and Economic Development in Africa', Journal of Economic Perspectives, 24 (3): 207–232.

BAYE, M., MORGAN, J. & SCHOLTEN P. (2007), 'Information, Search and Price Dispersion', in T. Hendershott (ed.), Handbook on Economics and Information Systems, 323-375.

DUFLO, E., GLENNERSTER, R. & KREMER, M. (2006), 'Using Randomization in Development Economics Research: A Toolkit', Working paper, M.I.T. (December 2006).

JACK, W. & SURI, T. (2011), 'Mobile Money: The Economics of M-Pesa', NBER Working Paper No. 16721 (January 2011).

JALAN, J., & SOMANATHAN, E. (2008), 'The importance of being informed', Journal of Development Economics, 87: 14-28.

JENSEN, R. (2007), 'The Digital Provide: Information (Technology), Market Performance, and Welfare in the South Indian Fisheries Sector', The Quarterly Journal of Economics, 122 (3): 879-924.

JEON, S. (2011), unpublished results from a survey of 1000 households in Kibera (January 2011).

KREMER, M., LEINO, J., MIGUEL, E. & PETERSON ZWAN, A. (2009, 'Spring Cleaning: Rural Water Impacts, Valuation and Property Rights Institutions' unpublished draft (August 2009).

TOYAMA, K. (2010), 'Can Technology End Poverty?', Boston Review (Nov.-Dec. 2010),

UMANDE TRUST (2007), 'The Right to Water and Sanitation in Kibera, Nairobi, Kenya: An Action Research Report', Nairobi, (December 2007).

WORLD BANK (2005), 'Rogues No More? Water Kiosk Operators Achieve Credibility in Kibera', Field Note, Water and Sanitation Program (June 2005).


1 For an introduction, see

2 USSD (Unstructured Supplementary Service Data) is a technology available on basic GSM phones. In Kenya, USSD is often used for tasks such as purchasing minutes from mobile service providers and paying utility bills. Unlike SMS, USSD opens a channel for two-way communication.

Table des illustrations

Légende Fig. 1: System overview
Fichier image/jpeg, 82k


Sangick Jeon is the M-Maji project lead, and a Ph.D. student in Political Science at Stanford University, where he is completing a dissertation project that explores strategies for supporting interethnic cooperation and realizing the productivity gains from diversity (e.g., in problem-solving and innovation).

Eran Bendavid is an infectious diseases physician and an assistant professor of medicine at Stanford University. His work explores the economic and political conditions that promote health improvements in developing countries. His latest writings evaluate the World Health Organization’s approach to treatment of HIV, and how the Global Gag Rule affected women’s reproductive choices in Africa. With the M-Maji group at Stanford, he explores how innovation and technological design can leapfrog institutional malaise in delivering clean water and increasing agency.

Joshua Cohen is Marta Sutton Weeks Professor of Ethics in Society, and professor of political science, philosophy, and law at Stanford University. He is author, most recently, of The Arc of the Moral Universe and Other Essays (Harvard U.P., 2011). He is also editor-in-chief of Boston Review, on the faculty of Apple University, and, with Ph. van Parijs as well as some other contributors to this volume, a member of the September Group.

Katherine Hoffmann is a Policy Associate with the Safe Water Program at Innovations for Poverty Action (IPA) where she works on clean water issues with a focus on Western Kenya. She holds a B.A in International Relations and Economics and an M.A. in International Policy Studies from Stanford.

Terry Winograd is Professor of Computer Science at Stanford University, where he co-directs the Human-Computer Interaction program and the Liberation Technologies program. He is a founding faculty member of the Hasso Plattner Institute of Design ( where he teaches a course on Liberation Technologies with J. Cohen. His books include Understanding Computers and Cognition (Addison-Wesley, 1986) and Bringing Design to Software (Addison-Wesley, 1993).

© Presses universitaires de Louvain, 2011

Conditions d’utilisation :