Kudu began with research looking at how to use mobile technology to make agricultural markets more effective in Uganda.
From 2010 we started looking at past price data for agricultural produce in Uganda, and noticed large price discrepancies which indicated problems with efficiency in the market. Farmers faced problems in finding buyers for their produce, being threatened with spoilage of their goods when no buyers could be found and often having little negotiating power. Traders also faced uncertainty in being able to locate produce, relying on word-of-mouth networks. Mobile price advisory services provided some help, particularly to farmers, but seemed to have problems with accuracy and timeliness. Conventional auction or listings services were not accessible to the majority of farmers and traders in Uganda who have basic phones with SMS functionality only.
The insight behind Kudu was that rather than providing a listing of items offered which buyers could bid on (known as a single auction), a different type of system called a double auction is more appropriate to users with basic phones. In this type of auction, buyers and sellers separately communicate their requirements and the prices they are willing to trade at. The system then periodically clears the market, matching compatible buyers and sellers. Therefore our users only have to send a single text message, and the system takes into account price, location and other factors to automatically find the best matches.
Google provided support for development in the form of a Research Award in 2011, and the system went live in 2012 as a free service. We began by holding meetings with groups of farmers in Bukomansimbi district and with traders in Kampala. By continuing such meetings in other areas, and combined with radio broadcasts and other forms of publicity, we have had an enthusiastic response from users and rapid user growth, and are matching increasingly large numbers of buyers and sellers across Uganda daily.Contact/enquiries
Richard Ssekibuule has had many years of experience in managing computing projects in Uganda, including directing the provision of ICT services at the College of Computing and Information Sciences, Makerere University. He has an MSc from Radboud University, Netherlands (2006) and a BSc from Makerere University (2002). He worked on the mechanism behind Kudu as part of his PhD research, and has also studied security in electronic marketplaces up to 2010 as a visiting postgraduate researcher at Radboud University.
John Quinn is a Senior Lecturer in the Department of Computer Science, Makerere University, Uganda. He holds a PhD in machine learning from the University of Edinburgh (2007), and a BA in Computer Science from the University of Cambridge (2000). Since 2007 he has worked in Uganda on applications of data science and artificial intelligence to problems in agriculture, health and resource allocation, and runs the Makerere University Artificial Intelligence in the Developing World (AI-DEV) research group.
Kevin Leyton-Brown is a Professor of computer science at the University of British Columbia. He holds a PhD and MSc from Stanford University (2003; 2001) and a BSc from McMaster University (1998). He works at the intersection of computer science and microeconomics, and also applies machine learning to improve algorithms for solving hard computational problems. He has co-written two books, "Multiagent Systems" and "Essentials of Game Theory," and ninety peer-refereed technical articles. His practical experience of auction systems includes working with Auctionomics on algorithms for the next round of FCC auctions, a huge-scale reallocation of US radio spectrum from TV to mobile; he has also consulted on problems related to auction design for Cariocas, Arbiba and TradingDynamics.