Tuesday, August 2, 2011

Crowdsourcing microtasks to mitigate poverty

The most novel talk at this week’s IPEG conference was one by Maradona Gatara on mobile crowdsourcing in Africa. It was fascinating on so many levels: on new approaches to crowdsourcing, on the role of mobile phones in economic development, on ways to tackling crippling poverty in LDCs, and even on the theory of the firm.

Maradona started with the explosive growth of mobile phones in Kenya — 100x from 2000 (200k) to 2010 (20m). However, unemployment is nearly 50% for a population of 38.5 million.

He described a crowdsourcing service (txteagle): imagine Amazon Mechanical Turk pushed to a mass LDC market with payments under a dollar. Txteagle was started in 2009 in Kenya and Rwanda, and now covers 17 African countries.

He listed two sample needs. One is medical transcription: 2 minutes of transcribing an audio snippet into a SMS message; Kenya has an advantage here because English is an official language. The other is software localization, building a vocabulary dictionary (e.g. for the 60 different languages in Kenya); the speaker is asked to translate a specific word.

Txteagle adds its value by partitioning a client task into microtasks, dispatches questions and collate the answers. The service does QC by comparing multiple solutions for the same problem, and also builds rating of the quality of the submitters.

In addition to microtasks, the other unusual wrinkle is that the service targets the unbanked — those who lack a bank account but have a cellphone. The service is built on a mobile banking platform (M-Pesa), created by Safarifone subsidiary of Vodafone, the largest European network operator. He estimated mobile banking is $7 billion/year in Africa, to the point that existing banks are partnering with M-Pesa to reduce cannibalization.

So translating one word might be KYD 0.30. He mentioned that an unemployed single mother of four could raise USD 7.50 (KYD 7.00) for three hours of work — enough to feed her family of five for 2-3 weeks. (Like many poor Africans, she doesn’t pay rent because she lives in a squatter community).

There is a long body of research on telecommunications infrastructure in LDC economic development, and a more recent literature on wireless infrastructure being more suitable than wired.

Now there’s the idea (also appealing to Nokia) that mobile phones provide banking of the unbanked. Mardona estimated that 364 million Africans will have cellphones by 2012 (not all of them poor). Obviously the other African operators will be providing their own banking services.

From the audience came a very provocative question from Sebastian von Engelhardt. Years ago, Coase said that firms form because it’s not practical to organize and monitor a series of small tasks. So to what degree do such microtask architectures push the natural boundary of the firm. (During lunch, we also debated what impact it has on the nature of employment — is it complementary or substituting for employment here in Africa.)

Maradona is now completing his master's thesis on what factors cause poor Kenyans to use (or not use) the service. I believe he’s also going to do some work to define the potential market, particularly in East Africa (Kenya, Rwanda, Uganda, Nigeria).

He’s also working to develop (or join) a consulting practice facilitating such mobile crowdsourcing. He expects to get interest both from firms seeking crowdsourcing and possibly governments/NGOs seeking to use these income streams to ameliorate unemployment and poverty.

No comments: