91av

Cellphones tap wisdom of the crowds

Thousands of people armed with cellphones across rural Kenya are combining their brainpower to boost their income
Cellphones are a lucrative opportunity for people living in rural Kenya
Cellphones are a lucrative opportunity for people living in rural Kenya
(Image: Nathan Eagle)

DAVID, a Masai herdsman from Kisumu in Kenya, answers a call on his cellphone. After listening to the message, he repeats a short phrase in his Masai dialect. He then listens to another short message, and repeats the new phrase. After 30 minutes, he ends the call, having earned enough for a week’s worth of personal cellphone airtime.

David is working for txteagle, a service that allows rural Kenyans to earn airtime and money by performing small tasks such as translation and transcription using their cellphones.

The service is based on crowd-sourcing, which relies on the power of thousands of people, each carrying out small tasks, to help solve much bigger problems. The idea is that while computers are very good at many tasks, there are still some activities, such as translation or image analysis, at which humans thrash them.

“Thousands of people, each carrying out small tasks, help to solve much bigger problems”

Nokia, for example, wants to provide cellphone interfaces in the 60 or so languages spoken in Kenya, but lacks the linguistic know-how to do so. So the company has begun using txteagle to recruit Kenyans to translate English words into local dialects. Contributors are sent text messages with the English words that need translation. The same word or phrase is sent to multiple users, and if a high percentage of people return the same answer, it is accepted by the system, says , founder of txteagle and a cellphone technology researcher at the Massachusetts Instituteof Technology.

The service rewards those who are correct more often than not by paying them at a higher rate. That is because the more trusted the contributor, the fewer redundant translation requests txteagle has to send out, saving it money.

Other applications for txteagle are also being developed. One involves training speech-recognition engines by asking people to listen to a phrase and then speak it back, to provide the software with examples of different accents and the many ways words can be pronounced. Another is transcribing speech into text: a user listens to a fragment of speech, transcribes it and sends it back as a text message. “There is such a huge market for medical transcriptions,” says Eagle. “If we can enable individuals living in villages to start doing that work, it would be a huge thing.”

He estimates that if the transcription service takes off, contributors could earn $3 to $4 for each hour’s work, a substantial amount of money in rural Kenya.

Paying people for their services in rural Africa – where there is no extensive banking or internet infrastructure – needed some innovation. This came just as Eagle was developing txteagle. Kenyan cellphone company Safaricom had a scheme in place which allowed people to exchange cellphone airtime in lieu of currency. “At the market, I could pay for my groceries by transferring airtime,” says Eagle.

Buoyed by the scheme’s success, in early 2008 Safaricom created , a mobile banking system that allows people to transfer money between M-PESA accounts using their cellphones. The money can be cashed at post offices.

In no time, Safaricom became one of the largest banks in east Africa and Eagle was able to pay txteagle contributors by crediting their M-PESA accounts. Later this year, he hopes to expand the service to countries such as Rwanda and the Dominican Republic. He also hopes to extend it to people who cannot afford a cellphone, by allowing community phones to be rented by people completing txteagle tasks.

That is, of course, if the scheme ultimately proves successful. , who investigates the use of computing to support economic development at the University of California, Berkeley, thinks txteagle is “a great model”, but he cautions that providing good quality translation and transcription within the limits of short text messages will be a challenge. And while Kenyans benefit from access to the M-PESA bank, “it’ll be interesting to see how the model works in places where such payment mechanisms are not in place, or where regulations prohibit it.”

But Raffi Krikorian, of the crowd-sourcing service Wattzon (see “The wisdom of crowd-sourcing”), says txteagle has a great opportunity to change the economic and telecoms landscape of east Africa.

The wisdom of crowd-sourcing

IN THE west, crowd-sourcing services are growing in popularity, helping people with everything from earning a little extra cash to reducing their carbon footprint.

Take Amazon’s , a set of tools that allows companies to create what are called Human Intelligence Tasks (HITs). Internet-savvy users can earn money by solving parts of these tasks, which can be anything from sifting through catalogues to weed out duplicate items, to upholding editorial quality by identifying language and images in blogs that violate a company’s guidelines. The system was launched in 2005. “Since that time, we’ve added a number of significant features, including the addition of rupee disbursements for India-based workers,” says Kay Kinton of Amazon Web Services in Seattle.

Environmentally friendly crowd-sourcing applications have also sprung up. , run by Synthesis Studios in Cambridge, Massachusetts, allows people to share information about their daily energy consumption. Tell Wattzon what you’ve been up to and it analyses your information and presents it back to you in a way that enables you to compare your energy usage with that of others in similar circumstances. Reducing your consumption reduces the carbon emissions you are responsible for, helping in the fight against climate change. “Energy, next to water, is the biggest crisis on the planet,” says Wattzon’s Raffi Krikorian, “so we want to make it really obvious how people themselves can start to make a difference.”

In a similar vein, helps people to anonymously share their spending patterns – including information from credit-card statements – so fellow users can identify potential savings. For example, another user may spot that you are buying your weekly groceries from an expensive supermarket, and suggest a cheaper store nearby. “It’s about mining a group of people for ideas that benefit the group as a whole,” says Krikorian.