Who uses mobile phones? If you ask Nathan Eagle, a researcher at Santa Fe Institute in the United States, he will tell you it is the developing world. As the majority of mobile phone subscribers live in the developing world, it made sense for him to pack his bags and move to a small village on the coast of Kenya when he was a research scientist at the Massachusetts Institute of Technology.
And there he came up with the idea of a service that helps people make money through their mobile phones, and created Txteagle (http://txteagle.com), which is set to come to India soon.
The service enables crowd sourcing through mobile phones. “People need to be aware that large chunks of work can be broken down in to micro-tasks that can then be distributed to people around the world,” said Mr. Eagle, in an interview to The Hindu, describing the shift in mindset corporations needed, to make use of crowdsourcing.
Any task that a human could outperform a computer was appropriate, and there is a wide range of such tasks including translation and labeling images. A farmer in Orissa who could translate some lines of a phone manual and SMS it. But how to ensure quality of the tasks?
‘Secret sauce’:
“This is our secret sauce a sophisticated machine learning [Artificial Intelligence] algorithm developed jointly at MIT and Harvard lies at the heart of our accuracy inference engines,” said Mr. Eagle. “It does three things – infers if a task has been completed accurately within a 99% confidence interval, estimates the expertise of our users in a wide range of application domains, and allocates tasks – sending each particular task to the most suitable active user.”
The idea used to determine accuracy was straight forward, he said. If multiple individuals were given the same task and asked the same questions, and if they provided the same response, then we could assume it is right, he said. And the service is active – the more the user uses it, the more it learns.
The clients pay for the completed tasks. “We pay our users and provide the operators with a fraction of the money their subscribers earn,” said Mr. Eagle, describing the revenue model. The payment to the users could be in airtime too.
Search for partners:
In India, there was a huge market fort mobile crowdsourcing, said Mr. Eagle. “The question is not whether or not there is a market, the question is whether there are clients.” Currently Mr. Eagle is scouting for partners in the BPO industry who are familiar with Indian outsourcing.
A certain degree of literacy was required for the tasks, said Mr. Eagle. An average user in a place such as Kenya who did not have any special expertise could earn 70 cents to $1.75 an hour for doing work, he said.
Companies were typically “extremely excited about a new and cheaper way of completing their tasks.” Initially, they were skeptical of the performance until they saw a demonstration, using a small subset, of how the tasks could be performed accurately, he said.
