Paid Crowdsourcing as a Vehicle for Global Development

By connecting remote workers to a global marketplace, paid crowdsourcing has the potential to improve earnings and livelihoods in poor communities around the world. However, there is a long way to go before realizing this potential. To date, most workers on microtasking platforms come from relatively well-off backgrounds, and there has been limited impact on low-income individuals. In this position paper, we outline a research agenda to extend the benefits of informal, paid microtasking to low-income workers in developing countries. This goal will require research along multiple fronts, spanning the crowdsourcing platforms themselves, their impact upon users’ livelihoods, and their scalability to large populations. While there are many challenges to overcome, the rewards are great. We believe that a new focus on low-income workers is critically important to unlock the potential scale and impact of paid crowdsourcing platforms. INTRODUCTION There are reasons to believe that paid crowdsourcing1 could be of particular benefit to low-income workers in developing countries. Unlike most employment opportunities, online marketplaces such as Amazon Mechanical Turk (MTurk) do not require geographic co-location between employer and employee. There is no formal contract needed to commence work; the only criterion for employment is the ability to complete the task at hand. In addition, the working hours are completely flexible, allowing workers to earn money whenever they have extra time. Workers with access to a mobile computer may even be able to earn money during their daily commute, or when they are idle during other jobs (e.g., a driver could work while waiting for a client, or a shopkeeper could work while the store is empty). For all of these reasons, one would expect online microtasking services to lower the barrier to entry for employment in low-income settings, potentially boosting the socio-economic standing of otherwise disadvantaged populations. In this paper, we consider paid crowdsourcing to be the provision of small tasks in exchange for monetary payment via websites such as Amazon Mechanical Turk, CloudCrowd, ShortTask, etc. While we are agnostic as to the platform used, we restrict our attention to the informal setting in which workers choose to work on one task at a time, rather than having a long-term contract with a company. But has paid crowdsourcing delivered on its potential to benefit those at the bottom of the economic pyramid? The answer today is both “yes” and “no”. On the positive side, it has been documented that over a third of workers on MTurk are based in India [4, 5], suggesting that the service can make inroads in a developing-country context. And in our own survey of 200 Indian Turkers [5], we uncovered several cases in which MTurk had fundamentally changed the livelihood of our respondents. For example, a 26-year old college graduate from Kolkata, India, describes how he came to earn $1860 per year on Mechanical Turk: I’m from a middle class family. After completing my degree I looked for job everywhere but failed. But when I found MTurk, it changed my life. It helped me a lot. However, despite such anecdotes, the impact of paid crowdsourcing in developing countries remains limited. While many Turkers are based in India, in our survey we found that these workers have a relatively high socio-economic standing: 80% have a Bachelor’s degree or higher, 92% have a PC and Internet connection in their home, and their median annual income is $2700. This is substantially better than the position of the average Indian. Only 6% of India’s workforce has a Bachelor’s degree, only 6% of households have a computer and Internet access, and average per capita income is $1100 per year [5]. To have an impact on poor communities, systems such as Mechanical Turk will need to be accessible to those with lower formal education (secondary and higher secondary schooling) and only basic skills with computers and the English language. Today, the benefits of paid crowdsourcing are largely out of reach for this population. In this paper, we outline an agenda to shape the evolution of paid crowdsourcing into a vehicle for socio-economic development. Our agenda is heavily influenced by our recent study, which examined the usability of Mechanical Turk for low-income workers in India [5]. The study found that while there are tasks on MTurk that are not beyond the cognitive capabilities of low-income workers, the language and user interface represent significant barriers for this population to earn money on MTurk. By simplifying the user interface and task instructions, and translating all content into the local language, we demonstrated that low-income workers can achieve much higher rates of task completion. Still, it remains challenging for such workers to reliably earn money on MTurk, due to costs of accessing computers and other barriers. Many questions also remain in scaling up our re-