Leveraging crowdsourcing for team elasticity: an empirical evaluation at TopCoder
Razieh Saremi, Ye Yang, Guenther Ruhe, David W. Messinger · 2017
There is an emergent trend in software development projects that mini-tasks can be crowdsourced to achieve rapid development and delivery. For software managers requesting crowdsourcing services, it is beneficial to be able to evaluate and assure the availability and performance of trustable workers on their tasks. However, existing rating systems are facing challenges such as providing limited information regarding worker's abilities as well as potential threats from workers' gaming or cheating the systems. To develop better understanding of worker performance in software crowdsourcing, this paper reports an empirical study at TopCoder, one of the primary software crowdsourcing platforms. We aim at investigating the following questions: How diverse are crowd workers in terms of skill and experience? How fast do crowd workers respond to a task call? How reliable are crowd workers in submitting tasks? And how much does CSD benefit schedule reduction? The main results of this study showed that on average, (i) 59% of workers respond to a task call in the first 24 hours, (ii) 24% of the workers who registered early will make submissions to tasks, and 76% of them exceeding the acceptance criteria, and (iii) an overall average of 1.82 schedule acceleration rate is observed through organizing mass parallel development in 4 software crowdsourcing projects. Such empirical evidences are beneficial to help exploring resourcing options and improve team elasticity in adaptive software development.