Detecting Crowdsourcing Impasses

Abstract

The adoption of crowdsourcing systems to support successful and effective collaboration among actors to perform a variety of tasks continues to grow. The success of such systems relies on actor willingness to contribute to task achievement in pursuit of a collective goal. This willingness may be negatively affected under certain circumstances. One area where this is the case is work progressing insufficiently or even grinding down to a halt. Such situations can be referred to as impasses and this has been captured more formally in the form of the Stasis pattern in previous work. In this paper, we first investigate the various forms stasis can take, next we characterize these forms in the form of a classification problem, and finally, detect their presence and predict the likelihood of their future occurrence in a practical setting. The latter is achieved through an exploratory case study involving computational analyses of development activities within an open-source crowd-sourced software platform (GitHub). Our findings contribute towards a rich understanding of stasis in crowdsourcing systems and how its various forms can be detected early (and thus mitigated) or even prevented altogether. As such, this work helps improve the chances of successful collaboration through the use of these systems.

Keywords

• Co-destruction
• crowdsourcing
• stasis
• detection
• prediction