Impression formation in online peer production

Abstract

Technology increasingly enables new forms of distributed and ad-hoc online collaboration, among them a form of large-scale, loosely coordinated, mostly voluntary work known as peer production. Some peer production environments now connect social media functionality directly to collaborative work artifacts, which provides participants with detailed information about other contributors' work history, interests and interactions.Activity visibility provides informational signals individuals can use to make inferences about important characteristics of the people they interact with. However, the impact of the increased variety of information about collaborators and potential colleagues in a peer production setting is not well known.Understanding how people form impressions of other contributors can inform the design of peer production environments. This thesis investigates the process and outcomes of using activity traces for interpersonal impression formation in online peer production. Finally, I present an experiment investigating how the visual presentation of activity history influences impressions of contributors and evaluation of work. In the study I vary the amount of detail and quality of work shown in an activity history and measure the influence on impressions and evaluation of work products. I find that greater detail enhances valence and persistence of initial impressions and bias towards an unknown worker as well as effort expended to correct the worker's output on the task. 4My thesis advances our understanding of when and how social networking information and activity traces influence the process of making sense of unknown contributors' inherent qualities, and how this relates to work-related decision-making in peer production. The thesis also informs design principles for showcasing individuals' activity history in collaborative production sites.5

Work Practices and Challenges in Pull-Based Development: The Integrator's Perspective

“…Furthermore, our survey brings additional insights to the insightful but smaller scale interviews that have been conducted by other researchers on the pull based model (e.g. [16], [11], [15], [14]). We welcome replications of this work; potential directions include replications with integrators that (1) use different (non-GitHub) repositories, e.g., Bitbucket, (2) work on private repositories, and (3) work on non-pull request intensive projects.…”

“…This is a complex process, involving both technical [6], [7], [8] and social aspects [9], [10], [11].…”

“…There is no empirical evidence on whether the developer's previous actions play a significant role in contribution assessment in the context of pull-based development; in fact, quantitative data from Gousios et al [1] suggest otherwise. Finally, Marlow et al [11] found that developers on GitHub use social signals, such as the developer's coding activity and the developer's social actions (e.g. following other developers), in order to form an impression of the quality of incoming contributions.…”

“…Some integrators also use interpersonal relationships to make judgements for the contributor and, by proxy, for their work. The process of impression building through social signals has been further elaborated by Marlow et al [11].…”

“…With the raise of transparent work environments [16], and based on previous work on the subject [9], [11], one would expect that the developer's track record would be used by the majority of integrators to make inferences about the quality of incoming contributions. Despite this, the track record is mostly used as an auxiliary signal; in both Figure 5a and Figure 6a, we can see that developers equally mentioned the track record as top and second criterion for quality evaluation and prioritization.…”