Crowdsourcing not all sourced by the crowd: An observation on the behavior of Wikipedia participants

Highlights

• Wikipedia articles are not always rightly sourced from the crowd.
• Wikipedia articles are often maintained by a dominant few.
• The dominance of the few participants becomes more severe as the crowd matures.
• Intrinsic motivation encourages Wiki participants to engage in revising articles.

Abstract

This study investigates the behavioral patterns of Wikipedia participants to obtain a picture of internal dynamics of the world's largest crowdsourcing platform. It observes the responses of people when “other” people enter a crowd where internal and external controls are mostly absent. From the analysis of 342 Wikipedia articles, this study shows that the overall tone of Wikipedia articles is mostly decided by a dominant few rather than by a trivial many, and such domination worsens as the number of participant increases and the article matures. This result contradicts a common belief on crowdsourcing that Wikipedia would reflect the voices of a vast majority, obtain a balanced solution, and attain democracy on the Internet. Therefore, this study contributes to the literature by analyzing how effectively Wikipedia functions as a crowdsourcing platform within the context. It also implies that developing a proper crowdsourcing strategy such as effective management of a platform is necessary, especially when an organization has a specific goal to achieve throughout a project.

Introduction

To understand the dynamics of a complex crowdsourcing platform, this study investigates the behavioral patterns of Wikipedia participants. We observe the responses of participants when “other” people enter a large crowdsourcing platform, Wikipedia, where internal and external controls are mostly absent. This observation would elucidate how the behaviors of crowdsourcing participants change as the crowd grows larger and subsequently provide insights into the internal mechanism of the complex crowdsourcing platform.

Nowadays, people are connected more than ever because of the Internet, which forces users to be part of an unexpected but fully functioning collaboration (Doan et al., 2011). For example, reCAPTCHA is a user-dialogue system that allows Project Gutenberg to digitize public domain materials, such as ancient books that cannot be read by optical character recognition (OCR) software (von Ahn et al., 2008). The system collects contributions from more than 100 million Internet users every day from websites such as Facebook, Twitter, and Craigslist.

Such strategic collaboration through the Internet has been conceptualized and developed into the term “crowdsourcing” (Albors et al., 2008). Crowdsourcing does not simply refer to the gathering of resources from numerous people; rather, the concept emphasizes the synergic and value-added effects created by resources collected from the “crowd.” Crowdsourcing differs from typical Internet-based collaboration in that the former emphasizes the value created from the “crowd” rather than from a “vital few” (Kittur, 2010).

As crowdsourcing phenomena continue to evolve in forms and functions, various types of crowdsourcing platforms emerge (Kohler, 2015). For example, one popular type of platform is tournament-style intermediaries for solution and idea crowdsourcing (Prpić et al., 2015a), such as Kaggle.com for predictive modeling projects, Threadless.com for product design, InnoCentive.com for research and development, and TopCoder.com for software development projects (Dissanayake et al., 2015). These platforms have established a large pool of self-selected participants, who are problem solvers rather than workers for hire. Organizations often commission crowdsourcing intermediaries as a paid platform service to acquire their own crowds (Prpić et al., 2015b).

As the crowdsourcing approach widely spreads, the dynamics inside a crowd is gaining attention from researchers and practitioners (Afuah and Tucci, 2012). When the number of participants of a crowdsourcing project increases, the intricacy of coordination among the participants also increases. The more openness crowdsourcing embeds, the more likely crowdsourcing be abused by people with malice or incompetence (Yasseri et al., 2014). Furthermore, crowdsourcing is concerned not only with ways to collect resources, but also with methods to create value through fair assessment and management of collected resources (Bonabeau, 2009). Therefore, crowdsourcing can be considered a double-edged sword that can either build knowledge or generate misinformation (Leimeister et al., 2009).

For these reasons, the current study investigates the basic behavioral patterns of crowdsourcing participants and selects Wikipedia as a crowdsourcing platform to explore. Wikipedia is selected because it is one of the largest crowdsourcing platforms with a notably low level of control and a high level of dynamics of people with pure motivation. While organizations actively utilize and attempt to control various crowdsourcing platforms such as Kaggle and Innocentive for their business interests and create their own proprietary crowds (Piezunka and Dahlander, 2015), people contribute to Wikipedia to achieve intrinsic motivation, such as seeking the truth, rather than extrinsic motivation, such as monetary incentives. Hence, Wikipedia, which is designed as an Internet-based control-free discussion agora, is suitable as a platform for testing the participating behaviors of crowdsourcing participants (Martinez and Walton, 2014). Observing the behavior of Wikipedia participants in various aspects will provide researchers and practitioners with an in-depth understanding of crowdsourcing at the principal motivation level.

This study is organized as follows. We first review the literature on crowdsourcing and Wikipedia as a crowdsourcing platform with low control. We then formulate three primary hypotheses on the behavioral pattern changes of Wikipedia participants as the crowdsourcing project matures. Next, we collect data from 342 Wikipedia articles and analyze the article creation and revision patterns to validate the hypotheses. Finally, we summarize our results and discuss implications.