WWW 2010 • Full Paper April 26-30 • Raleigh • NC • USA Volunteer Computing: A Model of the Factors Determining Contribution to Community-based Scientific Research Oded Nov David Anderson Ofer Arazy Polytechnic Institute of Space Sciences Laboratory, School of Business, University of New York University, University of California, Berkeley, Alberta, Edmonton, AB, Canada New York, NY, USA Berkeley, CA, USA [email protected] [email protected] [email protected] ABSTRACT number of individuals who volunteer their personal computers to Volunteer computing is a powerful way to harness distributed the initiative. The second pillar is participative: recruiting and resources to perform large-scale tasks, similarly to other types of retaining a large number of individuals to volunteer their community-based initiatives. Volunteer computing is based on computer resources to the project, and facilitating their continuous two pillars: the first is computational – allocating and managing contribution. large computing tasks; the second is participative – making large A decade since the launch of SETI@home, a flagship volunteer numbers of individuals volunteer their computer resources to a computing project, we are witnessing a substantial increase in the project. While the computational aspects of volunteer computing number and scope of volunteer computing projects [e.g. 3]. This received much research attention, the participative aspect remains rise represents a growing recognition in the potential of volunteer largely unexplored. In this study we aim to address this gap: by computing as a cost-effective approach for managing drawing on social psychology and online communities research, computationally-intensive tasks. However, while the we develop and test a three-dimensional model of the factors computational aspect of volunteer computing received much determining volunteer computing users’ contribution. We research attention [e.g. 2, 3, 4, 19, 30, 46], the participative aspect investigate one of the largest volunteer computing projects – remains largely unexplored. What do we know about the factors SETI@home - by linking survey data about contributors’ driving participation in volunteer computing projects? How can a motivations to their activity logs. Our findings highlight the large scale scientific project provide an environment that would differences between volunteer computing and other forms of encourage resource contribution from many volunteers? In the community-based projects, and reveal the intricate relationship present study we address these questions and investigate the between individual motivations, social affiliation, tenure in the factors that determine the contribution of resources to volunteer project, and resource contribution. Implications for research and computing projects. We draw on prior research in the area of practice are discussed. online communities and explore the similarities and differences between volunteer computing and other forms of community- Categories and Subject Descriptors based projects. H.5.3 [Information Interfaces and Presentation]: Group and Organization Interfaces — Collaborative computing; J.4 In recent years the internet and the web have emerged as powerful [Computer Applications]: Social and Behavioral Sciences. platforms for harnessing resources contributed by large numbers of geographically distributed individuals. Examples include General Terms Wikipedia – a user-created encyclopedia, open source software, Management, Human Factors. and other community based initiatives [1, 7, 10, 27, 37, 51, 55, 60]. Underpinning the sustainability of such initiatives is the Keywords willingness of people to contribute resources - knowledge, time Volunteer computing, online communities, citizen science, and skills – voluntarily [11, 13, 29]. In a recent Scientific BOINC, SETI@home, crowdsourcing, motivations. American article, Shadbolt and Berners-Lee [48] discuss advances in the study of the World Wide Web. They argue that “we need to 1. INTRODUCTION know why people who contribute content link it to other material. Volunteer computing is based on the use of computers Social drivers--goals, desires, interests and attitudes--are volunteered by the general public to carry out distributed fundamental aspects of how links are made. Understanding the scientific computing [2, 4]. It is a powerful way to harness Web requires insights from sociology and psychology every bit as distributed resources to perform large-scale tasks, similarly to much as from mathematics and computer science.” Indeed, studies other types of community-based initiatives, such as Wikipedia, of motivations have been helpful in explaining participation in Delicious, Slashdot or open source software development various forms of community based projects, such as Wikipedia projects. Volunteer computing is based on two pillars. The first is [37, 47] or open source software projects, [32, 45, 51]. By computational: managing large computing tasks by breaking them drawing on prior research on online communities, we investigate down to many small tasks, which are then allocated to a large the factors that determine the level of resource contribution in volunteer computing projects. In particular, we focus on the following questions: (1) what motivates participants to volunteer Copyright is held by the International World Wide Web Conference their computer resources? (2) what is the role of non-motivational Committee (IW3C2). Distribution of these papers is limited to classroom use, and personal use by others. factors, such as social affiliation and tenure in the project? and (3) WWW 2010, April 26–30, 2010, Raleigh, North Carolina, USA. how do these various factors interact to impact participants’ ACM 978-1-60558-799-8/10/04. contribution level? 741 WWW 2010 • Full Paper April 26-30 • Raleigh • NC • USA This paper contributes to the research and practice of volunteer motivated differently. Both these studies, however, did not link computing in two ways: First, we use theories of motivation and motivations and actual user contribution to explain the factors social participation to develop a novel framework of the factors determining contribution levels. determining volunteered computing (personal motivations, team affiliation, tenure). Second, we test our model and expose the Given the little work done on motivations in the context of relations between these factors and actual contribution levels. volunteer computing, we extend our review to neighboring areas. An area of research that is highly useful for understanding The setting for our empirical study is SETI@home1 (Search for volunteer computing contribution, is that of social media Extra-Terrestrial Intelligence), one of the best known volunteer contributors’ motivations. Mass collaboration of large numbers of computing projects, which is hosted at U.C. Berkeley. Established individuals distributed across time and space represents a new and in 1999, SETI@home harnesses the computing resources of over productive trend in the creation and dissemination of information 500,000 computers worldwide to collaboratively analyze radio [7, 60]. This phenomenon is characterized by distributed groups telescope data [7, 14], with the aim of detecting intelligent life of volunteers, who split up work into modular tasks [35], and are outside Earth. supported by systems that facilitate collective action and social interaction online [59]. Sustained contribution by individual More recently, this volunteer computing initiative was extended volunteers is critical for the viability of such communities [11, 13, through the creation of BOINC (Berkeley Open Infrastructure for 2 29]. Reflecting this, understanding the motivations of contributors Network Computing) , an open-source platform dedicated to has been viewed as critical for successfully managing and running volunteer computing projects. Currently, the BOINC sustaining web-based community-based projects [6, 12, 36, 47], platform, of which SETI@home is now part, serves over thirty and extensive research focused on contribution to a wide range of projects in various scientific fields, including astronomy (e.g. online communities [13, 24, 36, 42], and in particular, Einstein@home), climate modeling (e.g. climateprediction.net), communities where contribution is made by volunteering mathematics (e.g. PrimeGrid), biology and medicine (e.g. amateurs such as Delicious [33], Flickr [38], Twitter [58], Rosetta@home), and others. To contribute, a participant needs to YouTube [23] and Wikipedia [10, 37]. Extrinsic motivations for download a client application which is then used for managing the contribution in such communities - motivations that are volunteered computer’s allocated tasks. After the initial download instrumental and represent cases where an activity is carried out in and installation, contribution is done without any human order to achieve a separable outcome [44] - include improvement intervention, and without a need for the contributor to interact of skills [39] as well as recognition and enhancement of status with the system. A participant’s contribution level is determined [32, 45]. Intrinsic motivations on the other hand – which by the setting of her profile. The participant can set (and later emphasize inherent satisfaction from an activity rather than its change) her level of contribution in a number of ways, for consequences [44] - include altruism [58], fun [53], reciprocity example, by determining the amount of disc space, memory