Chandan Sarkar

Tracking website data-collection and privacy practices with the iWatch web crawler

In this paper we introduce the iWatch web crawler, a tool designed to catalogue and analyze online data practices and the use of privacy related indicators and technologies. Our goal in developing iWatch was to make possible a new type of analysis of trends, the impact of legislation on practices, and geographic and social differences online. In this paper we present preliminary findings from two sets of data collected 15 months apart and analyzed with this tool. Our combined samples included more than 240,000 pages from over 24,000 domains and 47 different countries. In addition to providing useful and needed data on the state of online data practices, we show that iWatch is a promising approach to the study of the web ecosystem.

Online and offline interactions in online communities

Online communities, while primarily enacted through technology-mediated environments, can also include offline meetings between members, promoting interactivity and community building. This study explores the offline interactions of online community members and its subsequent impact on online participation. We argue that offline interactions have a counterintuitive impact on online participation. Although these offline interactions strengthen relationships, these relationships undermine the communitys sustainability in terms of site participation. Participation has been defined as contribution of content to the online community. A multi-method analysis technique using content analysis, qualitative interviews, and server level quantitative data of users in Everything2.com supports our claim.

A quantitative explanation of governance in an online peer-production community

In this paper, we examine how user ratings of content produced for an online community are taken into account by administrators when they decide whether to delete content. Incorporating about 10 years of server data from the online peer-production community Everything2, we looked at how specific features of voting predicted deletion of posts. We found that not all types of voting are the same: negative voting of users was the strongest factor explaining deletion of a Write-up. Receiving a positive vote from a member with higher status decreases the chances of deletion, while receiving a positive vote from a user with neutral status has a very little effect on the deletion of content.

Engineering Sociability: Friendship Drive, Visibility, and Social Connection in Anonymous Co-Located Local Wi-Fi Multiplayer Online Gaming

In this study, the authors show that online multiplayer gaming via local Wi-Fi can be used as a “social lubricant” to increase social connections between co-located strangers at a cafe. In a field experiment in real world cafes, they recruited people who were sitting alone at the same cafe to play an online game together using iPads, from wherever each happened to be sitting. Some pairs could see each other; some were facing in opposite directions, some were in separate rooms. Visibility influenced how and how much players communicated outside of the game, but had no impact on enjoyment or self-reported experience of social connection. The authors measured “friendship drive” and found that social yearners were more interested in gaming with a stranger and more likely to hope to see their gaming partner again than were socially satiated players. Friendship drive did not impact communication between players or feeling social connection. DOI: 10.4018/jgcms.2012040101 2 International Journal of Gaming and Computer-Mediated Simulations, 4(2), 1-18, April-June 2012 Copyright

The Uncanny Valley Effect in Behavioral Targeting and Information Processing of Peripheral Cues

This paper examines how young adults process information related to privacy, and how that affects their attitude towards behavioral targeted advertising. Differences between computer novices and experts were examined based on the Elaboration Likelihood Model (Petty & Cacioppo, 1984), which argues that people who have the ability to process information do so differently than those who do not have the ability. Consistent with the theory, we found that computer novices were relying on peripheral cues to process information related to security due to their lack of knowledge. We also identified an “uncanny valley” effect where people liked customization of targeted advertisements, but then became uncomfortable if the advertisements seemed to know too much of their past behavior until the suggestions were perfectly aligned with their interests.

Remote Automated User Testing: First Steps toward a General-Purpose Tool

In this paper we explore options for conducting remote, unattended usability tests to enable users to participate in their own environments and time zones, including multiple users’ participating at the same time in remote, unattended studies. We developed a general purpose tool, and code-named it “TCA” (for “Total Cost of Administration”) to catalog and analyze database administrators’ behavior within software. In this paper, we present example findings from the data collected over a period of 6 months. We analyzed users’ deviations from the best paths through the software, in addition to collecting traditional measures such as time on task, error rate, and users’ perceptions, including satisfaction level. Further, we explore how this type of tool offers particular promise in benchmark studies, also collecting the ideal best-path performance that assumes error-free, expert user behavior, to compare to more real-world data collected initially in laboratory tests or over time in longitudinal field studies.

Exploring Social Structure, Participation and Knowledge Dissemination within an Online Game Design Community: Sploder

This study explores social ranks, participations and knowledge dissemination within an online game design community. We describe how social ranks have evolved and supported the overall participation over time. We argue that higher social rank members drive the community in terms of forming effective idea and knowledge sharing. A multi-method analysis technique using content analysis, qualitative interview, and server level quantitative data of users in Sploder.com supports our claim.

Effects of Editorial Feedback in a Peer-Production Online Community

In this paper, we examine how editorial feedback for an online community is taken into consideration by users when they contribute in the future. Analyzing server data from the online peer-production game community Sploder, we looked at how editorial feedback supported future content production. We found that receiving editorial feedbacks are likely to improve the quality of the contribution and reduce the amount of spam in the post. Editorial feedback will garner more comments and visits on articles.