Reverse engineering is a complex task. As with many other expert tasks, reverse engineers rely on colleagues and the broader reverse engineering community to provide guidance and develop knowledge necessary to achieve their goals. For example, it is common for reverse engineers to reach out for help to understand and effectively use new tools. Thus far, however, there has been limited investigation of the way knowledge is developed in this community and new tools are adopted. This paper takes a first step toward understanding reverse engineering community dynamics around tool adoption, using the release of the National Security Agency's Ghidra reverse engineering framework as a point of focus. In this paper, we review discussions about Ghidra to identify what features reverse engineers are most interested in, how reverse engineers develop knowledge about Ghidra together online, and whether these dynamics differ between forums. In total, we analyze 1590 reverse engineering discussions between 688 reverse engineers over 3 forums (i.e., Twitter, Reddit, and StackExchange). Our results suggest reverse engineers are most interested in features that allow them to customize Ghidra. We also observe limited evidence of collective sensemaking on the forums, with few reverse engineers participating in multiple discussions threads and most acting as either knowledge producers or consumers. Finally, we found that the forums operated similarly, but Twitter was most often used to announce information (e.g., tutorial links, tool overviews, vulnerabilities in Ghidra) and reverse engineers used StackExchange mostly to get support for specific problems. Reddit acted as a middle option. Based on these results, we make recommendations to improve reverse engineering tool development, improve community participation during adoption, and suggest directions for future work.

Most social media platforms record, display, and archive users' personal histories. This persistence of posts over time can be problematic, as users' self-presentation goals and network composition change, but old content remains. In this paper, we explore an alternative feature that provides control over content persistence. We present findings from interviews with 16 users of the popular Chinese social media platform WeChat Moments. We focused on Moments' Time Limit setting, which makes social media data ephemeral to audiences, but persistent to posters. Interviewees described changes in their self-presentation goals and social network composition over time and reported the Time Limit feature helped them effortlessly manage their desired self-presentation as they matured. Drawing on these findings, we discuss design implications for social media to facilitate greater control over content visibility and persistence, which may have significant benefits for social media users with large and diverse networks.

The rapid advancement of artificial intelligence (AI) is changing our lives in many ways. One application domain is data science. New techniques in automating the creation of AI, known as AutoAI or AutoML, aim to automate the work practices of data scientists. AutoAI systems are capable of autonomously ingesting and pre-processing data, engineering new features, and creating and scoring models based on a target objectives (e.g. accuracy or run-time efficiency). Though not yet widely adopted, we are interested in understanding how AutoAI will impact the practice of data science. We conducted interviews with 20 data scientists who work at a large, multinational technology company and practice data science in various business settings. Our goal is to understand their current work practices and how these practices might change with AutoAI. Reactions were mixed: while informants expressed concerns about the trend of automating their jobs, they also strongly felt it was inevitable. Despite these concerns, they remained optimistic about their future job security due to a view that the future of data science work will be a collaboration between humans and AI systems, in which both automation and human expertise are indispensable.

Q&A websites compile useful knowledge through user-generated questions and responses. Many Q&As use collaborative tagging systems to improve search and discovery while distributing the work of categorizing and organization throughout the community. Although early work on collaborative tagging questioned whether consistent categorization schemes could emerge from large groups with little to no coordination, empirical studies have found surprising coherence among users' tags. We build on this research by testing whether coherence emerges in tag usage on Q&As, a more challenging context, focusing in particular on mismatches in the specificity of tags (basic level disagreement). We found that some users shifted toward more specific tag usage over time slightly increasing conflict, but that moderators were instrumental in helping to resolve some of this conflict. This study highlights the importance of learning and moderation in the development of coherence in collaborative tagging systems.

The democratization of data science and open government data initiatives are inspiring groups from civic hackers to data journalists to use data to address social issues. The analysis of open government data is expected to encourage citizens to participate in government as well as to improve transparency and efficiency in government processes. Through interviews and survey responses we gathered information on forty projects that involved the analysis of open data. We found that collaborations were interdisciplinary, small in scale, with low turnover, and synchronous communication. Most of the projects asked exploratory questions and made use of descriptive statistics and visualizations. We discuss how these findings contribute to an understanding of the emerging practice of open data analysis and to a broader understanding of open collaboration.

Open innovation contests have been incredibly successful at producing creative designs and solutions. While participants compete for prizes in these contests, over half of contests conducted on online platforms allow participants to share ideas during contests, for benefits such as individual learning, community building, and cumulative innovation. Such sharing, however, is at tension with the competitive nature of crowd contests. To understand this tension, this study investigates community-level sharing of code on Kaggle, a contest platform for predictive modeling. Analyzing data on 25 contests in 2015 and 2016, we find that 10% of users shared code during contests, that participants doing medium well in the contest were the most likely to share code, and that sharing code improved individual, but not collective performance. These findings allow us to contribute insights about the participants, conditions, processes, and outcomes of community-level collaboration to both research on and design of open innovation contests.

Technology is changing the way in which scientists and mathematicians communicate. New platforms for scholarly communication enhance informal communication and imbue it with some of the functionalities of formal communication. A citation analysis was conducted to examine how content from one of these platforms, MathOverflow, has been cited and referenced within the mathematics literature. Citation patterns suggested that some authors were treating MathOverflow content as a legitimate source of scholarly knowledge. Some problems with references and attribution occurred, which can be addressed by changes in the design of these technologies. Developing better systems for scholarly communication can help advance open science.

Researchers and theorists have proposed that feelings of attachment to subgroups within a larger online community or site can increase users' loyalty to the site. They have identified two types of attachment, with distinct causes and consequences. With bond-based attachment, people feel connections to other group members, while with identity-based attachment they feel connections to the group as a whole. In two experiments we show that these feelings of attachment to subgroups increase loyalty to the larger community. Communication with other people in a subgroup but not simple awareness of them increases attachment to the larger community. By varying how the communication is structured, between dyads or with all group members simultaneously, the experiments show that bond- and identity-based attachment have different causes. But the experiments show no evidence that bond and identity attachment have different consequences. We consider both theoretical and methodological reasons why the consequences of bond-based and identity-based attachment are so similar.

The Internet has the potential to accelerate scientific problem solving by engaging a global pool of contributors. Existing approaches focus on broadcasting problems to many independent solvers. We investigate other approaches that may be advantageous by examining a community for mathematical problem solving -- MathOverflow -- in which contributors communicate and collaborate to solve new mathematical 'micro-problems' online. We contribute a simple taxonomy of collaborative acts derived from a process-level examination of collaborations and a quantitative analysis relating collaborative acts to solution quality. Our results indicate a diversity of ways in which mathematicians are reaching a solution, including by iteratively advancing a solution. A better understanding of such collaborative strategies can inform the design of tools to support distributed collaboration on complex problems.

We develop and evaluate a real-time language feedback system that monitors the communication patterns among students in a discussion group and provides real-time instructions to shape the way the group works together. As an initial step, we determine which group processes are related to better outcomes. We then experimentally test the efficacy of providing real-time instructions which target two of these group processes. The feedback system was successfully able to shape the way groups worked together. However, only appropriate feedback given to groups that were not working well together from the start was able to improve group performance.

Why do people contribute content to communities of question-answering, such as Yahoo! Answers? We investigated this issue on MathOverflow, a site dedicated to research-level mathematics, in which users ask and answer questions. MathOverflow is the first in a growing number of specialized Q&A sites using the Stack Exchange platform for scientific collaboration. In this study we combine responses to a survey with collected data on posting behavior on the site. User behavior suggests that building reputation is an important incentive, even though users do not report this in the survey. Level of expertise affects users' reported motivation to help others, but does not affect the importance of reputation building. We discuss the implications for the design of communities to target and encourage more contributions.

There are two perspectives on the role of reputation in collaborative online projects such as Wikipedia or Yahoo! Answers. One, user reputation should be minimized in order to increase the number of contributions from a wide user base. Two, user reputation should be used as a heuristic to identify and promote high quality contributions. The current study examined how offline and online reputations of contributors affect perceived quality in MathOverflow, an online community with 3470 active users. On MathOverflow, users post high-level mathematics questions and answers. Community members also rate the quality of the questions and answers. This study is unique in being able to measure offline reputation of users. Both offline and online reputations were consistently and independently related to the perceived quality of authors' submissions, and there was only a moderate correlation between established offline and newly developed online reputation.

In this note we introduce a new methodology that combines tools from social language processing and network analysis to identify socially situated relationships between individuals, even when these relationships are latent or unrecognized. We call this approach social language network analysis (SLNA). We describe the philosophical antecedents of SLNA, the mechanics of preprocessing, processing, and post-processing stages, and the results of applying this approach to a 15-month corporate discussion archive. These example results include an explicit mapping of both the perceived expertise hierarchy and the social support / friendship network within this group. Author Keywords social language processing, social network analysis, network