When Forcing Collaboration is the Most Sensible Choice

Abstract

Individuals share increasing amounts of personal multimedia data, exposing themselves (uploaders) as well as others (data subjects). Non-consensual sharing of multimedia data that depicts others raises so-called multiparty privacy conflicts (MPCs), which can have severe consequences. To limit the incidence of MPCs, a family of Precautionary mechanisms have recently been developed that force uploaders to collaborate with the other data subjects to prevent MPCs. However, there is still very little work on understanding how users perceive the precautionary mechanisms together with which ones they prefer and why. In addition, precautionary mechanisms have some limitations, e.g., they require linking content to the co-owners' identity. Therefore, we also explore alternatives to precautionary mechanisms and propose a new class of solutions-Dissuasive mechanisms-that aim at deterring the uploaders from sharing without consent. We then present a user-centric comparison of precautionary and dissuasive mechanisms, through a large-scale survey ( = 1792; a representative sample of adult Internet users). Our results showed that respondents prefer precautionary to dissuasive mechanisms. These enforce collaboration, provide more control to the data subjects, but also they reduce uploaders' uncertainty around what is considered appropriate for sharing. We learned that threatening legal consequences is the most desirable dissuasive mechanism, and that respondents prefer the mechanisms that threaten users with immediate consequences (compared with delayed consequences). Dissuasive mechanisms are in fact well received by frequent sharers and older users, while precautionary mechanisms are preferred by women and younger users. We discuss the implications for design, including considerations about side leakages, consent collection, and censorship. CCS Concepts: • Security and privacy → Usability in security and privacy; • Human-centered computing;

“I thought you were okay”: Participatory Design with Young Adults to Fight Multiparty Privacy Conflicts in Online Social Networks

“…One privacy-threatening behavior is the non-consensual sharing of co-owned content [70,98]; content that belongs to an uploader who shares the content, and to one or several data subjects depicted in that content. Such non-consensual sharing can create interdependent privacy conflicts [29,53] or so-called multiparty privacy conflicts (MPCs) [21,116,117].…”

“…They also have less of a tendency to tag their friends [95] or obtain their consent before sharing [117]. There are many reasons that data subjects do not approve of MPCs [21,117]. Some data subjects worry because they do not look good in the shared content, and others have concerns about private facts that can be deduced from the content, such as where, how, and with whom they were at the time the content was recorded.…”

“…1 The consequences of MPC can be severe. One out of every thirteen US adults deals with severe MPCs [21] that cause grave consequences, such as public shaming [69,124], cyberbullying [23], and discrimination [43]. The most severe case of MPC is revenge 1 Even in countries such as the United States, where photography in the public sphere can be taken and published without prior authorization [66], publishing someone's photo on social media can disclose private facts and could have negative consequences for the reputation of the data subjects [100].…”

“…The second stream includes studies that propose privacy-preserving solutions for OSNs. They provide either technical solutions [52,55,95,115] or dissuasive strategies stemming from persuasion theories [6,7,21,81]. These studies do not, however, place users at the heart of the design process, instead they propose solutions that do not necessarily consider users' real-life MPC experiences, attitudes, and behaviors in OSNs.…”

“…We believe these justice strategies have potential benefits for addressing MPCs, hence we use them as seeds, or inspirations, to guide us in our participatory design sessions. Until now, only a few of these strategies were used to manage MPCs (e.g., see the Law Threat strategy in [21], an approach based on deterrence). To the best of our knowledge, the use of these practices has not been studied in the MPC context.…”