Forgetting the Forgotten with Letheia: Concealing Content Deletions in Presence of Persistent Observers [C3F-592]
Mohsen Minaei, Mainack Mandal, Patrick Loiseau, Krishna Gummadi, Aniket Kate
Today most people are susceptible to oversharing their personal information publicly on social platforms. As a coping strategy, most platforms offer mechanisms allowing users to withdraw their posts. In fact a significant fraction of users exercise this right to be forgotten. However, these withdrawals leave users more vulnerable to privacy attacks: due to the now popular ``Streisand effect,'' users' attempt to reduce attention to sensitive posts by withdrawing those, can, in practice, attract unwanted attention to the posts from others (including curious friends, cyberstalking and even blackmailers). Therefore, users hoping to forget their posts are left with a ``damned if I do, damned if I don't'' dilemma. In the form of intermittent withdrawal mechanism, we propose a rather disruptive solution to this catch-22 problem. If the social platforms are willing to give up the uninterrupted availability of non-withdrawn posts, it is possible to provide a privacy guarantee to the withdrawn posts. Here, an adversarial observer becomes unsure if some post is permanently withdrawn or just temporarily unavailable for some duration, and at the same time overwhelmed by a large number of falsely flagged undeleted posts. We analyze large scale data about users' deletion patterns over Twitter and thoroughly investigate how to choose probability distributions for alternating temporary withdrawals and resurrections of non-withdrawn posts. We report the trade-off between privacy, utility and adversarial overhead in our proposed design under different settings.