Privacy is a widely studied concept in relation to social computing and sensor-based technologies; scores of research papers have investigated people’s ‘privacy preferences’ and apparent reluctance to share personal data. In this paper we explore how Ubicomp and HCI studies have approached the notion of privacy, often as a quantifiable concept. Leaning on several theoretical frameworks, but in particular Nissenbaum’s notion of contextual integrity, we question the viability of obtaining universal answers in terms of people’s ‘general’ privacy practices and apply elements of Nissenbaum’s theory to our own data in order to illustrate its relevance. We then suggest restructuring inquiries into information sharing in studies of state-of-the-art technologies and analyze contextually grounded issues using a different, more specific vocabulary. Finally, we provide the first building blocks to such vocabulary.