Abstract
Creating technology that monitors presence in shared spaces requires more than technical excellence, it demands careful attention to how individuals experience and interpret that technology. My experience developing the HoosStudying technical project, and conducting research on Carnegie Mellon University's deployment of Mites sensors for monitoring human presence in shared spaces, led me to a critical understanding: even with technology designed for privacy preservation, if trust, transparency, and the balance of power among institutions are not present, it may be perceived as a surveillance tool. The projects outlined here show that the line between acceptable monitoring of the presence of an individual in a given space versus an invasive act of surveillance of that individual depends not only on what is tracked, but how those tracking systems are understood and trusted by other stakeholders in the observation process. The HoosStudying technical project addressed a straightforward problem: UVA students waste valuable time searching for available study rooms during peak periods. We developed a real-time occupancy tracking system for identifying when there were free study rooms using mmWave sensors and ESP32 microcontroller-based technology to detect when humans were physically present in a given location, without recording image or audio data about those humans. The tracking technology allows students to locate available study rooms in real time via a Web-based application. From a technical perspective, we delivered an accurate, cost-effective, privacy-preserving solution for a verified student need. In researching the CMU Mites case for my STS project, I concluded that merely having technical protections for privacy does not mean users will accept them as such. Mites (which is the name of the system in CMU's TCS Hall) collects data about how many people are in the building and about the air quality, similar to HoosStudying. Mites carefully avoided collecting any audio or video of the occupied, and it processed data on the device and encrypted it before transferring it. The Mites installation led to great controversy amongst students and faculty at CMU; both felt they had been subjected to surveillance without sufficient consent, transparency, and understanding of how and why they were being monitored. Through Actor-Network Theory I examined how the controversy regarding Mites was not a result of the technology, but of the connections among the technology, CMU administrators, researchers, and building occupants. While CMU positioned Mites as a neutral device for measuring occupancy and environmental conditions, many occupants believed it was surveillance because they were not consulted and its purpose remained unclear. This disconnect showed that privacy is not simply a technical requirement but instead is a social expectation influenced by the situation, trust, and the power of the institution. My STS research was used directly to inform our approach to HoosStudying. We knew that the only way that students would trust our system is if we were transparent about what data we collect and how it benefits them. We talked to potential users as early as possible, to best understand their privacy concerns and determine their actual needs. We clearly explained that sensors detect presence but not identity. Ultimately, we framed HoosStudying as a solution for the students by the students (as opposed to an administrative tool for monitoring as that would demonstrate a lack of trust in student autonomy). The Mites deployment is a comparison of how two different ways of rolling out an identical sensing technology can be perceived in a completely different way depending upon the sociotechnical network within which it exists. The occupants of a facility see monitoring as being acceptable if they trust the institution, understand the purpose of the system, and are involved in the design process. Conversely, if a sensing technology is introduced without engaging the stakeholders or having adequate communication, the same privacy-preserving system could seem invasive to occupants. The work I am doing indicates that surveillance and monitoring differ based on how technology, organizations, and individuals interact and create that distinction. As universities deploy more sensor networks, they must recognize that privacy-preserving technology alone does not guarantee acceptance. In order for there to be truly accepted systems with privacy protections, there must also be transparency among stakeholders, meaningful participation, and institutional practices that foster trust. Engineers can design and develop monitoring technologies with human dignity as a priority by considering the responsive issues identified in conjunction with technical design.
