“Eavesmining”: A New Addition to the Digital Surveillance Lexicon
In the post below, Stephen Neville reflects on his article, “Eavesmining: A Critical Audit of the Amazon Echo and Alexa Conditions of Use,” which appeared in a recent issue of Surveillance & Society.
///
When reflecting on surveillance, what immediately comes to mind? Perhaps images of technology are conjured, such as CCTV cameras or facial recognition systems. Or perhaps George Orwell’s famous cautionary slogan from 1984 is evoked, namely, that “Big Brother is watching you.” Or one may recall Michel Foucault’s writing on the panopticon which theorizes how the symbolic gaze of the watcher serves to discipline the subjects of surveillance. Each of these examples reflects normative assumptions about surveillance as a visual practice that generates social problems of visibility and visuality — what can be seen and how things are made to be seen. As a result, it is remarkably difficult to discuss surveillance without relying on dominant visual cases, metaphors and concepts: this is problematic because it can prevent us from adequately questioning how particular modes of surveillance operate, thus limiting our understanding of the critical issues at stake.
It is remarkably difficult to discuss surveillance without relying on dominant visual cases, metaphors and concepts.
In my recently published paper, “Eavesmining: A Critical Audit of the Amazon Echo and Alexa Conditions of Use,” I provide the results of a case study of Amazon’s smart speaker and voice-activated personal assistant, highlighting some contemporary problems of auditory surveillance that are dissimilar from those posed by other media forms. Further, I reflect on the divergence of emergent sonic technologies from older mediated practices of eavesdropping, such as wiretapping, clandestine bugging, and other forms of targeted audio interception. I suggest that smart speakers and interactive digital persona, like Alexa, Siri and the Google assistant, expose the limitations of our surveillance vocabulary: after all, the interactive and monitoring affordances of these platforms do not simply reproduce processes of listening. Rather, they involve the combined use of microphones, digital sensors, signal processing algorithms, database systems, and data mining techniques. To characterize this unique mode of surveillance, I’ve coined the concept of eavesmining (eavesdropping + datamining), to capture surveillance mechanisms that simultaneously integrate acoustic inputs and digital infrastructure to monitor voices and other sounds within public and private environments.
The concept of eavesmining stresses the computational logic of emergent sonic technology while emphasizing that sound features unique qualities that can be analyzed for the sake of identification and discrimination. For instance, the human voice embodies a personally identifying biometric trait while the way in which one speaks can be used to problematically deduce aspects such as age, accent, gender, race, affect, mental health and physical health. This presents new problems of social sorting that warrant scrutiny, as applications of eavesmining technology proliferate and become increasingly normalized in our homes, vehicles, workplaces, public spaces, and in various other institutional and commercial environments.
The paper helps to think through these problems by grounding analysis in an empirical study of Amazon’s end-user agreements or “conditions of use” governing the Echo and Alexa platform. I trace the evolution of these documents from 2012–2019 to provide evidence of growing privacy and surveillance concerns and demonstrate how they are obfuscated by the illegibility of the documents. More generally, the article concludes by arguing that sonic forms of thinking can complement other approaches to surveillance studies.
