Making infrastructure visible: a case study of home networking

Technological infrastructure is often taken for granted in our day to day lives until it breaks down, usually because it invisibly supports tasks otherwise. Previous work in HCI has focused on how people react and deal with breaks in infrastructure as well as how to help people to fix or exploit these breaks. However, few have sought to understand how people will react to information about infrastructure when it is functioning normally or whether access to this information causes changes in how people engage with that infrastructure. The goal of my dissertation is to understand how people's engagement with infrastructure changes when they are provided with information about that infrastructures normal functioning, i.e., when this oftentimes invisible information is made visible. To achieve this goal, my chosen case study of infrastructure is the home network, a self-contained socio-technical system within the domestic space. Much like other infrastructures, the home network suffers from visibility issues when it is functioning properly or even when it breaks even though it has a distinct digital footprint—the collection of information about traffic between devices on the home network as well as information about the devices comprising the network. The digital footprint not only contains information about network use but it also reflects computing routines or everyday household activities and policies around the network, such as what resources the network uses (e.g., bandwidth or electricity) or who is using the network, when and for how long. The home network is an interesting case study because the digital footprint is still largely inaccessible to household occupants for four reasons. First, networked domestic technologies have become increasingly common and oftentimes are being integrated directly into the home's built environment. For instance, Ethernet cabling may be built into new homes, and in the event that connections are not working, this problem may be impossible to troubleshoot without tearing up walls. Other times, technologies may be totally wireless and difficult to visually assess for problems. Second, the devices making up the home network are distributed throughout the home making it hard to gather an overview of the network topology or configuration in one place. Third, many of the devices making up the home network have reduced interfaces such as a series of blinking lights which provide little information, particularly when problems occur. Fourth, even when tools exist to access glimpses of the digital footprint, such as network traffic visualizers, these are accessible and known only to the most technically able household occupants. Often, these technical occupants know where to look for this information and what to do with it when they get it, e.g., using “ping” to determine if the network is working. In all, these issues make it difficult to understand who is contributing to the digital footprint at any one time. It also remains unclear how we can use this information as a point of reflection on computing routines (e.g., how is the network used and what resources does it require) or for broader maintenance purposes such as troubleshooting the devices making up this network. In this dissertation, my aim has been to investigate how surfacing the invisible in the home can affect users' engagement with infrastructure. The document is structured as follows. I first describe the motivation for this work and previous related work on infrastructure and visibility. I then provide background information on my chosen case study, home networks, and discuss how the digital footprint has evolved over time. Next, I describe how the digital footprint remains largely invisible and the opportunities that exist for surfacing this information. I also describe the user-centered design approach I followed. Thereafter, I detail the empirical results that have informed my design of the Kermitsystem, the technology probe I used to surface aspects of the home's digital footprint. I implemented Kermit to determine how householders react to information about their home networking infrastructure when it is functioning normally. My aim was to determine whether this changes their engagement with that infrastructure. I describe the detailed design and requirements for Kermit and the pilot study and final field trial conducted in Spring 2010. I then discuss themes from the entire body of work and design considerations for making infrastructure visible, a concept I call inspectability. Finally, I summarize my overall contributions and suggest future research directions for achieving inspectability in the home via another concept I introduce called network-mediated sensing. I conclude that my research suggests making infrastructure more visible has benefits for helping users manage these complex systems of systems.