We present the design, preliminary deployment, and user evaluation of a Large Language Model (LLM)-based tool to improve privacy in online communications within grassroots and frontline service-providing community organizations that interact with clients as a part of their work. During coordination and service provision, personally identifiable information (PII) may be shared between workers, volunteers, and clients. Collaborating with the Seattle Community Network (SCN) volunteer organization, we built a tool that automatically redacts PII from client coordination messages shared in their public chat server, and reveals the PII only to authorized users upon request. User feedback on this early iteration reveals a high valuing of its privacy-preserving functions, but concerns around trusting LLM output, a requirement to self-host LLMs even on limited hardware, and potential frictions around redaction time and tool onboarding.

Networking research often abstracts away the people who build, operate, and experience the Internet. Yet, human decisions shape what gets measured, which problems are prioritized, and how solutions are deployed. This paper argues that such human influence is foundational and deserves methodological attention. To do so, we discuss three well-known qualitative methods and approaches: participatory action research, ethnographic methods, and positionality as concrete ways of engaging with the social and operational realities that underlie technical systems. These approaches formalize processes that are commonly implicit in networking research, help surface questions that cannot be answered with traces alone, and make space for voices often left out of the research pipeline or are inadvertently concealed due to the lack of formal procedures to include them in our research methods. Ultimately, we argue for a broader understanding of legitimate, and sometimes informal, contributions to networking research—one that better reflects the human element of how the Internet is structured and is experienced.

In tiny house villages (THVs), land ownership, local regulations such as zoning, and management frameworks affect how formerly unhoused residents can leverage and augment their built infrastructure with ''smart'' technologies. These dynamics influence THV residents' lived experiences, creating both constraints and opportunities for deploying networked sensors in support of communication, physical security, and shared governance. Through a series of participatory design (PD) workshops and visits with residents of two villages, we identified how residents balance privacy and security concerns when considering Internet-of-Things (IoT) sensor design and deployment. We find, for example, that residents impose constraints on designed sensors for the group's protection, such as a strict ban on camera-based visual or audio surveillance and a preference for local as opposed to cloud data storage. They also identify opportunities for diverse sensors and actuators to improve village accessibility and to alleviate resource sharing tensions. Meanwhile, differences between public and private land ownership directly impact the regulation and infrastructure possibilities for THVs, and historical and current zoning of land shapes the social problems that the community must navigate. Our findings deepen the CSCW research agenda by examining parameters for successful ''smart'' and data-driven technology interventions among low-resourced urban groups experiencing housing precarity and homelessness.

We present dAuth, an approach to device authentication in private cellular networks which refactors the responsibilities of authentication to enable multiple small private cellular networks to federate together to provide a more reliable and resilient service than could be achieved on their own. dAuth is designed to be backwards compatible with off-the-shelf 4G and 5G cellular devices and can be incrementally deployed today. It uses cryptographic secret sharing and a division of concerns between sensitive data stored with backup networks and non-sensitive public directory data to securely scale authentication across multiple redundant nodes operating among different and untrusted organizations. Specifically, it allows a collection of pre-configured backup networks to authenticate users on behalf of their home network while the home network is unavailable. We evaluate dAuth's performance with production equipment from an active federated community network, finding that it is able to work with existing systems. We follow this with an evaluation using a simulated 5G RAN and find that it performs comparably to a standalone cloud-based 5G core at low load, and outperforms a centralized core at high load due to its innate load-sharing properties.

The global community networking movement promotes locally-managed network infrastructure as a strategy for affordable Internet connectivity. This case study investigates a group of collectively managed WiFi Internet networks in Argentina and the technologists who design the networking hardware and software. Members of these community networks collaborate on maintenance and repair and practice new forms of collective work. Drawing on Actor-Network Theory, we show that the networking technologies play a role in the social relations of their maintenance and that they are intentionally configured to do so. For technology designers and deployers, we suggest a path beyond designing for easy repair: since every breakdown is an opportunity to learn, we should design for accessible repair experiences that enable effective collaborative learning.

Congestion control mechanisms, by which network users share constrained capacity on Internet links, are heavily studied in computer science. Such mechanisms are traditionally automated, assuming that users do not wish to be involved in addressing congestion. However, in community-owned and operated networks, users have control over daily operational choices. We explore the design of community-based congestion policies and mechanisms, through the lens of network capacity as a Common Pool Resource (CPR). Through a series of workshops and interviews in a rural community in Oaxaca, Mexico, we encounter design opportunities for new types of tools supporting communal network management. Participants expressed desires for preserving individual privacy while collecting longitudinal data to track the network's impact on the community, prioritization of high-value applications, equal link sharing between users, and human-mediated congestion management in lieu of automated enforcement. We report qualitative insights and offer design directions for future systems to address network resources in a manner compatible with Ostrom's principles for CPR governance.

Across the world, government websites are expected to be reliable sources of information, regardless of their view count. Interactions with these websites often contain sensitive information, such as identity, medical, or legal data, whose integrity must be protected for citizens to remain safe. To better understand the government website ecosystem, we measure the adoption of https including the "long tail" of government websites around the world, which are typically not captured in the top-million datasets used for such studies. We identify and measure major categories and frequencies of https adoption errors, including misconfiguration of certificates via expiration, reuse of keys and serial numbers between unrelated government departments, use of insecure cryptographic protocols and keys, and untrustworthy root Certificate Authorities (CAs). Finally, we observe an overall lower https rate and a steeper dropoff with descending popularity among government sites compared to the commercial websites & provide recommendations to improve the usage of https in governments worldwide.

Recent work has applied machine learning methods to automatically find and/or assess pedestrian infrastructure in online map imagery (e.g., satellite photos, streetscape panoramas). While promising, these methods have been limited by two interrelated issues: small training sets and the choice of machine learning model. In this paper, aided by the recently released Project Sidewalk dataset of 300,000+ image-based sidewalk accessibility labels, we present the first examination of deep learning to automatically assess sidewalks in Google Street View (GSV) panoramas. Specifically, we investigate two application areas: automatically validating crowdsourced labels and automatically labeling sidewalk accessibility issues. For both tasks, we introduce and use a residual neural network (ResNet) modified to support both image and non-image (contextual) features (e.g., geography). We present an analysis of performance, the effect of our non-image features and training set size, and cross-city generalizability. Our results significantly improve on prior automated methods and, in some cases, meet or exceed human labeling performance.

This paper analyzes the processes and challenges of technology repair in remote, low-income areas far from standard ICT repair infrastructure. Our sites of study are the fishing and farming villages of Dibut, Diotorin, and Dikapinisan in Aurora Province, Philippines, located in coastal coves against a mountain range. Residents are geographically isolated from urban areas, with the nearest peri-urban center of Baler a boat trip of several hours away, infeasible in some sea conditions. Unlike prior work in more connected rural areas, there are no local repair shops and device repair is uncommon, despite frequent breakage due to harsh conditions for electronics. The scarcity of local electronics repair limits technology access and leads to accumulation of e-waste. While prior work demonstrates that local electronics repair capability does arise in many rural areas around the world, we must also acknowledge that the successful emergence of this infrastructure depends on the intersection of many structural conditions and cannot be taken for granted. We present the material hardships of achieving local repair in terms of seams between heterogeneous urban and rural infrastructures, which illustrate the cove communities' marginality with respect to many forms of public infrastructure. However, intermittent and informal repair infrastructures based on trust relationships emerge to patch these seams in remote settings. We show how trust affects the way people dynamically construct repair infrastructure and why, based on their remoteness and the resulting value propositions of repair. Networks of trust between repairers, their clients, suppliers, fellow repairers, and certifying or training institutions crucially facilitate the movement of resources and expertise across the Philippines, but also reinforce the marginality of residents and repairers in the coves. Despite these structural challenges, local people are able to maintain a robust ecosystem for rural electrical line repair, from which we generalize the model of training grounds as a strategy for sustaining local communities of repair experts.

Repair and maintenance requirements limit the successful operation of rural infrastructure. Current best practices are centralized management, which requires travel from urban areas and is prohibitively expensive, or intensively training community members, which limits scaling. We explore an alternative model: crowdsourcing repair from the community. Leveraging a Community Cellular Network in the remote Philippines, we sent SMS to all active network subscribers (n = 63) requesting technical support. From the pool of physical respondents, we explored their ability to repair through mock failures and conducted semi-structured interviews about their experiences with repair. We learned that community members would be eager to practice repair if allowed, would network to recruit more expertise, and seemingly have the collective capacity to resolve some common failures. They are most successful when repairs map directly to their lived experiences. We suggest infrastructure design considerations that could make repairs more tractable and argue for an inclusive approach.

The radio interfaces and network architectures of WiFi and cellular systems are converging along many dimensions. While both systems are largely adopting the centralized architecture of traditional cellular deployments, this design comes with fundamental disadvantages that limit how these networks grow and develop. As a response, we present Distributed LTE (dLTE), an architecture offering the high radio performance of licensed and coordinated waveforms as well as the openness to organic expansion and growth of traditional WiFi. We challenge the assumption that good performance requires a centralized packet processing core, and propose hybrid approaches to coordination that prioritize system openness. We argue that dLTE is a particularly good fit for rural areas, where the LTE waveform is more appropriate than WiFi, yet it is uneconomical for centralized providers to deploy traditional cellular systems.