TPRC 2024: The 52nd Research Conference on Communications, Information and Inter

Link to paper

Abstract:
New legislation has prompted US regulators to develop processes by which crowdsourced Internet measurements can be used as part of a formal process to fix the National Broadband Map--the official record of where Internet access is (and is not). Currently, the Federal Communications Commission (FCC) Broadband Data Challenge (BDC) process assumes that the people who experience Internet inequities are willing and able to participate in this civic data work. However, there is a body of evidence that demonstrates that the demographics most likely to experience digital inequities are also unlikely to contribute to crowdsourced data collection efforts. Thus, it is critical to understand how the framing and design of this civic crowdsourcing work may prevent marginalized stakeholders from participating in efforts to rectify the National Broadband Map.

In this paper, we make three contributions. First we used the Theory of Planned Behavior to deductively examine six of the most common crowdsourced Internet measurement platforms to characterize how they may encourage (or hinder) people from marginalized groups from participating in a measurement-based challenge process. We focus particularly on the FCC Speed Test app, which is the official mechanism by which citizens can participate in the FCC BDC. Second, we conducted a case study to examine the incentives and barriers that influence citizen participation in the FCC BDC in a community in the Southwest (pop. 70,000). This involved conducting exploratory interviews with eight residents living in the more rural outskirts of the community and the development of a survey, which was completed by 36 residents living in the area. While the data is small, it provides useful insight into how different groups of people within a single community perceive of the challenge process with different priorities and concerns based on location, race, and ethnicity. Specifically, it reveals how non-white and Hispanic participants are much more likely to be concerned about issues of privacy when participating in crowdsourced data work. Finally, we conducted a measurement focus group in the same small metropolitan area, involving nine participants in a one-hour data collection walk around the neighborhood using the FCC Speed Test app. Through this activity, we found that the official tool for making challenges is relatively unappealing to citizens who are generally interested in participating in civic action, and needs to be substantially re-evaluated to be more engaging for a broader civic audience.


Through our analysis, we find that there are straightforward but important ways that crowdsourced measurement platforms could be designed in the future to support broader participation and the collection of more representative crowdsourced data sets. Moreover, we identify specific critiques with the current FCC BDC that will likely prevent participation by everyday citizens. Ultimately, for crowdsourced Internet measurement data to be meaningful and representative, regulators need to attend more critically and carefully to the design and deployment of crowdsourced measurement tools.

Link to paper

Abstract:
As the speeds of residential consumer Internet access services continue to increase (especially in the download direction), speed is becoming less of a barrier to satisfactory Internet service. Attention is now moving to other measures of quality for access services, including latency and its dynamic variation (jitter), and packet loss.
The objective of this paper is to present data on actual measured latency and jitter in the Internet today, based on measurements from five different measurement platforms. We have two goals for this paper. First, we relate the results to possible impact on user Quality of Experience (QoE). Second, and equally important, these comparisons let us uncover limitations and inconsistencies in the various measurement methods, and point out barriers to accurate data collection.

Link to paper
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4925783

Abstract:
This paper studies the construction and use of undersea internet cables. In our model, firms decide whether to invest in new cables and then country-level demand for data leads to flows across the cable network. Investment in any one market affects global data flows. We estimate this model using new data on cable construction and usage via moment inequalities. Decomposing growth in internet usage into growth in demand and improvements in the cable network shows that the latter is an important contributor. Our counterfactuals highlight the role of business stealing and network spillovers in generating an inefficient allocation of cables.

Link to Paper

Abstract:
How do untrustworthy, non-compliant, and otherwise dangerous certificates arise in the Web? What are the causes that underlie the issuance of these certificates? To determine the ground truth, we compiled reports of public key infrastructure (PKI) incidents that have resulted from Certificate Authorities' (CAs) issuance of non-compliant certificates from 2001 to December 2021 from reliable public sources and provide an analysis using qualitative coding of the CAs description of the reported incidents. Our data sources had to be public, reliable, impartial, and trustworthy. These requirements eliminated incidents published in media without proper sources, for example Medium blog posts. The backbone of our incident collection was Mozilla’s Bugzilla where we collected 597 incident reports. Our results combine both qualitative and quantitative analyses. We document the trends in incidents including causes and types. We identify the parties that have erred, the ways in which they have failed, the patterns of behavior among and between CAs. We enumerate the common recommendations where we concur with the literature, and make some of our own. We argue that there is a need for systematic improvement in PKI now, and this need will only increase as the interaction space for warnings and indicators decreases with IoT and embedded systems. We also discuss potential avenues for future work to prevent future incidents and detect problematic certificates before issuance.

Link to paper

Abstract
Society increasingly relies on the Internet as a critical infrastructure. Inter-domain routing is a core Internet protocol that enables traffic to flow globally across independent networks. Concerns about Internet infrastructure security have prompted policymakers to promote stronger routing security and the Resource Public Key Infrastructure (RPKI) in particular. RPKI is a cryptographic framework to secure routing that was standardized in 2012. In 2024, almost 50% of routed IP address blocks are still not covered by RPKI certificates. It is unclear what barriers are preventing networks from adopting RPKI. This paper investigates networks with low RPKI adoption to understand where and why adoption is low or non-existent. We find that networks’ geographical area of service, size, business category and complexity of address space delegation impact RPKI adoption. Our analysis may help direct policymakers’ efforts to promote RPKI adoption and improve the state of routing security.

Link to paper

Abstract
What is the role of broadband Internet in shaping political participation in Africa? This study leverages the staggered arrival of submarine Internet cables and the terrestrial backbone network in Africa to examine the influence of high-speed Internet on protests. Combining large-scale cross-country surveys on political behaviors and disaggregated data on conflict events, robust difference-in-differences estimates indicate a significant increase in both the likelihood of protest participation and the frequency of protests. This effect is particularly observed in countries that actively exercise political rights and civil liberties. The analysis explores two key mechanisms---information dissemination and coordination facilitation---with the latter being the predominant channel of explanation.