Searching for signal and borrowing wi-fi: Understanding disaster-related adaptations to telecommunications disruptions through social media

Disaster events can expose the vulnerability of telecommunications infrastructure to service disruptions. During these traumatic events, when connectivity is most needed, it sometimes takes days, or even weeks or months, for normal service to return. Affected people and communities attempt to adapt to these disruptions in creative ways, but this can lead to changing demands on other parts of the infrastructure. To understand the societal impacts of disasters and inform disaster preparation and response, it can be valuable to understand these behavior changes. In this research, we look to social media (Twitter) to provide insight into how people in Puerto Rico adapted to extended telecommunications disruptions after Hurricane Maria in September 2017. First, to address the challenge of limited signal within the noise of online discourse, we articulate an approach for using machine learning to detect adaptations to telecommunication disruptions in a massive Twitter dataset. Next, using a grounded approach, we developed and applied a qualitative coding scheme that revealed the different ways that people adapted to disruptions in cell service, Wi-Fi access, and electricity for their communication devices. Some of these adaptations demonstrate affected people's willingness to go to great lengths to access telecommunication services, and shifts in how people relied in new ways upon existing infrastructure - e.g. as schools become a place to charge devices. These findings offer empirical insights about how people adapt to telecommunications disruptions as well as methodological contributions around using social media as a signal for informing disaster research and response.

Automated summary

Disaster events can expose the vulnerability of telecommunications infrastructure, leading to prolonged service disruptions. Understanding how people adapt to these disruptions can provide valuable insights for disaster preparation and response. In this research, the authors used Twitter data to examine how people in Puerto Rico adapted to extended telecommunications disruptions after Hurricane Maria in September 2017. They applied machine learning techniques to detect adaptations to telecommunication disruptions in the Twitter dataset, and used qualitative coding to analyze the different ways people adapted to disruptions in cell service, Wi-Fi access, and electricity for communication devices. The findings highlight the willingness of affected individuals to go to great lengths to access telecommunication services, as well as shifts in reliance on existing infrastructure.