Resilience is often described by a number of other words beginning with ‘R’. Resistance or Robustness are two. These refer to infrastructure that is designed or an activity that is organised to avoid risk and withstand the expected impact or disruption. Redundancy and Replication are two more. These refer to equivalent or at least parallel alternatives that enable the same outcomes, even if not by the same means. In transport, it could be an alternative route or another mode that still enables access to the same destination. Finally, there is Response and Recovery, how quickly action is taken and life returns to either its previous ‘normal’ or stabilises into a new ‘normal’.
As I explored in my recently awarded doctoral thesis, Information and Communication Technologies (ICT) have a major role to play in resilience to severe weather events. Broadband networks are often more robust than transport infrastructure and access to some activities such as work can be replicated online. Opportunities for communication are also enhanced to speed recovery.
Yet the ability of ICT to play these roles may be constrained by the interactions between supply and demand, availability and quality. Broadband supply and availability can be limited if there are not the physical connections to support superfast services, or certain providers simply do not operate. Yet even where the supply is available, broadband quality is affected by increased demand – something the industry calls ‘contention’ or slowdown in busy periods. Usually this happens due to entertainment, and the daily peak tends to be in the evenings, when streaming video content is most popular. Other times, such as during major sporting events, capacity can be managed to some extent by Internet Service Providers.
In contrast, my research was designed to test if contention could be detected when internet activities are in unusually high demand because household members are unexpectedly home due to extreme weather events. Thus, I analysed data comprising 5 years of broadband speeds, from 2012-2016, in England and Wales, which was provided by Speedchecker Ltd, a private company that allows internet users to check their own broadband upload / download speeds, and stores every speed-check with a timestamp and geolocation. Using multilevel modelling to control for supply side variation, I regressed download speeds against daily weather records to assess how more severe weather affects contention experienced at the neighbourhood level. My results confirmed that winter weather and storm-level winds have significant, albeit small, negative effects on broadband speeds.
Yet these effects are likely to have been an underestimate – sensitivity testing suggested greater contention in subsets of the data designed to control for uneven growth in average broadband speeds or exclude spatial extremes in service. There are also divergences between weather parameters and weather impacts, with the latter dependent upon confounding variables like what type of transport infrastructure is affected and the length of advance warning. Also, the noise in this big data source and the subtleties of the assessment meant it was not easy to identify who was most likely to avoid travel risks and work from home during severe weather or where the most telecommuting activity occurs.
In contrast, we now have an ongoing disruption where everyone who can has been told to use ICT in order to be more resilient, avoid a very different type of risk, and replicate their productive activities as much as possible until we can recover. How is the quality and level of service of broadband in the UK coping with what is likely to be the most extreme level of demand it has ever experienced? I’m planning to find out!