Terminologies and Taxonomies of Internet Shutdowns
To view the first iteration of this paper please visit our GitHub page.
Internet shutdowns are an extreme form of internet censorship which has steadily increased since 2007. Despite their use as a tool for digital repression for over 15 years, there is uncertainty surrounding what an “internet shutdown” is and the types that exist. Additionally, as most taxonomies of shutdowns focus on the technical methods used, there is a gap for a taxonomy focused on non-technical measurements, such as duration and geographic impact. By reframing the terminologies used when discussing shutdowns and creating a taxonomy based on temporal and geospatial variables, this paper provides two novel frameworks which could be helpful for a non-technical audience. This upcoming paper will discuss shutdowns in the context of terminologies and taxonomies, respectively.
The Four Pillars of Internet Shutdowns
“The Four Pillars of Internet Shutdowns” refers to a new framework we look to propose which consists of three key points. Firstly, internet shutdowns are a sub-field of internet censorship and in turn, internet censorship is a sub-field of information controls. This framing situates internet shutdowns in the broader literature. Secondly, given the advocacy work done by various NGOs and the accepted use of “internet shutdowns” as a key concept within the digital freedoms community, we have opted to keep this term as an umbrella term to encompass the four key types of intentional internet disruptions we have identified. Lastly, the four key types of intentional internet disruptions we have identified are:
1) internet blackout,
2) network shutdown,
3) platform blockage,
4) and internet slowdown.
A Shutdown Taxonomy for Data Science Research
Various taxonomies for shutdowns or outages exist. For example those from Access Now’s report “A taxonomy of internet shutdowns: the technologies behind network interference”, Feldstein’s book “The Rise of Digital Repression“, Thousand Eye’s eBook “The Internet Outage: Survival Guide”, and West’s 2016 article “Internet shutdowns cost countries $2.4 billion last year”. Many of these taxonomies are focussed on the technical manner in which a shutdown occurred, e.g. via Domain Name System (DNS) manipulation, Deep Packet Inspection (DPI), or DoS Attacks. Rather than provide another technical taxonomy, we aim to propose a new taxonomy for the purposes of social data science research. Based on the aforementioned taxonomies, and taking inspiration from Marchant & Stremlau’s 2020 article “A Spectrum of Shutdowns: Reframing Internet Shutdowns From Africa“, we believe there is value in expanding the one-dimensional spectrum approach to a two or three-dimensional space. For example, but not limited to, spectrums surrounding reactive/preventative shutdowns, duration, networks affected and geographic scope.
Choosing three key variables, for example, geographic scope, duration and depth, with provide 8 quadrants or buckets to analyse shutdowns in a methodologically principled manner. For example, to classify a large geo-scope and long duration shudown as Type 1, or a small geo-scope, short duration shutdown as Type 5, and so forth. We acknowledge this oversimplifies the complexities of shutdowns, and that deciding the threshold for a small or large (or medium) scope is a difficult task. However, we believe this approach will provide more granularity than current research, which often groups all shutdowns together.
Our paper explored the terminologies and taxonomies of internet shutdowns and provided a brief historical context and review of existing approaches, proposing two novel non-technical frameworks. The first was the “Four Pillars of Internet Shutdown”, which described four archetypes of shutdowns that all fall under the umbrella term of “internet shutdowns”, namely 1) internet blackout, 2) network shutdown, 3) platform blockage, and 4) internet slowdown. The second described a systematic way to group shutdowns, for example, based on geographic scope, duration, or depth. This approach would enable, for example, computational social scientists to analyse between and within similar shutdowns in a methodologically principled manner. This paper provided a new addition to the literature, as it explored the grey areas between shutdown definitions and synonyms and went beyond the current “spectrum” approach to shutdowns.