The Cascading Technology Stack: A Multilevel Approach to Effective Deplatforming

• In Event: Lightning Round: Graduate Student Research: Ideas, Institutions, Coalitions & Movements

Abstract

Deplatforming is generally thought to reduce the amount of extremist content on mainstream social media. However, recent studies of content moderation show that this may not be an accurate representation of the entire information ecosystem (Ali et al. 2021; Chandrasekharan et al. 2017; Newell et al. 2016). Available evidence suggests that the rise of alternative social media platforms, like Gab and Parler, undermine content moderation functions of mainstream platforms by creating safe havens for extremist content (Jasser et al. 2021; Lima et al. 2018). If conventional deplatforming tactics at the platform-level lead to the migration of banned users to extremist-friendly alternative platforms, this then raises the question of what constitutes politically effective deplatforming? In this project, I use a series of case studies involving QAnon and the Stop the Steal (StS) movement to develop a model of deplatforming that considers the role of the technology stack in the cross-platform migration of extremist users.
The physical and digital infrastructure that the internet relies upon to operate is known as the technology stack. It is composed of websites, social platforms, cloud services, content delivery networks, domain registrars, internet service providers, and government/telecommunications hardware (Donovan 2019; Fowler and Alcantara 2021). These various layers of the tech stack operate in tandem to determine what kind of content is available on the internet and on what channels. Just as social media platforms can deplatform users, technology companies, such as Google, GoDaddy, and Amazon Web Services, at deeper levels of the tech stack can also suspend entire platforms, such as occurred when AWS suspended Parler from its web hosting service. Soon thereafter, Google and Apple App Stores also suspended support for the Parler app.
The paper proposed here develops a content moderation model that takes account the technology stack. When multiple layers of the technology stack engage with what is to be understood as norms noncompliant content, we refer to it as a tech cascade. Our reference to a cascade reflects one of the apparent features of multi-level content moderation: two or more layers of the technology stack tend to suspend noncompliant users and platforms in or near unison with one another. Thus, when technology companies at multiple levels of the tech stack deplatform extremist content and users, there is a greater likelihood that these extremists will not be able to find refuge in less-regulated corners of the internet.

Theoretical Framework
Our tech cascade model investigates the development of an alternative technology ecosystem. For decades, extremist groups have used social media to organize and grow their reach. In response, mainstream platforms have sought to deplatform them (Donovan et al. 2019; Benkler et al. 2018). In response, a countermovement has emerged intent on building an “alt-tech” system consisting of online video platforms, crowdfunding sites, web browsers, news aggregators, etc. (Donovan 2019). These technologies create a kind of parallel internet that caters to those whose views fall outside of what is allowed on mainstream technologies (Freelon et al. 2020). The overall goal of alt-tech is to create a copycat version of mainstream internet services so that the far-right remains resilient to deplatforming efforts (Letson and Squire, 2019).

Research Design of Case Studies
This paper is an extension of several on-going collaborative research projects I’m involved in as a co-author. They all entail the investigation of cross-platform migration within the StS and QAnon movements. The quantitative aspect of this paper relies on case studies conducted on platform migration rates across Twitter, Reddit, Parler, and Gab. To do this, we first subsetted our Twitter and Reddit datasets to users who posted about either QAnon or StS. We then use Twitter and Reddit’s academic APIs to identify which of these users have been deplatformed. Next, we take the list of deplatformed users and match them to users on Parler and Gab that have the exact same username.

Early Findings and Next Steps
We have used this research design on Twitter users tweeted about StS between 2018 and 2021 and migrated to Parler after being banned by Twitter. Our initial analysis found only 525 migrants, which we hypothesize was due to Parler being suspended by AWS servers three days after the insurrection. This is a prime example of how delplatforming occurring at the cloud services level of the tech stack eliminated Parler’s ability to harbor extremists after January 6th.
The next stage in the platform migration research process is to use this same research design for QAnon. The results from both QAnon and StS analyses will form the quantitative foundation for my deplatforming model. At the core of this analysis will be how content moderation occurring at different levels of the tech stack impacts the cross-platform migration rates of these groups.

Author

• Lorcan Neill, George Washington University