Deplatformization and the governance of the platform ecosystem

Abstract

This article analyzes deplatformization as an implied governance strategy by major tech companies to detoxify the platform ecosystem of radical content while consolidating their power as designers, operators, and governors of that same ecosystem. Deplatformization is different from deplatforming: it entails a systemic effort to push back encroaching radical right-wing platforms to the fringes of the ecosystem by denying them the infrastructural services needed to function online. We identify several deplatformization strategies, using Gab as an example of a platform that survived its relegation and which subsequently tried to build an alternative at the edge of the mainstream ecosystem. Evaluating deplatformization in terms of governance, the question that arises is who is responsible for cleansing the ecosystem: corporations, states, civil society actors, or all three combined? Understanding the implied governance of deplatformization is imperative to assess the higher stakes in future debates concerning Internet governability.

Introduction

On 6 January 2021, a cascade of digital events ensued after a mob, incited by former President Donald Trump, invaded the Capitol in Washington and temporarily halted the democratic process to confirm the 2020 election results. Trump was permanently banned from Twitter and his accounts on Facebook, Instagram, and YouTube were suspended; social media services such as TikTok and Twitch (owned by Amazon) followed suit by imposing a similar injunction. The US President’s banning from social media platforms was a significant intervention in one person’s access to those specific networks. What happened next, though, was an intervention in the configuration of the platform ecosystem and arguably in the governance of the platformized public sphere. In the days following the siege, Twitter also suspended accounts linked to QAnon, frequented by many Trump supporters. And before Trump’s followers could flock to one of the rising radical right-wing platforms, Parler was banned from Google’s and Apple’s app stores, while Amazon Web Services (AWS) pulled the rug from underneath it. Parler was left frantically looking for a new infrastructural refuge, as other platforms, including Gab and BitChute, had experienced before them. ACLU-lawyer Ben Wizner commented in the New York Times that, while he understood the curtailment of a specific user from a platform, “we should recognize the importance of neutrality when we’re talking about the infrastructure of the internet” (Nicas and Alba, 2021).

In just over a week, a high-profile act of deplatforming—“the removal of one’s account on social media for breaking platform rules” (Rogers, 2020: 1)—was followed by “deplatformization”: an effort to push back encroaching extreme (right-wing) platforms to the fringes of the ecosystem by denying them access to the infrastructural services needed to function online. In this article, we aim to analyze the deplatformization of radical right-wing platforms as an implied governance strategy utilized by major tech companies to keep the platform ecosystem clean and orderly, while consolidating their power as designers, operators, and governors of that same ecosystem. The next section will discuss the concepts of platformization, deplatforming, and deplatformization in an attempt to capture the techno-economic dynamics of the ecosystem’s operating powers (Poell, Nieborg & Van Dijck, 2019). While big tech companies wield their hegemonic position as private gatekeepers when deplatforming account holders, they ostensibly act in the public interest when “cleansing” the ecosystem by forcing entire platforms to the fringe.

The dynamics of deplatformization are the subject of section “Deplatformization strategies: the case of Gab.” We identify several strategies that result in pushing far-right, radical platforms to the edge, while fortifying the hierarchical powers of mainstream platforms and their control of the ecosystem. Three such strategies are blocking access to distribution, demonetization, and the disabling of infrastructural services. We will explore these efforts by zooming in on Gab, one of the well-known fringe platforms that has survived several rounds of deplatformization since its inception in 2017. Several other fringe platforms, including Parler and BitChute, experienced similar convolutions. Section “Building a parallel ecosystem?” will focus on extreme platforms’ responses to deplatformization, highlighting which tactics they deploy at the edge of the ecosystem (Donovan et al., 2019). Again taking Gab as an example, we will discuss how this platform, severely weakened by Big Tech, fights a struggle for survival in the margins of the ecosystem where it tries to penetrate the alternative infrastructure set up by alt-tech communities and where the same deplatformization mechanisms seem to apply.

The dynamics of deplatformization between mainstream and fringe, and between fringe platforms themselves, prompts the question, raised in Section “Deplatformization as implied governance of the ecosystem,” of (implied) governance: who is responsible for governing the platform ecosystem and who controls its infrastructure? In the past, frictions between mainstream and fringe have always been contests for hegemony and ruling power, but the impact of private tech companies on the technical architecture of the ecosystem is much more profound due to its global reach and infrastructural reign. As Oxford-based philosopher Luciano Floridi (2021, np) argued in response to the Capitol siege aftermath, private tech companies have grabbed a role which is of crucial public interest, “since they decide what may or may not happen in the infosphere and hence in the lives of billions of people.” We argue that understanding the implied governance strategy of deplatformization, as well as its effects at the fringes of the ecosystem, is imperative to assessing the higher stakes in future political debates about the Internet as a public infrastructure.

Deplatforming, platformization, and deplatformization

Before introducing the concept of deplatformization, we first need to reiterate the notion of deplatforming (Rogers, 2020). This definition requires some unfolding: who or what is deplatformed, according to which and whose rules, for what type of violation, and to what effect? “Deplatforming” most commonly applies to stopping a single user account (e.g. Donald Trump) by a social network (e.g. Facebook). Over the past years, a number of controversial content creators have been removed from one, several, or all social media platforms. Mainstream social media platforms (MsSM), such as Facebook, Instagram, Twitter, YouTube alongside a number of smaller ones, including Twitch, TikTok, Reddit, and Snap, have increasingly cracked down on users violating their policies. Each platform has a different set of rules, reified in its Terms of Service (ToS), to define what is allowed and which violations warrant which sanctions. Deplatforming may be the ultimate step in a tiered moderation process. A platform usually first issues warnings, for instance, by flagging or removing pieces of content or by suspending someone’s account temporarily, before deciding on a permanent ban. Deplatforming account holders who do not abide by the rules is indeed a powerful instrument, one that has real consequences for those who are excluded from access to valuable online social circles (Gillespie, 2018, 177).

What types of violations have warranted deplatforming? Until 2016, it was primarily a strategy to force a content creator to abstain from posting illegal content, such as pornography, terrorist threats, or copyright infringements, which are against the rule of law; later it was also applied to disputable claims of pirated content, nudity, or content transgressions that were simply against a platform’s ToS. In the slipstream of the US and UK election processes, MsSMs also began to take more responsibility for the removal of fake news and disinformation as well as for inflammatory, discriminatory, and hateful messages posted by far-right and alt-right communities (“toxic content”). In other words, the act of deplatforming gradually covered a wider range of materials and applied to the broader media and cultural setting of which social media are a part, providing further proof of how online and offline contexts are intimately entangled.

Looking at the effects of deplatforming, we can see it has a substantial impact on ousted individuals and user communities. Over the past few years, notorious account holders such as Alex Jones or Milo Yiannopoulos lost their amplifiers in the online realm after being banned from MsSM (Krauss, 2018). On forum-structured social media like Reddit, entire discourses and communities have been effectively deplatformed. For instance, when Reddit closed down some hate-riddled subreddits in 2015, offended users either left Reddit or moved to other subreddits. Researchers who examined these migrations on Reddit found a significant decrease in extreme speech in the platform as a whole (Chandrasekharan et al., 2017). Other researchers have investigated the migration of users deplatformed by MsSM to platforms where they find their expressive needs more in line with the rules set by their new hosts, for instance, Gab, Parler, BitChute, Telegram, 4chan, and 8chan. Swiss communication scholars, Aleksandra Urman and Stefan Katz (2020), studied the “waterbed effects” following banned social media users to Telegram, where they tend to re-establish their dominance that existed on the platforms from which they were banned, leaving both researchers to wonder whether this is the best way to “curb their influence” and “prevent users’ radicalization” (Urman and Katz, 2020: 15). Dutch media scholar Richard Rogers (2020) concludes that deplatforming helps mainstream platforms like Facebook and Instagram to purify their channels from vitriolic language and rule-violating speech, but is less effective on YouTube and Twitter. As Rogers (2020) rightly asks: “Is it [deplatforming] indeed a viable means to detoxify mainstream social media and the Internet more broadly, and/or does it prompt the individuals to migrate to other platforms with more welcoming and ‘oxygen-giving’ extreme publics?” (p. 215).

To answer this question, we propose the concept of deplatformization. Although deplatforming may be a part of deplatformization, they are not the same, so it is important to clarify the distinction. Whereas deplatforming denotes when users or channels lose their profiles on social media platforms, often due to the dissemination of content that violates the Terms of Use, deplatformization applies to tech companies’ efforts to reduce toxic content by pushing back controversial platforms and their communities to the edge of the ecosystem, denying them access to basic infrastructural services needed to function online. For instance, pulling an online cloud service (e.g. Microsoft Azure) from a specific platform (e.g. Gab) because it offers a haven to far-right hate communities, is a tactic that goes beyond a “content moderation strategy.” Deplatformization not only impacts the position of single platforms, but also affects the dynamics and infrastructure of the ecosystem as such—an ecosystem that is hierarchical and proprietary in nature. Let us elaborate on each of these aspects in more detail below.

Deplatformization can hardly be understood without referring to platformization. In earlier work, we defined the term platformization as “the penetration of the infrastructures, economic processes and governmental frameworks of platforms in different societal sectors and spheres of life” (Poell, Nieborg & Van Dijck, 2019: 5). It is important to note that platformization refers to a dynamic rather than to platforms as objects (Van Dijck et al., 2018). The platformization dynamic happens in a platform ecosystem—a corporate space that in the Western hemisphere is dominated most notably by the Big Five tech companies known by the acronym of GAFAM (Google, Apple, Facebook, Amazon, Microsoft). These competing and cooperating tech companies have succeeded in building a sociotechnical infrastructure on which they rely for their own financial health and global reach (Dolata and Schrape, 2018), but on which also entire public sectors and public communication spheres have become dependent (Mansell and Steinmueller, 2020; Napoli, 2019). Several scholars have envisioned this ecosystem as a “stack” or metaphorical “tree” that constitutes a hierarchically organized ecosystem with various layers to facilitate the connection between a large number of data flows (Van Dijck, 2020; Bratton, 2016). The Big Five have the unique leverage to control the ecosystem’s connectivity by operating more platforms across the stack or tree, integrating them both horizontally and vertically. For instance, while Facebook deploys several other social media networks across the same layer (Instagram, WhatsApp), the firm also operates one of the largest online advertising platforms, allowing it to integrate data flows horizontally and vertically. In a similar vein, Google is able to pull the strings of its large video streaming service (YouTube), whereas it also operates a cloud service, a pay system, a search engine, and one of the two major app stores. Apple offers the other major app store, as well as a crucial operating system and pay service. Amazon, which also owns Twitch, is the largest owner of cloud services (AWS) while Microsoft not only operates a cloud service (Azure) but also a major social network (LinkedIn) and has a dominant share in cloud-based office software.

The platformization dynamic is vital to explain the difference between “deplatforming’”and “deplatformization.” First, while deplatforming applies mostly to specific actors on specific platforms—for example, blocking account holders who produce toxic content from social media networks—the concept of deplatformization applies more broadly to the denial of infrastructural services deeper down the stack, for example, browsers, cloud services, pay systems, app stores, advertising services, domain name systems, and so on. Deplatformization impacts a platform’s ability to operate as a part of the ecosystem—an existential threat to its survival. In addition, deplatformization affects not just single account holders because they break the rules of one platform; instead, platform operators are denied infrastructural services tout court because they cannot (or refuse to) keep their channels clean.

Second, while deplatforming is mostly the result of one platform’s act of self-governing its own online space, deplatformization efforts refer to a much broader attempt to control the ecosystem’s hierarchical structure by responding to each others’ actions. The ensuing power struggle exposes inherent tensions (1) between mainstream and fringe platforms and (2) between fringe platforms themselves, often exposing clashing commercial and public interests. Deplatformization effects thus ripple across the ecosystem, from the core to the edges. Indeed, the growing number of deplatformed account holders stems from mounting public pressure on tech firms to keep their social platforms clean from illegal and toxic content. Following the public demand for self-governance while also trying to avoid stricter state regulation, each MsSM platform has installed a different set of rules to define what is allowed and which violations warrant deplatforming. Although there is some overlap between their policies, there is no agreed upon set of rules that governs the platform ecosystem as a whole.

In contrast to deplatforming, deplatformization is arguably a response to a growing need to view the platform ecosystem as a collective realm where private business and societal interests constantly overlap and clash. On the one hand, mainstream platforms are heavily invested in designing algorithms that amplify and broadly spread popular content to keep users on their sites longer; content containing disinformation and extreme messages is proven to spread faster and attract more users which forms the core of their business models (Vosoughi et al., 2018). On the other hand, big tech firms need to operate their various platforms within legal limits to keep regulators at bay and to remain attractive to a large user base. Radical and alt-right platforms, for their part, need mainstream platforms to spread their messages, generate revenue, and secure growth while also satisfying their core base of extreme content generators (Conway, 2020). Deplatformization hence involves a delicate balancing act between mainstream and fringe platforms, accommodating contradictory commercial interests while responding to public calls for an online infrastructure that is governed transparently and fairly. As of yet, there is neither any agreed upon set of rules by which the platform ecosystem is self-governed, nor any (supra-)national regulation to govern it as a public infrastructure. Therefore, we ask whether deplatformization is an implied governance strategy to detoxify the ecosystem of radical and toxic content or, more than that, is it a strategy to control hegemony in the online infrastructure as such?

In order to better understand the strategies and dynamics of deplatformization, we will examine the specific case of Gab in the following two sections. First, we will analyze deplatformization dynamics between mainstream platforms and Gab, reconstructing the sequence of unilateral cancelations of collaborations, partnerships, and customership of fringe platform Gab.com from the infrastructural and supporting services it depends on. In section “Building a parallel ecosystem?,” we will study the effects on Gab in their ensuing struggle with other fringe platforms at the edge of the ecosystem. This two-tiered reconstruction is based on publicly accessible news articles on Gab reporting their clashes with platform providers, found by crossreferencing and through the Web archive, Gab’s own public communication on this process, and official public documentation of tech companies like SEC filings and developers’ pages.

Mainstream versus Gab: deplatformization strategies

The platform Gab was launched by Andrew Torba in 2016 as a microblogging service, an alternative to Twitter that also featured an email-service, text messaging, and a web browser. Under the flag of free speech and defending a free flow of information, Gab attracted extremists—everything from the alt-right, far-right, and neo-Nazis to QAnon and conspiracy theorists. Since its inception, Gab has offered a refuge to voices banned from mainstream social media, such as Twitter, Facebook, and YouTube—a “platform for the deplatformed.” Over the years, Gab’s user base has benefited significantly from MsSM’s mounting efforts to purge extreme voices from their premises, such as Yiannopoulos’ and Jones’ ban from Twitter in, respectively, 2016 and 2018; both subsequently resorted to Gab. But while Gab gained in terms of user numbers, it lost in terms of networking effects—its ability to benefit from connectivity with other platforms. Constraining actions mounted particularly after the Pittsburgh synagogue shooting in October 2018, as it turned out that the antisemitic perpetrator of this attack had announced his plan on Gab.

It is easy to block a single accountholder from a specific platform; it is more complex to deplatform a platform from an ecosystem. You cannot merely suspend a platform from all sites, but mainstream actors (big tech platforms) can consistently and systemically undermine a platform’s material basis for connectivity, hence blocking its entry to data flows that function like oxygen. For optimal functioning, Gab is dependent on partnerships with various platform operators that are either owned by or partner with big tech companies, or that are in other ways reliant on GAFAM’s infrastructural services. The deplatformization efforts in Gab’s short history are too numerous to recap here, but we distinguish three specific strategies that involve services from all layers of the stack: (1) blocking access to networked distribution, (2) demonetization, and (3) disabling infrastructural services.

First, blocking access to networked distribution is a strategy that impedes a platform’s ability to attract users to its platform and to publish its content widely. For the purpose of attracting users through mobile devices, Gab is largely dependent on app stores. Gab was banned from Apple’s app store soon after its launch in 2017, while Google, after first admitting Gab in its Play store, quickly followed Apple’s decision to ban Gab on account of violating Google’s hate speech policy. Without the possibility to download the Gab App or Gab Premium from the app store, Gab’s reach was limited to its own website. However, access to this website was also curtailed in various ways. In order to draw attention to its content, Gab heavily relied on Twitter, where they had a GetOnGab account. In 2017, Twitter cut off Gab’s access to the Twitter API without specifying their motivation, leading Gab to say that this banning was politically motivated. In 2019, in reaction to the restrictive comment sections used by newspapers, Gab added an in-browser extension named Dissenter, enabling a censorship-free comment section on top of any web address. The feature led to a considerable increase in users, but it was soon disabled by Google Chrome and Mozilla Firefox—the most popular mainstream browsers. Later, in 2020, the Dissenter browser was built on the open-source code of Brave—a strategy we will return to in the following section.

A second deplatformization strategy is demonetization, which can be achieved in various ways, such as taking away a payment service or denying access to fundraising activities. To start with the latter, because Gab positioned itself as an “ad-free social network” its monetization abilities came to depend heavily on online crowdsourcing efforts and subscription payments from users for Pro and Premium accounts. Gab’s partnership with StartEngine, a securities brokerage firm that helps companies prepare regulatory filings and sell investment shares to the public, got crippled after the Pittsburgh shootings, when they removed Gab from their crowdsourcing platform (Dougherty and Hayden, 2019). The first major financial service to withdraw their service was PayPal; the company terminated its relationship with Gab in 2018, based on its review of user accounts that engage in the perpetuation of hate, violence, or discriminatory intolerance. Having been cut off from payment systems, Gab could no longer sell merchandise nor transact direct donations from users or sponsors. In early October 2018, Stripe—a payment processing platform allowing merchants to accept credit card payments on their apps and websites—also suspended Gab’s account. After it became clear that maintaining a business relation with regular online financial services was no longer possible, Gab turned to alternative payment processing services based in crypto-tokens such as Coinbase—a money exchange—and Cash App—a payment processing platform, whose founders include Twitter’s CEO Jack Dorsey. When they also refused to service Gab, the platform partnered with Second Amendment Processing (SAP), a payment company that soon folded after it came under scrutiny for suspicious transactions. In other words, demonetization efforts pushed Gab to the edge where it had no choice but to deal with semi-criminal partners, resort to semi-legal services, or build its own infrastructure.

The third strategy to push Gab away from the mainstream ecosystem was to disconnect the platform from infrastructural services further down the stack, including domain registrars, cloud analytics, and storage services. As far as domain registrars are concerned, in 2017 Gab was initially hosted by AsiaRegistry, but a dispute over their demands forced Gab to move to a new domain host (Robertson, 2019). After GoDaddy, the largest ICANN-accredited registrar in the world, terminated their relationship with Gab, they finally partnered with Epik—an alternative domain registrar that has become famous as a refuge for extreme right voices; Epik purchased webhosting platform Sybil Systems in 2019 as well as several other related services. For its data storage and analytics, Gab has been ousted by all mainstream webhosting services, including Microsoft Azure (in 2018) and AWS (in 2019). Since then, Gab is no longer hosting its service in the cloud but has moved to renting hardware in an undisclosed data center, according to the Wall Street Journal (McMillan and Tilley, 2021). However, Epik is known to indirectly rely on AWS, which it uses to host many of their DNS (domain name system) servers; it is currently unclear whether Amazon still facilitates Epik in its cloud services.

Gab’s 4-year history divulges a sustained pattern of being denied access to mainstream distribution networks, online monetization abilities, and infrastructural services; as a result, Gab is consistently pushed aside by mainstream platforms and their dependent partners. Indeed, the same convolutions also apply to other platforms, including Parler and BitChute. Parler, founded in 2018 by John Matze and financed by Rebecca Mercer, positioned itself from the beginning as an “alt-tech” social network boasting 4 million users at its peak. The platform gained a major boost in 2019 and 2020, after Twitter flagged some of Donald Trump’s tweets. Its apex of popularity came in the week after the 2020 elections, when Trump began to spread the “stop the steal” mantra and many politicians migrated to the platform (Otala et al., 2021); the Parler app was downloaded 1 million times and rose to the top of both app stores. Two days after the Capitol Siege, Apple removed Parler from its App store, after having issued a warning that was dismissed. A number of smaller services followed Apple’s example, and on 9 January 2021, Amazon denied Parler access to its cloud service AWS, arguing that Parler’s denial to curb violent content provided a real risk to public safety. Parler responded by filing a lawsuit against Amazon for breach of contract and defamation. It took Parler 1 day to transfer to Epik as a domain registrar and to move their cloud services first to the Russian-owned company DDoS-Guard and subsequently to SkySilk Cloud Services—a web infrastructure company based in Los Angeles with a dubious history (Brodkin, 2021).

Another case exemplifying similar deplatformization patterns is BitChute, an alternative video-sharing platform founded in 2017 by Ray Vahey in the United Kingdom. Its growth was slightly bigger than Gab and the two platforms partly overlap in their user bases. For users banned from YouTube, BitChute typically provides an alternative channel. Due to BitChute’s permissiveness of extreme hateful and conspiracy-laden content, Twitter began blocking posts linked to the site, limiting their access to distribution channels. BitChute was banned by PayPal in 2018, causing the demonetization of its users who financed their videos by linking to fundraising websites such as SubscribeStar and cryptocurrency processors. In January 2019, BitChute moved their domain services to Epik after being banned by major infrastructural services.

We may conclude here that deplatformization indeed affects extreme platforms’ access to the ecosystem as such. It should be emphasized, though, that despite tech companies’ enforcement of various deplatformization strategies—blocking access to networked distribution, demonetization, and disabling infrastructural services—connections between mainstream services and radical, controversial platforms are not completely cut off. Four American researchers found that, despite YouTube’s efforts to deplatform extreme-right users who subsequently moved to BitChute, there are still substantial links between the two platforms: over 25% of URLs found in BitChute’s video descriptions point back to YouTube (Trujillo et al., 2020). In many cases, they are pushed back to, but not over the edge; they remain connected to the centralized ecosystem, so mainstream operators can still strategically profit from their traffic. “Fringed” platforms, for their part, are subsequently forced to recalibrate their position not just vis-a-vis the mainstream but also vis-a-vis other fringe players. The next logical step in this inquiry is to look at the impact of this push. How did Gab respond? Did it build a parallel, alternative network infrastructure at the edge of the mainstream ecosystem? And how did other fringe platforms respond to Gab’s efforts to mobilize other decentralized forces at the fringes of the ecosystem, both their ideological allies and their opponents?

Gab versus fringe: building a parallel ecosystem?

Several researchers have warned about the danger of alt-right and far-right platforms building a media ecosystem outside the mainstream, which is highly interlinked internally and stands far apart from other media in terms of technical and ideological connections (Benkler et al., 2018). The threat of such parallel ecosystem lies in its ability to isolate the far-right from the rest of the ideological spectrum, resulting in “asymmetric polarization” that “complicates the process of governing ideologically diverse polities” (Freelon et al., 2020: 3). In order to understand this potential peril, we look at Gab’s move away from the mainstream toward entrenching itself in a new online environment at the edge.

To build a viable alternative ecosystem, we assume that a platform like Gab needs concurring partners operating from the same technological and/or ideological premises. However, it is not clear whether fringe platforms regard each other as rivals competing for the same user base or as partners working toward a shared political goal. Gab certainly keeps looking for like-minded allies, but it also competes with Parler, Telegram, 4Chan, 8Chan, and other smaller platforms who are fishing in the same user pond (Zannettou et al., 2018). For instance, after the attack on the US Capitol, Gab boasted how it profited from the Big Tech companies’ assault on their alt-right competitors Parler, BitChute, and Telegram, claiming a 40% increase in traffic and “gaining 10,000 new users per hour” on 9 January (Stimson, 2021). At the same time, we witnessed how competing networks also resorted to the same infrastructural services, such as Epik, the domain host that explicitly welcomed extremist platforms, including 8chan, BitChute, and later Parler, in addition to hosting right-wing publishing platforms, for example, The Daily Stormer. Epik typically underscores Gab’s avowal of “free speech, individual liberty and the free flow of information online” (Gab, 2016), defending its position to host extremist content platforms from its principled stance and welcoming all views without bias or predilection. And yet, even though they resort to infrastructural services that support their ideological view, Gab can hardly claim the successful development of a “parallel alt-right ecosystem” (Roose, 2017).

What is interesting, though, is to trace how Gab strategically changed its motivation to create an alternative ecosystem from free speech toward a new narrative that champions the rhetoric of decentralization, openness of software, and user privacy and agency—a rhetoric that is commonly deployed by its ideological opponents (Squire, 2019). Besides partnering with Epik for its hosting services, Gab also turned to open-source tools to pursue its goal to become independent of the mainstream GAFAM-system, for instance, by forking the free and open source-based Brave web browser (Jimenez, 2019). Guardian reporter Jason Wilson investigated how far-right supporters, over the past years, increasingly moved to open-source platforms to evade censorship and create a truly decentralized online space. For instance, Luke Smith, an influencer with a history of promoting right-extremists, monetizes a channel on YouTube where he offers tutorials on how to use open-source software applications “encouraging viewers to detach themselves from Silicon Valley’s products” and “pushing users in the direction of decentralized, resilient social media platforms” (Wilson, 2021). The moment of this narrative shift is not a coincidence: extreme-right platforms resort to decentralized open-source systems at a time when the Big Tech companies are under mounting attack by conservative and progressive politicians alike for their unfettered control over online connectivity, but for very different reasons. Whereas conservatives argue that GAFAM-platforms have too much power to censor their right to free speech, progressives contend they are not doing enough to keep their channels clean from toxic content and to guard the legal and civil boundaries of online public communication.

In light of this narrative shift, it should not come as a surprise that Gab put up a sustained effort to forge itself onto Mastodon, a free and open-source social networking service similar to Twitter, whose design offers the possibility of running “nodes”; each node defines its own code of conduct, ToS, privacy options, and moderation policies—much like platforms. Mastodon is part the Fediverse (“federated universe”): a network of independent social media sites that wants people to freely connect across different decentralized, nonprofit, self-governed platforms (Kwet, 2020). Fediverse partners—which, besides Mastodon, also include PeerTube, Diaspora, and Friendica—rely on standardized shared protocols (e.g. ActivityPub) so users can interact with other users on different platforms. Since no one owns them, neither governments nor corporations nor even their own user groups can stop these platforms from developing their own peer-to-peer networks. The Fediverse partnership has so far been the most concerted effort to build an online counter-space that stands apart technically, economically, and ideologically from the GAFAM-nucleus.

For right-wing, free speech-incentivized platforms such as Gab, open-source and decentralization principles offer the possibility to wrap their own goals in the cloak of an ideologically opposite narrative. After Gab got kicked off hosting services and app stores in 2019, it tried to rebuild itself on Mastodon’s software to the great dismay of many Mastodon user groups who pride themselves in holding higher standards for filtering hate speech than MsSM. The “forking nodes” offered by this platform’s design favor a self-directed moderation approach; and since indeed no one owns, controls, or supervises all other nodes, the decentralized Fediverse philosophy was put to test. In response to the Gab invasion, which felt like a Trojan horse to its inhabitants, user group administrators started to prevent Gab content from spilling into other nodes by blocking all its posts. Evidently, Gab did not much care for this type of “internal deplatforming.” After all, they did not land on Mastodon for its Fediverse connections, who were not exactly its ideological allies anyway. Neither were they after Mastodon’s infrastructural services; as we explained in the previous section, Gab had already taken charge of its own domain registration and payment processing.

The one attractive Mastodon feature for Gab to become parasitical upon was its potential to indirectly offer access to mobile online traffic via mobile apps featured on Mastodon’s central home page (Robertson, 2019). Although blocking one specific node’s access to common features was against Mastodon’s (and the Fediverse’s) open principles, this is exactly what happened. Indeed, it took a concerted effort by Mastodon’s user communities to keep Gab at bay; the hostile takeover attempt forced Mastodon to imitate the deplatformization strategies of the very ecosystem it tried to escape. Interestingly, the struggle to design and build a self-governed, decentralized alternative to the centralized, economic powers of mainstream gatekeepers reveals the same intricacies as the larger struggle between GAFAM-platforms and alt-right platform communities. If anything, the reconstruction of Gab’s deplatformization and its subsequent efforts to stay afloat as a part of the ecosystem divulges the interdependence of mainstream and fringe platforms as a part of the same online dynamics.

Gab’s response to deplatformization teaches us an important insight about the platform ecosystem as such: it operates not as a constellation of separate sociotechnical universes, but as an online infrastructure where all platforms are inextricably intertwined and mutually dependent. Which brings us back to our main questions concerning deplatformization and its governance: who is responsible for governing the platform ecosystem—keeping it clean from toxic content—and who controls it as an infrastructure? And is deplatformization an implied (self-)governance strategy to detoxify the ecosystem of radical content or is it rather a strategy to control the hierarchical online infrastructure?

Deplatformization as implied governance of the ecosystem

Before addressing this layered question, let us go back to what happened after the Capitol siege on 6 January 2021, as explained in the “Introduction” in terms of deplatforming and deplatformization. The decision to deplatform former president Trump, taken by social media companies—including Twitter, Facebook, and YouTube—triggered disputes not just about the justification of his banning, but more profoundly about the extent of big tech corporations’ governance power over the ecosystem as such (Hitkul, Prahbu, Gujathakurta et al., 2021). As private companies, they are perfectly in their right to tidy their own walled gardens; each company creates and enforces its own ToS to decide what content is permitted on their platform and what is not, applying a form of self-governance in what is de facto a private online space. Some have argued that Facebook and Twitter, due to their size and gatekeeping abilities, carry a larger responsibility than smaller platforms because they form obligatory entry points to global public squares, a function that comes with societal duties (Lane, 2019). In fact, companies such as Facebook have responded to recent calls for corporate responsibility and transparency by introducing formal structures to review and repeal content moderation decisions, resulting in the Facebook Oversight Board (FOB) established in 2020. The FOB reviews moderation decisions, including the deplatforming of account holders. While some legal scholars have disputed the institutional design of this quasi-judicial body and its process of arbitration (Golia, 2020), others call it “an ambitious experiment . . . that should neither be hailed as a comprehensive fix nor dismissed as an inconsequential façade” (Douek, 2020, 10-11).

But while deplatforming could arguably be regarded as the responsibility of private companies operating specific social networks, deplatformization goes well beyond the jurisdiction of a single platform company, thereby exposing the growing accountability gap between individual platforms and the ecosystem as such. The decision to ban Parler from Apple’s app store and Google’s Play store, after the storming of the Capitol, resulted in “arbitration conversations”: both tech companies required Parler to adjust their content moderation policies and enforce the removal of hate speech according to their required ToS for app distribution. Apple announced that they were ready to reinstall Parler in April 2021, while Google is still undecided. Meanwhile, Amazon’s decision to pull Parler’s web services ended up in court, where both companies fight Parler’s right to define its own moderation policies vis-a-vis Amazon’s carrier obligation to provide basic cloud services to every platform. Like in the Gab case, Parler’s cut-off from basic online services is the result of a sustained effort by at least three Big Tech platform owners to police the “grounds” of the larger ecosystem.

As argued in the earlier sections, deplatformization dynamics inevitably show that the ecosystem should be considered as a hybrid space, containing both private and public property. This raises two questions: what kind of space is the online space between and beyond each tech company’s walled gardens? And how and by whom should it be governed—by corporations, states, or as a sort of commons?

Deplatformization in the form of denied access to infrastructural platforms, such as cloud service, app stores, web browsers, and so on, happens in a gray area where responsibility is unregulated. In this space, there are many different sets of rules (ToS) issued by different platform operators and relating to different types of services. The patchwork of all these ToS and their enforcement practices currently informs the implied governance of the ecosystem in a hierarchical fashion, so that the “companies that are deep in the stack may become the ultimate arbiters of what content gets to stay online” (Donovan et al., 2019: 62). As said before, the influence of tech companies lies neither in their sheer size or reach, nor just in their capability to decide who gets access to networking and monetizing services and who does not; most of all, it lies in their ability to control who stays connected to the core infrastructure and who is relegated to the periphery. Deplatformization efforts analyzed in this article divulge a strive toward general rule-setting power, an ambition that extends the dominion of individual platforms and applies to the online infrastructure as such. Rule-setting power (Castells, 2009) is increasingly inscribed in the ecosystem’s evolving architecture; the implicit governance rules are neither subject to universalistic democratic principles, nor subject to scrutiny by independent oversight institutions, whether national or supranational regulatory agencies (Srnicek, 2017).

The question of who should be tasked with cleansing the ecosystem to keep it uncontaminated from pollution by toxic speech, to protect its inclusiveness, and to prevent fragmentation and polarization, is currently hotly debated in political and academic circles. Platform operators who currently set the rules for deplatforming often take it as their corporate responsibility to protect their users as consumers; they also typically defend their wider deplatformization acts in the name of a larger public interest to protect “citizens” and the “public sphere.” The latter task has historically been assigned to states and governments, but the fact that most online platforms now operate globally has complicated the ecosystem’s governability (Schlesinger, 2020). Most poignantly, the original decentralized nature of the Internet’s technological architecture stands in sharp contrast to the current centralized governance of the hegemonic platform ecosystem (Bimber and Gil de Zuniga, 2020; Pasquale, 2017). The “infosphere,” as Luciano Floridi (2021) calls this space, should be governed neither by private companies nor by states, but as a commons structured by “transparent rules, legally grounded on all human rights and on human dignity, to avoid arbitrariness, unaccountability, abuse, and discrimination” (p. 4). The current situation, as we argued above, allots the power to deplatformize to the mainstream actors, and yet deplatformization dynamics extend beyond the mainstream to infiltrate the platform dynamics at the fringe of the ecosystem. Hence, platform governance needs to account for the larger infosphere in which all kinds of different deplatformization efforts take shape—between mainstream and fringe or between fringe platforms.

It is beyond the scope of this article to sketch new governance models for the platform ecosystem or its wider infosphere—a challenge that has been picked up by a number of colleagues (Fuchs, 2021; Zuckerman, 2020). Instead, we want to stress the potential of utilizing deplatforming and deplatformization as key concepts for assigning governance responsibility. The problem with the current ecosystem is that while the production of online content and its moderation happens decentralized via individual platforms, the distribution of content is increasingly centralized as a result of the corporate control over the ecosystem’s infrastructural design (Sekloca, 2019). Therefore, platform functionalities that support the mass distribution of content across the ecosystem—particularly via infrastructural services such as cloud storage, web browsers, or app stores—could be regulated differently from social networks policing their own content moderation. More importantly, tech firms operating both social network platforms and infrastructural services hold vertically integrative control over data flows; they have the power to define who stays at the center and who gets relegated to the fringe. For instance, while Amazon has pulled cloud services from Parler, it simultanesouly decides to keep hosting Epik, while also operating Twitch—a major social media platform that still allows extremist voices to post video content and monetize it (Browning, 2021). In other words, Amazon positions itself as an entrepreneurial facilitator, judge, and juror of toxic content all at the same time.

It is particularly this kind of corporate power concentration across the ecosystem that necessitates the regulatory intervention of national and supranational bodies in the platform ecosystem. Given the common architecture of the Internet, no platform operator can act in isolation; neither should a closed system of corporate platform operators be allowed to govern its collective infrastructural design without public oversight. Calls for stronger government intervention and public oversight have been rising both in the United States and in Europe (Ghosh and Couldry, 2020; Seyfert, 2021; Suzor, 2018). However, in democratic countries, governments and regulators cannot be the final arbiters of all public speech, just as corporations cannot be the sole gardeners of public space beyond their walled gardens. British communication scholars, Dutton and Dubois (2015), have suggested novel ways of “pluralistic accountability” to shape the governance of the Internet in such a way that civil society actors play a substantial role in its policing. Such concepts deserve to be researched more carefully. After all, governing the platform ecosystem is ultimately not a technical matter but a (geo-)political and economical concern.

Conclusion

In this article, we have argued that the concepts of deplatforming and deplatformization help articulate the need for acknowledging different levels of governance in the hierarchical ecosystem. In our analysis of Gab, we have shown how deplatformization strategies deployed by tech companies are not just manifestations of technological prowess but are just as much economic strategies propelled by clashing ideological narratives. Mainstream platforms that have the ability to cut off extreme platforms’ access to data flows and connectivity—the online equivalents of oxygen and electricity—claim to operate a neutral infrastructure in the public interest without accountability to that same public. By the same token, Gab’s response to the deplatformization strategies imposed on them by MsSM was not simply a technological response—to build a decentralized alternative system with like-minded allies—but also an attempt to infiltrate in a decentralized community of alt-tech platforms operating from an ideological opposite contention, forcing them to utilize similar deplatformization tactics as mainstream actors. Technical, economic, and political arguments are indivisibly intertwined in the conquest for hegemony in this new frontier of the digital ecosystem (Elishar-Malka, Ariel & Weimann, 2020) Indeed, deplatformization strategies are weaponized in the struggle to control common spaces—the spaces between private platforms that glue together the ecosystem as a whole—hence signaling the higher stakes in this geopolitical fight. Something we have not explored in this article is how deplatformization strategies may also be deployed by big tech companies between themselves as economic tools to fight the cut-throat competition to reign the digital infrastructure.

Finally, the realization that deplatformization strategies reach beyond the proprietary platform ecosystem into the wider “infosphere” or “global public sphere”—spaces that also encompass legacy media, information, and communication systems—is an important note to end with. Such broader concepts raise the awareness that sustained access to our global communication ecosystem requires the involvement of local, regional, national, and supranational communities. The responsibility over the hygiene of our common online public space—an infrastructure that is used by billions of people across the globe—is daunting, and therefore, it cannot be left to a handful of corporations or a handful of nations. Indeed, the practices of deplatformization require transnational and transcorporate regulation and oversight, including the possibilities for appeal, to protect the Internet as a common infrastructure and prevent these strategies to become tools in the hands of authoritarian governments or corporate elites. Just as the reduction of carbon dioxide emissions is a joint responsibility of nations, corporations, and civilians to save our planet from environmental destruction, the effort to maintain a clean online ecosystem, carried by rationality and mutual respect, has become an urgent global concern. Understanding how deplatformization works and what mechanisms it entails may hopefully add to a more sustainable concept of Internet governability.