Twitter and Turkey: Social Media Surveillance at the Intersection of Corporate Ethics and International Policy

ABSTRACT

The Arab Spring was a period of optimism when researchers saw social media platforms as a catalyst for advancing democracy worldwide. But Twitter and its peers are corporations that prioritize monetary and political gains. Through digital-ethnographic and archival research on Turkey, we found that the technology social media companies promised would enhance democracy may hinder it by providing authoritarian regimes with additional surveillance/repressive tools. We contribute to social media and surveillance studies by analyzing these tools through the lens of policy and neoliberalism. We further challenge monolithic understandings of neoliberalism and narratives on the positive impact of social media.

In the 2010s, both Google (Project Dragonfly) and Twitter (Country Withheld Content [CWC] Rules) demonstrated their willingness to comply with undemocratic regimes, despite controversy: these decisions have disturbing implications about the role of undemocratic policy among tech corporations and social media ecosystems. Social media had acted as a catalyst to express human rights in the early 2000s, with Twitter and Facebook at the forefront of movements such as the Arab Spring and various Occupy movements. However, the special rapporteur of the UN Human Rights Council expressed concerns at that time that social media would gradually become blocked, filtered, bottle-necked, and therefore, censored¹; their fears were eventually realized in government backlashes against social platforms. Turkey under the Justice and Development Party (JDP) is a standout example of the consequences of the criminalization of Internet-based speech, as well as a media autocracy of corporate and state powers.

We examined how the JDP's increasingly authoritarian tendencies led them to eventually control Twitter during the political uprisings between 2013 and 2020, and the role of Twitter's own neoliberal tendencies and policy changes. In this vein, we conducted digital ethnography and ethnography in and of archives²: we traced the use and perception of Twitter in Turkey as well as Twitter's own shifting practices and rules from 2013 to 2020, using official Twitter documents and reports, news coverage, and semistructured interviews with dissidents who used the Twitter platform. Based on Twitter's yearly transparency reports, we created a chart that shows censorship of Twitter content in Turkey from 2013 to 2018 and key developments in politics and the economy during that period. Furthermore, we surveyed the history of media censorship in Turkey and its relationship to recent political shifts in Turkish political atmosphere.

Through our analysis, we found neoliberal tendencies in what Twitter chose to report, and not report. More specifically, Twitter is shifting focus from the rights of individual users to those of business and political interests. Our analysis revealed intricate corporate–state relationships between Twitter and Turkey; as this relationship strengthened, so did censorship. More specifically, we found that Twitter's own policies opened the door to repression of dissidents, and the Turkish government then walked through. Consequently, after Twitter implemented nation-specific content moderation between 2013 and 2014, Turkey accounted for an overwhelming majority of censorship requests to Twitter between 2015 and 2017. This period also marked a period of major political conflicts including elections and a coup attempt. In the face of global criticism over online crackdowns and resistance from Twitter, the JDP no longer relies as strongly on legal requests to Twitter. However, it has employed other mechanisms such as hiring trolls, throttling, and arresting critics due to their Twitter posts.

In this study, we contribute to conversations on content moderation and media surveillance by studying the changing face of state-corporate relations and focusing on the relationship between Twitter and Turkish politics. We look at sociopolitical and socioeconomic changes, both locally and globally. This study sheds light on the intertwined relationship between social media and local and global politics through the framework of neoliberalism as an ideology.³ We contribute to a history of contradictions between social media policy and practice as well as the decline from open platforms to the more censored spaces of today.⁴ We also contribute to the history of the relationship between those policies and international law, and how these factors contributed to Turkey's current social media space.⁵ This study, then, challenges both (1) simplistic narratives about the positive role social media plays in contributing to liberal and democratic ideals and (2) monolithic understandings of the concept of neoliberalism.

It is crucial to define “neoliberalism” and articulate why we prefer it over “late capitalism” and “globalization” as our framework in this study. While “late capitalism” is used to refer to the transformations associated with capitalism, “neoliberalism” refers to a “thought collective” as a result of ideological and philosophical movements that aimed to oppose state-centered planning and socialism and differs from classical liberalism. However, neoliberalism “seems to mean many different things depending on one's vantage point.”⁶ For instance, Chakravartty and Schiller approach the concept from an economic standpoint to study the emergence of digital capitalism.⁷ They argued that radical changes in ownership and control due to capitalist agenda led to neoliberal newspeak and digital capitalism. Andrejevic, meanwhile, takes a broader perspective and sees the influence of neoliberal practices in new operations of power, surveillance, and censorship in a world of algorithms.⁸ In this study, we primarily draw from Ganti and Treanor's definitions: neoliberalism is both “a prescriptive development model that defines very different political roles for labor, capital, and the state compared with prior models, with tremendous economic, social, and political implications”⁹ and an ideology that prioritizes market exchange as “an ethic in itself, capable of acting as a guide to all human action and substituting for all previously held ethical beliefs.”¹⁰ Building on Ganti and Treanor, we see neoliberalism as an ideology that permeates not only social structure and institutions, like politics, media, and economy, but also everyday social interactions. Treating neoliberalism as an ideology in this study provides us the broader framework that allow us to see how it shapes the way governments, social media platforms, corporations, and individuals as users of social media platforms and political agents construct their ethical worlds and subjectivities as neoliberal agents. Furthermore, this approach allows us to move beyond narrow definitions of neoliberalism, which tend to see it as a fixed set of attributes and qualities that give rise to homogenous outcomes in each country and corporation where its logic is implemented.

Twitter: From Optimism to Skepticism

During the era of social media optimism, researchers saw these platforms as decentralized tools contributing to the implementation of democratic principles and liberal reforms. They highlighted the vital role these platforms played in social movements and political activism: marginalized groups deployed them in circumventing bans, evading surveillance, and disseminating information, all of which led to the formation of networks of counterpublic spheres.¹¹ In this regard, studies on Arab uprisings showed that protestors mobilized better on the ground as they turned to social media platforms to connect with others, disseminate information, and access accurate knowledge. In their study of the role of Facebook in the Chilean protests of 2011, Valenzuela and colleagues demonstrated that social networking sites “emerge as resources that may create the kinds of collective experiences that are necessary conditions for successful protest movements.”¹² Similarly, Juris, studying the relation between the use of social media and public space within the #Occupy Everywhere movements, illustrated that the turn to social media helped protestors better organize and bring diverse groups from diverse backgrounds into actual physical spaces, which might “help to ensure the sustainability of the #Occupy movements in a post-eviction phase.”¹³ Additionally, these scholars argued that success was due to the lack of governmental control over these platforms at the time.

However, scholars working on social media platforms increasingly show how social media companies emerged as global corporations that prioritize monetary gains. At the same time, they show how states increasingly come up with new ways to control and censor social media use. For instance, Yesil analyzed how the combined state and private elite used public media initiatives to control the narrative and frame themselves in a positive light.¹⁴ Bulut looked at legal structures governing social media inside Turkey, and how Turkey began to co-opt Twitter, which had once identified itself as pro-dissident, by employing threats and incentivizing the social media giant with access.¹⁵

Current literature on media autocracy, censorship, and surveillance focuses on the contradictions of social media companies advertising to investors and users that they are places to freely share content while downplaying how that content is moderated and why. Gerrard, for example, analyzed content moderation as a signal to investors that they are doing “something” to deal with their many controversies, rather than an effective or fair means of governance.¹⁶ Generally, social media companies try to apply, encode, and review moderation consistently across borders. Content moderation, however, happens within existing sociohistorical paradigms that influence how these policies perform,¹⁷ which we will address in this study.

Media Censorship in Turkey: From Traditional Media to Twitter

Political and media elites in Turkey had always controlled and censored media in some form to maintain the status quo and their economic interests.¹⁸ However, since 2002, new political and economic developments in Turkey under the JDP's uninterrupted tenure turned media control and censorship toward “a model of neoliberal media autocracy.”¹⁹ This process was primarily intensified after the reelection of the JDP in 2007; it gained new momentum during and after the Gezi Park protests and corruption scandals of 2013. More recently, the coup attempt of 2016 has brought this media autocracy to a new level. In addition, in the aftermath of the failed coup attempt, we began to see the state trying to establish a new form of relationship with its citizens. Bulut and Can demonstrated this point by analyzing the televised disclosures of former Gulenists. They argue that citizenship bonds are becoming “affectively redefined along the lines of disclosures and apologies rather than political and legal accountability” in a polarized political atmosphere constituted through fear and anxiety.²⁰ Yesil argues that it is important to see the JDP's media capture at recent critical events as a continuation of their earlier policies and practices instead of seeing it solely as an authoritarian turn.²¹ Nevertheless, it is not only the new economic and political transformations in Turkey that contributed to the creation of a media autocracy in Turkey but also the neoliberal trends of global media and politics. Consequently, we analyze the parallels and overlaps between the political and economic interests and transformations of Turkey and Twitter in the last decade.

Turkey has had a history of media censorship since its foundation, although the methods have shifted throughout its history. Researchers in the communication and information policy sphere have long noted the Turkish government's unique form of media control and censorship. However, this media control and censorship entered a new era when Turkey moved to become a neoliberal state. Turkey's neoliberal revolution began with Turgut Ozal's economic reforms in the 1980s. It shared with other countries worldwide a promise of a reduction in politics and the advancement of a market-based and consumption-driven ideal.²² In Turkey, the neoliberal reforms failed on both fronts: A few big corporations bought up any available smaller media companies, and the government targeted the remaining media outlets when they presented themselves as enemies. By the 1990s, the majority of Turkish media outlets were owned by big businesses, leading to the creation of media empires. From the Uzan-run private television company to Dinc Bilgin's Daily Sabah, the Turkish government replaced the leadership of these remaining conglomerates with government loyalists, making Turkey's media landscape a unique combination of government and private interests. These new media conglomerates have suppressed journalists' critical stances through editorial censorship mechanisms.²³

In the aftermath of their election victory in 2002, the JDP started to form its own government-friendly media with the support of like-minded businessmen (e.g., via favoring government bids to them). It also began to apply systematic pressures on existing mainstream media through judicial suppression, tax fines, accreditation discrimination, cutting advertisement options, and so on. Such pressure started to bear fruit, as seen during the 2013 Gezi Park protests, when some widely watched TV channels such as CNN aired documentaries on penguins while the country was shaken by protests against the government's environmental policies. This trend continued further after the corruption scandal of December 2013. JDP continued pressing the remaining critical media outlets and now controls about 90% of media outlets.²⁴

Due to this consistent decay in media freedom, critics started to turn primarily to Twitter. Ozduzen and McGarry engage closely with Twitter for political practice in Turkey.²⁵ They found visual and textual expression of social movements critical to the collective memory of the Gezi Park protestors and “rematerializing” the Gezi protests as a leaderless, empathic, and collective movement. Pro-government users and trolls however, attempted to destroy the narrative of the protest by characterizing the mobilization as a centralized riot full of backward “baby killers” and gun grabbers. Furthermore, pro-government users characterized protestors as standing in the way of Turkey's bright future, and this messaging had potential to stifle the Gezi Park protest as a rallying cry for those seeking progress and agency.

Government opponents continued to utilize these platforms during the corruption scandals of 2013, and the Twitter and YouTube ban in 2014. Until 2014, the Turkish government saw Twitter as an unreliable foreign tool that should be avoided. They initially minimized its impact by throttling Internet traffic, threatening to shut it down, and eventually, arresting critics. However, Turkey began to take a more active role in regulating Twitter later in 2014. Specifically, right after the leak of corruption tapes about President Erdoğan's inner circle, they implemented a new Internet law (5651) that gave them the right to block the URLs of websites that did not comply with the state's Telecommunication Information Directorate (TIB), setting off a new stage of state-corporate relations. As we detail below, these efforts eventually became more sophisticated, to the point where the JDP decided to hire an army of Twitter trolls to create its own campaigns and influence public opinion. Trolling in Turkey is decentralized, and done opportunistically by the pro-government public. These troll groups believe they benefit JDP leadership by providing quick and low-cost surveillance and punishment of antigovernment Twitter users. These tactics do not appear to be directed primarily by JDP; however, Saka found many indirect connections to party officials, as well as quotes from high-level JDP officials wishing to weaponize the troll groups.²⁶

Methodology: Understanding Content Moderation and State-Corporate Relations

In this study, we employ mixed methods, primarily from anthropology, in collecting and analyzing the data to understand the intertwined relationship between the Turkish government and Twitter. Initially, we began conducting digital ethnography to participate and observe Twitter in Turkey in 2013. However, our online participant observation in Twitter by 2014 led us to move away from ethnography as “participant observation” toward ethnographic modes of reading and hence to ethnography of/in archives, seeking to understand the past and present of Twitter and Turkish media and relate them to sociopolitical and economic changes over the last 20 years (vis-a-vis social, political, and economic changes in Turkey).²⁷ In our ethnography of archives, we looked into the implementation of deregulatory reforms in the 1990s, and the reunification of mainstream media under the JDP in the early 2000s. Next, we collected policy documents, news coverage, and journal articles to understand Twitter's history from its founding, to its brief status as a free speech platform during the Arab Spring, to the start of its regulatory reforms immediately after.

Furthermore, we conducted discourse analysis of the shifting narratives and ideologies of Twitter and JDP leadership during major political events in Turkey from the 2013 Gezi protest until the 2020 coronavirus outbreak by drawing on legislative and policy documents, media coverage, transparency reports of Twitter, and social media field notes. However, as Zeitlyn argued, we read archives “against the grain.” That is, we subvert the mode of understanding these documents from those who produced them toward those whose lives have been impacted by them.²⁸

Moreover, we not only participated in online platforms but also among recently displaced communities, primarily in the United States. This field work is part of a connected project that aims to understand recently displaced individuals in the context of Turkey's increasingly authoritarian regime. For the purposes of this article, we chose a diverse group of past and current influential Twitter users (mainly Turkish and Kurdish journalists and activists who had recently been displaced and sought political asylum in Europe or in the United States) from a larger pool. We identified 20 influential Twitter users by looking at their number of followers and frequent Twitter use. Then we conducted semistructured and open-ended interviews with this diverse group (by gender, ethnicity, occupation, education level, year of displacement). About half of the interviews were conducted face-to-face and the other half online. We asked them “What is Twitter?,” “How can you define Twitter?,” and “Have you seen a change in the way you see and use Twitter?.” In open-ended interviews, we also simply discussed what they had been through recently and whether they had seen a change in the way they define themselves. We bolstered these interviews with existing interviews on Twitter use from online reports from organizations like Human Rights Watch.²⁹

The Emergence of Twitter as a Platform and as a Corporation

Social media platforms reflect their designers' ideologies and the cultural milieux where they were created; however, this does not determine how they are deployed as they move from their birthplace to other geographical locales.³⁰ Thus, it is crucial to look at the ways they interact with local cultures, ideologies, and governments. Users in new locations are not the only ones who shape new social media tools through interaction; social media companies also shift their own identities and policies as they interact with local governments. In addition, states change their policies and laws to deal with these emerging tools and their affordances. To better explore these intertwined relations, it is crucial to first examine the evolution of Twitter as both a platform and a corporation. We do so by analyzing Twitter's rule and policy alterations— and hence identity shifts—in the name of legality.

However, it is also important to discuss what Twitter gains by defining itself as a “platform.” This is a powerful discursive strategy that Gillespie has theorized as allowing companies to advertise themselves to different stakeholders while hiding tensions between them.³¹ Depending on which stakeholder the company is engaging with, the metaphor shifts between raised platforms that empower users to speak, platforms from which advertisers can sell, and technology platforms on which clients can build. As Gillespie argued, it is important to look at the terms the social media companies employ and hope to be judged by and the roles they aim to play. To examine the roles that Twitter aims to play, we analyzed not only the terms Twitter employs to construct its identity but also its actions and practices.

Twitter began as an open platform where users—including stakeholders from government, academia, private industry, and media—could interact with each other without gatekeepers (Excerpt 1). Twitter did not have any official guidelines from its founding in 2006 until 2009.

However, at the beginning of 2009, they created a simple ruleset of 568 words with boundaries against impersonation, violence, copyright infringement, unlawful use, serial accounts, malware/phishing, spam, and porn. Even from the beginning, Twitter did not specify social/issues (which received only 22 words) but instead prioritized more technical concerns such as spam (130 words). When Twitter explained that “You may not use our service for any unlawful purposes or for promotion of illegal activities,” they did not elaborate on illegality, or how that term would interact with geography. This problem would highlight itself post-Arab-Spring.³²

Twitter used the Communications Decency Act (CDA) to defend its actions before, during, and after the Arab Spring (Excerpt 2). Senator James Exon designed the CDA in 1995 as a response to guardians who feared that online pornography was too easily accessible to children. The CDA made it illegal to knowingly expose minors to pornographic material. However, some intermediaries, such as Internet service providers, could not reasonably be aware of everything on their platform. As a response, Congress added Section 230. Most critically, CDA 230 provided legal protections for website providers in a kind of safe harbor clause outlined in section C. Congress called this section “protection for Good Samaritan blocking and screening of offensive material”; social media companies use this section as a defense against legal harm to this day. Section C further specified: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”³³ This allowed websites to treat themselves as platforms for speech, not publishers of speech, as Gillespie pointed out.³⁴ Sans legal liability, Congress allowed online companies to create their own ethical arrangements, and to restrict constitutionally protected speech. The CDA further solidified a multistakeholder Internet, that is, an Internet-governance model of control over the web by the government, the private sector, and civil society. Flew, Martin, and Suzor comment that the provisions in CDA 230 have been invaluable to the growth of social media, but they have also been the source of the greatest platform controversy. Social media's massive share of the media landscape exposes the companies to constant public shocks. During every public shock, both the public and the more regulated traditional media outlets question why social media companies have special privileges and different standards. Furthermore, CDA 230 creates international tensions because US freedom of speech laws do not apply in other countries. This means that companies must juggle different requirements without splintering their US-based platforms into multiple websites.³⁵

The Tweets Must Flow: Twitter during the Arab Spring

From 2010 to 2011, Twitter put fear into traditional institutions of power. Social media shocked the world as a new peer-to-peer media environment for transmitting news faster, and more unpredictably, than in the past. Instead of news slowly filtering from a large media group in one country to a large media conglomerate in another, users in Egypt, Syria, and other nations talked directly about their experiences with users in the United States, Tunisia, and elsewhere. Twitter offered a previously unattainable and uncensored view.³⁶

Protestors used social media as megaphones to the outside world with hashtags in different languages.³⁷ In addition to newsrooms' calm discussions of evidence of police abuse and war crimes, one could now see the person on the receiving end of the violence projecting themselves by cell phone camera, screaming in terror as the events unfolded. This more uncensored and shocking coverage contributed to massive international pressure from governments around the world. Potentially most frightening to political powers was that they could do nothing to stop the flow of information other than to shut the Internet down: in 2010, social media platforms were still based entirely out of the United States and protected by CDA 230.

Twitter reacted positively but also cautiously to the Arab Spring. At the end of January 2011, Biz Stone, the cofounder of Twitter, wrote: “The Tweets Must Flow” (see Excerpt 3). Here, they reaffirmed their belief in freedom of expression on behalf of Twitter, and in a clear reference to the Arab Spring, they stated that “Some Tweets may facilitate positive change in a repressed country.”³⁸

Twitter maintained that they remove only illegal tweets and spam, and that they try not to remove tweets based on even controversial content. However, Twitter hinted that there may be limits to that expression and concluded that they will try to notify users if a court demands that they turn over information.³⁹ The language of appeasement stands in stark contrast to Twitter's more liberal standpoint during the Arab Spring. It starts to become clear here that Twitter is not simply a free speech platform, but a business that follows rules and regulations with hopes of maintaining profit.

Identity Crisis: Twitter's Business Model Opens the Door to Government Manipulation

Twitter's cooperation with political elites begins with its business model. Twitter makes 86% of its more than $3B revenue through advertisements.⁴⁰ Twitter must get as many eyes on as much content (and, most importantly, ads) as possible by allowing users to seek out content that interests them, and by encouraging them to share their content with other interested parties with no subscription cost. Similar to YouTube's business practices,⁴¹ Twitter also offers advertisers access to that same platform and to the platform's users for a cost. When Twitter filed an amendment to its S-1 with the US Securities and Exchange Commission (SEC) in preparation for an initial public offering, it explained its advertisement business model.⁴² By 2017, Twitter earned over 77% of its revenue through advertising services.⁴³ Although users are critical, their business model does not rely on input from individual users per se as much as revenue from local advertisers paying for access to those users. We cannot say that advertising initiatives cause more users to come to the platform; however, Twitter provided a chart to the SEC that shows correlations between targeted advertisements and monthly active users on their platform.

Twitter's business model put the company into an early awkward position in the United Kingdom. In the context of international law, one of Twitter's founders, Alex Macgillivray, said in March 2011, “We tell them, we're a U.S. company, we have C.D.A. 230 here, and you're welcome to come and try your hand at suing us here.”⁴⁴ This began to change as the company expanded its market, and came under increasing international pressure to play by local rules. In 2011, Twitter attempted to open an office in the United Kingdom, which would benefit the company with servers in Europe and increased revenue.⁴⁵ In May, however, Twitter fought through controversy when thousands of users repeatedly exposed a previously anonymous footballer involved in a UK court case.⁴⁶ The Culture Secretary at the time, Jeremy Hunt, accused Twitter of “making an ass of the law.” The UK government then took Twitter to court to compel the company to identify the users spreading the individual's name; in response, Twitter announced that it would notify users before handing over any identifying information.⁴⁷

In the end, the Liberal Democrat MP John Hemming broke the court order, stating that it would not be practical to imprison the 75,000 Twitter users who had named the player.⁴⁸ However, this case set the precedent that Twitter could and should follow local legal norms.

The Tweets Still Must Flow?: Twitter Adds Geography-Based Stipulations to Freedom of Expression

Gillespie notes that social media companies have become deeply invested in presenting themselves in a positive light to critical governments' stakeholders, and in lobbying for a friendly business environment.⁴⁹ We can observe such an entanglement in The Tweets Still Must Flow, a statement posted by Twitter itself. When asked about legal issues, Tony Wang (Twitter's UK general manager) told observers at the March 2012 Guardian Changing Media Summit that Twitter is “generally neutral toward content created by its users, and that Twitter has rules in terms of what is and is not allowed.” Mr. Wang continued, “Generally, we remain neutral as to the content because our general council [sic] and CEO like to say that we are the free speech wing of the free-speech party.”⁵⁰ In contrast to Mr. Wang's statement, Twitter posted “Tweets Still Must Flow.”⁵¹

Twitter stated that the only way to abide by different national norms had been to remove content globally; they now compromised with a policy to withhold content in specific countries (see Excerpt 4). This shows how Twitter worries more about the market space, via a neoliberal logic that prioritizes states as entities over individuals and hence human rights.

Although tweets did flow freely during demonstrations such as the Gezi Park protests in 2013 in Turkey, Twitter's relationship with the government changed in 2014 with often contradictory yet mutually beneficial policies. Through these changes, Twitter made their most impactful steps toward a neoliberal arrangement of corporate–government relationships. By hiding behind the term “platform,”⁵² they made their move to exist as a corporation in certain countries, rather than a platform for the public. The marginalization and/or full elimination of opposing views in Turkish Twitter led to increasing dominance of neoliberal newspeak in Turkey, as foreseen by Chakravartty and Schiller, who voiced similar concerns for digital capitalism.⁵³

A New Relationship with Twitter: The Intersection of Economics, Law, and Security

Twitter pays taxes where its servers are located, which did not include Turkey. Amid leaks of documents embarrassing to the JDP, President Erdoğan took the opportunity to accuse Twitter of tax evasion and justified a temporary shutdown of Turkish access to the site.⁵⁴ He had previously referred to Twitter as a menace to society; just before shutting it down in 2014, he declared, “The international community can say this, can say that. I don't care at all. Everyone will see how powerful the Republic of Turkey is.” Moreover, he began demanding that Twitter set up a local office and follow Turkish law (Excerpt 5).⁵⁵

After the press conference, government-linked Turkish Internet providers changed Twitter's IP address to redirect to a government-run website⁵⁷; Turkish users were then treated to a statement from Turkey's telecommunications regulator describing four court orders as the reason why the government blocked the site and identifying the move as a protective measure. For example, BTK, a Turkish Telecom regulation and inspection body, said, “Because there was no other choice, access to Twitter was blocked in line with court decisions to avoid future victimization of citizens.”⁵⁸ By framing its action at the intersection of economics and security, the JDP reused narratives from international agreements drawn up at the World Summit on the Information Society (WSIS) and World Conference on International Telecommunications (WCIT), which established that governments have sovereign rights over Internet users located within their borders.

The JDP reaction publicly backfired, but the controversy also resulted in a more cooperative relationship with Twitter. Despite JDP declarations, Twitter joined with Turkish legal experts to file a lawsuit, successfully arguing that shutting down Twitter was against the Turkish constitution. Additionally, in a rare act of post-2011 defiance, Twitter not only publicly denounced the government's actions, but announced that users could still post to Twitter via text messaging despite the ban (see Excerpt 6).⁵⁹

In April 2014, Colin Crowell, Twitter's head of Global Public Policy, met with Turkish officials. As a response to the Turkish government's demands, Crowell never promised to open an office. He stated: “We didn't agree to open an office; our decisions to open offices around the world are based upon whether the underlying economic climate justifies it.”⁶⁰ Twitter appointed a country coordinator in 2015 and stated that they would be more understanding and would react faster to court decisions (Excerpt 7).

To address the tax loophole, the government implemented Presidential Decision No. 476, which withheld taxes on online advertisements in accordance with the already established Article 11 Paragraph 7 of the tax procedural law, which set a 15% tax rate. We do not know how effective the laws are in getting taxes out of Twitter, but closing the previous tax loophole creates a relationship where money will flow both ways, drawing the company closer to the government's interest.

However, more recently Twitter agreed to establish a legal entity in Turkey and announced its decision on March 19, 2021 (Excerpt 8). They used the phrase “legal entity” instead of “office,” even though they used the word “office” when they rejected Turkey's request in 2014. Moreover, they stated that Twitter would remain available for all who use it in Turkey. Establishment of a legal entity in Turkey would ensure that Twitter comply with the government's requests faster. However, we will see its ramifications more clearly in the coming years.

Governance: WSIS and WCIT

The events described earlier were not simple interactions between Twitter and Turkey; they stem from a much longer history of Internet governance that began with CDA 230. What played out were struggles between the older multistakeholder model (bringing together government, the public, business, and academia) and the government-driven multilateral model. In addition to CDA 230, there were two other major agreements affecting Internet governance: WSIS and WCIT, mentioned briefly earlier (Excerpt 9).

The largely US-controlled Internet of the 1990s did not remain that way, as international government bodies began to take notice of the importance of the Internet. In 2005 the UN endorsed a WSIS, where hundreds of stakeholders gathered in Tunisia for the Tunis Agenda for the Information Society (Excerpt 10). Here, the group created a plan focused on “financial mechanisms for bridging the digital divide, and on problems relating to Internet governance.” The 2005 WSIS recognized the importance of the financial sector in developing information and communications technology (ICT), but it also called for governments to become more involved in regulation and mitigation. Additionally, they recognized that “central responsibility for coordination of public financing programs and public ICT development initiatives rests with governments.”⁶³

Section 35-a declared that governments had sovereign rights over Internet-based policy issues. By turning the Internet into a sovereignty issue, the WSIS aimed to create digital borders in what was still a unified net. Most importantly, the 2005 WSIS created the Internet Governance Forum (IGF), an international body that would break the unofficial regulatory monopoly of the United States in the Internet governance space. The IGF could continue the WSIS mission of discussing Internet governance to “foster the sustainability, robustness, security, stability, and development of the internet.”⁶⁵

Seven years later, in the wake of the Arab Spring, world leaders gathered in Dubai for the first WCIT, where they strived to place the role of Internet governance more firmly on the shoulders of governments.⁶⁶ WCIT 2012 was the first negotiation of international telecommunications regulations since 1988. Nestled in the preamble, between innocuous commitments to improve access to the Internet for peoples with disabilities and improving ICT network efficiency, was Resolution 3 (“To foster an enabling environment for the greater growth of the internet”) (Excerpt 11).

For Turkey

In signing the International Telecommunication Regulations (ITR) of the WCIT (Dubai, 2012), the delegation of the Republic of Turkey:

In the declarations and reservations section of the final acts (Excerpt 12), the Turkish government confirmed its pro-multilateral stance. Moreover, the government declared its right to approach Internet governance in terms of its best interest and it reserved further rights to change its agreement to the ITR as it feels necessary; this cemented the JDP occupation of Twitter.

WCIT Goes into Effect

In 2015 this agreement went into effect; a spike in court-ordered requests to Twitter followed. Most critically, Turkey accounted for about 90% of all court-ordered takedown requests sent to Twitter. In response to the JDP's abuse of CWC rules to attack journalists, Twitter advocated for verified journalists and news outlets in courts, and filed legal objections for targeted Turkish journalists when “possible under Turkish law.”⁶⁷ None of the objections prevailed; this should come as no surprise, as “unlawful use” was still defined as violating the user's local laws.

In 2016, the JDP made critical changes to Turkey's state of emergency laws. As Zingg points out, by adding language to include antiterrorism, these efforts supplanted Turkey's existing freedom of speech laws.⁶⁸ “Terrorists” are not treated as kindly as “citizens” and security takes priority. States of emergency always precede large online crackdowns, and the term “security” was used in justifications for later Internet shutdowns in the Kurdish regions of the country. Additionally, Twitter continued to add rules, with clarifications on multiple account abuse⁶⁹ and private information.

The multiple account abuse rule ensures the criminalization of anyone wrongfully taken off the platform who attempts to continue to communicate, not only the trolls that the policy is aimed at. Additionally, Twitter specified certain information as private and confidential such as credit card numbers, street addresses, intimate photos, and national identification numbers⁷⁰; this appears harmless, but as we have shown with many of Twitter's previous policies (e.g., the soccer scandal), implementation of these policies varies across geographical and cultural contexts.

Twitter Now: Strategies in Content Moderation

The current definition of “unlawful use” sounds more like the original 2009 version, wherein Twitter demands that the user does not use Twitter for illegal activity, but they still do not specify type or geography here. However, we note that Twitter demands that users comply with “all applicable laws,” and that CWC rules still exist. In terms of private information, the 2016 rules still apply, but now they add that what constitutes private information depends on local laws. Currently, Turkish users experience six forms of content moderation and censorship that come mainly from the government, but increasingly from Twitter as well: CWC, terms of service (TOS), throttling, trolling, arrest, and shutdowns. All of these strategies help the Turkish government create their own Twitter, different from its use in other locations. More specifically, they not only help marginalize different voices in Twitter through creating fear and polarization but also help block and eliminate them totally.

CWC: Content That Is Available Everywhere, Except Where It Counts

In 2014, Twitter used its CWC rule for the first time in Turkey.⁷¹ CWC is Twitter's attempt to abide by “applicable local laws,” and CWCs can result in specific tweets or entire accounts being withheld. Content is typically only withheld in the jurisdiction of the reporting country; for example, if Twitter is responding to an account in Turkey, they will make that content unavailable in the country, but international users can still see it. Twitter then notifies users of withheld content with a grayed-out tweet or account page. Twitter is semitransparent by providing some detail on why they withheld the content: either in response to a legal demand such as a court order or based on local law.

Analyzing Twitter's transparency reports covering governments' legal requests for online censoring or content removal, we found that Turkey took up the bulk of total censorship on Twitter from 2014 until 2018 and accounted for the majority of reported government censorship worldwide (up to 90% of withheld accounts) during that time. Censorship and removals correlated with major political events (see Figure 1) and Twitter's compliance rate (accounts and tweets withheld) reached its peak when the company employed a rhetoric of cooperation with the JDP. Twitter's compliance was highest after the first shutdown in 2014 but appeared to decline as the JDP more regularly made moves that could affect Twitter's ability to stay present online (such as extended throttling and shutdowns). Another possibility is that the JDP did not need to work through Twitter as much with throttling, shutdowns, and trolling as options.

Additionally, Twitter sends details of CWC requests to Lumen, an organization that collects and displays legal requests for the removal of online content. Lumen receives PDFs of each request and further details about the sender. However, even this level of transparency is limited by local law. In interviews, several Turkish Twitter users voiced how they have been affected by this type of censorship. For instance, when we asked Turkan, a journalist, if Twitter is an open platform for all voices, marginalized or powerful, she suggested that:

Terms of Service

TOS removals are a method by which content is removed according to Twitter's corporate rules. Twitter added TOS requests to their transparency reports in 2016, and within two years, accounts removed for violating Twitter's TOS exceeded accounts removed due to legal requests by 1096 (see Figure 2). Turkey accounted for the majority of TOS requests, court orders, and other legal demands after the coup attempt that marked a critical turn in Turkish history.

As we have highlighted throughout, the Twitter rules changed over Twitter's history. With vague definitions of violence and terrorism, governments may report content to Twitter accordingly. As Twitter takes up a larger role as censor, TOS requests may have the benefit of being less politically damaging to both Twitter and the host country, as the TOS can have the appearance of fair, universal, and corporate standards.

Throttling and Shutdowns

Bandwidth throttling is intentional traffic shaping by Internet service providers (ISPs) to control the speed of Internet service. Typically, there is only so much bandwidth on a network, and ISPs must use some throttling so that there is enough for everyone.⁷² In Turkey, however, ISPs are beholden to the BTK (the regulating body), which is beholden to the government. Censorship by throttling has the benefit of maintaining the operability of businesses and infrastructure while slowing the virality of hashtags and news.

Turkey Blocks found evidence of throttling of social media and other services during politically sensitive times. For example, access to Twitter was throttled in October 2015 during the Ankara bombings of a pro-Kurdish peace rally that killed 90 people; some users could still access Twitter, but in a limited fashion (see Excerpt 14).⁷³ Moreover, the JDP used throttling during the coup attempt in July 2016, during which Twitter officially commented on the issue and users reported problems with posting (Excerpts 13 and 15, @policy, 2016).

Unlike throttling, which is targeted and allows limited communication, a shutdown (such as the one in 2014) takes an entire area offline and prevents all communication. Shutdowns are much rarer, and they cost money in lost business; however, Turkey Blocks has noted a few cases in addition to the 2014 ban:

A timeline of Internet shutdowns in Turkey during the 2016 period. The JDP enforced all shutdowns in Kurdish minority regions of the country. Designed by the authors.

Populist Trolls: Government-Led Polarization on Twitter

Besides all other means, the JDP also began employing trolling (focused harassment meant to discourage a particular group or person) to energize its supporters, strike fear into dissidents, and harass activists and journalists. Trolling happens within the context of increasing populism in Turkey, and the trolls use many of the simplified, extreme, yet effective tactics that populism is known for.⁷⁴ Many authors state that the JDP utilizes an army of paid trolls to perform polarizing tactics on Twitter such as conducting Twitter campaigns or digital “lynching” of opposing political parties, LGBT groups, and other undesirables.⁷⁵ Additionally, by making policy appear popular to the tune of thousands of supporters, the trolls contribute to a silencing self-censorship effect on dissidents, who fear triggering a tidal wave of violent and defamatory messaging, effectively halting undesirable political conversations. One of our interlocutors described the unease she felt as a journalist:

Another Kurdish journalist, Dogan, discussed the increasing role of trolls over time and how trolling became a powerful tool for the authoritarian regimes:

Furthermore, trolling tactics unite scattered networks of supporters into collective actions⁷⁶ and act as a polarization tool among different groups. In our interviews, Levent, an academician and activist residing now in the United States, pointed out this polarization:

Self-Censorship and Twitter-Related Detentions, Trials, and Arrests

In addition to throttling, trolling, and blocking Twitter, the JDP has made Twitter-related arrests since the 2013 Gezi Park protest.⁷⁷ These arrests intensified during major political crises in Turkey (i.e., corruption scandal of 2013, coup attempt of 2016, and Syrian invasion).⁷⁸ By 2015, the Turkish government formed a social media task force to quickly monitor social media accounts and bring criminal charges against users. By 2017, these numbers began to be published in The Interior Ministry of Turkey website. According to weekly arrest reports, Turkey arrested around six individuals for their social media use each week in 2017; this number reached 232 in one five-week period (April–May 2017).

By 2018, the Interior Ministry of Turkey began to inform the public about social media-related arrests in public press conferences. For instance, during the military incursion of Turkey in Syria, Turkey's Interior Ministry reported that they detained 648 people for tweets criticizing this move between January 20 and February 26, 2018. More recently, they revealed that they detained 410 Twitter users for their provocative tweets about the 2020 coronavirus outbreak. The 2017 Reuters Institute for the Study of Journalism Report⁷⁹ not only demonstrated that use of Twitter and Facebook have declined in Turkey, especially after the coup attempt; they also argued that this decline is conceivably due to increasing government surveillance and censorship. All the individuals we interviewed also supported this report's findings. For instance, Banu, an academician residing in the United States, voiced her concerns as follows:

Similarly, the Human Rights Watch report argued that Turkish users now think twice before they post something to Twitter.⁸⁰ A feminist activist, Aysel, summarized this point well in our discussion. When we asked her if Twitter is a free platform for everyone, she looked at us, laughed, and sarcastically expressed:

Conclusion

By studying the dialectical relationship between Turkey and Twitter as they simultaneously negotiate their identities, we have demonstrated that it is crucial to understand these relations vis-à-vis past and current social, political, and economic changes. Consequently, we found that the recent economic and political restructuring associated with neoliberal policies in Turkey has reshaped politics and ideas about the state, citizenship, and human rights. As Turkey became more of an authoritarian state, this has shaped the Turkish government's perceptions of Twitter and increasing control of Twitter. Not only did Twitter enter Turkey at a critical historical moment, but Turkey has also changed their Internet laws and policies to become a more autonomous and authoritarian state in response to these “external threats” to their sovereignty. Our analysis showed that neoliberalism did not promote market freedoms or bring more equality in Turkey. Instead, it promoted the conditions for the growth of a militarized and surveillance state. This is hence evident in the way Twitter is perceived and controlled by the Turkish state.

Championed at first as an open platform for marginalized voices, Twitter optimism has faded and given way to skepticism. In our analysis, we demonstrated that the restructuring of social media industries in recent years, particularly major transformations in their political economies, made it easier for increasingly authoritarian governments to manipulate them. Twitter's rule structure fits into a trend of market-oriented policies and allows the company to operate both in authoritarian and democratic contexts by appealing vaguely to illegality. The vague notion of illegality that Twitter set up at its inception continued to make the company fail as a platform for the users, as what was considered illegal changed with context and geography. Analyzing Turkey and Twitter as agents that articulate and mediate neoliberal ideologies and practices, we illustrated that neoliberalism is not a monolithic ideology and process but instead takes specific forms as it interacts with local histories and politics, hence it is important to study its material effects. Furthermore, our analysis showed that Twitter is not a single unified platform. Instead, Twitter takes different forms as it interacts with different cultures and states with different laws and ideas about human rights, citizenship, and freedom of speech. Twitter's own business ethical model, favoring neoliberal sensitivities and centering around market-based principles, allows identity fluctuations between a liberation tool and a surveillance tool, depending on the context. Consequently, Twitter is increasingly becoming a new surveillance tool for the Turkish state.

FOOTNOTES

BIBLIOGRAPHY