Policy-making by tweets: discursive governance, populism, and Trump Presidency

ABSTRACT Experience in various countries demonstrated that populist leaders enfeeble democracy. Once elected, populist leaders concentrate power in their hands while undermining horizontal checks on their power. By drawing upon the Trump presidency in the U.S., this article reveals one of the dynamics in which populist leaders bypass institutions of horizontal checks in policy-making. It argues that populist leaders use social media platforms to disseminate discourse to convince people that a certain course of action is necessary and thereafter bypass formal institutions in policy-making. Trump used discourse first to discipline the federal bureaucracy, second to roll back Obama-era social and environmental regulations, and third to reorient the US migration policy. His discourse became pervasive thanks to his efficient use of Twitter, which allowed him to achieve political change without going through formal institutional channels.


Introduction
The world has witnessed a global decay of liberal democracy (Appadurai, 2017;Mounk, 2018). Liberal institutions, long understood to be constraints on abuse of power, have been repurposed to serve populist leaders in countries as diverse as Turkey, Hungary, India, and the United States (Muller, 2016). Common to all varieties of populism is a belief that liberal democratic institutions are inconsistent with the people's will (Seligson, 2007). When in office, populist leaders use a variety of mechanisms to concentrate power in their own hands. One particular mechanism they use is discourse, which is a critical asset for populist leaders since they rely on unmediated communication and public statements instead of formal policy processes (Jansen, 2011;Weyland, 2017). Formal policy processes entail accountability of politicians to both the legislative and judicial branches of the state. They follow institutional routes of governance rather than endorsing an unmediated communication between the leader and the public. However, populist movements benefit from discourse, as spoken or written public statements legitimate and animate political action in populist mobilisation (Jansen, 2011). They can use discourse to convince people that a certain course of action is necessary or appropriate (Schmidt, 2008), which will then be used to bypass formal institutions in policy-making.
Drawing upon the discursive governance framework developed by Korkut et al. (2015), we argue that populist leaders use direct mechanisms of governance informed by narratives, leitmotifs, and strategic metaphors embedded in political language, as tools to bypass formal institutions in policy-making. Populist leaders use informal communication channels, that is social media in our case, to disseminate certain ideas and discourses that shape the collective rationality of the general public (Korkut & Eslen-Ziya, 2016). Social media's mass network involving billions of people provides populist leaders with a direct channel for appealing to the people without any mediation (Gerbaudo, 2018). At the same time the informal and sustained communication through social media with 'the people' demotes and enfeebles political institutions making them almost irrelevant, and appropriates policy-making to the discretion of the populist leader (Korkut & Eslen-Ziya, 2016). We define these populist leaders as political entrepreneurs.
We therefore depart from the assumption that a fundamental element of populism is the direct communication that political entrepreneurs establish between themselves and the public. To this extent, we explore how social media facilitates such communication as it removes institutional barriers that may otherwise have had a sway over the nature and course of typical communication between the electorate and the elected. Departing from Searle's (2010) interpretation of speech acts as the basis of all institutional realities, we present the construction of language with strategic slogans and subsequent framing with the use of these slogans as crucial tools to inculcate ideas in line with the goals of the political entrepreneurs in charge. Thereafter, as new political problems arise, discourses in circulation provide 'templates' for how they should be handled. In the U.S. case, we explore how Trump created meta-narratives for how present and future political problems were to be handled so that all subsequent politics were to serve these meta-narratives. Ultimately, Trump's discourses became the social representation of these meta-narratives and social media availed this representation by circulating his discourses to U.S. publics directly and without any deliberation that institutional checks and balances would have otherwise provided.
By relying on shared social representations as the basis for political judgments, creative slogans in circulation become accessible to a public that has little knowledge of or interest in the actual debate (Lau & Schlesinger, 2005). Politicians choose representations to bolster their cause, to mobilise 'attentive publics', and hence, to control the political agenda (Yishai, 1993) by means of defining political issues as existential meta-problems that might also sound unfamiliar for their audiences (Mehta, 2010). We reflect on metanarratives that social media avails to this extent. Meta-narratives as catch-all phrases have been an extremely useful tool for providing solutions to the very issues Trump has depicted to do with the institutional basis of the U.S. polity. That is how, in time, certain belief systems became ubiquitous as normative yardsticks, sufficiently powerful to constrain public perceptions of what is legitimate, 'acceptable' and 'good' (Korkut et al., 2015). Thereafter, political entrepreneurs use these collective rationalities to accrue support for political action without going through the formal policy-making channels such as debate at the parliament or following legislative procedures. Henceforth, rather than explicit mechanisms of political change via public policies, discursive governance provides the implicit mechanisms that political entrepreneurs adopt and exploit as they purport to have established a direct link between themselves and their electorate.
That is why, we follow the theory of discursive governance to analyse the policymaking style of Donald Trump. Trump used discourses as tools of direct communication availed by Twitter to create audiences that would demand first disciplining the federal bureaucracy, second rolling back Obama-era social and environmental regulations, and third reorienting US migration policy to serve his political ambitions. While there has been earlier research elaborating on the role that discourse has gained in policymaking processes, our article first examines how this discourse became pervasive thanks to Twitter. Second, our article notes that the weakness in the existing research thus far is the failure to clarify the mechanism that discourses play in effect to governance, and advances Twitter as the missing mechanism to this extent. Twitter's decision to ban Donald Trump from the platform in the final days of his presidency underscores the importance of understanding the role of this particular mechanism in Trump's approach to governance. Third, we contend that partisan loyalty weakens institutions of horizontal accountability, such as the judiciary and legislature, empowering the political entrepreneur to command through populist discourse with little challenge.
In what follows, the first section provides the context of the Trump presidency. We argue that Trump exposed structural weaknesses of the much vaunted U.S. system of 'checks and balances', which was susceptible to populist capture. The second section sets out the theoretical framework upon which this article draws. The third section, by referring to four cases, illustrates the ways in which Trump disseminated discourse through social media in order to bypass hitherto formal institutions in policy-making. The last section summarises the main argument of this article and makes some concluding remarks.

The context
The U.S. political system is vaunted for its federalism and separation of powers, which ought to militate against populist capture. Yet, the effectiveness of these safeguards has been overstated. The U.S. political system has seen a growth in centralisation through presidential authority over the past century (Milkis & Jacobs, 2017). The growth of the public sector in the twentieth century produced an 'administrative presidency', with the president sitting atop a two-million member federal workforce, led by roughly 4000 political appointees of his/her own choosing. While the president faces institutional checks (e.g. Senate confirmation of political nominees), partisan loyalties have diminished their effectiveness (Lee, 2016). The separation of parties matters more now than the separation of powers does. The ultimate sanctionimpeachmentis not a serious threat to the executive insomuch as the Senate's supermajority requirement for removal demands bipartisan cooperation. Similarly, when the judiciary is filled with actors sympathetic to the president's mission, their ability to 'check' the president diminishes. Left without serious horizontal constraints, the president can direct the federal bureaucracy with relatively little challenge.
When Trump was elected in 2016, worried onlookers took solace in the idea that U.S. political institutions such as the Senate and the federal judiciary would mitigate the president's worst excesses. 1 Indeed, a number of institutional actors tried to thwart the president's worst initiatives. Advisers in the executive branch refused to put pieces of paper in front of the president to protect US trade agreements. 2 Federal courts blocked Trump's travel ban, which was designed to limit arrivals from Muslim-majority countries, in spring 2017. The courts struck down the ban again, after a revised plan was put into effect in autumn 2017. 3 In the Senate, John McCain, with other Republicans Lisa Murkowski and Susan Collins, gave the decisive vote against Trump's repeal of the Affordable Care Act (Obamacare) in December 2017. 4 Initially, it appeared that the executive, legislative, and judicial branches were doing their 'job' in constraining an errant president. 'Trump is a systemic stress test', wrote Charles Krauthammer in Washington Post early in Trump's presidency, 'The institutions of both political and civil society are holding up well'. 5 Jack Goldsmith in the Atlantic agreed: 'The courts, the press, the bureaucracy, civil society, and even Congress have together robustly enforced the rule of law'. 6 Over time, however, Trump navigated these checks more deftly. For instance, while not repealing the Affordable Care Act, Trump undermined it in key ways. He reduced the penalty for not having insurance to $0, effectively eliminating the 'individual mandate', which supported the whole system. His administration re-crafted the travel ban to withstand the Supreme Court scrutiny by including, tokenistically, a couple of non-Muslim countries.
Over time, the judicial branch became less effective in constraining Trump's actions. When it came to the Senate's role in confirming the president's judicial appointments, Trump and Senate Leader Mitch McConnell were in an 'expedient pact' (Hacker & Pierson, 2020). McConnell maintained discipline within the Senate Republican caucus, gave reassurance with donors, and showed rhetorical loyalty to Trump in exchange for Trump's appointment of conservative federal judges who were acceptable to the Republican senators. By the end of his presidency, Trump had appointed over 230 judges to the federal judiciary. His most significant prizes were three new Supreme Court justices, a firstterm record last surpassed by Nixon in 1971.
Having overcome and outmanoeuvred horizontal constraints on his power, Donald Trump set about refashioning the executive branch. Trump used discourse to achieve this, repeatedly disseminating messages through Twitter that were contemptuous of expertise, scornful of bureaucratic and legal procedures, and dismissive of norms. Even more dramatically, as we will show, Trump directly intimidated members of the executive branch through menacing tweets.
Trump's presidency was marked by a mass departure of career bureaucrats. While every new administration sees some churn in the federal workforce, resignations usually come from those who were appointed by the previous administration, rather than non-political government workers and career bureaucrats. Yet, in the first two years of the Trump presidency, more than 1600 federal scientists left their jobs. One in five of these positions were simply left vacant by the Trump administration. Around 700 of these departures took place in the Environmental Protection Agency, which under Trump became a vehicle of environmental deregulation. 7 Other departments saw an exodus of their expert staff. Two-thirds of the Agriculture Department's Economic Research Service, which advises farmers on the economic impact of externalities such as drought or trade disruption, resigned their positions. 80% of researchers at the National Institute of Food and Agriculture left their jobs under Trump. 8 These departures were strategically provoked by Trump. Rather than sacking government experts directly, he used discourse to create a climate, which was inimical to the agencies' traditional goals. The hollowing-out of the executive branch in turn enabled Trump to pursue his policy agenda with less resistance. In doing so, Trump minimised bureaucratic autonomy (Carpenter, 2001), and drew executive branch departments more closely under his personal orbit.
For those civil servants who remained, the imperative to support the president's personal agenda became paramount for agency survival. This transformation of state power was most notable in the Department of Justice's Office of Legal Counsel (OLC). In spite of its prosaic name and small staff (just about 20-25 lawyers and clerks), the OLC is the most important legal office within the executive branch. Its role is to be consulted by executive branch officials to confirm the legality of planned government actions and policy initiatives. All presidential declarations, executive orders, proclamations, and signing statements are reviewed by the OLC to ensure that they are consistent with the president's statutory and constitutional constraints. If the OLC says an action was contrary to statute or was unconstitutional, then administration officials would revise the policy before putting it into effect. The OLC provides what it calls 'controlling advice', which by convention presidents respect as if it were legally binding.
Under the Trump presidency, the role of the OLC was transformed. The OLC retrospectively acted to find legal justifications for Trump's policy announcements. Trump made declarations on Twitter and in other public fora, which were not cleared in advance with the OLC. We argue that this is discursive governance made manifest. Established institutions were repurposed for Trump's goals to give substance to discourses circulated via Twitter but without any fundamental shift in their structure, once Trump made his views public. The OLC lawyers devised legal rationale, however strained, for whatever action the president wished to take. Erica Newland, who left the OLC two years into the Trump administration, revealed that the OLC repeatedly conjured spurious legal arguments to defend Trump's statements and actions. The Atlantic reported, 'They were using their legal skills to launder his false statements and jury-rig arguments so that presidential orders would pass constitutional muster'. 9 Matthew Collette, who worked in the Justice Department for thirty years, described the OLC's actions as 'twisting legal views to fit the personal views or needs of the President'. 10 Shalev Roisman, a former attorneyadviser in the OLC, observed sullenly, 'OLC has been on the frontlines defending some of the Trump Administration's most politically fraught policies'. 11 The OLC defended the 'Muslim travel ban', which was struck down twice by the courts as unconstitutional. 12 The OLC drew up legal rationale for the policy of separating children of immigrants from their parents at the border. In fact, the OLC operated at break-neck speed to defend the president's Twitter announcements, even when they were of dubious constitutional and legal validity. As will be discussed below, the OLC scrambled to defend an out-of-the-blue ban declared by President Trump on Twitter prohibiting transgender members of the military. When Trump fired off a tweet at 10:06pm' 13 in April 2020 declaring that he would freeze most legal immigration in the United States, it was reported that the OLC 'sprung into motion'. 14 When Trump's views came into conflict with other agencies in the executive branch, the OLC defended the president against them. For example, the Food and Drug Administration (FDA) had declared certain drugs unsafe to be used for lethal injection in federal executions. Trump declared his intention to restart federal executions anyway. The OLC issued a legal opinion declaring that the FDA could not regulate drugs used for the purposes of execution, enabling the executions to move forward. 15 Consequently, after 17 years without any federal executions, Trump allowed thirteen executions to take place in his final months in office.

Discursive governance as a populist style of governance
Politics is constituted in language and the use of language has the ability to foster the effects of authority, legitimacy and consensus (Chilton, 2004quoted in Wang, 2017. Therefore, language is among the primary resources for politics (Wang, 2017). Indeed, Schmidt (2008Schmidt ( , 2010 defines politics as a vigorous arena where discursive interactions prompt actors to refine, reframe and reinterpret their ideas. In their turn ideas, defined as beliefs held by the people (Korkut & Eslen-Ziya, 2011), employ several functions in governance (Zurnic, 2014). Ideas help political entrepreneurs to define an issue as a priority. Ideas also shape the content of the policy, which addresses the policy problem. Last but not the least, political entrepreneurs use ideas to underline strategies in order to ensure support for their policies. When systematically used in a discourse, ideas facilitate change by shaping and convincing public opinion that policy change is necessary (Zurnic, 2014). Political leaders formulate their discourses not for an audience, but for the consumption of the general public (Pehlivanli-Kadayifci et al., 2020). Accordingly, discourse is a communicative action: it is the active process of conveying ideas to the public where the agent uses persuasion to create shared understandings and build consensus (Schmidt, 2008(Schmidt, , 2010. It is 'institutionalized structures of meaning' channelling political thought and action in certain directions, which is then used to persuade the public of the necessity and appropriateness of a specific course of action (Schmidt, 2008).
As a set of policy ideas and values and as a process of interaction, discourse is a tool for policy formulation and policy communication (Schmidt & Radaelli, 2004). It enables politicians to establish a dominant narrative of an issue, which determines the parameters of the discussion thereon (Busby, 2016). Discourse also helps politicians to alter other actors' perception of the policy problems and influence their preferences (Schmidt & Radaelli, 2004). Therefore, it can be used to affect the course of political events in a particular way, to legitimate political actions, to develop or reshape political identities, and to frame the political discussion (Schmidt & Radaelli, 2004). Political entrepreneurs use discourse in shaping socio-political meaning of an issue, as they frame the same issue in different ways to varying audiences and handle the very same issue at hand considering its audience (Fairhurst, 2009). They also use discourse to actively construct a situation or context, which in turn delineate or legitimate options available to deal with the situation (Grint, 2005).
Discursive governance framework, which refers to implicit mechanisms of governance resting on narratives, leitmotifs, and strategic metaphors in political language, is a useful tool to understand how political entrepreneurs bypass formal institutions in policymaking (Korkut et al., 2015). We propose that this gains further relevance in populist politics. Political entrepreneurs use ideas and discourse to shape the collective rationality of the general public (Korkut & Eslen-Ziya, 2016), causing a shift in the conditions of debate in favour of the political entrepreneur. We furthermore argue that the medium through which political entrepreneurs communicate their discourse with the general public matters because every communication medium has different physical, psychological, and social features that are distinct and fixed, which shape how users of that medium process information (Ott, 2017). Social media, with its vast network and ability to place the focus on the individual leader, provides political entrepreneurs with a platform where they can shape and dictate the content of their communication and disseminate discourse without any inhibition (Enli & Skogerbø, 2013;Gainous & Wagner, 2014).
Political entrepreneurs benefit from informal forms of communication and increasingly turn to social media platforms (e.g. Twitter) to disseminate their discourse, and to sustain their dialogue without mediation of established institutions. They use slogans, symbols, and metaphors to shape political and social representations within the public sphere in accordance with their own political goals (Korkut & Eslen-Ziya, 2016). This sustained communication with the general public demotes or enfeebles political institutions, restraining policy-making to the guidance of the political entrepreneur (Korkut et al., 2015;Korkut & Eslen-Ziya, 2016). This one-way communication from the political entrepreneur to the public does not necessarily inform the general public about the evolution of policies (Korkut & Eslen-Ziya, 2016). Instead, ideas and discourse produced by the political entrepreneur and disseminated through social media create publics amenable to new policies. Eventually, political entrepreneurs generate public support for their policies and political aims without introducing contentious, costly, and time-consuming institutional changes. Social media provides the exact means for this, as it allows politicians to produce messages; to shape and dictate the content of these messages and to distribute these messages to the public without any filtering (Buccoliero et al., 2020;Gainous & Wagner, 2014). Furthermore, social media enables political leaders to construct and negotiate their image as they see fit (Buccoliero et al., 2020). Accordingly, social media gives even more power to political leaders and 'Tweeting to Power' is now the de facto standard (Gainous & Wagner, 2014). These features of social media make them particularly ideal instruments for political entrepreneurs.
At this stage, a short discussion on populism is also needed. Populism is a communication phenomenon whose appeal is driven by the articulation of antagonistic categories (i.e. the elite vs. the people) (Mudde, 2017) and a transgressive delivery of its messages (Bucy et al., 2020). Populism is hybrid and parasitic in form. What unites different varieties of populism is the assertion that people's will can be expressed without mediation of institutions (Molyneux & Osborne, 2017). This is because populism is a moralistic imagination of politics where political entrepreneurs claim to be the only representative of the people (Muller, 2016). Political entrepreneurs, as self-appointed advocates of the people, prefer 'direct, unmediated access to the people's grievances' (Kriesi, 2014, p. 363). Accordingly, they use social media in order to circumvent the mainstream media or other gatekeepers (Engesser et al., 2017). Twitter is a well-suited means for populist discourse because, as a mode of communication, it allows political entrepreneurs to connect to their audience through simple, impulsive and incivil language (Ott, 2017). Especially in unsettled times, 16 where the general public is more receptive to alternative narratives (Krebs, 2015), social media such as Twitter offers political entrepreneurs with the platform to communicate their narrative to their audience directly and without any filteration (Buccoliero et al., 2020).
Because political entrepreneurs argue that nothing should constrain the will of the people (Kyle & Gultchin, 2018 quoted in Rockman, 2019Rockman, , p. 1549, they reject the idea of horizontal accountability. Instead, they prefer vertical accountability where they only answer to the people through elections or plebiscites. Populism, therefore, includes a core belief that liberal democratic institutions are in conflict with the people's will (Seligson, 2007). These institutions are designed to limit the executive, who is elected by the people. The result of this line of thought is an attempt to concentrate power in the executive, who 'represents the true will of the people' (Levitsky & Loxton, 2013;Rockman, 2019;Soare, 2017). Thereafter, when in office, populist entrepreneurs attack liberal institutions (Levitsky & Loxton, 2013) in order to enhance executive power (Rockman, 2019).
Political entrepreneurs exhibit strong discursive features (Bartha et al., 2020) especially when they are in office. This is because populism is hostile to institutionalised policymaking processes (Bartha et al., 2020). Their dislike for existing rules of the game (Molyneux & Osborne, 2017) generates a policy-making style, which excludes formal institutions and legal constraints (Ostiguy, 2017;Rockman, 2019;Soare, 2017;Weyland, 2017). What happens is as follows. When populists are in office, their ideas and discourse prevail over formal institutions and the pre-existing legal framework, often-forcing formal institutions to follow the new discourse informed by populism. They mobilise ideas and discourse to bypass formal institutions and foster populist style of governance. Therefore, populist governance first evolves into discursive governance and second replaces and bypasses formal institutions and legal frameworks that have hitherto initiated policy-change.
Their adopted mechanism is as follows. Through discourse, political entrepreneurs present a populist Manichean metanarrative where the society is divided between the pure people and the corrupt elite (Mudde, 2017). In the populist metanarrative, the establishment and institutions are presented as extensions and power hubs of the corrupt elite. Accordingly, their presence and continuing power interferes with the supremacy of the will of the people. Political entrepreneurs then raise questions about a previously unvisited issue (agenda-making), and situate this issue within the metanarrative that they have already set. They present this issue in an unmediated manner to consolidate a due audience (re-framing). Turkish politics represents a typical case for the policy-making style that we illustrate. In Turkey, Erdoğan contextualised the aging Turkish population in his metanarrative. He presented the West and the domestic opposition as 'them' conspiring against the nation (metanarrative) that is 'us'. He also suggested that the aging population is a threat to the nation (agenda-making) and argued that population control is a conspiracy of the elite, who is alienated from their own culture (re-framing) (Korkut & Eslen-Ziya, 2016). Notably, before Erdoğan's intervention, population control was not an issue in Turkish politics. He produced slogans such as 'at least three children', establishing a new norm. Hence, Erdoğan monopolised the discussion of the issue. His policy proposal became the new norm in population politics. In the final stage, political change is introduced without making any institutional changes. In Turkey, there has been no change in regulations, which allows abortion up to 10 weeks. Nevertheless, after Erdoğan's discursive intervention, public hospitals became highly unlikely to carry out abortion unless serious health complications develop during the pregnancy (Korkut & Eslen-Ziya, 2016), seriously limiting lower and lower-middle class women's access to birth control. In view of the Turkish example, let us now visit the Trump presidency, political institutions and the unfolding of discursive governance.

Discursive governance and the Trump Presidency
Trump introduced discourses to bypass institutions of horizontal accountability, and fell out of formal policy-making mechanisms as a result. It is quite clear how his metanarrative evolves to produce a subsequent political agenda. Through unmediated communication with the public thanks to his efficient use of Twitter, Trump offered solutions to problems that he himself flouted in the first place. The weakened and reordered federal institutions were later bandwagoned to his discourse to endorse his agenda. As Trump tweeted, his policy advisors scrambled to come to his support, even if this meant inventing subsequently crude legal justifications that defied established norms.
This process was detectable across a wide range of policy fields. Metanarratives of the Trump presidency that certain tropes such as 'Drain the swamp', 'Build the Wall', 'Law and Order', 'Rigged Election' pervaded began as campaign slogans. Others were developed by Trump in response to major events during his presidency, namely 'Stop the witch hunt'.
We examine four case studies of discursive governance in the Trump presidency to elaborate on political appointments, rolling back social and environmental regulations, and migration policy. These four cases have been chosen due to the prominence of the metanarratives throughout Trump's presidency. The Trump Twitter Archive was used to search for key terms, named in the respective case studies, to construct the discursive timelines described as analytic narratives, an approach which 'pays close attention to stories, accounts, and context' while extracting 'explicit and formal lines of reasoning' (Bates et al., 1998, p. 10). The cases have been selected not for the purposes of hypothesistesting but for process-tracing. Therefore, they do not need controls of similarity or difference (Gerring, 2007).

Swamps and witches
Trump was the first U.S. president without prior political or military experience. A New York property developer, who had built up a business brand in the private sector, he was the consummate outsider. As he ran for president in 2016, Trump portrayed Washington, DC as full of corrupt insiders who had 'rigged' the system against the citizens. His Democratic opponent Hillary Clinton had spent nearly her entire professional life working in or around politics. She was the wife of a former president, and she had served as a Senate aide, U.S. senator, and Secretary of State. For her sins, Secretary Clinton was christened 'crooked Hillary' by Trump. 17 As Election Day approached, a new trope passed Trump's lips: 'Drain the swamp', a reference to the political class. Trump first tweeted the phrase on 31 October 2016 before a rally in Macomb County in Michigan. Trump subsequently tweeted about the phrase 36 times. Initially, commentators who gave Trump the benefit of the doubt supposed that the president might institute some kind of ethics in government reform. The BBC ran an article in October 2016, which explored various reforms, which Trump could implement, including a ban on lobbyists, an end to revolving door appointments, campaign finance reform, and term limits. 18 Trump implemented none of these reforms. Yet, he was successful in establishing the metanarrative that Washington was corrupt and that its 'establishment' needed to be purged.
The 'Drain the swamp' narrative gave rise to a related, but more personally focussed, slogan, which emerged later in the Trump presidency ('Stop the witch hunt') (agendamaking). What had initially been interpreted as an agenda for government ethics reform had evidently been re-framed into an agenda of self-preservation. The political change offered by Trump was to fire the people who did not agree with him. Below, we explore the sacking of federal government employees to demonstrate how Trump disciplined the federal administration, configured by discursive governance.
On 6 January 2017, weeks before his inauguration, Trump was briefed by FBI Director Comey that the Bureau was investigating ties between members of the Trump campaign and Russian intelligence services. Three months later, in response to media reports, the House of Representatives Intelligence Committee held hearings on Russian election meddling. At these hearings, Comey confirmed publicly the ongoing investigation. Trump, watching the proceedings on television, provided a live Twitter commentary, which was described as 'troll[ing]' in tone. 19 Two months later, in May 2017, Comey found himself before the Senate Judiciary Committee. Comey expressed his view that Russian President Putin wanted Trump to win the election because 'he wasn't Hillary Clinton, who Putin hated and wanted to harm in any possible way'. 20 Within the week, Trump wrote to Comey to tell him he was 'not able to effectively lead the Bureau', firing Comey 'effective immediately'. 21 Comey was temporarily replaced by his deputy Andrew McCabe. Concerned that a new Trump appointee would end the Russia investigation, McCabe requested that the Department of Justice take over, by appointing a semi-autonomous special counsel to oversee the investigation. The Attorney General, Jeff Sessions, recused himself from making a decision, saying that he was compromised because he had been a Trump campaign advisor who had met with Russian officials. The decision to appoint a special counsel fell to Deputy Attorney General Rod Rosenstein, who then appointed Robert Mueller to conduct the investigation.
Trump tweeted that Sessions's recusal had been 'an unforced betrayal of the President of the United States'. 22 In spite of having called him a 'great man' earlier, Trump turned on Sessions and began to accuse him of being soft on the 'swamp creature' Clinton. Trump tweeted on 25 July 2017, 'Attorney General Jeff Sessions has taken a VERY weak position on Hillary Clinton crimes'. 23  Sessions's removal of McCabe was not, however, enough to save the embattled Attorney General. In spite of having tried to correct his 'mistakes', Sessions had been unable to 'end the witch hunt'. In August 2018, Trump tweeted, 'This is a terrible situation and Attorney General Jeff Sessions should stop this Rigged Witch Hunt right now, before it continues to stain our country any further'. 27 He accused Session of not having proper control over the Justice Department, tweeting that same month, 'he doesn't understand what is happening underneath his command position'. 28 Sessions had lost the confidence of the president, and on 7 November 2018, Trump tweeted that Sessions would be removed from his role and replaced by his own chief of staff. 29 Trump's use of Twitter to shame or attack his own officials set an example for other members of the federal bureaucracy. It was rational for employees to take extra caution not to challenge him or to implement policies that would upset him, not wanting to be the next McCabe or Sessions. Their acquiescence to Trump's declarations is an example of an 'effectiveness trap'. First identified by James Thomson, the effectiveness trap refers to the belief among officials that in order to remain effective at their job in the long run, they should not challenge in the short term. 30 Short-term challenge could result in termination or being side-lined, which serves no one's interests. Thomson observed that foreign policy officials failed to challenge political declarations, which slid the US further into war in Vietnam because they were worried that if they spoke up, they wouldn't be listened to when truly egregious decisions were made. He called it, 'The inclination to remain silent or to acquiesce in the presence of the great mento fight to live another day, to give on this issue so that you can be "effective" on later issues'. 31 But when the time comes to fight on the big issues, too much has been compromised, and the political entrepreneur has already won out.

Red tapeenvironmental and social regulation
In the 2010 midterm elections, Republicans gained 63 seats in the House of Representatives, ending four years of Democratic control in Congress. For the rest of Obama's presidency, the Republicans held their majority in the House, limiting his power to legislate. To achieve any policy reform, Obama was forced to use his role as the head of the executive branch to issue executive orders, which offered guidance on how the federal government carried out its statutory responsibilities. Many of these executive orders related to the president's discretion over safety and environmental standards.
Trump portrayed these regulations as excessive government 'red tape' which was killing jobs and destroying the 'American way of life'. These were tied to Trump's claims that there was a 'deep state', which was self-serving, unaccountable, and uninured to policy change or reform (metanarrative). Trump tweeted over 30 times about the 'deep state' while president. He sometimes used highly emotive language in describing these shadowy figures. In March 2020, Trump tweeted, 'We have a long way to go. There are still some very bad, sick people in our governmentpeople who do not love our Country (In fact, they hate our Country!)'. 32 Trump made clear his response to this apparent problem in his first year in office, when he stood next to piles of paper stacked taller than his height ('regulations'), and a red ribbon ('red tape') which he cut with a pair of golden scissors (agenda-making). Trump's aide Chris Liddell unfurled a long list of regulations relating to highway and road construction. 'Chris is not tall enough for this chart, and neither is anybody else', Trump observed. He said the length of the chart 'really explains what a disaster' the federal regulatory regime had been before he became president. Trump was most incensed by environmental regulations introduced by Obama. He rolled back more than 100 different Obama regulations on the environment. 33 Trump also removed a number of regulations relating to workplace safety and hygiene standards. These included removing regulations designed to prohibit factories and power plants from dumping toxic, arsenic-filled waste in water supplies. 34 To achieve his goals, Trump tweeted angrily at federal agency staff to ignore or bend normal regulatory processes ('red tape') to achieve Trump's desired policy outcome. For example, during the coronavirus pandemic, he attacked the Food and Drug Agency (FDA) for acting too slowly in approving COVID-19 treatments. Trump became a great advocate for blood plasma treatments although the medical evidence was not conclusive. He issued three tweets about plasma during the pandemic, and even tweeted a video in July 2020 claiming that plasma treatments helped to heal millions of U.S. citizens. 35 Trump held an event at the White House on 30th July calling for plasma treatments: 'So if you've had the virus, if you donate, it would be a terrific thing. We really need donations of the plasma. To those that have had the virus, you've gotten through it, and I guess that means you have something very special there'. 36 Trump re-framed the FDA's caution in approving the therapeutics as an example of the 'deep state' working against him. Typically, the FDA is responsible for conducting or approving medical trials to certify the safety and effectiveness of medical treatments. Trump grew angry with the time these trials were taking. He tweeted on 22 August 2020, 'The deep state, or whoever, over at the FDA is making it very difficult for drug companies to get people in order to test the vaccines and therapeutics. Obviously, they are hoping to delay the answer until after November 3rd. 37 This accusation was consistent with Trump's re-framing of the 'deep state' as consisting of anti-Trump bureaucrats, who wanted to remove him from office. Trump ended the tweet with the twitter handle of Stephen Hahn, the head of the FDA, an unsubtle instruction for Hahn to take action before the November 2020 election.
The following day, a Sunday, the FDA issued an emergency authorisation for convalescent plasma as a treatment for  Senior scientists in other agenciesnotably Dr. Fauci, director of the National Institute of Allergy and Infectious Diseases, and Dr. Collins, director of the National Institutes of Healthexpressed concern that the data did not justify the authorisation. 39 Other experts worried that the failure to follow proper agency. Trump, however, was delighted with this political change, a solution, which he had himself proposed over Twitter. He thanked FDA Director Hahn, 'The FDA really stepped up'.

Build the wallmigration
Trump launched his quest for the presidency on 16 June 2015. In his announcement speech, Trump infamously declared, 'When Mexico sends its people, they're not sending their best. They're not sending you. They're not sending you. They're sending people that have lots of problems, and they're bringing those problems with us. They're bringing drugs. They're bringing crime. They're rapists'. 40 These words became core to the construction of Trump's metanarrative that US national sovereignty was under threat by the illegal entry of migrants over the US-Mexico border. In the final presidential debate against Clinton in October 2016, Trump described 'illegal immigrants' as 'bad hombres' (Verney, 2018). Thereafter, 'illegal migration', which was not considered to be a priority on the US political agenda, was publicised (agenda-making) and discursively re-framed as a vital issue by Trump (re-framing). He asserted that U.S. immigration and asylum laws needed to be tougher. His proposal for political change entailed deporting 'illegal immigrants' already in the U.S. and increasing security at the U.S.-Mexico border to make border crossings more difficult. The central political change to this apparent problem was the construction of a border wall. Trump repeatedly told audiences as he campaigned for president that he would find a way for Mexico to pay for the wall.
The Mexican government had no interest or obligation to fund the border wall. Trump needed to support the project using federal funds instead; however, Democrats in Congress repeatedly held up appropriation bills that contained border wall funding, even being prepared to shut down the federal government in the winter of 2018/19 in order to block Trump's key policy. On Christmas Eve 2018, Trump's family had left Washington, DC for the warmer climes of his Mar-a-Lago resort in Palm Beach, Florida. At 12:32 pm, he tweeted, 'I am all alone (poor me) in the White House waiting for the Democrats to come back and make a deal on desperately needed Border Security. At some point the Democrats not wanting to make a deal will cost our Country more money than the Border Wall we are all talking about. Crazy!'. 41 Five hours later, Trump issued a new tweet to announce, 'I am in the Oval Office & just gave out a 115 mile long contract for another large section of the Wall in Texas'. 42 Nothing had changed in the situation in Congress. The government remained shut down and no new funding for the wall had been passed. Instead, Trump had begun to draw down funds from the military and repurpose them for wall construction.
Executive branch budgets, which are passed by Congress, are not endless repositories of cash which the president can repurpose for his own ends. The Anti-Deficiency Act specifies that presidents cannot use funds for projects which have not been given congressional approval. Yet, in a sign of the subservience of the executive branch to the president's declarations, the Office of Legal Counsel backed Trump's reallocation on the basis that the border project was a 'national emergency' and constituted a 'military construction project'. Trump transferred over $18 billion, without congressional approval, to fund the construction of his wall. 43 A plain reading of the Anti-Deficiency Act would empower the OLC to advise that this spending was not legal. Margulies defines these departures from unwritten norms as a form of 'insular lawyering' that offers a 'lifeboat' to a political principle, regardless of the legal soundness of the act (2019, p. 1915).

War on Christianitytransgender ban
Trump was keenly aware that evangelical Christians form an essential element of his electoral coalition. In 2016, one in two voters for Trump was a white evangelical. 44 Trump also made historic gains in the traditionally Democratic voting bloc of white Catholics, which helped to tip key states including Ohio in his favour. Accordingly, Trump provided more substantive support for his Christian constituency than to nearly any other group in the electorate.
Trump portrayed Christianity and its values as under siege from the Obama administration and its mission of 'political correctness' (metanarrative). When he was running for president, Trump tweeted, 'Christians need support in our country (and around the world), their religious liberty is at stake! Obama has been horrible, I will be great'. 45 In his election campaign, Trump retweeted a number of pro-Trump Christian accounts, which made this point. He retweeted one person, 'Donald Trump will be greater president than Ragan [sic]. Trump will set the button for morality, Christianity'. 46 Another tweeter declared, 'What frightens them abt Trump is his totally NON POLITICAL CORRECT natur-e∼Donald is shaking up the house!'. 47 Trump has portrayed himself as the protector of the Christian community (re-framing). He retweeted one tweeter, who said, 'only our Donald will save Christianity. Christ bless our land'. 48 A vote for Trump was a vote to save Christianity, as one account which Trump retweeted declared in all-caps: 'NOW'S THE TIME 4 CHRISTIANS TO GET OUT THE VOTE BIG TIME TO SAVE CHRISTIANITY ONCE & 4 ALL. TRUMP'. 49 Trump tweeted his appreciation to his base of white evangelicals, 'Have great love for the evangelicalsgreat respect for you' 50 and 'I am especially grateful for the tremendous support I have received from the Evangelicals'. 51 At the core of this response to Obama's 'war on Christianity' was Trump's belief that Obama's 'politically correct' decisions needed to be reversed (agenda-making). A totemic policy agenda to reverse was Obama's support for LGBT groups. In June 2016, President Obama ended the ban on transgender military personnel, six years after he had signed into law the repeal of the ban on homosexual soldiers. Just six months into office, Trump decided to re-impose the ban (political change). On 26 June 2017, Trump tweeted, After consultation with my Generals and military experts, please be advised that the United States Government will not accept or allow transgender individuals to serve in any capacity in the U.S. Military. Our military must be focused on decisive and overwhelming victory and cannot be burdened with the tremendous medical costs and disruption that transgender in the military would entail. Thank you. 52 The tweet reportedly came as a complete surprise to the president's military experts. It was sent three weeks after the Defense Secretary Mattis had announced a six-month review of opening the military to transgender U.S. citizens. Mattis had been on holiday at the time of the tweet, and the Pentagon was clearly caught off guard. The Defense Department spokesperson Jeff Davis simply stated, 'We refer all questions about the President's statements to the White House'. Yet, the administrative state was already clicking into gear to defend and implement the president's policy declaration. The following month, Trump signed a memorandum instructing Defense Secretary Mattis to implement the ban within six months. The ban was subjected to numerous court challenges (e.g. Stone v Trump, Karnoski v Trump, Stockman v Trump), yet the policy was eventually validated by the Supreme Court. The institutions had, once again, empowered discursive governance rather than impaired it.

Conclusion
One of the ways in which political entrepreneurs consolidate their power and undermine democratic institutions is through discursive governance. As political entrepreneurs have contempt for institutions, they attempt to bypass formal institutions and legal frameworks that define their operation. They benefit from discourse to sidestep formal institutions of policymaking. Under political entrepreneurs, discursive governance replaces and subsequently bypasses formal institutions by encroaching a change on the established political mechanisms. To this extent, they use social media. Political entrepreneurs use social media as an instrument to create an audience equally both within the state as well as the public sphere. Our article followed Twitter as a tool to disseminate discourse and create a public amenable to policies and ideas proposed by a political entrepreneur. By drawing upon this theoretical assumption, this article studied the Trump presidency and four cases of discursive governance that it involved. Table 1 presents our cases and findings.
As cases of discursive governance, we examined how Trump first disciplined the federal bureaucracy, second rolled back Obama-era social and environmental regulations, and third reoriented US migration policy. Our article, while showing the role that discourse has played under Trump's presidency, also examined how his discourse became pervasive thanks to Twitter. Our case studies do not conclusively 'prove' that the tweets 'caused' policy change. Such direct, observable evidence, especially close to the end of the Trump presidency, would be difficult to come by. However, our cases provide substantive evidence to draw the plausible inference that Trump's tweets led to policy-making. Our process tracing method shows the process by which Trumps' tweets were soon followed by policy change, personnel change, or a legal rationale being produced. This is the reverse of the usual policymaking process, by which executive actions receive prior clearance and examination by relevant stakeholders in the federal government, not least department secretaries and the Office of Legal Counsel. The fact that the FDA changed policy about blood plasma treatments for Covid-19 on a Sunday through an emergency directive one day after President Trump had sent a threatening tweet that copied the name of the FDA director is not conclusive that Trump's tweet caused the policy change, but taken in the context of the analytic narratives, alongside the other cases, a clear pattern emerges.
Using the administrative state to expand presidential power is nothing new in the U.S. politics. Indeed, presidents will seize as much power 'as they think they can get away with' (Howell, 2013, p. 134). Presidents operate within a highly institutionalised setting and that the need for power is built into the institution itself. President George W. Bush was accused of contorting OLC guidance for his own aims, especially with respect to the use of torture and eavesdropping in the 'War on Terror'. What sets Trump apart from previous presidents is not his attempts to centralise or politicise the bureaucracy, but the mechanism by which he does so. This was thanks to his style of discursive governance that his unprecedented Twitter use facilitated. By using Twitter instrumentally, he repurposed and transformed the executive branch bureaucracy. No president in recent times had publicly 'trolled' officials until they resigned or changed policy, in spite of scientific, legal, and other forms of evidence. Trump achieved policy change through deploying the demotic potential of the US presidency: ad hoc policy announcements and tweets which then needed legal justification, fait accompli policy announcements and direct actions of dubious legality. Trump's transgender soldier ban was tweeted while Secretary of Defense was on vacation and while an official review had not yet been completed. Yet, before any of these formal processes had been completed, Trump's tweet was sufficient to reorient the executive branch to change the policy according to the president's wishes. Our take on discursive governance has implications for other countries as the populist wave shakes the foundations of liberal democracy and as Twitter and other social media increasingly gain prominence in facilitating communication between the political entrepreneur and the public. President Trump's most ardent opponents did not appreciate the readiness of large numbers of U.S. citizens to accept and celebrate his disregard for constitutional conventions. Still, in 2020, Trump garnered more than 74 million votes, which was nearly half of the voting U.S. citizens. This amounts to 17.8% increase in the votes of Trump after the 2016 presidential elections despite the economic recession and poor handling of the pandemic by Trump. Similarly, Turkish president Erdoğan has consistently defied constitutional conventions and used discursive governance to inititate policy change. Yet, opinion polls demonstrate 49% job approval for Erdoğan 53 in the midst of an economic crisis characterised by depreciation of Turkish lira by more than 200% since 2018. Hence, for populist party voters, institutional policy-making and even facts on the ground might mean less than what scholars expect, as political entrepreneurs has been largely successful to protect their support base. They have continued to use social media to communicate their message, sustain their interaction with their base, and discursively govern when in the office. Thus, the important question is how to protect institutions and the rule of law against relentless attacks of political entrepreneurs. One possible answer is to impose a control on social media accounts of political entrepreneurs as social media have become important means in discursive governance. This method, however, as expressed by the Twitter boss himself, can 'fragment the public conversation' and set a precedent with possible negative repercussions 'for an open and free Internet'. 54 The next challenge is then to strike the balance between open public discussion on social media and controlling political trolling by political entrepreneurs. Notes 1. Douglas, Lawrence. 2017. "Why Trump Wants to Disempower Institutions that Protect the Truth." The Guardian, February 7; Levitsky, Steven, and Daniel Ziblatt. 2016. "Is Donald Trump a Threat to Democracy?" New York Times, December 16. 2. https://thehill.com/homenews/administration/404933-woodward-gary-cohn-pulled-lettersoff-trumps-desk-withdrawing-us-from. 3. https://www.nytimes.com/2017/10/18/us/travel-ban-blocked.html.