What authoritarianism is … and is not: a practice perspective

This article highlights three main problems with current conceptualizations of authoritarianism: they constitute a negative or residual category, focus excessively on elections and assume that authoritarianism is necessarily a state-level phenomenon. Such ‘regime classifications’ cannot help us comment intelligently on public concerns that politicians like President Rodrigo Duterte, Prime Minister Narendra Modi, Prime Minister Viktor Orban or President Donald Trump are essentially ‘authoritarian’ leaders. This article proposes that, in order to provide political scientists with better tools to distinguish between contemporary threats to democracy and interpretations imbued by left-liberal prejudice, authoritarianism studies must be reoriented towards studying authoritarian as well as illiberal practices rather than the fairness of national elections alone. The article defines and illustrates such practices, which exist in authoritarian, democratic and transnational contexts. Comparative analysis of authoritarian and illiberal practices will help us understand conditions in which they thrive and how they are best countered.

whether there are such things as 'everyday acts of authoritarianism' or 'autocratic leadership' in a democratic society, and if so what they would look like. They have developed sophisticated analyses of the quality of democracy, and some warning signals of 'democratic backsliding' into authoritarian rule. 5 But theywe-lack the vocabulary and the tools to provide a clear, research-based analysis of these apparent phenomena of authoritarianism and illiberalism within established democracies. They have much to say on why leaders like Duterte, Modi, Orbán or Trump get elected, but very little on how to evaluate what they do once in office. Nor can they respond to public accusations that the IMF or the WTO are unaccountable and need to be democratized, 6 or that digital surveillance such as that practised by the US National Security Agency (NSA) and revealed by the Snowden leaks is authoritarian in an Orwellian sense. Yet without such analysis, without really understanding what authoritarianism or indeed illiberalism might look like in a democratic or transnational context, we are in the dark as to what the exact problem is, what the current trends are, and how those trends might relate to other recent tendencies such as populism, xenophobia and nativism.
Why have political scientists developed such blinkers, and how can we set about taking them off ? In the academic literature, 'authoritarian' means two quite different things. In comparative politics, it refers to a regime that does not organize periodic free and fair elections. Such regime type classifications tell us that leaders such as Duterte, Modi, Orbán and Trump were all (relatively) freely and fairly elected, so that unless they dissolve parliament or steal elections, their respective regimes do not merit formal classification as authoritarian. In political psychology, authoritarianism is about the psychological profile of people characterized by a desire for order and hierarchy and a fear of outsiders. Authoritarian personality theory can tell us about likely correlations between holding what they call authoritarian values and voting behaviour. 7 But it does not aim to investigate what the leaders elected by these 'authoritarian' voters do once in office.
Neither regime classification nor authoritarian personality theory helps us to comment intelligently on the concern that Duterte, Modi, Orbán and Trump may be 'authoritarian' or 'illiberal' leaders. Can there still be substance, a substance we political scientists are failing to grasp, behind such verdicts by political commentators, journalists or activists? We currently lack the tools to distinguish between tangible threats to democracy and interpretations imbued by left-liberal prejudice, because we have failed to define or operationalize 'authoritarianism' or 'illiberalism' in ways that relate to the commonsense meanings journalists and citizens are 5 Although recent assessments agree that ' freely using. We should be able to judge the 'authoritarianness' of governments not solely by how they came to power, or by the supposed personality traits of the electorate, but also by what they do once they are in power. I will argue that, instead of focusing exclusively on authoritarian regimes or authoritarian personalities, we should study (that is, define, operationalize, observe, classify, analyse) authoritarian and illiberal practices. A focus on practice has the additional advantage of helping us go beyond a single-state context and recognize such phenomena as transnational illiberalism or public-private authoritarian partnerships. Below I will demonstrate what analytical challenges we face, discuss how political science came to be so myopic in its study of authoritarianism, and suggest how a practice perspective can provide more socially relevant and resonant understandings of authoritarianism and illiberalism. I will define authoritarian practices as patterns of action that sabotage accountability to people over whom a political actor exerts control, or their representatives, by means of secrecy, disinformation and disabling voice. These are distinct from illiberal practices, which refer to patterned and organized infringements of individual autonomy and dignity. Although the two kinds of practice often go together in political life, the difference lies in the type of harm effected: authoritarian practices primarily constitute a threat to democratic processes, while illiberal practices are primarily a human rights problem. 8 I end with a few words on what needs to be done in order that political scientists can better detect and analyse authoritarian and illiberal practices, and thus better advise their societies when their democratic foundations appear to be under threat.

Diagnosis of disciplinary blinkers
To show the mismatch between political science tools and real-world politics, let us first consider a country, Venezuela. In 2006, Hugo Chavez was re-elected as president by a landslide, verified by international election monitors. Having convincingly won a referendum held to abolish the limit of presidential tenure to two consecutive terms, Chavez stood again in 2012 and won again by a significant margin. During Chavez's entire period in office, opposition leaders were harassed and the work of civil society activists and journalists was hampered; some were imprisoned. After Chavez's death, his successor, Nicolas Maduro, narrowly won the presidency in a poll held in 2013 amid electoral violence. In December 2015 the opposition won parliamentary elections and took its seats, but since then Maduro has dissolved the National Assembly and replaced it with a new Constituent Assembly after elections boycotted by the opposition. 9 Political scientists have struggled to classify Venezuela. Cheibub, Gandhi and Vreeland, using data up 8 Ultimately, sustained illiberal practices may also come to constitute threats to the democratic process, and conversely, subversion of the democratic process typically also comes to threaten the autonomy and dignity of the individual. Nonetheless, there is analytical utility in distinguishing between the two categories on the basis of the primary form of harm. 9 Michael Shifter and Ben Raderstorf, 'Venezuela after the Constituent Assembly: a darker phase in the crisis', Foreign Affairs, 1 Aug. 2017, https://www.foreignaffairs.com/articles/venezuela/2017-08-01/venezuela-afterconstituent-assembly. Now let us consider a less obvious case, a policy rather than a country: specifically, the policy of rendition. By this I refer to the secret detention and interrogation of so-called enemy combatants, led by the US Central Intelligence Agency (CIA) but implemented by a variety of national military, police and secret service agencies, in a variety of countries, from 2002 to 2008. 11 It is now generally acknowledged that the rendition programme violated the human rights of the prisoners involved, who were almost without exception non-Americans. The US administration kept the policy secret even from members of the intelligence committees in Congress. 12 The Bush administration that was ultimately responsible for the policy was of course democratically elected. It may not have been fully aware of or in control of the CIA policy. It would be peculiar to label the Bush administration an authoritarian regime because of the rendition programme. Should we draw the opposite conclusion, that because the government was freely elected, it cannot merit the label 'authoritarian', even if democratic oversight was clearly being subverted in relation to rendition? And should we ask only whether the rendition policy was subject to oversight by representatives of the American people? Or are the prisoners themselves and their representatives (for instance, their lawyers) also relevant to determining what counts as authoritarian? What about other nationals on whose soil secret detention and torture took place? What makes this example noteworthy is that the CIA may have extraterritorialized its practices precisely in order to thwart national democratic and judicial oversight. 13 The failure of political science to offer clear answers to such questions is due to three main problems with our current way of thinking. The first is that authoritarianism is in fact a negative category without a definition of its own. The second 10 Jose Antonio Cheibub, Jennifer Gandhi and James Raymond Vreeland, 'Democracy and dictatorship revisited', Public is an excessive focus on elections, at a time when the relationship between voting in elections and actual influence on policy-making is widely doubted by citizens and political scientists alike. The third is the assumption that authoritarianism is a structural phenomenon located only at the level of the nation-state.
The vacuum at the core A peculiarity of the study of authoritarianism is that it does not start with a definition of its own main subject. The term first received extensive conceptual attention, as a category in between totalitarianism and democracy, in Linz's 1975 classic Totalitarian and authoritarian regimes. 14 Linz set the tone for many subsequent studies that characterize authoritarianism first as the shortfall of democracy, and second as a container concept that is given content only in its subcategories. 15 Great analytical work has been done on some of these subcategories, 16 but they do not help much in defining 'authoritarianism' as such.
When attempting to investigate all authoritarian regimes, not just subsets, authoritarianism scholars still fall back on classic definitions of democracy, as formulated by Schumpeter or Dahl, to delineate their field. Schumpeter called a democracy an 'institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote', or even more briefly, a 'free competition for a free vote'. 17 Authoritarianism, then, is characterized simply by the absence of free and fair competition. Dahl's ideas underpin a more comprehensive definition, according to which democracy is not purely about elections, but also entails respect for freedom of expression, access to information and freedom of association. 18 By extension, authoritarian regimes are those that fail to organize free elections and fail to respect these freedoms. The Dahl-based formulation gives a little more information about what authoritarianism is actually like, as well as giving more flesh to in-between types, sometimes called hybrid regimes or defective democracies. But 'the core is still a vacuum', 19 since the definition continues to rely on an absence, on lack of elections and freedoms, rather than on a positive definition of what authoritarianism actually is and what it does.

Looking beyond elections
The presence or absence of free and fair elections is considered the primary touchstone of whether a state is authoritarian or democratic. But this reification of elections, never entirely unproblematic, is less so today than ever. Scholars of authoritarianism have devoted much energy to one side of the problem, namely the prevalence across the world of states that hold elections with some element of competition but with what Levitsky and Way have termed an 'uneven playing field'. 20 But this literature on 'real but unfair' elections remains insulated from recent work on the flaws and limits of elections in established western demo cracies. Scholars of authoritarianism do not appear to take the analytical connection with democracy, upon which their negative definitions rest, completely seriously. Leading scholars on authoritarianism such as Cheibub, Gandhi and Vreeland write that 'elections allow citizens to influence policy by their control over leaders', 21 while Geddes, Wright and Frantz assert that in democracies, 'a ruling coalition of 50 percent (plus) of voters can tax those outside the coalition to distribute benefits to those inside'. 22 But their dummy-variable categorizations of authoritarian and democratic states, which hinge on contested elections, do not begin to test whether citizens are actually enabled to influence policy or organize distribution. These authors have a blind spot for the widespread scepticism, among scholars of western democracies and their general publics alike, as to whether elections actually are a vehicle for policy change in response to popular demand. 23 While scholars of democracy disagree on the extent of, and reasons for, growing public distrust of politicians and political parties in recent decades, they generally agree that it is a real phenomenon. 24 Declining voter turnout, and more lately, the turn to populist candidates and parties have been attributed to precisely this scepticism. Indeed, some authors have begun to argue that there may be a convergence between what were formerly starkly different authoritarian and democratic national governments. 25  and authoritarianism scholars should cease to identify authoritarianism merely by their absence. Instead, we should contemplate what elections originally stood for in the democratic/authoritarian divide: that is, the accountability of rulers to the demos. Indeed, accountability, rather than elections, was the core concept of democracy developed by Schmitter and Karl in the context of the post-1989 democratizations: 'Modern political democracy is a system of governance in which rulers are held accountable in the public realm, acting indirectly through the competition and cooperation of their elected representatives.' 27 In a later piece, Schmitter explicitly dropped the word 'elected' before representatives, opening the way to the inclusion of more informal types of representation as accountability mechanisms. 28 A focus on accountability could still include elections as a frequent-and usually to some extent valid-mechanism of accountability, but it would not conflate an indicator with the category.
Much has been written about alternative forms of accountability in the absence of elections, especially at levels other than the state. Such forms are often identified at the local level, where the mechanism enabling accountability may be informal institutions, civil society, the media, or indeed the central state itself, which may turn to local accountability structures as a means of solving its own principalagent problem vis-à-vis local officials. 29 Similarly, there is a literature on accountability via civil society at the transnational level. 30 The depth and significance of these alternative forms of accountability are, as they should be, much contested. The point here is not to identify what types or conditions of accountability might count as sufficiently democratic, but instead to pinpoint what would count as definitely authoritarian. The 'traffic light' conception of states being democratic when they hold free and fair elections and authoritarian in all other cases does not help us. Instead, we should look for an active practice of disrupting or sabotaging accountability, rather than absence of free and fair elections, as the core feature of authoritarianism. Such sabotage will manifest itself in political practices, not necessarily in constitutional arrangements.  From national regime types to political practices A third blind spot in authoritarianism research has been its failure to notice the impact of globalization on politics. Authoritarianism research overwhelmingly presupposes that the relevant arena for studying politics, authoritarian or democratic, is national. This was not always so. The founding fathers of authoritarianism and democracy studies had a much broader focus. Harry Eckstein and Ted Gurr, who stood at the cradle of the Polity project, aimed originally to identify 'authority patterns of social units', which could in principle include any unit ranging from the nuclear family to the international organization. 31 Robert Dahl, too, in his early work addressed conditions for democracy in a 'social organization', 32 which was by no means necessarily a national state. The dominance of the state in the political imagination, together with a quantitative predilection for country-year units, may explain why foundational ideas on authoritarianism and democracy were narrowed down to an exclusive state focus. Today, this narrow focus gets in the way of addressing some of the most urgent citizen concerns of our time.
Scholars of the democratic West, as well as of developing countries, have extensively researched how, why and to what extent 'the autonomy of democratically elected governments has been, and is increasingly, constrained by sources of unelected and unrepresentative economic power'. 33 This conundrum was strikingly illustrated by the Greek debt crisis. The Greek people repeatedly had the opportunity to choose between different parties in free and fair elections between 2011 and 2015, and made different choices at different times. But even after the radical left-wing party Syriza won a landslide victory on a platform of renegotiating the country's debt repayments, Greece's negotiating position did not substantially alter, and Syriza eventually accepted terms that kept the austerity measures largely intact. National elections were of limited relevance to the imposition of austerity policies on the Greeks in recent years, since the real source of the policies was not national. It was, rather, the so-called Troika (the European Commission, the European Central Bank and the IMF), which held no voter mandate. The Greek situation may be an extreme case, and some may find David Held's conclusion, that 'some of the most fundamental forces and processes which determine the nature of life-chances within and across political communities are now beyond the reach of nation states', 34 overstated. But the claim that state autonomy has been diffused, and that the international system has moved towards multilevel, sometimes overlapping or competing, governance arrangements, has been affirmed in many strands of contemporary political science literature, including public policy, International Relations, political economy and democratic theory. If we take this claim seriously, the question naturally arises whether and how new forms of authoritarianism may manifest themselves at levels below, above or beyond the state. To give but one example, the EU famously suffers from democratic deficits. But does it follow that it is, or can be, authoritarian? Or that the loan conditions of the IMF, or the arbitration decisions of the WTO, which have often been called unaccountable, are authoritarian? Even to be able to answer such questions, we need to think of authoritarianism in such a way that the label could in principle apply to transnational governance arrangements, but this would not automatically follow from the absence of elections.
In sum, to understand contemporary politics we need a definition of authoritarianism that is substantive and dynamic rather than negative and systemic; that focuses on sabotage of accountability rather than the quality of elections alone; and that lends itself to assessing political institutions within, below or beyond the state. Consequently, a practice-oriented definition, rather than a systemoriented definition, is better suited to understanding authoritarianism today, and to answering urgent questions from society about it.

A practice perspective
Practices are, simply put, 'patterned actions that are embedded in particular organized contexts'. 36 According to Theodore Schatzki, one of their prime theorists, 'practice approaches can … analyze (a) communities, societies, and cultures, (b) governments, corporations, and armies, and (c) domination and coercion as either features of, collections of, or phenomena instituted and instantiated in practices.' 37 Philosophers, sociologists and cultural theorists have all turned to conceptualizing 'practices', in slightly different ways, for slightly different reasons. I will highlight here just two important and almost universally agreed features of practices.
First, practices are much more than the action or behaviour of an individual, but much less than a state structure. A focus on practices allows a shift away from designating only 'regimes' as authoritarian, recognizing that in contemporary politics, governance arrangements can be more fluid. In this way, we can begin to imagine (and hence identify defining features of ) authoritarian practices occurring in India, the United States or the EU. At the same time, practices do not narrow the focus to the individual. 38 While political science may be too concerned with state structures, in common parlance we sometimes fall into the opposite trap, referring to individuals like Modi or Trump as if they were all-powerful and uniquely responsible for all political life inside and emanating from their respective states. 36  The global digital surveillance programme of the US National Security Agency (NSA), made public through the Snowden revelations, nicely illustrates what constitutes a practice. For a number of years, the NSA gathered massive amounts of data primarily on non-US citizens through various methods, including siphoning data from land and undersea cables, ordering companies to share metadata, using malware and pressuring vendors to install 'back doors' into their products. This practice was not associated specifically with one administration: while various sub-projects such as XKeyscore and PRISM appear to have been initiated under George W. Bush, 39 they continued under the Obama administration, and the 2008 FISA Amendment Act that authorized the NSA, in principle, to monitor electronic communications of foreigners abroad, was renewed in 2012. 40 The programme was sustained for years, well documented and to some extent transnational, with the British Government Communications Headquarters and the Australian Signals Directorate being particularly close collaborators. 41 Hundreds of people have been involved in its implementation. 42 This brings us to the second useful commonality in practice theory: its emphasis on organizational and social context. According to Schatzki, 'a practice is a set of doings and sayings organized by a pool of understandings, a set of rules'. 43 This chimes with what we know from case-studies of authoritarian regimes. People do not obey an isolated dictator out of pure fear, or collaborate with him out of pure greed or hunger for power. They develop common understandings of how things are done within their social context, whether they are true believers in the government's legitimation narratives, or just pragmatists, or somewhere in between. Indeed, while practice theory is rarely explicitly invoked in authoritarianism studies, 44 many excellent country or area-focused studies implicitly take a practice-centred approach. Stern and O'Brien, for instance, find that politicized Chinese citizens are constantly receiving and interpreting 'mixed signals' about what is and is not permissible, an observation which suggests that 'the Chinese state, even at its most repressive, is not as single-minded as it is sometimes portrayed', but instead consists of a 'hodgepodge of disparate actors' with very different ways of operating. 45  cases, make a careful distinction between 'the machinery of the state' and 'its operators', 46 and argue that strong state institutions can be a remarkable resource for effective authoritarian practices, for instance by ruling parties. Heydemann and Leenders insist that a shift from holistic regime analysis to investigating the authoritarian practices of judicial, social policy or religious institutions in Syria and Iran is necessary to analyse what they call 'recombinant authoritarianism'. 47 When considering the possibility of 'authoritarianness' in Hungary or the United States, too, we must not get obsessed with the personalities of Orbán or Trump alone, but equally consider the indispensable 'doings and sayings' of clusters of politicians, civil servants and public figures, at different levels, who are associated with them. It was the common understanding, within and beyond the intelligence community, about what constituted necessary and permissible data-gathering for national security that made the NSA's surveillance practice possible.

Authoritarian practices
What, then, are authoritarian practices? There is a risk of stretching the term to encompass all political phenomena that have a negative impact on people's lives, including discrimination, violence, corruption or inequality. That would be analytically unhelpful. As I have suggested above, we should refocus our understanding of authoritarianism from failure to hold elections to sabotaging accountability. A closer look at the meaning of accountability itself should clarify what would constitute sabotaging it, and why it matters. According to a parsimonious and widely cited definition, 'accountability is a relationship between an actor and a forum, in which the actor has an obligation to explain and justify his or her conduct, the forum can pose questions and pass judgment, and the actor may face consequences'. 48 The reasons for valuing accountability tend, when inverted, to shine a light on what most of us would intuitively label 'authoritarianism', and why we consider it a normative problem. According to Rubenstein, fundamentally, 'accountability enables-more precisely, it helps to constitute-non-domination'. 49 She goes on to enumerate its virtues: increasing substantive and procedural rule-following by political actors, promoting the preferences and civic virtues of those to whom accountability is rendered, and providing useful information to everyone concerned. Bovens similarly discerns a democratic aspect, a constitutional aspect and a learning aspect to accountability. 50 Authoritarian practices enable domination: they entail substantive and procedural rule-breaking, interfere with the preferences and inhibit the civic virtues of those to whom accountability is owed, and strictly control information flows. 46  Authoritarian practices of sabotaging accountability should not be equated with a mere lack of accountability, which may be caused by lack of capacity or may be institutional. Politically, one might like to call this 'authoritarianism' to make a point, but analytically it would lead us back to a negative definition. Investigating active practices of accountability sabotage as the core of authoritarianism, by contrast, is particularly relevant today, because, unlike a few centuries ago, discourses and institutions of accountability are now ubiquitous, and often mimicked. Most authoritarian regimes today have a parliament, a constitutional court and perhaps even pseudo-pluralist media, making analytical tools that can discriminate between imperfect accountability mechanisms and actual subversion of accountability all the more important.
Not all types of accountability are relevant to authoritarianism. Deriving from the Latin auctoritas, meaning 'authority', authoritarianism presupposes power. More specifically, it presupposes a social unit where there is control by some over others. So (in terms of practical power if not necessarily constitutional arrangements) the relationship between the actor and the forum that is being sabotaged is one of downward accountability, not upward or bureaucratic or peer accountability. Thinking of authoritarianism in terms of accountability sabotage makes it possible to escape the term's conventional 'othering' subtext. Actually existing democracies are themselves forms of domination and are never fully accountable. Democracies may also need to be democratized, even when they are not in a process of 'backsliding'. 51 A focus on actual control over others also makes it possible to discern transnational practices without resorting to far-fetched notions of accountability sabotage vis-à-vis all world citizens. People over whom state actors exert control are usually either nationals or people within that state's borders, but occasionally there is physical control, and there can be accountability sabotage, without either. In the rendition example, the CIA in collaboration with other security services clearly had control over the 'unlawful combatants' who were neither US citizens nor within US borders. In such a case, accountability is owed, and may be sabotaged, vis-à-vis US citizens who have a right to know what their state agencies are doing, the rendition victims themselves and their representatives, and residents of states on whose territory secret detention and torture took place. Another example would be the extraterritorial practices of authoritarian regimes such as Eritrea, Iran, Syria and various central Asian states, which involve physical and digital harassment of their critics in the diaspora. 52 To recapitulate the argument so far, then: in order to fulfil today's analytic needs, a definition of authoritarian practices should (a) be able to approach authoritarianism as a substantive phenomenon, rather than just a shortfall of democracy; (b) step away from elections as the primary touchstone; and (c) capture the character 51 I wish to thank one of the anonymous reviewers for formulating this point. 52  of authoritarianism when no longer necessarily embodied in and exercised by national governments of states alone. Authoritarian practices presuppose a downward relationship, where the political actor engaged in them has control over the people affected. Drawing on these criteria, I define an authoritarian practice as a pattern of actions, embedded in an organized context, sabotaging accountability to people ('the forum') over whom a political actor exerts control, or their representatives, by disabling their access to information and/or disabling their voice.
As represented in figure 1, authoritarian practices are manifestations of the prevention of dialogue between power-holding actor and forum. Keeping actions and decisions secret from the forum precludes dialogue by disabling its access to information. It should be clear that not all forms of secrecy in politics constitute sabotage of accountability. Under certain circumstances, political secrecy can be legitimate, provided that the procedure for determining exceptions to publicity should itself be public. Confidential sharing of information with designated representatives of the forum can also be a legitimate alternative to full publicity. 53 Most parliaments, for instance, have a secret intelligence committee, where a select group of parliamentarians will be briefed on intelligence matters not openly discussed with other representatives, let alone the general public. There are undoubtedly hard cases, but for discerning authoritarian practices, the appropriate focus is on the easy ones: a pattern of disabling information, not exceptional incidents or well-regulated secrecy bound by transparent procedures. The CIA's rendition programme would meet this test: on a number of occasions, the Bush administration refused to hand over relevant memorandums and other information requested by members of House and Senate Intelligence Committees, or denied they existed, and in 2005 failed to release many relevant documents in response to a court order. 54 Access to information is also obstructed when a forum is deliberately given inaccurate information. Of course, politicians spin, twist and deflect from the truth all the time. But a pattern of disinformation is more than an occasional gloss on the facts. A one-off political lie does not constitute an authoritarian practice, but a pattern of inaccurate information by a number of people in authority on the same issue at different times could well do. For instance, President Trump's assertion that there was a record attendance at his inauguration should not be considered sustained or consequential enough to count as a pattern. But the allegations that millions of illegal migrants had fraudulently voted in the US presidential elections would be an example of a more sustained pattern. These allegations were first aired during the Trump election campaign, 55 and then repeatedly voiced by the President himself, 56 by his spokesman, 57 and by the White House senior adviser Stephen Miller; 58 finally they were made the subject of an investigation headed by Vice-President Mike Pence. 59 The contemporary relevance of authoritarian practices in a time of 'post-truth' politics and alternative facts need hardly be stated. As shown schematically in figure 1, both secrecy and disinformation disable the communication flow from power-holders to the forum.
The other form of accountability sabotage is disabling voice. This disrupts the dialogic flow in the other direction, from the forum to the actor. Critical questions may be discouraged, and questioners intimidated, penalized or bought off. Or it may be that criticism, 'passing judgement' on the conduct of the actor, is obstructed. This particular form of sabotaging accountability is most readily recognizable to those who study authoritarian regimes: we tend immediately to think of free and fair elections as the means of passing judgement, and thwarting them as authoritarianism.
But a voice can be much more than a vote. Passing judgement does not happen only at the ballot box, but can also take the form of journalism, NGO reports, sermons or rap songs. Disabling the forum from passing judgement can be manifested as interference in free and fair elections, but also as censorship of, or arbitrary interference with, critics of a particular action or decision. The questioners and critics may be ordinary people or professional questioners and critics, such as parliamentarians, journalists, human rights defenders or other activists. Or they may even be internal critics, such as actual or potential whistleblowers.
Hungary's media law of 2010, which established a government-controlled oversight body, provides a good example. The institution does not formally engage in censorship, but has the authority to levy prohibitively high fines on radio and 55  television outlets. Overseen by a party-controlled body with broad powers, and in combination with biased tendering procedures, it has had a chilling effect on Hungarian media. 60 The mass cancellation by the Indian government of permits for NGOs to receive foreign funding licences is another example of a measure aimed at disabling critical voices. 61 And Venezuela has done both: under Chavez it nationalized television stations and adopted restrictive media laws, as well as prohibiting foreign funding of NGOs, and under Maduro it has continued and sometimes extended these practices. 62 All these examples have been well documented, and lawyers have pointed out how these practices are in breach of the respective countries' international human rights obligations. But political scientists have tended to look at such practices primarily in relation to the freedom and fairness of elections, or to ideologies such as populism. They do not scrutinize them as potentially 'authoritarian' practices in and of themselves. The accountability sabotage framework allows us to do so. It does not systematically classify governments or institutions as either 'authoritarian' or 'not authoritarian', but applies the term instead to specific practices, which may be more or less endemic to the overall mode of governance.

Illiberal practices
Some readers may at this point note the absence of elements that they were expecting in a definition of authoritarianism, such as the violation of human rights. Just as some political scientists find it useful to distinguish between plain democracy and liberal democracy, I make a distinction between authoritarian practices and illiberal practices. It may seem academic in the face of threats to democracy, civil society and human rights to make such theoretical distinctions. But it is important precisely when liberal alarm bells are ringing and emotions are running high to make clear analytical distinctions, the better to understand the phenomena we are witnessing.
Authoritarian practices, at their core, are about accountability sabotage. Sometimes, they violate individual political rights. But a whole range of other rights are liberal rather than political. According to Fareed Zakaria, constitutional liberalism 'refers to the tradition, deep in Western history, that seeks to protect an individual's autonomy and dignity against coercion, whatever the source, state, church, or society'. 63 One may question Zakaria's assertion about the depth of this tradition in the West, and its assumed absence elsewhere, in the light of colonial history International Affairs 94: 3, 2018 as much as contemporary developments, but it is analytically useful to identify illiberalism as a phenomenon distinct from authoritarianism. At the level of states, the distinction can also further our analysis of so-called 'hybrid' regimes. It is now well understood that hybridity is not unidimensional, various authors having made a distinction between an electoral and a liberal dimension. 64 But hybridity on either of these dimensions is still measured against democracy, and so classified as underperformance, rather than being thought of in terms of active practices, and studied exclusively at the national level.
I define an illiberal practice as a pattern of actions, embedded in an organized context, infringing on the autonomy and dignity of the person. Belonging to the class of illiberal practices are patterns of interference with legal equality, legal recourse or recognition before the law; infringement of freedom of expression, fair trial rights, freedom of religion and the right to privacy; and violations of physical integrity rights.
The delineation between authoritarian and illiberal practices is not a mirror image of the distinction often made between free and fair elections and the separation of powers. The separation of powers serves the same purpose as free and fair elections: accountability. Subversion of the separation of powers, just like election fraud, is therefore authoritarian. Various forms of bypassing parliament, either through secrecy or quite openly via unmandated presidential decree, should also (when patterned) be considered authoritarian practices because they disable one of the most consequential forms of passing judgement, parliamentary oversight. The same is true when judicial oversight is thwarted, corrupted or ignored. Figure 2 shows how authoritarian and illiberal practices are distinct but overlapping categories. In particular, violations of freedom of expression are authoritarian practices, because they block the accountability dialogue. At the same time, they are illiberal practices because they infringe on the autonomy and dignity of the individual.
Needless to say, illiberal practices are no more or less reprehensible or consequential than authoritarian practices. The distinction is analytical. Authoritarian practices aim to shield power-holders from accountability. Illiberal practices can have many purposes, which may include suppressing the voices of those who constitute a threat to power-holders, but may also be designed to promote an ideological project, or even to carry out the will of the majority. Filipino President Duterte's profoundly illiberal endorsement of killing drug users is a case in point. 65 It is not plausible that drug users constitute a particular threat to Duterte's power. Rather, the idea that drug users are dangerous and less than human apparently enjoys considerable support in the Philippines, and Duterte's stance may have helped him win the presidency. Unless allegations of drug use are being turned against government critics, it cannot be assumed that the illiberal practice of endorsing the murder of drug users has an authoritarian intent. Rather, it 64 See e.g. Mikael Wigell, 'Mapping "hybrid regimes": regime types and concepts in comparative politics', is a typical manifestation of populism, defined as majoritarian anti-pluralism. 66 Populism can lead to illiberal practices, but it is by no means their only cause. While the Russian government's repressive measures against homosexuality, for instance, may stem from populist motives very similar to Duterte's, the regime's repeated imprisonment of Alexei Navalny is an authoritarian bid to silence a popular, perhaps even populist, oppositional voice. At times, there may be a causal connection: illiberal practices often provide an incentive for authoritarian practices. The rendition programme, for instance, was first and foremost an illiberal practice: so-called 'unlawful combatants' were subject to arbitrary detention and torture. However, its practitioners, aware of how controversial the torture programme would be, took pains to keep it secret-an authoritarian practice. At other times, illiberal practices, especially when sanctioned by the majority, may be engaged in quite openly, and criticized openly, as exemplified by the drug-user killings in the Philippines, or by Sheriff Joe Arpaio's treatment of prisoners and racial profiling in Arizona.

An urgent agenda
Many political scientists have been asking themselves recently whether they should do things differently: take a more explicit stance on political developments they find worrying, engage more actively in public debates, pepper their dry analyses with emotive and close-to-home examples so as to increase their impact, or find ways to counter attacks on scientific knowledge and scientists. These are all worthy and necessary initiatives; but we should also re-examine our  core business, and consider whether our analytical tools are still fit for purpose. Conceptual innovation might not look like a political intervention, but it is. The most important contribution social scientists can make to society is to do what they do best: make systematic observations, abstract from what they are seeing, then again operationalize from the abstractions, classify and analyse, to answer the descriptive, causal and normative questions of their day. Authoritarianism studies began with scholars like Karl Popper and Hannah Arendt, and later Juan Linz and Guillermo O'Donnell, analysing the horrific developments in their own societies with a view to learning how to counteract such trends. 67 It turned into the professional study, from the vantage point of the West, of political systems other than, and considered inferior to, our own. In an endeavour to revitalize authoritarianism studies by going back to first principles, I have defined and illustrated authoritarian and illiberal practices, and suggested that they are theoretically distinct, even though in reality the two will often coincide. Not just illiberal, but also authoritarian practices can occur regardless of how the political actor(s) got into their position(s), since authoritarianism is not defined as a lack of free and fair elections.
A practice-based definition allows us to observe that the Venezuelan government engaged in extensive authoritarian and illiberal practices under Chavez, while at the same time acknowledging that he was repeatedly brought into power by popular vote. Similarly, we can postulate without logical contradiction that while Prime Ministers Modi and Orbán and Presidents Duterte and Trump appear to have been freely and fairly elected-and could be re-elected-their governments engage in authoritarian and/or illiberal practices.
I have also shown that the practice of keeping illegitimate secrets is entrenched in the security agencies of the United States, regardless of which party or president is in power. This is not intended to point the finger particularly at the US, but rather to show that authoritarian practices in specific issue areas or by specific agencies can persist within established democracies. Indeed, as the critical security literature has long recognized, what I have termed authoritarian and illiberal practices frequently coincide with the invocation of security concerns. 68 Such practices require political science analysis, in terms of their authoritarianness, even when there is no imminent threat of full-scale regime change.
This conceptual essay is just a beginning. Authoritarian and illiberal practices must be better operationalized, classified and compared, and causal connections established with other phenomena, if we are to suggest ways of responding to them. Redefining authoritarianism and illiberalism from a practice perspective allows us to bring back home the knowledge we have developed about how Anastassia Tsoukala, eds, Terror, insecurity and liberty: illiberal practices of liberal regimes after 9/11 (Oxford: Routledge, 2008). The practice approach remains agnostic as to whether securitization is the cause or only the justification of these practices, or both at once.