Between militant democracy and citizen vigilantism: Using citizens’ assemblies to keep parties democratic

Abstract

The essential role of parties in democracies makes it important to keep them democratic. This article argues for sortition-based citizens’ assemblies (CAs) organized in and by civil society to formulate democratic standards for political parties to follow, to evaluate them individually and to criticize them publicly if they do not. This is a third and potentially complementary way to keeping parties democratic, placed between militant democracy on the one hand and citizen vigilantism on the other. Militant democracy is challenged by the fact that few democratically problematic parties are ostensibly anti-democratic and therefore likely to fall under the legal criteria for issuing party bans and other legal sanctions. Militant democratic measures are also likely to be ineffective and are vulnerable to abuse. Citizen vigilantism, whereby active democratic citizens take on the responsibility for protecting democracy, deals better with the ambiguous nature of democratically problematic parties but suffers from a lack of democratic authorization and clear standards of critique. While not perfect, the proposed model remedies many of the shortcomings of both approaches. Contributing to an emerging literature on CAs as instruments in the protection of democracy, the article evaluates the model’s normative justifiability, feasibility and likely effectiveness.

I. Introduction

Parties are an important part of the infrastructure of democracy (Müller 2022; Rosenblum 2010). They are the key agents of democratic government; they are involved in the election process; they (ideally) offer citizens comprehensive views of the common good of society, which can integrate citizens’ dispersed preferences; they translate electoral majorities into governing majorities; and they act as executers and controllers of government power (Herman 2017; Kitschelt 2008; Sartori 1976; Schattschneider 1942). They contribute to the reproduction of liberal democracy when they engage citizens with the exercise of democratic power through elections, public debate, deliberation and reciprocal justification, and when they exhibit public commitment to the protection of minority rights, to political toleration and pluralism above and beyond the immediate observation of the rules of the game (Herman 2017; Moore 2023; Muirhead and Rosenblum 2020). This means it is important whether parties are democratic or not. It is also important that they aim to justify policies to all citizens, respect the rights of their opponents and political pluralism, and do not seek to monopolize political power and dismiss and exclude their competitors as enemies. Anyone concerned with democracy should thus be concerned with the democratic nature of parties and with how to keep them democratic (Muirhead and Rosenblum 2020; Waldron 2016). In fact, many of the current threats to liberal democracy arise from political parties. This includes both explicitly anti-democratic parties and parties that are not explicitly anti-democratic, but nonetheless have problematic views of key democratic rights (below we refer to them as non-liberal-democratic parties, or NLDPs). Such parties are likely to damage democratic culture, institutions and rights.

Adding to an emergent literature on the topic, this article discusses the justifiability, feasibility and potential effectiveness of using citizens’ assemblies (CAs) organized in and by civil society as an instrument for keeping parties democratic. Such assemblies should formulate democratic standards for political parties to follow, and evaluate whether individual parties meet the standards and publicly criticize those who do not. Such a strategy thereby aims to keep political parties democratic through ‘naming and shaming’. This approach is a third way between two competing approaches. The first is the militant democratic approach, whereby the public institutions (the state) issue bans against and impose other restrictions upon parties, which pose a threat to democracy. Militant democracy is problematic to the extent that it restricts the democratic freedoms of some citizens. It is likely to be ineffective and is vulnerable to abuse. A key challenge is that many democratically problematic parties are not explicitly anti-democratic, but rather have an ambivalent or ambiguous commitment to democracy – sometimes referred to as ‘far right lite’ parties (Daly and Jones 2020) – and it is therefore difficult or even impossible to produce sufficient evidence for their threat to democracy to pass a legal verdict. Moreover, although courts are supposed to be independent from politics, they risk being perceived as instruments of the governing parties to strike down competitors and to restrict the political rights of political opponents. In addition, and conversely, militant democratic measures may indeed be abused if the interpretation of democratic norms and principles is too closely tied to a very specific institutional configuration barring democratic contestation and innovation and/or if courts are subjected to the control of non-democratic governing parties and instrumentalized in efforts to pacify political opponents. The second approach is citizen vigilantism, whereby groups of citizens who are concerned about the democratic credentials of certain parties unilaterally decide to fight them with all legal – and sometimes illegal – means available. Despite all good democratic intentions, this approach suffers from the lack of democratic authorization and clear standards for critique. It is in the nature of civil society organizations and liberal political cultures that citizens and their organizations are free to criticize each other for not sticking to democratic principles and norms. However, after a certain point, reactions of civil society actors will have important consequences for the targets of their protests. Without a common ground for deciding what is democratic and what is not, and an agreed procedure for reacting towards transgressions of democratic norms, civil society actors may thus unduly impact and restrict the freedom of parties that has been defined by constitutional norms and constitutionally structured democratic procedures (cf. Berkey 2021).

This article will proceed as follows. Part II reviews the various elements of a democratic system in a broad sense that they are either intended to keep parties democratic or tend to have democracy-promoting consequences. Part III discusses why regulating parties and/or influencing them in particular ways might be problematic from the viewpoint of the freedom of association and other democratic freedoms. It addresses the criticism of the militant democratic approach and points to a particular challenge faced by many democracies, namely that few parties are clearly and explicitly anti-democratic and therefore it is difficult for public institutions operating according to the principle of the rule of law to have them declared undemocratic and to subject them to restrictions in the protection of democracy. Part IV considers the pros and cons of vigilantism, according to which the main responsibility for protecting democracy belongs to active and engaged groups of citizens. This defence may be based on both legal and illegal means, the latter being the most problematic aspect of this approach as it entails the violation of rights. Part V introduces the use of CAs as an alternative way of keeping political parties democratic. It discusses their justifiability, feasibility and potential effectiveness. The discussion aims to show that while CAs are no panacea, they can constitute a promising third way to keep parties democratic if they are introduced at the right time and therefore should be considered seriously as a stabilizing part of democracy’s infrastructure.

II. How democratic systems (aim to) keep parties democratic

There are several elements of constitutional systems that are intended to keep parties democratic and may potentially do so. In some ways, distinguishing those elements of a constitutional and legal system that set up a democratic system and those that protect it requires drawing a line that may be difficult to discern in practice – especially as those elements that exist to make a well-functioning democracy possible (such as civil and political rights and the electoral system) simultaneously exist to bring the system into existence and to ensure its functioning. And ensuring the functioning of the democratic system also means protecting it to some extent by defining the acceptable forms of action within the democratic system.

Within this overall system of democratic government, political parties occupy an ambiguous position. While most constitutions acknowledge the nature of the political system as a multi-party system and some even discuss the role of political parties as fundamental to the ‘formation of the political will of the people’ (Article 21(1) Basic Law of Germany), how political parties generally fit into the constitutional scheme is in the process of being theorized (Khaitan 2020). In general, constitutional systems fall somewhere on a spectrum between liberal models that leave parties free to handle their internal affairs as they please and a more regulatory model where the state imposes certain requirements on the internal functioning of the party. Belgium, France and Italy, for example, have no law regulating parties and the electoral law either has no definition of a party or has a very minimal requirements to register as a political party in terms of electoral requirements. Other European democracies, such as Germany, have extensive and detailed regulations regarding the definition of parties. Where party laws exist they aim, broadly speaking, to achieve five purposes: (1) they define what a party is and how it acquires legal recognition as a political party; (2) they regulate the types of activities in which political parties may engage; (3) they set norms that need to be followed by parties; (4) they set sanctions for parties for various breaches; and (5) they regulate the functioning of funding in various respects (Bértoa, Piccio and Rashkova 2014: 127–34; Karvonen 2007). Many of these requirements go towards establishing a framework that makes party politics possible by instituting the necessary processes and mechanisms for operationalizing a party-political system for and between elections. Some of the criteria also explicitly protect the democratic nature of that system. In this part, we briefly consider the main forms of party regulations that aim to explicitly keep political parties democratic.

One of the most common ways by which parties are regulated is through requiring them to adhere to certain criteria before they can be registered as political parties. Some countries have very minimal requirements that require a party to register to be able to show that it has a certain degree of support, often by producing a list of signatures (such as Belgium), while others require some other minor details such as a logo of the party to be registered (Italy). The German Party Law (Parteiengesetz) imposes several requirements on political parties by requiring that a political party adopt a constitution that specifies a great range of elements of the party’s functioning (Articles 6–17 of the Party Law). While the German criteria are detailed and require the party to adhere to some form of internal party democracy, in terms of commitments to values, the Party Law only refers to the power of the state to ban political parties in terms of the party ban and the way in which it is to be executed (Article 33). Several countries include commitment to democratic values specifically in the requirement to be registered as a political party. In Finland, for example, to be entered into the party register, ‘the rules and regulations of the association [must] guarantee that democratic principles are abided by in the decision-making and activities of the association’ and it is necessary that ‘the association has drafted a party program, which takes up the goals and principles followed in the national activities, whose final objective is mentioned in the rules and regulation’ (Articles 2(3) and 2(4), Puoluelaki, 10.01.1969/10).

The regulation of party funding and expenditure is a complex matter, which needs to account for a variety of sources of funding and different contexts in which it can be spent. In terms of private funding, most countries impose limits on party donations received from foreign and anonymous sources to prevent these from exercising undue, possibly anti-democratic, influence. In addition to private funding, many countries have adopted public funding for political parties to support their work, reduce their reliance on private funding and reduce corruption. While it is important that funds are allocated to political parties on an equitable basis (although what exactly that basis is may vary), at least Germany has enacted a procedure through which the party funding of an anti-democratic party can be withdrawn. The decision to amend the Basic Law was taken after the Federal Constitutional Court refused to ban the NPD (BvB 1/13) and was overwhelmingly approved by the Bundestag. The amendment to Article 21 now permits the withdrawal of party funding from parties that threaten the ‘basic democratic order’ of the German state, which can be applied even where the party ban provision poses too heavy a procedural burden.

There are, of course, a whole host of other measures, especially in criminal law, that can either directly or indirectly aim to keep parties democratic. One such measure could be the investigation by the Bundesamt für Verfassungsschutz (BfV) to bring the Alternative fúr Deutschland (AfD) right-wing political party under surveillance. The move to do this was blocked twice by German courts, which decided that the surveillance would be prejudicial to the party and would interfere with the formation of the democratic will. Ultimately, surveillance was permitted as the court found that members of the more radical wing and the youth wing remained active in the party even after the radical wing had been disbanded. Thus the BfV could investigate the finances and eavesdrop on members of the AfD. While there may be no immediate legal consequences from the surveillance, it is – as the courts noted – an action that affects the democratic standing of the AfD.

Beyond these state regulations, political parties have various tools at their disposal to hold each other to democratic standards. Other political parties may be able to influence the democratic nature of specific political parties through their interactions with them at risk of embracing undemocratic norms (Akkerman, de Lange and Rooduijn 2016; Downs 2012). While political parties do not have formal influence over one another, the way in which the political arena is structured gives competing political parties a position that they may leverage to hold other parties to democratic standards. The study of responses to a variety of parties that may be considered at risk of developing undemocratic tendencies – for example, right-wing parties such as the AfD or (other) populist parties – is steadily increasing and reveals several ways in which competing parties can respond to parties that are at risk of turning undemocratic (Bourne 2023; Heinze 2022). They include public criticism and ostracism on the one hand, and inclusion and the promise of collaboration on the other. Depending on the context, they create incentives for staying committed to democratic norms and principles.

The possibilities of political parties to hold others to democratic standards is further increased where formal relations exist between them, as is the case with party groups in the European Parliament. The saga of the Hungarian governing party FIDESZ and its relationship with its party group, the European People’s Party (EPP), in the European Parliament might illustrate this point, albeit from the perspective of missed opportunities rather than of successful maintenance of a party’s democratic nature. FIDESZ’s efforts to subvert the Hungarian democratic system and to replace it with one that imposes no checks and balances on the government and that has moved control of many key sectors of society, including the media, large parts of the economy and social and cultural institutions, to the ruling party and its sympathizers and their negative impact on the Hungarian constitutional system have been thoroughly analyzed. The European Union has several tools at its disposal that could be used to address this FIDESZ-led democratic backsliding, yet for a number of reasons (Kelemen 2020), the European Union has not acted particularly forcefully against Hungary. One reason for this is arguably the membership of FIDESZ of the EPP and the protection it affords to FIDESZ, through having, for example, voted against sanctions against Hungary. Yet the EPP has the tools to at least try to keep FIDESZ from undermining Hungary’s democracy, or to push FIDESZ to restoring democracy in Hungary, or at the very least not prevent the EU from bringing sanctions against Hungary. These tools are chiefly suspending FIDESZ’s membership and expelling FIDESZ. While the EPP did ultimately employ both these devices, it did so very late – almost ten years into the process of backsliding. Speculating on what the EPP might have been able to achieve in terms of hindering the FIDESZ led degradation of Hungarian democracy is a mug’s game, but the example shows an instance where a party group at the European level might have been able to rein in the anti-democratic tendencies of the ruling party (Kelemen 2020; Wolkenstein 2022a).

Thus, while the democratic system implements a system that heavily relies on political parties acting in a democratic manner, they are quite limited in what they can do to bring that situation about through constitutional means. While some political systems enact requirements for parties, such as committing to democratic values before they are allowed to register or receive funding, such requirements are rare and reflect the difficult tradeoff between an open party system that forms a key part of the democratic will formation necessary for a representative democracy and protecting the democratic system from parties that may abuse the democratic system to gain power and subvert the system.

III. Militant democracy and democratic freedoms

Broadly speaking, militant democracy ‘refers to the idea of a democratic regime which is willing to adopt pre-emptive, prima facie illiberal measures to prevent those aiming at subverting democracy with democratic means from destroying the democratic regime’ (Müller 2012: 1253). The term itself was coined by German émigré lawyer and political theorist Karl Loewenstein in two articles in the mid-1930s after the governments of several European states were taken over by autocrats who won elections (Loewenstein 1937a, 1937b). To be sure, measures that resembled those advocated by militant democrats pre-existed before Loewenstein’s account, albeit in more modest forms, but it was after World War II that his ideas found broader support.

These days, militant democracies come in many shapes and sizes. We cannot go through the deep ethical arguments about limiting democracy to protect it (for this, see Kirshner 2014; Rijpkema 2018), nor can we give a detailed account of all the ways in which militant democracy has been operationalized. Instead, we will try to show why the main tools of militant democracy – party bans and bans on hate speech – are insufficient in the face of contemporary NLDPs (see also Landau 2013: 192–93; Sajó 2019).

Party bans are probably the best-known tool of militant democracy, but in order not to interfere with the legitimate formation of the people’s democratic will, party bans, if not deployed in tightly defined circumstances (and according to some, even then), may lead to severe restrictions on democracy. A range of different models for operationalizing party bans exist, which political scientists and comparative constitutional lawyers have analysed extensively (Bourne 2012; Cappoccia 2013; Fox and Nolte 1995). For our purposes, it should suffice to note that the general problem faced by party bans in dealing with NLDPs tend to arise from these parties expertly hiding whatever anti-democratic impulses they may have.

We can use the German case to illustrate this point. The Basic Law institutes a system for banning political parties that ‘by reason of their aims or the behaviour of their adherents, seek to undermine or abolish the free democratic basic order or to endanger the existence’ of Germany (Article 21(1) of the Basic Law). The process for banning a party is closely circumscribed, and applications for a party ban can only be brought by political organs (Article 43(1) of the Bundesverfassungsgerichtsgesetz). The decision to ban a party lies with the Federal Constitutional Court (FCC). In a handful of decisions in the 1950s, the FCC created several criteria that it would consider when deciding whether a party was seeking to undermine the free, democratic order of the Federal Republic. Each of these criteria sets a high standard before a party can be banned as unconstitutional. The first relates to what is considered the democratic basic order, which includes respect for human rights, popular sovereignty, the separation of powers, free and fair elections, judicial independence, multi-party democracy and the rule of law. The party in question must seek to actively ‘undermine or undo’ the basic democratic order. Simply ignoring it is insufficient; rather, the party must take an openly hostile attitude towards it (BVerfGE 2,1; BVerfGE 5, 85). The severity of these restrictions to democratic rights posed by a political party ban is captured by such a narrowly tailored provision permitting the bans.

While such a restrictive approach to party bans is needed to protect the democratic rights of citizens and to protect against abuse by government against competitors, it means that it is an efficient tool against those parties that are explicitly against the fundamental basics of democratic government. In practice, it also means very few parties fall foul of its requirements. In Germany, the system has been used twice in the 1950s against the successor party of the Nazi Party, the Sozialistische Reichspartei (BVerfGE 2,1) and against the Communist Party (BVerfGE 5, 85). More recent attempts have failed because they were brought against entities that were either not parties (the Freiheitliche Deutsche Arbeiterpartei, a far-right organization that was banned by the Minister for the Interior, BVerfGE 91, 276) and more recently the Nationaldemokratische Partei Deutschlands (NDP), which was found to be too small to pose a threat to the democratic order and therefore should not be banned (BVerfGE 144, 20; Molier and Rijpkema 2018). There are systems that differ from the German version to some degree, but these too tend to ban relatively few parties (Bourne 2012).

A significant further concern is that parties that are banned can reconstitute themselves in a new form and pursue their agenda while at least superficially adhering to democratic values. The story of the Turkish Welfare Party illustrates this point well. The Turkish Constitutional Court banned the explicitly Islamist party only for the party to be reincarnated with a more secular appearance and ultimately as the current ruling party, AKP, which has effectively turned Turkey into an authoritarian state (Çalışkan 2018). The Turkish case illustrates a more general problem with party bans, namely that contemporary parties tend to be less explicitly opposed to democracy than the parties for which these provisions were designed. Rather, threats to the democratic form of government come from parties that have an ‘ambivalent’ or ‘ambiguous’ commitment to democracy (Daly and Jones 2020). These ‘far-right lite’ parties tend to operate on ‘partially detoxified’ platforms that keep them far below the threshold of the criteria for an unconstitutional political party. Although these parties may well pose as big a threat to democracy as anti-democratic parties, the rule of law considerations involved would make it impossible to apply the standards of militant democracy to them. Indeed, this short overview on party bans is primarily intended to demonstrate the limits faced by its tools when dealing with contemporary NLDPs. The point is not to deny their potential importance for protecting democracy, but rather to show the limits they face in the current political climate.

A similar point emerges about hate speech bans. Although bans on anti-democratic speech and speech that promotes hatred against certain groups of people is often part of militant democratic systems (Bhagwat and Weinstein 2021: C5-P65-6), the limits this poses to actors within NLDPs are questionable. First, because of the weight of political speech in a democratic society, the speech that can be captured by hate speech regulations is limited to instances that clearly meet the criteria for hate speech. This leaves plenty of space for NLDP politicians to spread similar messages through lawful means. Second, a prosecution for hate speech may increase support for the politician and their party, either directly or indirectly, for reasons that are still unknown (Jacobs and van Spanje 2020). As in the case of party bans, hate speech bans may perform an important part in removing some of the most egregious anti-democratic elements from the political system, yet because of the interests protected and the ability of politicians to circumvent these through legal means, they only provide a limited form of protection.

IV. Citizen vigilantism

The notion of ‘far-right lite’ parties (Daley and Jones 2020) points to the existence of parties that are not fully committed to the principles and institutions of liberal democracy without necessarily being outright against them. They may also not fully realize how their commitments differ from commitments to liberal democracy. As mentioned, here we call them non-liberal-democratic parties (as opposed to anti-liberal-democratic parties). In circumstances when the commitment to liberal democratic norms is incomplete, as is the case when NLDPs attract a large following and have gained government power or significant political influence, the main responsibility for keeping political institutions democratic lies with the public institutions themselves. They are tasked with acting on behalf of society and have the legal, political and economic means available to do so. Of course, it may be objected that if NLDPs have a large following, this implies a democratic majority preference for a different kind of democracy – perhaps an illiberal democracy – and the pursuit of such a responsibility would therefore be paternalistic and illegitimate. Space does not allow for a longer discussion of this point. We posit that liberal democracy – interpreted sufficiently abstractly as a set of principles that integrate public and private autonomy – is the most convincing conception of democracy available and that, for example, illiberal democracy implies the removal of important preconditions for the independent and unforced participation of all citizens, including the minority, in democratic decision-making, and that it is therefore unclear that it would be part of anyone’s democratic rights (powers) to fully remove those rights of others (Brettschneider 2009; Habermas 1996; Vinx 2020; Wolkenstein 2022b).

Nonetheless, there are still two challenges involved in leaving the protection of democracy to public institutions. First, it is problematic and likely too ineffective for public institutions to intervene, and there would be the suspicion that interventions are partial to those in power, as we argued in the preceding sections. Second, when NLDPs attain government power, it is less clear that public institutions even have the capability of intervening to the benefit of liberal democratic institutions and rights. Not least in such a situation, different actors would have a natural duty to seek to re-establish liberal democratic institutions, including political parties, international organizations and ordinary citizens (Olsen 2022, 2023). Indeed, it would be seen as part of citizens’ obligations to contribute to maintaining democracy by practising active citizenship directed against the sources of the republic’s corruption and decay (Rawls 1971; Simmons 2010).

Practising active citizenship is not uncommon. Social movements are formed to counter political parties that are considered problematic from a democratic perspective. To illustrate, recent examples include the many civil society initiatives against the populist Polish government, reactions against Alternative für Deutschland (AfD) and the Lega in Italy. In Poland, civil society organizations employing primarily legal and non-violent forms of action have joined forces with European-level actors (the EU Commission, European parties and members of the EP and European NGOs) to pressure the populist Polish government to undo changes to the legal system that have undermined the independence of Polish courts, and with it the security of rights of Polish citizens (Moroska-Bonkiewicz and Domagała 2023). Recently, these efforts have seemed to pay off, especially in connection with the leverage that the EU institutions have managed to generate in the context of the post COVID-19 recovery plan for Europe (NextGenerationEU), the EU budget and the so-called rule of law conditionality (Zalan 2022).

The Italian Sardines was a social movement that was formed precisely with the aim of reducing support for the populist Lega. It was deliberately anti-populist in its purpose and style and organized public protests to show strength and demonstrate an alternative to Lega’s populist online (SoMe) activity and rhetoric. The targets of the Sardines were not only the Lega. Their main ambition was to (re)activate the liberal democratic commitments of the (presumed) majority of citizens and to reinforce a liberal democratic political culture and thereby diminish the political and electoral success of the Lega. Although the Sardines might have seemed threatening to some due to their sheer numbers, as a physical force, their main means of influencing was through social pressure in the more or less overt ‘naming and shaming’ of those inclined to support the Lega as well as by trying to convince the general audience that the style, policies and wider political goals of the Lega went against the fundamental principles of the Italian republic. The Sardines were a rather clear example of an anti-populist movement established with the sole purpose of reducing the influence and electoral victories of the Lega. Succeeding in its immediate goal, the movement ceased its activities. The Sardines worked purely within the legal framework and their liberal democratic rights, such as the freedom of speech, assembly and association. On first impression, it is difficult to think that the Sardines and their activities would in any case be problematic, except if you find the supporters of Lega and other problematic NLDPs unjustly marginalized through their campaigns (Campo 2023). ¹

Other forms of active citizenship directed at parties are not quite as peaceful and civilized as that of the Sardines and Polish civil society. The reactions against the AfD in Germany have been varied; however, some of them entail the use of ‘intolerant means’ to pressure and to some extent intimidate AfD members and supporters to leave the political arena. Examples are the Berlin Antifa groups, which have applied physical pressure on potential conference venue owners/administrators to have them deny populist access to their facilities. Antifa groups also publicly display members of the AfD on posters. Other German more mainstream civil society actors use less drastic means, but similarly aim to pressure the AfD out of the political arena – for example, by creating legal barriers to the utilization of venues that carry a historical and democratic meaning in society. The Hambach Castle Foundation, administering the Hambach Castle used for political events, thus sought to exclude the AfD by including derogatory clauses in the leasing contract. Particularly after a few controversial statements from the AfD, which for some bring back memories of Germany’s totalitarian past, a key aim of some civil society actors is to completely marginalize the AfD politically (Laumond 2023).

The defence of democracy by active citizens in civil society has some immediate advantages compared with the militant approach, and more generally compared with the defence of democracy by public institutions. First, it does not face the problem of applying strict legal criteria to parties that are not openly anti-democratic. As discussed above, the ambiguous nature of many problematic parties’ commitment to liberal democratic principles makes it difficult to make party bans and other rights restrictions apply. Similarly, the approach avoids applying state coercion in the limitation of the freedom of association. State restrictions on the freedom of association is prima facie problematic from a liberal democratic point of view. The state should guarantee the autonomy of civil society, which arguably is where parties ultimately are rooted. The civil society defence of democracy is also less liable to being perceived as a partial intervention against some parties to the benefit of other parties – that is, those in power. Or at least it does not imply using government power to intervene to the benefit of specific parties or political positions. And it does not entail the same risk of abuse.

Positively, leaving the defence of democracy to active citizens has the advantage that they do not have to rely on ‘formal’ proof of clear anti-democratic convictions or intentions before they can react. This is different from official state institutions – not least courts – which must be able to document clearly without a doubt that a party is problematic from a democratic perspective. Of course, the objection here could be that civil society actors also would have to stick to clear criteria for judging whether specific actors are democratically problematic. They cannot just accuse parties wantonly for being anti-democratic. However, civil society actors can go after a hunch, call out prima facie problematic viewpoints, statements and policy proposals, and try to get behind public rhetoric and hypocrisy of parties. They are in a better position to ‘call the bluff’ (Tuovinen 2023).

Furthermore, due to their autonomy from the state civil society, organizations cannot easily be captured, controlled and brought under the command of central government actors in the same way that public institutions can when governments change. They remain in place when or if populist parties gain government power, either alone or as part of a coalition. Moreover, their members are generally able to establish new organizations in the unfortunate event that hostile government actors should have been able to bring their existing organizations under their control. When not over-stepping the boundaries of the rights of problematic parties and their supporters, civil society actors do not rely on coercion and/or violence; instead, their primary means are deliberation, persuasion and creating social pressure through naming and shaming, and through the strategic use of their own rights – for example, boycotts and strikes.

However, the civil society approach suffers from the risk of vigilantism where action is taken against a presumed ‘culprit’ based on an unreliable and partial procedure – or without any procedure at all. First, civil society does not have a democratic procedure to set out with authority which norms should apply for parties. The norms pertaining to a political culture often lack the precision that would unequivocally show when they are transgressed. Indeed, the content of such norms is part of what is disputed in politics. Still, the lack of common standards to which a critique can refer is problematic. Second, there is no commonly agreed, reliable and impartial procedure established for identifying breaches of these norms. In short, there is a lack of clarity regarding when exactly a party has transgressed norms and which procedure is reliable enough to decide whether those norms have in fact been violated.

This criticism of citizens’ vigilantism would seem to demand something very close to the procedures of liberal democratic institutions according to which the legislature sets the legal norms and the judiciary applies them to individual cases through reliable legal procedures. It would thus seem that if civil society should – so to speak – act as surveyor of the democratic nature of parties, then it would lose the nature of being made up of informal institutions. It would also mean that we would lose some of the benefits of the civil society approach to keeping parties democratic, listed above. In short, it would cease being an alternative to the state-based militant democracy approach.

Nonetheless, in the following we will propose a third way based on the civil society approach, which reinforces elements of democratic authorization and procedural reliability. This approach draws on the idea of employing CAs organized by civil society to set democratic standards for parties and to evaluate whether parties meet these standards. We have no direct experience with such a set up. Instead, we base our proposal on the existing knowledge of CAs and similar experiments where ‘ordinary citizens’ have been brought together to discuss policies and present their findings to the public. It gives us some idea of the model’s feasibility and likely effectiveness.

V. Citizens’ assemblies to keep parties democratic

As mentioned, in proposing that CAs play a key role in protecting democracy, we are adding to an emerging literature. Thus, Müller (2022: 10) briefly argues that CAs recruited by lottery could ‘play a constructive role in debating the line between democratic essentials and more or less contestable claims about justice’. The point is that a

shared view of essentials might make it easier, then, to perceive genuine attacks on them – and hence make democratic self-defence on the ground more effective, as it is not just based on elite deliberations, for example in relation to institutional design such as electoral districts, party funding and the number of MPs in parliament in relation to which parties themselves are not the most likely impartial decision makers. (Müller 2022: 10)

Müller does not think that CAs should be involved in deciding, for example, whether a party should be banned. And he worries that CAs might be captured by the state bureaucracy and/or strong economic interests if they get too involved in actual decisions about whether individual parties stray from democratic essentials and/or are serving longer terms (cf. Landa and Pevnick 2021).

Stone and Malkopoulou (2022) are less reluctant to give CAs a significant role in the defense of democracy. They point out that CAs would not be tied to specific parties and could not in the same manner as parliaments and/or executives, and even the courts, be instrumentalized by some parties (the government) to repress their opponents. They want to leave the question of how precisely to defend democracy to the CAs themselves in the spirit of respecting the latter’s democratic autonomy. However, they do discuss the pros and cons of three different models, which give democracy protecting CAs increasing powers.

The first model entails that CAs should be able to propose general measures or concrete actions to be undertaken by other bodies, such as the legislature and (constitutional) courts. They suggest, much in line with our proposal, that it ‘could even be convened by a civil society organization in an unofficial capacity’ (Stone and Malkopoulou 2022: 9). In the second intermediary model, the role of the CA is to ratify measures suggested by other bodies, parliaments or courts. In the strongest version, the role of the CA

would be to decide and introduce specific democratic self-defense measures. There should be no limit on the types of law and policies it could initiate, provided these are constitutional and justified as measures to defend democracy from threats posed by its internal enemies. Again, such a decision may seem quintessentially political, but given the democratic credentials of ACs, fewer questions would be raised by such a body taking over such decision-making power than a judge. (Stone and Malkopoulou 2022: 10)

Stone and Malkopoulou (2022: 10) seem to prefer the more moderate role for the CA because the strong version entails the risk that other powerful political and economic actors will try to influence its work (cf. Landa and Pevnick 2021).

Our proposal has considerable overlaps with both Müller’s (2022) and Stone and Malkopoulou’s (2022) ideas but is nonetheless distinct. Contrary to their proposal, we want to place the democracy defending CA squarely in civil society and limit its modus operandi to the use of the word – that is, the reiteration of democratic standards for parties and the evaluation and expression of criticism of specific parties. It does not entail its involvement in the initiation, ratification and execution of other legal, cultural, social and political measures. Thus it also places itself midway, as it were, between the state militant democracy and citizens’ vigilantism. First, in relation to the legalistic approach taken by traditional militant democracy, the model is based on the social and political pressure that comes from formulating public criticism. As discussed above, the legalistic approach seems too rigid. The criteria of validity of a legal decision are more demanding than a public criticism, which works poorly in conjunction with parties that are likely not to be unequivocally anti-democratic. Still, for the criticism to have maximum effect, it is important that the body that specifies the standards and applies them to individual cases is considered impartial. Impartiality is considered central to the effectiveness of ‘naming and shaming’ mechanisms (e.g. Sedelmeier 2017; van Erp 2021). To ensure impartiality – also at the level of public perception – and to avert the risk of capture of government institutions, we propose that the CAs should be organized and ideally funded by civil society itself through voluntary contributions or fees, which the CAs can collect themselves. Especially if NLDPs gain power and start working on problematic changes to the political institutions to secure their power beyond the next election, the CA will be difficult (but not impossible) to reach and if eventually it is ‘reached for’, the bad intentions of the NLDP government will be clear. There can never be full certainty that politicians will not try to interfere with CAs of this nature.

Thus, in addition to organizational autonomy as part of civil society, our ideal CA model has the following features:

1. 1. The members of the CA should be randomly selected by using a lottery. This secures that it consists of a representative sample of the population. ² The lottery should be stratified so that it is possible to take into account that people belonging to specific groups might have a higher-than-average tendency to decline participation. It is important that the CA is not skewed towards participation of those with more time and resources on their hands.

2. 2. It should elaborate the general norms that parties should follow. Studies suggest that recommendations of CAs have higher impact if there is an approximated consensus on their recommendations (Gastil et al. 2016; Lafont 2017). This suggests that decisions on standards (first formulations and subsequent changes/adjustments) should be based on supermajorities. The internal workings should be based on alternations between discussions in smaller groups and in plenums. Here it is important that all participants experience efficacy as one source of motivation (Fishkin et al. 2021). The CA’s work should be assisted by a secretariat, which can prepare information material for developing standards and it should be able to call in people with expertise on the issue – for example, experts on constitutional law and democratic theory. The use of experts should, however, be modest in the sense that is important that the CA is not, and is not seen as, the mouthpiece of some (academic) elite having a specific politico-cultural profile. Studies reveal that CAs and mini-publics are not necessarily controlled by professionals, experts and/or external actors, but are able to ‘decide for themselves’ (Fournier 2011: Ch 6).

3. 3. The CA should evaluate whether all parties meet the standards in their statements, policies and actions in the country within each year of the election period. Practical considerations suggest that members are subdivided into smaller groups of jury size (perhaps twelve or thirteen people) when they decide on evaluation of specific parties. In a sense, this would mirror the ‘separation of powers’ between the legislature and the judiciary and the distinction between a discourse of justification and a discourse of application (Habermas 1996: Ch 6). Again, research might speak in favor of supermajorities for deciding on specific verdicts. The deliberations might have to be secret while ongoing, but the decisions should be based on clearly stated reasons.

As mentioned, the proposal combines elements from both Müller (2022) and Stone and Malkopoulou (2022). We will now in turn address its general justifiability, its feasibility and its likely effectiveness.

In terms of justifiability, the model is fully consistent with democratic principles. If CAs were legislating, issues concerning lack of responsiveness and accountability would arise. It is difficult for citizens to mobilize against CAs selected on the basis of sortition, and sortition involves problems of rendering CAs accountable (Abizadeh 2021: 798). However, if sortition secures proportional representation of all groups in society, and if its deliberations are public and allow people to follow its reasoning (Abizadeh 2021; Lafont 2017), then there are no democratic problems with this kind of assembly. It does not legislate or produce legal verdicts. Instead, it exercises its power through the social pressure that is created from the formulation of standards and the evaluation of parties based on those standards. Moreover, its comparative legitimacy is stronger than that of other institutions (Waldron 2005). It is independent from public authorities. It deals with issues pertaining to the democratic and political system (and not specific policies) and thus issues on which neither government nor political parties are in a good position to produce impartial views. In fact, CAs have been used to deliberate on the reform of electoral systems in Ireland and the future of democracy in the United Kingdom (Joint Committee on the Constitution 2010; Renwick et al. 2022).

Existing research on CAs, mini publics and deliberative polls helps us to estimate our model’s feasibility and likely effectiveness. Feasibility relates to the question of whether it is possible to garner support for CAs of this nature and get them to function properly when established (cf. Gilabert and Lawford-Smith 2012). Effectiveness relates to whether CAs will influence parties, and more generally society, to stay democratic.

Feasibility discussions always involve some level of conjecture. One important question is whether it is likely that there is any civil society actor who would adopt the idea to establish such a CA. If they were such a good idea, one might expect that they would already exist. While this is true, one could cite the push that social movements and NGOs have made for more civil society participation in politics generally – for example, through the institutionalization of ‘citizen initiatives’ at both the national and the European levels (e.g. Christensen 2019; Longo 2019; Schiller and Setälä 2012). One might also refer to the increased use of CAs, mini-publics and deliberative polls in general (Fishkin et al. 2021). Also in favour of the feasibility of CAs of the nature we propose is the ubiquity of the support for the democratic principle of legitimacy. This support makes it difficult for parties and citizens (civil society actors) to argue against a body put in place to watch over democracy. One factor that that could work against it is that many parties are likely to be less interested in being monitored by an external body. They might argue that such a body represents undue interference with democratic liberties or that it amounts to censorship. Still, the fact that the assembly will be based in civil society and not created as a government body, and that it does not govern or pass verdicts based on coercion, is likely to make this criticism less convincing. There are several cases in which CAs or mini-publics have been supported by parties and legislators to work on topics relating to the democratic system (for a recent review, see Farrell and Stone 2020). Thus, it is likely that the CAs we propose would be able to obtain support from both political parties and civil society.

Of course, in contexts where NLDPs have gained a large following and there is a strong support for non-liberal-democratic regimes, it is unlikely that there will be support for the creation of CAs that monitor the democratic nature of parties. Thus, here the assumption is that the creation of CAs takes place under fairly ideal conditions – that is, the very conditions that CAs are supposed to protect. CAs should not be seen as miraculous instruments that independently would be able to reverse and/or re-establish democracy under non-ideal conditions characterized by a very low level of commitment to liberal democratic norms and principles by a large section of society.

The next question regarding feasibility is whether the model could function reasonably well if established. There are four important issues here. The first is that many might refuse to serve in the CA because it conflicts with their other obligations and interests. Some people might feel over-burdened by becoming a member of a CA. This means that there should be sufficient resources available to remunerate members of CAs and that their work should be organized in such a way that it can be combined with most jobs. It is important that it is not a particular type of person who does not join and that recruitment is structured so the CA becomes statistically representative of the population. This also means that it would be wrong to think that CAs are inexpensive. The costs of arranging shorter-term CAs have, in the United States amounted to US$1–2 million – a considerable amount. In addition, there will be expenses for staff and communication. Still, any developed society would be able to bear the costs.

A second concern would be that on average the skills of the people who are selected for CAs are not good enough for formulating standards and for carrying out reviews. One aspect of this is that for people to have confidence in CAs, they will – on average – need to have confidence that the members of the CAs know what they are doing. This confidence will be important for the legitimacy and the effectiveness of any criticism that the CA may direct against concrete parties.

The third important feasibility issue is that people with ambivalent commitment to democratic principles, who may or may not identify with/support NLDPs, will become members of CAs. This could mean that they would perceive the role of the CAs as illegitimate and rather use the CA as platform to undermine its purpose. A fourth problem might be that the CAs could not agree on any evaluation guidelines – or, for that matter, on passing any judgement (produce a coherent evaluation) on specific parties. The worst-case scenario is that CAs become fundamentally polarized and not able to function because they can neither produce guidelines nor reach any decisions.

Obviously, there is no guarantee that CAs will work according to their purpose, and there might be unexpected long-term dynamics of having CAs as part of democracy’s institutional landscape. However, experience with recruiting to CAs and similar events suggests that it is possible to recruit a statistically representative sample of people. Experience also speaks in favour of members of CAs being able to produce outputs of a rather complex nature – for example, proposals for referendums and policy recommendations (see Farrell and Stone 2020). With regard to the ability to reach agreements, available studies tell us that deliberation among people with strong initial disagreements tends to produce depolarization – either two-sided or one-sided – among participants and is thus promising when it comes to reducing polarization (Fishkin et al. 2021: 1478). Similarly, there is a good chance that being selected for the CA would help create a common identity among CA members that would lessen their original ties to different groups outside the CA (Myers 2022). To this should be added that it is only coincidental if CA members chosen by sortition have strong organizational ties to NLDPs. Further, it would seem legitimate to introduce a rule that members that publicly deny the purpose of the CAs can be excluded from them. However, prudence might speak against exclusion. Exclusion can easily be framed to the detriment of the public perception of the CAs. With regard to ostensive anti-democrats, it is unclear why they would join the CAs in the first place.

The effectiveness of CAs will depend on their recognition by all parties as institutions, which ‘quality seal’ parties should obtain to be considered democratically legitimate. ³ Part of the mechanism of naming and shaming is that actors – even those who might not believe in certain principles and values – do not think that they can afford being seen not living up to them (Risse, Sikkink and Ropp 1999). So the model relies partly on the civilizing force of hypocrisy (Elster 1995). It does not require full conviction about the stipulated democratic standards among all parties, but a relatively high degree of acceptance is necessary to guarantee the effectiveness of the CA in the political sphere.

A necessary precondition for the effectiveness of a model based on naming and shaming is that the CA becomes a ‘center of attention’ for media and public debate. Some research suggests that CAs or mini-publics suffer from invisibility – that is, that few people are aware of their existence and their work (Curato, Vrydagh and Bächtiger 2020). This means that the CA would have to put serious effort into becoming known to the wider public, preferably making sure its members participate in relevant debates in society. In the end, the visibility of the CA will rely on the editorial decisions of different media outlets and citizens’ interest in relating to its work, both of which can never be guaranteed. However, the increased possibilities for online communication and sharing of information make CAs less dependent on specific media outlets and increasingly able to communicate to the general audience directly, especially if they are assisted by trained media staff. Given substantial funding, the CA could even disseminate its evaluations to households directly.

An additional aspect to being effective is that citizens find the standards of CAs relevant to the evaluation of parties and do not see them as imposed on the political community by some extraneous body. Some authors worry that CAs continue to be detached from the citizenry and that citizens do not see their various contributions to political processes and decisions as valid and as applying to them if they have not been the topic of a wider public debate to which CAs are responsive. CAs should not become ‘deliberating experts’ in an institutional setup that somehow presupposes ‘blind deference’ by citizens to their conclusions and that denies the importance of citizen participation in democracy (Lafont 2019; Pateman 2012). These worries emphasize the connection between effectiveness and legitimacy, and the significance of securing the publicity of CAs’ work. The ability of citizens to discuss and provide input into the work of the CAs will be of some importance. Existing empirical studies may lead to scepticism about the abilities of CAs to engage with citizens in public debate and they serve as warning against undue optimism. However, they do not identify insurmountable constraints against the possible success of CAs (cf. Gilabert and Lawford-Smith 2012). Moreover, the CA in our model should not be seen as the solution to comprehensive systemic problems of current representative democracy such as insufficient inclusion, lack of high-quality public deliberation and active participation by citizens. The CA is supposed to play a limited function as an impartial evaluator some steps removed from the political process. It is only indirectly involved in making political decisions and the process of governing. It might therefore make do with less in terms of citizens’ attention and engagement than if it were to contribute to the solution of these comprehensive problems.

Of further importance to effectiveness is the (perceived) impartiality of CAs’ work, their standard-setting and individual evaluations of parties. One formal precondition for this might be that the CAs do not regularly change the standards of evaluation but maintain the same standards over longer periods of time. This speaks in favour of reviewing standards with set intervals of, say, ten years. This is supported by the idea of introducing the CAs in the first place, namely that it should reduce arbitrariness in evaluating whether parties meet liberal democratic requirements.

It is even more important that CAs remain independent from government and state institutions in a publicly transparent way. They should avoid becoming, or being perceived as, the instruments of political parties in power. This also applies to parties more generally. The CAs cannot be seen as an ally of specific parties. Such a reality or perception may become a fact if close attention is not paid to how, for example, administrative staff are recruited. The CAs are likely to attract people who have a special interest in politics, and thus people with party memberships. Although problematic from the viewpoint of discrimination based on political opinion, a rule might be that you cannot be employed if you have any party membership.

In the same vein, it is a problem for the effectiveness of CAs if they are perceived as illegitimate by segments of society, not least NLDPs and their supporters (cf. van Erp 2021). Again, it does not seem relevant to worry about ostensive anti-democratic parties since they clearly would be unconcerned about being labelled as non-democratic. However, as mentioned above, legitimacy is important to secure the effectiveness of a naming and shaming mechanism as few people probably feel ashamed of being criticized by a body that they do not feel is legitimate. A critical aspect of this is that in situations in which, say, populists have gained the majority, there is a likelihood that there is also a majority of the members of the CA that will support a populist party and it is therefore unlikely that the populist party that does have a problematic democratic practice will ever be criticized by the CA. This latter problem points to the importance of timing, and perhaps to some of the contradictions that we also find regarding the creation of other forms of protection of democracy, namely that if you need them, you cannot get them because there is too much conflict and support for non-democratic actors and regime forms, and if you can get them because there is a widespread support for democracy, then you don’t really need them (Müller 2016: 253). Admittedly, there are cases where things will be beyond repair. However, it is also important to underline that deliberation in CAs might have an independent effect, especially because many people see themselves as democrats. Deliberation may structurally push CA members towards producing more impartial views and adopting commitment to the norms of democratic deliberation, including respect for disagreement and pluralism (Button and Garrett 2016; Lindell 2023; Mercier and Landemore 2012; Muradova 2021; Setälä and Herne 2014). ⁴ One study of citizens with populist sympathies who participated in CAs concludes that they seem to be satisfied with the results (Jacobs 2023). Moreover, it should not be taken for granted that people support populist parties for their democratically problematic reforms of political institutions such as those we have seen in Poland and Hungary. Their support may be based on certain substantive policies on, say, economic conditions for families or farmers.

We do not have many concrete studies of the potential support for CAs among people who are likely to support democratically problematic parties. However, if we take populists as an important current example and accept as a premise that populists are relatively easy to identify and categorize, then the relevant results are a little mixed. One study demonstrates that populists’ support of democratic institutions depends on the kind of populism to which they are committed. Populists are generally wary of representative democracy and tend to be anti-pluralists. However, while right-wing populists are in favor majoritarianism, left wing populists favor deliberation (Heinisch and Wegscheider 2020). Obviously, the perceived legitimacy of the CA will to some extent depend on how it is perceived as an institution by parties such as populists. One study suggests that populist parties tend to relate opportunistically to CAs and their outcomes. Their evaluation depends on whether they agree with them or not (Jacobs 2023). On the other hand, given its anchorage in civil society rather than in formal political and legal institutions, it has a higher likelihood of being considered as legitimate than any state institution. In addition, the fact that the CAs consist of a ‘representative sample’ of the population rather than ‘an elite’ of trained experts will confer legitimacy on it in the eyes of anti-elite/establishment parties and their supporters, especially if efforts are made to make it transparent that recruitment is by lot and is not dependent on the political, economic and social connections and status of members, as is often the case with elected representatives. CAs generally receive more support by people who are dissatisfied with the political system as it is and scepticism towards CAs decline with people who have more knowledge how they are recruited and how they are internally organized (Talukder and Pilet 2023). Moreover, in the case of populists, it is important to note that they often, or generally, consider themselves to be democrats (Zaslove et al. 2021), and therefore identify with the very values that the CA in our model is meant to protect. More broadly, the experiences with CAs seem to be that their work is generally taken seriously and thus can have an impact on the surrounding society, even if their recommendations and concrete proposals are not always followed (Farrell and Stone 2020; cf. Goodin and Dryzek 2006).

One question that our discussion could raise is why we do not want to rely on more traditional actors such as the media, think-thanks, ombudsmen or independent public agencies (similar to human rights agencies or anti-discrimination bodies). Here it is important to note that our argument is not that actors or institutions of this more traditional kind should not have any role in keeping parties democratic. Our argument is rather that CAs can avoid certain disadvantages of both militant democracy and the civil society approach while combining some of their advantages. That said, CAs also have some comparative advantages over more traditional actors and institutions.

Roughly speaking, the media and think thanks are civil society organizations, which to different degrees can scrutinize and critique parties for lack of democratic commitment. Many media outlets are constrained by publicist norms. Similarly, many think thanks are constrained by epistemic (or scientific) norms. However, none has a clear impartial stance and they do not necessarily work on the basis of a set of norms for the evaluation of parties and their democratic commitments that have been validated by a representative institution. Of course, some media outlets and think thanks quite explicitly endorse rather particular views of society.

Ombudsmen might be organized by civil society, but they are public institutions as often as they are independent agencies. Even if they do not necessarily produce legal verdicts, but rather express criticisms on the basis of which other actors may or may not act, ombudsmen and independent agencies are in many ways analogous to the courts and will often have to live up to the same legal standards when they issue statements and make decisions. Similarly, like the courts, they are liable to be perceived as (and become) the instruments of the government.

Another concern could be that our CA model lacks teeth. As discussed above, the effectiveness of the model relies on certain preconditions, which will not be present in all contexts. The CAs might be strengthened to the extent that they are in regular contact with one another across national boundaries (if we imagine that it becomes a generally adopted model) and build bridges to international or supranational organizations. Both things might help to garner international attention and bring in the international community if important parties, in particular governing parties, are turning away from democracy. However, civil society organizations cannot be equipped with more powerful means without ceasing to be civil society organizations. Moreover, as explained above, the risk of abuse entailed in militant democratic measures as well as the nature of the problem at hand (democratically ambiguous parties vs explicitly anti-democratic parties) speaks against full-on militancy.

VI. Conclusion

Contributing to recent literature on the role of citizens assemblies in keeping parties democratic, we have argued for a model placed midway between the legalistic militant democratic model and the much less formal vigilantism of active citizens in civil society. The model seeks to avoid the drawbacks of both. The legalistic militant democratic model suffers from rigid legal criteria for banning and/or excluding parties, perceived partiality on the side of public authorities controlled by governing parties and the risk of abuse. Civil society vigilantism similarly suffers from arbitrariness and the lack of clear criteria of evaluation of the parties that become the targets of civil society actors. The strength of the militant democratic approach is its efforts – in the best cases – to develop clear criteria of evaluation and sanction. The strength of the civil society approach is its increased flexibility and – in the best cases – the absence of the employment of violence/coercion in an attempt to keep parties democratic. CAs anchored in civil society, producing and iterating democratic standards for political parties to follow and expressing criticism by way of yearly evaluations of all parties, will place political parties who profess democratic values under pressure to stay democratic. Its criticism will have a high level of legitimacy, cannot be suspected of being under government control, will remain in place if government should be overtaken by NLDPs and seeks to achieve compliance with liberal democratic principles and norms only by way of the word. The proposed model does not necessarily conflict with other models, which have a more comprehensive catalogue of measures to be employed in the defence of democracy and might be seen as a part of such models or complementary to them.