The Derailed Promise of a Participatory Minipublic: The Citizens’ Assembly Bill in Flanders

This article shows how the principles of participatory deliberative democracy can serve as a guide for the institutional design of minipublics, while also discussing the obstacles such proposals are likely to face in becoming realised in practice. It does so by discussing the case of citizen-initiated citizens’ assemblies in Flanders, Belgium. This case represents an ambitious proposal that combined elements of petition, deliberation, public consultation and parliamentary deliberation to generate a robust deliberative system. Yet in the end it was soundly defeated in parliament. By studying the institutional specifics of this proposal as well as the macro-deliberative circumstances that led to its failure, this article presents a nuanced picture of the promises and pitfalls of institutionalising deliberative minipublics. It concludes with a call to ‘deliberative activism’.


Introduction
Under what conditions can deliberative minipublics empower the macro-public of citzens? This is one of the issues that political theorist Cristina Lafont tackles in her book Democracy Without Shortcuts. Deliberative minipublics aim to create a space where a diverse group of randomly selected citizens can reason together about an issue of public concern under circumstances specifically designed to enable high-quality deliberation (Goodin & Dryzek 2006;Smith & Setälä 2018: 300). In the past few years, the literature on minipublics has moved on from studying them in isolation to discuss their role in the wider deliberative system (Mansbridge & Parkinson 2012). Some make a case for giving minipublics binding decision-making powers (Guerrero 2014;Landemore 2013). Others, however, find this approach suspicious, as empowering minipublics to make decisions on behalf of others implies that the vast majority of citizens will have little choice but to passively accept the minipublic's judgement. Consequently, political theorists such as Lafont, Marit Hammond, Simone Chambers and John Parkinson prefer a participatory conception of deliberative democracy which eschews the minipublic shortcut, and instead emphasises the transformation of actual public opinion as a non-negotiable prerequisite for democratic self-government (Chambers 2009;Hammond 2019;Lafont 2019;Parkinson 2006). This participatory stance does not mean that minipublics have no role to play in the deliberative system. Indeed, one of the important insights provided by Lafont's new book is precisely that, under the right conditions, minipublics can become a resource for citizens which empowers them and aids them in their democratic tasks, rather than a shortcut which silences them.
This article contributes to this debate by providing an empirical case of a deliberative minipublic that was characterised by a strong participatory dimension, from its formation to its conclusion. I present the case of the citizeninitiated citizens' assemblies in Flanders -an ambitious proposal that combined elements of petition, deliberation, public consultation and parliamentary deliberation, among others, to generate a robust deliberative system. I describe the ways in which this minipublic's design brought Lafont's prescriptions on the roles of minipublics to life through key design principles. I conclude, however, on a more sceptical note. In the end, this proposal was soundly defeated in parliament, even though a large majority of Flemish political parties had expressed support for the idea of a minipublic in some shape or form. I argue that the gaping distance between the proposal's ambitions and its ultimate fate in parliament shows that, while participatory principles can provide a consistent guide when designing minipublics, the institutionalisation of such designs will face formidable obstacles in the real world. Overcoming these obstacles will require a broad coalition of academics, activists and citizens committed to seeing these principles realised in practice. The transformation of actual public opinion is thus not only the end-goal for a participatory approach to deliberative minipublics, but one of its preconditions as well.
against it (Böker 2016;Lafont 2015;Pateman 2012). In response to these critiques, some have argued not only that minipublics should be given a permanent place within the larger deliberative system, but that they should be given a clear mandate to make binding decisions, either as the sole decision-making authority (Guerrero 2014;Landemore 2013;Van Reybrouck 2016) or through a bicameral solution combining sortition and election (Gastil & Wright 2019).
This view, however, is contested in deliberative theory. Lafont put forward arguably one of the most powerful critiques against granting decision-making powers to minipublics. The crux of Lafont's argument is that empowering deliberative minipublics to decide public policy by themselves would force non-participants to blindly defer to the judgement of the minipublic. It is supposed to represent 'what they would have decided if they had deliberated', even though from the perspective of an individual citizen there is no reason to assume that this is the case. Since minipublics tend to use some form of non-unanimous decision-making, there is simply no way to know whether a non-participant would have ended up in the majority or in the minority after deliberating, since both majority and minority went through the same learning process and can therefore claim to represent what the public 'would think' after deliberating (Lafont 2019(Lafont , 2015. Yet arguments for an empowered use of minipublics dismiss this potential disconnect between the broader public and the minipublic by presenting the latter as the embodiment of the ' considered will of the people'. 2 Meanwhile, the idea that people might have good reasons not to agree with the majority of a minipublic is reduced to ignorance: they did not think through the issue as the participants in the minipublic did and should simply blindly defer to the judgement of 'their better selves'. Rather than a viable means of achieving democratic selfrule, an empowered minipublic could act as a dominating force hiding behind a democratic visage -the ultimate usurper, as Lafont puts it (2019: 118).
This posits a challenge for institutional design. Can minipublics be institutionalised in such a way that their design simultaneously addresses the weaknesses of a purely ad hoc use of minipublics, and takes into account the normative arguments against fully empowering them to make binding collective decisions? Can minipublics have some form of empowerment so that they are not irrelevant, but also be strongly connected to the wider public so as to live up to deliberative democracy's participatory ethos? I argue that the citizen-initiated citizens' assembly in Flanders addressed many of these dilemmas. Before I discuss the design in greater detail, however, it is important to be clear about what exactly the institutionalisation of minipublics entails (for a different, but somewhat overlapping set of criteria, see Courant 2018). The criteria I propose are threefold.

1) Minipublics should be permanent
In order to achieve the status of institution, rather than being seen as an innovation to deal with extraordinary circumstances, deliberative minipublics would have to become part of the 'normal' way of doing politics. A straightforward way to address the minipublic's ad hoc and top-down character is to take the decision to organise a deliberative minipublic (at least partly) out of the hands of government by specifying the situations where minipublics are to be used (Setälä 2017: 853). Examples might include institutionalising minipublics as a regular part of inquiry procedures in parliamentary committees, establishing a permanent minipublic which can set its own agenda, oras in the case of the Flemish citizens' assembly -creating the possibility for citizens to demand that a minipublic be held on a particular topic (OECD 2020).
Of course, this does not mean that going forward every minipublic needs to have such an explicit legal basis. There will continue to be a legitimate role for ad hoc minipublics. It is also possible for minipublics to achieve the status of institution without any such legal basis whatsoever. This situation could emerge when the initial (governmental) decision to organise an ad hoc minipublic creates a precedent for the continued use of these forms of public consultation in the future. To the extent that this precedent effectively constrains the freedom of governments when deciding whether or not to organise a minipublic -for example because the public now expects important constitutional reforms to be prepared by a citizens' assembly -the effect might be very similar to a legal requirement (March & Olsen 2008). This presents a different road to institutionalisation where deliberative democrats capitalise on political opportunities to establish a ' deliberative foothold', in the hope that this will in turn create future opportunities for an even greater use of deliberative minipublics.

2) Minipublics' design should be relatively stable
Because of their ad hoc nature, the design of deliberative minipublics often varies from case to case. While this allows for experimentation and the discovery of new best practices, at the same time it undermines the legitimacy and status of deliberative minipublics from the perspective of the wider public. As Lafont argues, in order for citizens to be able to place their trust in deliberative minipublics, they first need to have a readily available understanding of what a deliberative minipublic is, how it functions, and what is to be expected of its outcomewhich is impossible if their design is subject to constant change (Lafont 2019: 139;Niemeyer & Jennstal 2018: 340). At the same time, this lack of stability provides room for governments to manipulate the conditions under which deliberation takes place, and to actively steer the process in a direction that is favourable to its own positions (Curato & Böker 2015). I do not doubt that most governments will organise a deliberative minipublic in good faith, but as long as they retain their ad hoc nature, deliberative minipublics cannot guard against these potential anti-emancipatory tendencies (Böker 2016: 12). Stable procedural rules are therefore essential to ensure the deliberative quality of minipublics over time. Still, the requirement of stability should not be interpreted too strictly. Even institutionalised minipublics should retain some flexibility to allow the process to adapt to changing circumstances. Yet a number of minimal requirements should remain fixed so as to ensure the overall integrity of the minipublic as an institution.

3) Minipublics should be autonomous
For minipublics to make the transition from innovation to institution, they should be able to assert a degree of autonomy vis-à-vis other institutions. Deliberative minipublics should be given enough powers so that the participants remain in control of their own deliberation and the process is protected from being co-opted by its organisers. Possible means to achieve this aim could include the right of participants to invite additional experts or to expand the agenda of the minipublic if they deem this necessary. Without such options, it would be all too easy for organisers to stonewall deliberations, forcing citizens to discuss some unimportant side-issue while the really pressing issues remain beyond their grasp. Under such conditions, public deliberation becomes little more than a PR operation intended to create the illusion of democratic engagement -or what Bächtiger and Parkinson have called 'wallpaper democracy' (2019: 82).
However, the fact that institutionalised minipublics need to achieve a degree of autonomy vis-à-vis other state institutions should not be taken to mean that they should assert this autonomy against the broader public (via the so-called shortcut approach). The requirement for autonomy concerns the internal workings of the minipublic and is a characteristic of institutionalisation, without which a minipublic would simply be an add-on to existing institutions. The relation of the minipublic to the wider public is a different question altogether and concerns the position of the minipublic within the deliberative system. Some designs for institutionalised minipublics stress the independence of minipublics from the broader public sphere. Philip Pettit's ' depoliticisation' strategy is a prime example, since under his proposal it is precisely by isolating the minipublic from the idiosyncrasies of the public sphere that minipublics are presumed to be able to deal with morally contentious issues such as criminal law reform (Pettit 2004). However, it is also possible to design a minipublic so that it is open to input from the wider public, while at the same time shielding its deliberations from illegitimate interference by other state institutions. The proposed citizens' assemblies in Flanders offer a good example of how this balance could be maintained.

Citizen-initiated Citizens' Assemblies in Flanders
Let us now turn to the Flemish case in some more detail. In 2014 the Belgian autonomous region of Flanders received a number of new powers to shape its own political decisionmaking process (as did the other federated entities in Belgium). This led the Flemish parliament to undertake a broad-ranging debate on democratic renewal which took place over the course of 2017 and the beginning of 2018. As part of this debate, the Green Party introduced a Bill which sought to make it possible for 80,000 citizens to demand a citizens' assembly on any topic belonging to the competencies of Flanders (hereafter called the Citizens' Assembly Bill). 3

Grassroots agenda-setting
Robert Dahl considers ' control of the agenda' as one of the defining criteria of democracy. He argues that 'the members must have the exclusive opportunity to decide how, and if they so choose, what matters are to be placed on the agenda' (Dahl & Shapiro 2015: 38). A healthy deliberative system should therefore include processes that enable citizens to actively raise claims for political action from the bottom up (Bachtiger & Parkinson 2019: 100). The Citizens' Assembly Bill's proposed form of citizen-initiated citizens' assemblies would have realised this aim by combining two political instruments, the agenda initiative and the deliberative minipublic.
The agenda initiative amounts to a collective right of petition. By gathering a specified number of signatures (80,000 in the Flemish case), citizens can require parliament to consider a proposal. The agenda initiative contributes to the deliberative system by encouraging the active involvement of civil society, and -at least in theory -promoting the responsiveness of empowered decisionmaking institutions towards discourses circulating in the public sphere (Setälä & Schiller 2012: 2, citing Dryzek 2000. However, petitions alone do not create room for citizens to actually deliberate on these proposals. Once an initiative is successful, it is generally up to parliament to deliberate and formulate a response. The next step in the citizen-initiated citizens' assembly addresses this weakness. It combines the sense of agency and citizen action embodied in the initiative process, something that is lacking in sortition (Abizadeh 2019), with the deliberative qualities of the minipublic. During the preliminary phase of signature gathering, the citizeninitiated citizens' assembly encourages civil society actors and social movements to mobilise citizens in support of their proposals, thus contributing to a lively public sphere. In the second phase of the process, before these proposals are formally submitted to parliament, they are subject to the scrutiny of a broadly representative sample of citizens, which has the power to amend them.
An important caveat to make with regard to this proposal is that, while the citizen-initiated citizens' assembly as a whole would serve an agenda-setting role by transmitting new issues, ideas and recommendations to parliament, the original initiative comes from outside the citizens' assembly. Every successful petition would result in a new citizens' assembly being convened (with new members), whose agenda would be limited to that specific petition. In this regard, a citizen-initiated citizens' assembly would operate very differently from other proposals which seek to endow deliberative minipublics with an autonomous agenda-setting power independent of the wider public, such as in the case of most proposals for a randomly selected legislative chamber (Gastil & Wright 2019).
There are several reasons why this connection to the wider public through the instrument of petition is preferable to an autonomous agenda-setting role for the citizens' assembly. Fundamentally, this institutional set-up offers a participatory approach to deliberative agenda-setting. Here, Cristina Lafont's work is instructive. She reminds us that while minipublics can play a useful role in a democratic system, one should take care not to conflate the random sample of citizens which constitutes a minipublic with the public as a whole (Lafont 2019). Even the best selection methods cannot guarantee that a random sample will include (equally) all the ideas and perspectives that make up society. Moreover, as previously mentioned, it is impossible to ascertain whether an individual citizen would share the majority judgement of the minipublic, so that individual citizens cannot be expected to blindly defer to the minipublic's decision. This is as true when deciding which issues require political action most urgently as it is for agreeing on substantive policy recommendations. In other words, public agendasetting cannot be delegated in its entirety to a deliberative minipublic without losing out on the value of selfgovernment which is central to a participatory conception of deliberative democracy.
The Citizens' Assembly Bill addressed these fundamental democratic issues by creating a right for every citizen to gather signatures and place their issues on the assembly's agenda. To be sure, whether citizens -individually or (more likely) in cooperation with others -would be able to make use of this right will largely depend on the macro-deliberative qualities of the public sphere and how inclusive it is. At face value, however, the requirement of 80,000 signatures does not seem excessively high in the Flemish context. 4 Additionally, the Bill ensured that the assembly would remain open to the wider public and be 'sensitive to the interests, values and ideas of the citizenry' (Lafont 2019: 103) by requiring that the assembly's deliberations are preceded by an extensive public consultation. To this end, the proposal suggested that every citizens' assembly would be accompanied by an online discussion platform open to all Flemish citizens. The results of this online discussion would be summarised and presented to the citizens' assembly at the outset of its proceedings.

Discursive filters
Of course, simply because citizens put forward a petition, or raise new issues through the consultation procedure, does not mean that these ideas should be treated as the law of the land without further scrutiny. What is needed then is an additional filter, one that is not based on numbers or signatures but allows the widest possible range of perspectives and viewpoints to be represented, before these are structured and narrowed down in a discursive process. It is precisely this role which the citizen-initiated citizens' assembly would be able to fulfil. It would act as something of a ' discursive filter' by passing calls for political action through a ' deliberative sluice'.
The gathering of signatures forms a first test, where the initiative's sponsors need to be able to show that their proposal addresses a felt need among a significant part of the citizenry. This is important since a deliberative system's capacity to respond to these calls for action is necessarily limited in terms of time as well as the financial means available. There is thus a need for a procedure through which we can distinguish between those claims, ideas and proposals that enjoy a degree of support in society that warrants that they be taken seriously by the formal institutions of the state, and those that do not. The signature threshold is such a ' claim-laundering' process (Cheneval & el-Wakil 2018: 300). This is then followed by a second, deliberative test: does the proposal hold up after being scrutinised by a diverse group of citizens over several days of deliberation? Only those proposals that pass both tests will be submitted to parliament. It is important to note that this second test is not intended to make up for cognitive incompetence on the part of the initiative's many sponsors. Rather, its aim is to correct for possible exclusions, and thus to ensure the equality of citizens in the political agenda-setting process.
This idea of the deliberative assembly acting as an additional discursive filter on top of the signature threshold can be seen in the institutional specifics of the Citizens' Assembly Bill, and most clearly in the composition of the assembly and the special role given to the sponsors of the initiative. According to the Bill, the assembly would include one to five of the initiative's sponsors in addition to the other participants. This last addition is specific to the citizen-initiated nature of this deliberative assembly. The special status of the proposal's sponsors gives institutional recognition to their claim to represent the voice of at least 80,000 of their fellow citizens. The initiative's sponsors are thus given the right not only to address the deliberative assembly -like other interest or activist groups -but to actively participate in its deliberations. The automatic inclusion of the initiative's sponsors in the deliberative assembly ensures that the petitioners' views are taken up in the process, while the fact that they only occupy five seats out of more than a hundred makes it obvious that they will still have to try and persuade the other members of the assembly.
If they succeed in this endeavour, however, the fact that the initiative-takers are supported in their demands by a broadly representative sample of the population would give more credence to their claim to represent the public voice, and not just a vocal minority. Compared to the aforementioned agenda initiative, the government would therefore be in a better position to assess calls for political action. This is particularly so because the Citizens' Assembly Bill proposed a mixed composition in which 75 seats would be occupied by randomly selected citizens and 25 seats would be assigned to members of the Flemish parliament. The presence of these elected representatives forms an instance of designed coupling (Hendriks 2016), and seeks to prevent a disconnect between the citizens' assembly and the formal sites of decision-making (Setälä 2017). Of course, this mixed membership brings with it the risk that the proceedings of the minipublic would be dominated by the voices of professional politicians. Whether this risk manifests itself would to a large extent depend on how deliberations are moderated, and whether moderators take active steps to secure the equal voice of citizens (Arnold et al. 2019). The Bill offered an additional guarantee, however, in the sense that decisions require a double majority both within the panel as a whole and within the group of 75 randomly selected citizens taken separately. In this way, the citizens are guaranteed a veto right vis-à-vis the professional politicians within the assembly, and the risk of elite domination is significantly reduced.
Even if the government were ultimately to dismiss the assembly's recommendations, it would not be without effect. Because of the broad representativeness of the assembly, its demands would no longer be able to be simply dismissed as unrepresentative or misguided. The citizen-initiated citizens' assembly therefore not only forces petitioners to appeal to mutually acceptable reasons as to why collective action needs to be undertaken (Gutmann & Thompson 1997: 55), once their demands have passed through the discursive filter of the deliberative forum, it also forces governments to provide the same kind of reasons -not just excuses -why such action should not be taken. Deliberative assemblies can thus strengthen democratic control over the public agenda through what Cristina Lafont has called the 'vigilant' use of deliberative minipublics (2019: 154): they draw the attention of the public to instances where existing policy clashes with the considered opinion of a diverse group of their fellow citizens, signalling that on these issues the political system is most likely insufficiently responsive. The public availability of this information can in turn serve to ' draw additional support from the general public toward social and political groups mobilized against whichever forces are impeding the proper flow of influence between the enacted policies and the processes of citizen opinion and will formation in which citizens participate' (Lafont 2017: 97).
Perhaps just as important is how the interaction between outside groups such as those referred to by Lafont and a minipublic can help to overcome one of the biggest weaknesses that deliberative minipublics suffer from: their lack of visibility (Rummens 2012). It has been argued that what minipublics lack most is the sort of narrative structure that enables non-participating citizens to easily grasp the different positions and stakes at play in the discussion. While this is true if one considers the minipublic as an isolated event, it is not necessarily so from a participatory perspective. When -as suggested by Lafont -a particular group chooses to integrate the fact that their policy position was supported by a minipublic into its communications with the public, it is up to them to weave this information into their own narrative and to place it within a broader context. While the motivation to do so may be self-interested, such actions have the potential to increase the reach of a minipublic's conclusions and thus contribute to a proper functioning of the broader deliberative system. The Citizens' Assembly Bill strengthened this visibility by requiring that the assembly's recommendations be debated by parliament in a public hearing to which the citizens who participated in the citizens' assembly are also invited. This provides an additional moment of accountability which could be picked up by the media or (more likely) the initiative's sponsors and circulated in the public sphere.

Practical Challenges to Participatory Minipublics
The promise of the Citizens' Assembly Bill failed to materialise. It was soundly defeated in the plenary vote, with only the Green Party voting in favour. This might come as a surprise to some observers. Recently, Belgium has been considered a trailblazer in democratic innovations (OECD 2020). Two of Belgium's federated entities, Brussels and Ostbelgien, have recently moved to institutionalise minipublics in their own decision-making process. Whereas Ostbelgien opted for a system which could be called 'bicameralism light', since it foresees the creation of a standing Citizen Council with its own agenda-setting powers along with committee-like Assemblies to deliberate on specific issues, Brussels chose to incorporate the random selection of citizens into a new type of hybrid select committee, composed of both elected representatives and ordinary citizens. In contrast to these successes, the failure of the Citizens' Assembly Bill in Flanders exposes some of the lingering issues when attempting to institutionalise minipublics.

Transforming fashionable ideas into institutionalised practice
Democratic renewal features highly on the political agenda in Belgium. While the challenge of reconnecting citizens with the democratic process is not unique to Belgium, it is often felt as particularly acute for this specific polity. Not only is Belgium a deeply divided country where political parties and the media system are split along linguistic lines -so that there is little in terms of an overarching Belgian public sphere -but its political system is characterised by strong executives, weak parliaments and highly centralised parties (so much so, that it has even been termed a 'particracy') which leaves little room for citizen participation (Dewachter 2014). Over the years, numerous ideas have been suggested to address this ' democratic deficit'. Most of the time, however, these ideas do not get past the stage of general statements in party manifestos or the media.
A brief review of the history of democratic innovations in Belgium is useful to illustrate this point. In the early 2000s government coalitions both at the federal and regional levels showed enthusiasm for instruments of direct democracy such as referendums and popular initiatives. However, few steps were taken to actually implement these proposals. 5 Interest in democratic reform picked up again in 2011, when a number of academics and public intellectuals managed to shift the debate from direct to deliberative democracy. They did this through the organisation of the 'G1000', a grassroots minipublic where 700 randomly selected citizens convened for a single day of deliberation while a smaller panel of 32 citizens (drawn from this larger group) elaborated on the G1000's suggestions and issued specific policy recommendations. The G1000 and the media attention it received offered politicians an interesting exemplar of democratic innovation. 6 The idea was then picked up by numerous political parties (including the Flemish ones), and deliberative minipublics became an important element of their discourse (Jacquet et al. 2016). Democratic innovations now feature explicitly in the election manifestos of several political parties, which often include calls to incorporate some form of deliberative minipublic in decision-making. However, just as was the case in the 2000s with direct democracy, the radical nature of these ideas stands in stark contrast with political reality. Most of these proposals lack any form of serious legislative follow-up (Vandamme et al. 2019: 125): they are enthusiastically launched in the media only to be quickly forgotten afterwards. The Ostbelgien and Brussels examples are the exception to this rule.
This recurring pattern raises the question to what extent (Flemish) political parties genuinely seek to give citizens control over policymaking through deliberative minipublics. The rejection of the Citizens' Assembly Bill is a perfect illustration of the 'participatory dissonance' that these parties suffer from. While they may be formally committed to increasing citizen participation, when push comes to shove they are unwilling to actually share power with citizens (Hendriks & Lees-Marshment 2019). In the same session in which the Bill failed to secure the necessary votes, the Flemish parliament nearly unanimously voted in favour of a non-binding resolution calling upon the Flemish government to engage in 'participatory experiments' and to actively include citizens in policy discussions. 7 The resolution explicitly recommended a number of democratic innovations to the government so that it could achieve this aim, including participatory budgeting, deliberative polling and citizens' assemblies. Little has come of this, however, and no steps have been taken as of yet to actually put this resolution into practice. This leads one to believe that (in Flanders at least) minipublics are still seen first and foremost as public relations tools rather than as a means to empower citizens. This impression corresponds with the results of a number of other case studies on deliberative democracy in practice (Johnson 2015;Parkinson 2006). This should be particularly worrying to deliberative democrats -such as Lafont -who stress precisely the contestatory role of minipublics.

Public pressure and accountability
The failure to institutionalise the Citizens' Assembly Bill is not only attributable to parliament, however, but represents a broader failure to hold politics to account: elected politicians felt free to support the idea of deliberative democracy in public, only to vote against a proposal that sought to bring this idea into practice. When the Bill was defeated, they suffered few consequences. The vote generated little media attention, and most people were not even aware that the proposal was on the parliamentary agenda in the first place.
This offers an important lesson to deliberative democrats that 'neither ideas nor institutions are self-implementing' (Barber 1984: 263). While the G1000 has been lauded for taking deliberative democracy out of academic circles and making it the subject of public debate in Belgium (Felicetti & della Porte 2019;Jacquet et al. 2016), as a stand-alone event it could not keep politicians accountable for what happened to these ideas afterwards. Although minipublics can contribute to the spreading of deliberative norms by acting as ' exemplars' of deliberation for non-participating citizens (Curato & Böker 2015;Niemeyer 2014), this ' deliberative capacity'-building or ' deliberative norm'building role is unlikely to be successful in realising the institutionalisation of minipublics in the first place. This strategy is bound to run into a paradox, for in order for minipublics to have this desired macro-effect requires a degree of habituation with minipublics which presupposes that they have already acquired the status of institution; a chicken and egg problem if you will (see also Niemeyer & Jennstal 2018: 431).
The G1000 did have a lasting effect in a different sense, however. The organising committee behind the G1000 which consisted of artists, academics and public intellectuals continued to convene after the G1000 itself was concluded. This became a platform to campaign for more deliberative democracy in Belgium, which was one of the driving forces behind the implementation of a system of citizens' assemblies in the small German-speaking community of Belgium, the so-called 'Ostbelgien Model' (Caluwaerts & Reuchamps 2018). Indeed, institutionalising minipublics demands ' a political movement composed of committed democrats who understand themselves to have an interest in [its realisation]' (Barber 1984: 263). In Flanders in contrast, the proposal remained disconnected from this broader movement, stood isolated and was easily defeated in parliament as a result.
Public pressure can come from different spaces. Social movements and activist networks are obvious agents to exert public pressure. Although social movements have for a long time been derided by deliberative democracy scholars as partisan enclaves of like-minded activists, these enclaves provide both the motivational and organisational foundations necessary for sustaining the political campaign to institutionalise minipublics over the long stretches of time it will take to see those proposals realised (Fraser 1990; also see White & Ypi 2016). In a way, if, as Jürgen Habermas argues, ' communicative power is exercised in the manner of a siege ' (1996: 486), then social movements provide both the troops necessary to maintain the siege, and the commitment and determination to see it through.
Of course, citizens campaigning for more deliberative democracy may sometimes find allies in elected officials, civil servants or other actors operating from within the political system (the Irish assemblies mentioned earlier being a good example). Yet this is unlikely to prove a reliable strategy towards institutionalisation. As the literature on electoral reform has shown, reforms running against the self-interest of incumbents require external pressures to be successful (Bowler & Donovan 2013: 23-25). The same applies to deliberative minipublics. If we wish such institutionalised minipublics to play the contestatory or vigilant roles defended by Lafont by creating a space for citizens to contest and scrutinise public policy, then the support for such reforms will probably have to come from outside the legislature to force 'politicians to acquiesce in reforms that, on the whole, they still do not want' (Renwick 2010: 167).
While most deliberative democrats now agree that social movements have an important role to play in a wellfunctioning deliberative system, and while deliberative democratic scholarship has now started to seriously study social movements (della Porta & Doerr 2018; Felicetti 2017), the social movement and deliberative minipublics literatures remain separate to a large extent (for an exception, see Felicetti & della Porte 2019). Most often, studies on the institutionalisation of deliberative minipublics focus on the supply side of the debate. They concern themselves with the nitty-gritty of institutional design: how citizens will be sampled, how briefing materials will be elaborated and by whom, how deliberations will take place, and how the minipublic will interact with other institutions (see, for example, Gastil & Wright 2019). These are all important questions and we have learned much over the years from the empirical experimentation with different institutional set-ups. Yet these types of questions form only half the picture. They should be complemented by a ' demand-driven approach' to deliberative minipublics (Felicetti & della Porte 2019: 146), which extends the focus from the micro-deliberative specifics of this or that proposal to include the study of the macro-deliberative politics necessary to see these proposals realised in practice.

Conclusion
The story of the Flemish Citizens' Assembly Bill is one of cautionary optimism. It is optimistic, because it shows how normative theories of deliberative democracy can serve as a practical guide when designing democratic institutions. It presents a picture of how minipublics can be adapted to serve the goal of democratic self-government without falling into the trap of the micro-deliberative shortcut that Lafont warns us about. At the same time, its ultimate failure constitutes a warning. The types of minipublics that are most promising for a participatory conception of deliberative democracy, namely those that lend citizens a voice to criticise, contest and challenge the status quo, will face the most uphill struggle to become institutionalised. If deliberative democrats want to see the promise of the deliberative wave realised, they cannot limit themselves to elaborating new institutional designs, but should also study (and support) the macro-political struggles and engage in the ' deliberative activism' (Lafont 2019: 146) necessary to see those designs realised in practice.

Notes
1 A third referendum to lower the age-limit for the position of president was held at the same time as the gay marriage referendum, but this proposal did not receive a majority of the votes. 2 Lafont is right to highlight the troubling similarities between the logic of ' embodiment' which sometimes characterises arguments for an empowered use of minipublics and populism (Lafont 2019: 121). For a similar argument as applied to referendums, see Van Crombrugge (2020). 3 The author was not directly involved in the drafting of the Bill, but did provide legal advice and testimony to the Flemish parliament on this matter. 4 By way of example, in 2009 -in the context of a local referendum in the Flemish city of Antwerp -a group of citizens called Ademloos ('without breath') was able to gather 66,158 signatures to demand a referendum on the city's plans to build a new highway bridge close to the city centre (https://www.demorgen.be/ nieuws/66-158-handtekeningen-voor-referendum-oosterweel~baf24a78/, last accessed 14 January 2020). It is difficult to provide examples for the region of Flanders as a whole since there exists no comparable right to referendum or petition at the Flemish level. 5 The exception to the rule is the Region of Wallonia, which introduced legislation in 2018 to make it possible for 60,000 citizens to demand a non-binding referendum: Décret special instituant la consulation populaire (Special decree instituting the popular consultation). 6 At the time, Belgium was going through a deep and prolonged constitutional crisis. After the 2010 elections, it took more than 541 days to form a new government -giving Belgium the questionable honour of holding the world record in 'government formation'. This sorry state of affairs threw the failure of traditional representative institutions into stark relief, and explains the considerable media attention for the G1000: it presented an alternative vision of what democracy could be like, at a time when many people had become disillusioned with traditional institutions. 7 Resoulutie betreffende burgerparticipatie (Resolution concerning citizen participation).