In recent years we have witnessed a radicalization in the academic debate on the use of sortition in democratic systems. Before, scholars assumed that randomly selected minipublics would play a limited advisory role and focused mainly on the potential ‘uptake’ of the minipublic’s recommendations in the macro-political process, which was itself understood in traditional parliamentary and electoral terms (Fung 2003; Goodin and Dryzek 2006). Emboldened by political experiments with minipublics, such as the G1000 in Belgium, the Irish Constitutional Convention, or the Icelandic Constitutional Council, and frustrated by enduring problems with electoral institutions, several authors now take things further. They advance the daring idea that minipublics should not merely provide input for the macro-political process but should, instead, be empowered to make the political decisions themselves.

Some authors still tread carefully. Jane Mansbridge (2020, p. 21), for instance, claims that empowering minipublics would fall ‘completely within the parameters of democratic action’, but acknowledges that the impact would be unpredictable and cautions to move ‘slowly and incrementally’. Mark Warren (2020, p. 87) believes that minipublics can be ‘trustworthy agents’, but also suggests that citizens should defer to minipublics only for unimportant issues. More ambitiously, John Gastil and Erik Olin Wright (2018), David Owen and Graham Smith (2018), and Arash Abizadeh (2021) propose a bicameral system in which an elected chamber is complemented by a sortition chamber. Some scholars go further still and claim that, ideally, democracy could do without elections and rely mainly or even exclusively on sortition for decision-making purposes. Examples are Alexander Guerrero (2014), David Van Reybrouck (2016), and Hélène Landemore (2020).

In this paper we argue that the rise of this lottocratic paradigm is problematic from a democratic point of view. We agree that minipublics might, under certain circumstances, enrich public debates. But the proposal to give them real power and make them carry the weight of democratic legitimation is based on a misguided interpretation of the democratic ideal. Minipublics are allegedly democratic because they provide a descriptive representation of the people: the selected members are ‘like us’ and will therefore make decisions that we can endorse. However, as Hanna Pitkin (1967, pp. 89–90) has already emphasized, descriptive representation offers only a partial understanding of representation: it ‘cannot account for (…) authoritative action that is binding on the represented’ and has no room for ‘accountability, (…) leadership, initiative or creative action’. This concern should be taken seriously. Democracy is the collective self-government of a group of autonomous citizens. The mechanism of descriptive representation is simply too weak to form the core of this process. In the felicitous phrase of Cristina Lafont (2020a), a randomly selected assembly constitutes a shortcut that undermines rather than facilitates collective opinion and will formation.

Before proceeding, two important methodological remarks are in order. The first concerns our conception of democracy and the normative criteria that guide our comparison. We fully endorse the systemic turn in deliberative democracy (Mansbridge et al. 2012). Democracy is not a set of norms that pertains merely to a society’s central decision-making body. What is really of interest are the democratic qualities of a political system as a whole, including for instance the interactions between formal institutions and the wider public sphere. Since we subscribe to the deliberative paradigm, we are obviously concerned with the system’s capacity to generate ‘reasonable’ outcomes. At the same time, we agree with political theorists such as André Bächtiger and John Parkinson, who argue that deliberative democracy should not merely be ‘deliberative’ but also ‘democratic’ and hence be concerned with additional elements such as political equality and popular control (Bächtiger and Parkinson 2019, pp. 1–14).

Although there is currently no general agreement about the proper criteria for a comprehensive assessment of a deliberative democratic system, we believe that its most important qualities can be captured in three dimensions. The epistemic dimension concerns the rational quality of the decision-making process: have all potential concerns and perspectives been taken into account and transformed into an impartial decision that gives equal concern to all citizens affected? The power dimension concerns the distribution of power: is power distributed equally and do citizens have genuine control over collective decisions? The motivational dimension concerns political integration: can citizens identify with the collective decision-making process and are they willing to accept its outcomes as legitimate?

Secondly, it should be emphasized that our analysis belongs to the field of normative political philosophy. We compare an idealized electoral system with an idealized lottocratic system by looking at their normatively salient features and their democratic potential. Of course, reality has a way of falling short of ideal-types. Actual electoral systems are notoriously deficient, and it is these very deficiencies that usually motivate the quest for lottocratic alternatives. The dire state of US democracy in particular is often the foil for arguments in favor of sortition.Footnote 1 There are, arguably, democratic regimes that fare much better than the US system in terms of, for instance, elite capture or the polarization of the media. But these regimes might come with other problems, such as populism or a declining belief in democracy. There is no point in denying or downplaying these widely documented problems. At the same time, when contemplating lottocratic alternatives, we should be careful, as Malleson (2018, p. 402) has rightly emphasized, not to compare the empirical reality of actual electoral systems with a future ideal of sortition.Footnote 2

One option would be to compare what actual electoral and actual lottocratic systems can achieve. This is not easy. As lottocratic systems have never been tried in full in modern society, the empirical data for such a comparison are lacking. Of course, informed speculation about the non-ideal working of lottocratic systems might be possible. For instance, advocates of sortition have asked whether allotted chambers could be vulnerable to capture by economic or other elite interests (Guerrero 2014; Owen and Smith 2018). Yet the results of such speculations remain hypothetical and controversial.Footnote 3

As normative theorists, we pursue the other option and focus on a comparison of ideal-types. We believe that such an approach is justified because our analysis demonstrates that the problems with the lottocratic model already arise at the level of ideal theory. The main problem lies with the conception of descriptive representation at the core of the sortition system (Guerrero 2014, Landemore 2020). Whereas advocates of the sortition model argue that the members of the allotted chambers are ‘like us’ and, therefore, allowed to act as trustees who can make decisions on our behalf, Landa and Pevnick (2021, pp. 49–50) rightly state that the lottocratic model thereby expects the citizens at large to simply defer to the allotted assembly to deliberate and to make its decisions in its stead. From a democratic perspective, however, this deference is deeply problematic. At the epistemic level, it precludes deliberation as a genuinely collective learning process in which the public at large participates. In terms of power, it fails to provide genuine accountability mechanisms allowing citizens to check what goes on in the sortition chamber and control collective decisions. In motivational terms, the lack of connections between the sortition chamber and the wider public inevitably leads to disaffection on the side of citizens and undermines the perceived legitimacy of the system. In contrast, electoral mechanisms allow for a richer form of interactive representation in which the representatives are not merely trustees but play a pivotal role in a much larger collective decision-making process. In this process, a collective understanding of the common good is jointly constructed. In epistemic terms, we are now dealing with a collective learning process in which all relevant concerns are included. In terms of power, elections allow for genuine authorization and accountability. In terms of motivation, the means of participation and control sustain a sense of identification and preserve the perceived legitimacy of the system.

Elections as Interactive Representation

In this section we argue that traditional representative institutions based on parliamentary elections offer a particularly well-suited infrastructure for the institutionalization of the ideal of deliberative democracy. We substantiate this claim by looking in turn at all three dimensions: epistemic, power, and motivational.

Elections: The Epistemic Dimension

According to deliberative theorists, democratic decision-making is primarily an epistemic process aimed at constructing the best possible interpretation of the general interest (Habermas 1996). Democracy is not a mere matter of aggregating preferences but of reaching insight through the mutual exchange of reasons. Importantly, this process ought to be inclusive. The citizens’ political autonomy requires that all their concerns and ideas be taken into consideration: every citizen should be accepted as a unique and irreplaceable source of potentially relevant arguments. The subsequent exchange of arguments should be guided by the force of the better argument, that is, participants should be sensitive to the arguments presented by other participants. This implies that the democratic process is necessarily transformative: citizens who participate in a deliberative process will change their views in light of the views of others.

Contrary to what is sometimes claimed, we believe that the traditional mechanisms of representative democracy are well suited to fulfill this epistemic promise. Building on Jürgen Habermas’s well-known reconstruction (Habermas 1996), we bring out their epistemic qualities by modeling representative democracy as a combination of an informal public sphere and formal political institutions, with a variety of semi-formal layers in between.

The informal public sphere, understood as the set of all spaces where citizens can discuss matters of public concern, ensures inclusiveness: here, all citizens can raise all possible considerations. This informal sphere is linked up with semi-formal actors (interest groups, NGOs, labor unions) who pick up the concerns that circulate and voice them eloquently and efficiently. They also provide protected forums where arguments can be developed and perfected and they give permanence to ideological positions so as to inform and inspire citizens. Most importantly, they provide a stable link with formal political institutions. Through a well-organized civil society, considerations coming from the informal public sphere can influence the formal decision-making process, with parliament as the typical focal point where the process of collective self-legislation reaches its ultimate stage. In parliament, politicians collect and condense the most convincing and relevant considerations into coherent positions. The party context in which politicians operate, is particularly helpful in this regard. The presence of multiple parties on the political stage gives ideological clarity to the societal debate as it makes visible how opinions and concerns are interwoven with a more encompassing political outlook and how these different outlooks are related or differ.

The debate between politicians ideally strikes a balance between the considerate spirit of deliberation and the more adversarial spirit of partisanship. Parliament can provide a well-structured context for treating significant societal differences in an argumentative manner. But the disagreement between politicians should not be too courteous. As Bernard Manin explains, today’s advocates of deliberation often overlook the benefits of strong oppositions. A vigorous confrontation between opposing views can actually lead to better collective decisions: ‘to hear contrary opinions is necessary for discovering the right answer’ (Manin 2017, p. 41). Moreover, reducing complex problems to a sharp choice between a limited number of policy options helps to engage citizens.

If the adversarial staging of societal problems encourages participation, it is not only because it reduces the cognitive burden on citizens, but also because the clash between political personalities lends itself particularly well to transmission through mass media.Footnote 4 Mass media operate according to a ‘dramaturgical’ logic: they follow the rules of staged dramas. Hence, the core arguments of important disagreements need to be delivered by recognizable faces, in eloquent discussions and speeches, with symbolic gestures, and with plenty of attention for the psychological dynamics of even the most rational debates. The captivating effect is of course increased by the fact that the politicians on stage are elected by the citizens themselves. Every voter has a direct personal stake in the debate, leading to greater identification.

Nor is this engagement of the citizens—even if it is only a mental engagement—something that we can dispense with. Deliberation should be a transformative process, in which the citizens’ ideas and interests are not treated as invariant inputs but change in light of the arguments of others. This transformative effect can only be achieved if society’s most momentous debates are visibly staged, i.e., if everyone is exposed to the main arguments in play. Visibility is crucial here (Rummens 2012, 2016). It is not enough that governmental processes are transparent in the sense that citizens have the formal right to access all relevant materials. If the decision-making process is to captivate all citizens, the core of the debate needs to be dramatically staged in a way that is straightforwardly visible to all. The citizens’ role as spectators to the political drama might seem passive, but this is misleading. It informs them about opposing views; it creates opportunities to discuss with friends and family; and, ultimately, it helps them to make up their mind when they go voting. In other words, the collective learning process that deliberative democrats are hoping for, finds its condition of possibility in a space where society’s most important disagreements are made visible and comprehensible to all of its members.

Note that we are here understanding representation as an interactive process, engaging citizens, civil society actors, and politicians. The concerns that appear in informal debates can be challenged but also amplified in the semi-formal sphere and might be adopted by political representatives in the formal arena. Inversely, when these representatives put forward certain views and claim to speak in the name of those whom they represent—i.e., when they put forward ‘representative claims’ (Saward 2010)—they might be challenged, but their claims might just as well be accepted and hence bring citizens to new views and thus transform the public’s opinions. Parkinson accurately writes that this turns responsiveness into ‘mutual co-creation’, with shared ‘responsibility between citizens and representatives’ (Parkinson 2012, p. 164).Footnote 5

Elections: The Power Dimension

Democracy is an ongoing exchange of arguments. But it is, at the same time, a process in which power is circulating. Some of that power is, as Habermas calls it following Hannah Arendt, communicative: ideas that convince by the sheer force of the better argument become more powerful as their adherents increase. The power of shared convictions emerges partly in informal settings but equally flows through official political channels. Here, power also assumes a more formal shape: persons and institutions are allotted various amounts of decisional authority and they exercise it in acts of legislation, administration, and government. Most theories of deliberative democracy are relatively silent about this institutional circulation of power. This is unfortunate, as this circulation needs to be carefully regulated. It is true that this process cannot be uncoupled from discursive interactions and permanently depends on communicative achievements. But the procedural rules concerning the allocation and the exercise of power are normatively important too and contribute to the democratic legitimacy of collective decisions (Geenens 2007). Ultimately, the entire process of decision-making and government is democratic only if it is formally authorized and controlled by the citizens themselves in a way that respects their standing as free and equal members of the political community.

In representative democracies, the voting mechanism—one person, one vote—plays a crucial role as it formally allows all citizens to participate in the decision-making process. Most prominently, it enables citizens to select representatives who speak in their name, or, more broadly, in the name of the political community. The power to appoint representatives is an important one and elections are democratic precisely because they distribute it in a strictly equal manner: the input of every citizen is recognized as irreplaceable and is given equal weight. Voting procedures cannot substitute for the process of rational justification, but they do provide a unique moment where the political or ideological orientation of every individual is taken into account and is considered equally relevant. If citizens autonomously appoint representatives through voting, the resulting body visibly reflects the scale of positions and perspectives which they believe to be politically significant at this specific moment and does so in proportion to the actual distribution of these views among the citizenry. The mechanism of proportional representation scores particularly well on this point.

That citizens can select their own representatives follows naturally from their political freedom. Electing politicians is, after all, an act of authorization: one authorizes someone else to speak in one’s name. Actual decision-making power is thereby bestowed on concrete persons. The moment of election reflects the basic normative given that in a democracy no one is naturally entitled to hold power over other citizens. The source of all decision-making power lies with the citizens themselves. Hence, it is they alone who can authorize certain co-citizens to exercise it on their behalf. And there seems to be no reason for distributing this authorizing power unequally. To the contrary, an equal distribution is an important expression of the citizens’ equal status.

The voting mechanism also helps citizens to hold their representatives accountable. It follows from the citizens’ political autonomy—and from the very concept of authorization—that the people whom they authorize to legislate and govern can be held accountable for their acts by the citizens themselves. This is why members of government are, in parliamentary systems, answerable to the people’s representatives. All governmental decisions can be publicly discussed and criticized in parliament, where those responsible are to appear and justify themselves. Again, this is an adversarial and highly visible process, driven by the dynamic relation between government and opposition: the politicians currently in power find themselves opposed by other politicians, who have a tangible incentive to point out all potential faults and weaknesses. In a further step, representatives are responsible to their voters. A well-functioning public sphere permanently exposes representatives to the informal pressure of an inquisitive and critical public opinion. And this informal pressure is underpinned by a more formal mechanism: citizens can reward or punish their representatives in the next election. In this regard, first-past-the-post is often said to score better because of the well-defined relation between representatives and citizens. In any case, a strong chain of accountability between citizens and decision-makers is a necessary component of collective autonomy. And note again that accountability depends on visibility —the public staging of the entire process—and, ultimately, on the formal mechanism of voting.

Elections: The Motivational Dimension

Democracy not only requires that decisions are made in a sufficiently epistemic manner and that formal political power is regulated in a sufficiently egalitarian way. It also depends on motivational elements. The ideal of self-authorship of the law implies that citizens can recognize collective decisions as their own decisions. In Habermas’s words: the law ‘can preserve its socially integrating force only insofar as the addressees of legal norms may at the same time understand themselves, taken as a whole, as the rational authors of those norms’ (Habermas 1996, p. 33). We discuss the motivational dimension separately for reasons of analytic clarity, but in fact it cannot be separated from epistemic and power considerations. It is the actual process of deliberative justification that creates, for the citizens, the presumption that collective decisions are rationally justifiable and hence worthy of recognition. And it is their real control over the decision-making process that gives them a sense of ownership over all collective decisions.

On the epistemic side of things, the public practice of justification is crucial if collective decisions are to be accepted by all those concerned. In this process citizens hear the arguments of others, submit their own arguments (or hear them submitted by representatives), and possibly change their views in light of newer or stronger arguments. Of course, there always remains an unconvinced minority. But as many authors, from John Stuart Mill to Bernard Manin, have emphasized, even the outvoted minority has a lot to gain from the preceding deliberation. Although they might disagree with the reasons that justify the majority decision, they have been duly informed about these reasons. The minority members, Manin rightly observes, ‘were therefore treated with the respect owed to autonomous beings’ (Manin 2017, p. 46). The fact that some political actors have publicly articulated the minority’s position in a well-argued manner constitutes an additional moment of recognition: the minority’s ideological outlook is heard and taken seriously by their co-citizens, even if they remain unconvinced. Moreover, a democratic decision is, in principle, never final. Even if a position is outvoted, it does not become illegitimate. The debate can always be reopened, and the minority can try again to convince the majority. A multi-party system offers a clear advantage in this regard: all political positions remain visibly present on the political stage, even the outvoted ones, giving these positions a form of public recognition and signaling that the current majority is only a temporary one. Minority parties and civil society organizations thereby act as epistemic reservoirs, in which arguments that, for now, have been defeated are preserved and possibly improved in order to, perhaps, convince more people in the future (Rummens 2012, p. 34). More generally, a publicly staged debate about our collective future—even a very conflictual debate—can have a unifying effect: rather than seeing ourselves as mere individuals, we come to see ourselves as political actors involved in a collective undertaking for which we are jointly responsible.

In terms of power, the procedures that distribute, allocate, and control formal political power have equally important motivational effects. Elections, which create a chain of authorization and accountability between voters and representatives, promise to citizens that it is they who, ultimately, control collective decisions and hence strengthen their sense of joint authorship. Universal suffrage is valuable precisely because it is responsive to the political view of every individual citizen. It does so by assigning every citizen, at least in principle, an equal share of influence in the decision-making process. This equal distribution of formal decision power is a visible embodiment of political equality (Mansbridge 1997, p. 412; Richardson 1997, p. 358; Geenens 2007). A voting procedure that is responsive to the input of all citizens and weighs their influence equally, creates a unique moment of mutual recognition: citizens are respected as equal members of a self-governing collectivity rather than as mere objects of legislation. This means that election procedures have an important ‘symbolic’ quality (Lefort 1988): the collective ritual in which everyone, politicians included, passes along the ballot box publicly expresses the idea that in a democracy no one has a privileged claim to power. This moment of equal recognition is especially important for the outvoted minority which remains unconvinced by the winning arguments and hence will not see in these arguments sufficient ground to justify the decision. The voting procedure offers additional ground here. The equal distribution of decision power helps the dissenter recognize the legitimacy of the winning decision. One could even argue that the numerical reference to the vote is a highly respectful way of treating those who disagree. Rather than saying that their proposal is rejected because we are certain that it is wrong, we simply say that it is, for the time being, outvoted. This is the intuition that Bernard Williams was after when he defended the procedural dimension of democracy. Rather than announcing ‘that the other party was morally wrong’, the outcome of a political procedure only indicates ‘that they have lost’ (Williams 2005, p. 13).

Sortition as Descriptive Representation

We now turn to the normative credentials of a deliberative system that centers on randomly selected assemblies rather than an elected parliament. No such regime has ever been implemented, and the theoretical proposals that circulate vary in terms of institutional design. For our analysis, we have in mind a system that assigns legislative power fully to minipublics appointed on the basis of sortition. Our assessment does not depend on the further institutional specifics of such a lottocratic system.

Sortition: The Epistemic Dimension

A key selling point of sortition assemblies is their epistemic potential. Minipublics embody the hope of a controlled environment where political emotions can be checked, partial interests kept at bay, and, consequently, the force of the better argument prevails. Practice has certainly confirmed that, under proper circumstances, ordinary citizens can deliberate about complex issues and come up with well-argued views and proposals. For this reason, we agree that minipublics can sometimes provide interesting contributions to the political debate. Recent proposals, however, upgrade minipublics from possible contributors to a shortcut for the entire democratic process. This move seems problematic to us. The deliberative ideal requires that democratic decision-making be inclusive and transformative. But can minipublics live up to these standards?

This is highly doubtful. Minipublics are primarily designed for internal deliberativeness and not for outside interaction. Hence, it is unlikely that they can maintain the dynamic feedback loop with the wider public sphere that is needed to constitute a genuinely collective opinion and will formation (Malkopoulou 2015). As many authors admit, minipublics lack the adversarial dynamics between identifiable political leaders that is needed to attract media attention (Parkinson 2005; Mansbridge et al. 2012, p. 22; Van Reybrouck 2016, pp. 125–126). Career politicians have many vices, but they are often eloquent and inspiring. They can conduct public debates in ways that captivate and energize ordinary citizens. It seems unlikely that randomly selected citizens are capable (or willing) to take up this task. This is in any case not the task that advocates of lottocracy give them. Precisely because deliberation in the allotted chamber is supposed to operate merely on the basis of the force of the better argument, its participants have to be shielded from pressure and undue exposure. As a consequence, the internal workings of the minipublic are bound to remain invisible. Debates within minipublics will hardly capture the citizens’ attention and will not be able to give focus and structure to the wider public debate. A minipublic becomes, in effect, a black box, disconnected from civil society where, in actual democracies, a large part of the epistemic work is being done (Rummens 2016). Generating arguments, perspectives, and proposals is a demanding process that requires ingenuity and perseverance. Feminist or ecological perspectives, for instance, do not fall from the sky but emerge over years and through the sustained efforts of activists in a wide variety of civil society organizations.

In consequence, a lottocratic system would have difficulties to sustain a collective deliberative process that is genuinely inclusive and transformative. Inclusiveness would be impossible to guarantee because the complex web of communicative channels that links citizens, civil society organization, and formal decision-makers is missing. Advocates of lottocracy assume that all relevant perspectives and arguments can be collected by sampling a diverse set of individual citizens. But this is not enough. Ordinary citizens need to have routes to raise new concerns and press decision-makers to take them into account. And the wide field of civil society organizations, which normally detect, group, flesh out, and express the concerns of citizens, needs to be linked up with decision-making centers. Think again of feminist or ecological perspectives. There is no guarantee that the knowledge and sensibilities generated in these movements will be present in the minipublic (not even if the assembly is, for instance, adequately stratified in terms of gender), nor can they simply be ‘injected’ into the minipublic (for instance by organizing hearings with civil society organizations). In fact, we cannot even assume that civil society would continue to thrive in a lottocratic system (cf. Mansbridge 2020, p. 21). Civil society organizations exist because they can change things by generating public pressure on elected politicians. If representatives are unsusceptible to electoral incentives, the channels for such influence become highly uncertain.

A lottocratic system might fare even worse with regard to the transformative aspect. Advocates of lottocracy argue that the process of transformation happens within the minipublic itself.  James Fishkin (2009, p. 28) writes that the outcome of minipublics can be seen as ‘representing the considered judgments of the people’. But this is overly optimistic. Transformation cannot be reduced to a learning process experienced by selected participants; it is a society-wide process. To make society aware of the cogency of feminist or ecological concerns, for instance, abundant contestation and argumentation is needed, pursued over decades. In this process, citizens critically test these new claims and gradually incorporate them into their own worldviews. The complexity of such issues requires an ongoing dynamic feedback loop between citizens, civil society, and decision-makers in which arguments, demands, and, ultimately, proposals for legislation are continuously circulated back and forth.

Advocates of lottocracy regularly mention the importance of contributions from the wider public and civil society (Guerrero 2014, pp. 177–178; Gastil and Wright 2018, pp. 311, 320–321; Landemore 2020, p. 148), but often this seems mere lip service.Footnote 6 Tellingly, none of them, even those with a genuine commitment to civil society, such as Mansbridge (2019, 2020), have concrete proposals for a permanent feedback loop between the wider public sphere and the minipublic. To the contrary, many authors insist that if minipublics are endowed with actual decision-making power, they should be protected from outside interference (Owen and Smith 2018, pp. 424–430; Fishkin 2019, pp. 82–86; Mansbridge 2020, p. 21; Abizadeh 2021, p. 801). Minipublics are epistemically valued because they are closed environments where participants are shielded from lobbyists, interest groups, and the ‘wild’ emotional debates going on in the public sphere. As the stakes become higher, outside groups have a much stronger incentive to put pressure on individual members (Umbers 2021, pp. 327–330). One could protect participants from this influence by using a system of fast rotation (Owen and Smith 2018, pp. 425–428) or secret ballots (Gastil and Wright 2018, p. 319). But such measures would make the interaction between the minipublic and the wider audience even more difficult. We believe that the trade-off between internal deliberativeness and outside interaction poses a genuine sortition dilemma that cannot be resolved within the lottocratic paradigm.

Sortition: The Power Dimension

The absence of strong channels of interaction between the minipublic and the wider public sphere also creates problems from a power-distributive perspective. How could a lottocratic system allow for genuine popular control by the citizens (cf. Parkinson 2006, p. 33; Malkopoulou 2015; Malleson 2018, pp. 406–408)?

Minipublics are often hailed for distributing political power more equally than electoral systems, in which elites have the upper hand. The claim is that citizens are now truly equal because they have an equal probability of being selected (Guerrero 2014, p. 169; Gastil and Wright 2018, p. 307; Landemore 2020, p. 90; Abizadeh 2021, p. 797). But this is problematic, not only because the probability is too low to be relevant but also because power is about real possibilities for action and influence rather than mere probabilities. The lottocratic regime cannot provide such possibilities and instead turns virtually all citizens into powerless bystanders who have to passively wait and see if and when they perhaps might get a chance to participate in the legislative process.

In an electoral system, by contrast, all citizens can exercise at least some form of control over the legislative process. Elections already provide a moment in which all citizens exert their political autonomy by authorizing representatives to act on their behalf, a moment where the decision-making power is officially distributed in an equal way. This gives citizens the possibility to consciously steer the collective decision-making process in a specific ideological direction (liberal, conservative, ecological, …) as the law-giving body literally represents the proportional strength of these different positions in the population.Footnote 7 In addition, politicians are constantly held accountable through the pressure and arguments coming from the public sphere. Citizens who want to exercise their political autonomy more actively, can engage in public debate, raise awareness, mobilize other citizens, launch petitions, and target their representatives. They can also join or create civil society associations and they can even decide to enter the formal political sphere.

Lottocrats have several replies to this worry about popular control. Landemore (2020, pp. 87, 108) argues that the problem of authorization can be solved by organizing a majority vote about the introduction of a lottocratic regime. This is a strange claim. It recognizes that in a democracy no one is naturally entitled to hold power over others and that the exercise of power needs to be legitimized by the citizens themselves through majority voting, i.e., through a procedure that distributes decision-making power in a formally equal way. Moreover, it suggests that the people could decide to permanently give away their autonomy through one single decision. From a democratic viewpoint this amounts to a performative contradiction since genuine popular control needs to be a structural part of the regime, not a one-time event.

A second response focuses on the possibility of deliberative accountability. Landemore (2020, p. 203) argues that ‘representatives would feel morally and institutionally compelled to provide good reasons for their decisions’. Similarly, Sintomer (2018, p. 353) states that ‘citizens who take part in a minipublic feel accountable to the wider public they represent’. Mansbridge (2019, pp. 197–203) expects that in empowered minipublics ‘spokespeople will emerge’ and that ‘some representatives will take upon themselves the task of describing, explaining, and justifying the decisions of the assembly’ both in the media and to the informal constituencies which she believes will form around some of these representatives. But this might intimidate many representatives as it requires skills and personalities for which they were not selected (Malleson 2018, p. 408; Mansbridge 2019, p. 200). More fundamentally, this type of accountability is too weak and noncommittal. The minipublic’s members have no electoral incentive for speaking out publicly. The media will not be very interested because there is no dramaturgical potential. Nor will the citizens themselves be interested since sortition forces them into a merely passive role. If they disagree with decisions that were made, there is no use in criticizing representatives or in threatening to withhold the vote which they no longer have. Nothing they can do will have any impact whatsoever on this or any future decision.

A third response by lottocrats relies on the claim that minipublics ‘mirror’ the public at large. This is where the notion of ‘descriptive representation’ becomes important. Warren, for instance, argues that deference to minipublics as information proxies or even decision proxies is warranted because they are descriptively representative. Participants in the minipublic can be trusted because they have the same interests and values as the citizens they represent (Warren 2020, pp. 86–87). Similarly, Malleson (2018, pp. 406–407) writes that a sortition chamber would include people that are ‘like me’ in the sense that they share ‘my values and principles’.Footnote 8

Unfortunately, this alignment of interests and values remains contingent since it is not based on an explicit relation of authorization. Instead, the relation of representation has been engineered: stratified sampling should ensure that all possibly relevant groups, based on criteria such as gender, age, ethnicity, or socioeconomic status, are adequately represented. The problem is, of course, that the relevant criteria will always be ‘politically contested’, as Gastil and Wright (2018, pp. 312–313, 318) acknowledge. This is not, as they appear to think, a minor problem of institutional design. The social engineer (the statistician, sociologist, or demographer) cannot, prior to the democratic process, decide on the basis of purportedly scientific arguments what constitutes a faithful mirroring of the demos. This amounts to a premature foreclosing of the democratic process that is incompatible with the autonomy of the citizens. The citizens’ political freedom includes the right to decide which considerations are salient in appointing a representative, and these considerations might well be political or ideological rather than descriptive.

Moreover, if we do not appoint our own representatives, it is unclear why they would defend our interests. As Lafont rightly argues, the independence of allotted representatives—cherished for epistemic reasons—is incompatible with a genuine relation of accountability:

There is no sense in which female participants are supposed to defend the views of women or Californians the views of other Californians. They participate as individual citizens with total freedom to express whichever views and opinions they have (…). But, for that very same reason, they are in no way accountable to citizens outside the minipublic. (Lafont 2020a, p. 119)

A fourth response centers on the inevitability of deference and the need for procedural control. The complexity of contemporary government requires a division of labor: citizens cannot but defer to bureaucratic agencies, judicial institutions, and advocacy groups (Warren 2020, p. 85; Mansbridge 2020, p. 16). According to Mansbridge (2020, pp. 19–21), such deference is unproblematic as long as citizens retain a procedural capacity for control. With minipublics, this requirement is met to the extent that citizens can ‘monitor them and keep the residual power to revoke them’. Similarly, Landemore (2020, pp. 203–204) argues that accountability can be strengthened on the basis of citizen’s initiatives and rights of referral, enabling citizens to trigger referendums on new proposals or existing laws. Yet repeal procedures come at a late stage, are highly cumbersome and are a drain on a democratic system. Moreover, they assume that citizens adequately monitor the legislative process. As argued, the lack of visibility and accountability makes it harder and less attractive for citizens and civil society to do so. Electoral systems perform better in this regard. They also involve a division of labor. As Habermas (1996, pp. 352–359) notes, parliamentary politics often operates in a routine mode whereby administrators and politicians produce legislation without much interference from the public. But in salient cases the switch to a different mode of politics is easily made. A well-aimed op-ed or a critical news item can push issues into the limelight and generate a visibly staged debate in which the wider public can contribute and influence the outcome. This is inconceivable with minipublics, where the institutional infrastructure needed for such interactive representation is simply unavailable.

Sortition: The Motivational Dimension

Motivational issues are interwoven with epistemic and power issues. Deficiencies in these latter dimensions therefore translate into motivational problems. As minipublics cannot sustain a genuinely collective epistemic process and lack visibility, citizens will often not be able to understand how decisions could be justified. And in the absence of authorization and accountability mechanisms, citizens will feel excluded and will lack a sense of ownership over the decision-making process.

Warren (2020) argues that citizens would defer to minipublics as ‘trusted decision proxies’ because they appreciate that its members are people ‘like us’. As Lafont (2020b, p. 101) rightly remarks, such deference remains ‘blind’ because citizens have no idea how the final decision has come about. Moreover, there is a high chance that citizens will disagree with the outcome, despite the minipublic’s purported resemblance to the population at large (Lafont 2020a, pp. 101–137). Absent a proper mediatization of the deliberative process—something nearly impossible given the lack of dramaturgical dynamics—the minipublic remains, to outsiders, a black box that produces singular decisions (Rummens 2016). Therefore, as Bächtiger and Parkinson (2019, p. 96) rightly conclude, ‘the “recommending force” of a claim emerging from a minipublic is a great deal weaker than many deliberative scholars assume’. Especially when the stakes are high and citizens are directly affected by the decision, blind trust and deference are unlikely to prevail.

Obviously, electoral systems are also confronted with citizens who dislike certain decisions, but the mechanisms for dealing with these disagreements, as explained above, are much richer. The visible staging of the debate enables people to see which arguments are used by whom and which arguments ultimately make a difference. If, in the end, a citizen still disagrees, she has at least seen that the arguments she finds important have been publicly voiced and received a fair hearing. This acceptance is also easier because the current minority remains visibly present on the political stage and might be the future majority. Moreover, citizens need not stay resentful on the sidelines: they can always step in and participate in civil society or in party politics themselves in order to convince a wider audience.

This contrasts sharply with the bleak prospects for ‘popular control’ in a lottocratic regime. Citizens would have little formal or informal influence over sortition members. Malleson (2018, p. 407) rightly observes that in a sortition system citizens would be ‘disenfranchised from the political process’ as they lack the means ‘to transform their dissent into political power’. He adds that this might incite frustrated citizens to try and get what they want through extra-constitutional and possibly violent means. Abizadeh (2021, p. 798) even writes that ‘a purely sortitioned legislature is a potential recipe for civil war’ because without elections the means for more peaceful forms of mobilization and protest are disabled. The prospect of civil war seems remote, but we agree that the disenfranchisement entailed by strictly descriptive representation is cause for serious concern. Politics is not simply about reason: it is also about emotions and identification. If all goes well, the visible staging of conflicts enables us to peacefully channel societal disagreements and to maintain citizens’ identification with the political process as a whole. The scholarly debate about populism brings a similar message. People turn to authoritarian populists-recognizable public figures promising to ‘take backcontrol’—precisely because political institutions have become too depoliticized. Democracy requires a vibrant confrontation of ideas, giving people real political choices and making sure that oppositional voices are heard and recognized. This is not what lottocracy would bring. It would, instead, exacerbate depoliticization and strengthen the feeling that citizens are no longer in control of their own society.

Concluding Remarks

In this paper we have compared ideal-typical versions of a system based on elections with one based on sortition in light of their epistemic qualities, their means for popular control, and their ability to ensure the adherence of citizens. And we have argued that the electoral system is superior in all three dimensions. By way of conclusion, we would like to formulate three remarks concerning the limitations of our approach.

We recognize, first of all, that actual electoral systems fall severely short of the idealized model sketched above. An analysis of how and to what extent these shortcomings could be overcome is beyond present means and purposes. Our main aim in this contribution has been, instead, to question the widespread suggestion that problems with existing democracies should be solved or bypassed by simply transitioning to a lottocratic alternative. Our reconstruction of the ideal-type of electoral democracy serves as a reminder of its normative promises. When these promises are not realized, reforming electoral democracy and pushing it closer to its ideal form might well be the wiser option. In fact, rather straightforward and well-known institutional means are available to improve the quality of the democratic process in an electoral context.Footnote 9 The fact that these institutional changes are often not immediately politically feasible poses a formidable challenge but, here as well, sortition does not make for an attractive alternative as the political feasibility of institutionalizing a lottocratic system seems more questionable still. We believe that electoral democracy remains an ideal well worth fighting for and should remain at the center of our efforts to reinvigorate democracy.

Secondly, it should be emphasized that our arguments target lottocratic systems in which legislative power is fully in the hands of an allotted chamber. Nothing we have said directly precludes, for instance, the use of minipublics in a merely advisory role. When used to enrich public debate—for instance, by formulating policy options or arguments pro and contra certain proposals as in the Oregon Citizens Initiative Review (Warren and Gastil 2015)—rather than to shortcut it, they can meaningfully enrich the democratic process (Lafont 2020a). Yet we believe that the potential impact of minipublics on the public debate should not be overestimated. Their lack of visibility implies that their output will always be hard to diffuse in the public sphere. Mixed minipublics, which include elected politicians, might perhaps partially alleviate this problem but they would, at the same time, bring new risks. If politicians feel tied to their outcomes, these mixed minipublics would again function as shortcuts for, rather than contributions to, the collective decision-making process.

Although our critique of lottocracy does not directly apply to hybrid models in which an elected and an allotted chamber coexist and share legislative power (Gastil and Olin Wright 2018; Owen and Smith 2018; Abizadeh 2021), we believe that our arguments will prove relevant for such proposals.Footnote 10 This deserves a much more extensive treatment than can be provided here, but we tentatively submit that hybrid models face a dilemma depending on how the powers between the two chambers are distributed. If the sortition chamber has strong veto rights and de facto control over the legislative process, the problems we have discussed will arise here too. If, on the other hand, the sortition chamber has limited powers and is mostly an advisory organ, it might not have real impact and instead become a burdensome and possibly superfluous addition to what remains, in essence, an electoral system.

A third and final remark concerns the status of our criticisms. As explained at the outset, we believe that our arguments demonstrate that the problems with sortition arise already at the level of ideal theory. In the epistemic dimension, the limitation of deliberation to a randomly selected set of citizens prevents the emergence of a collective, society-wide debate that lives up to the normative criteria of inclusion and transformation. In the power dimension, sortition fails to distribute decision-making power in an equal way and does not provide mechanisms of authorization or accountability. It thus fails to recognize citizens as equal holders of popular sovereignty with meaningful control over collective decisions. In the motivational dimension, lottocratic deference threatens the perceived legitimacy of the system. It presupposes a blind trust that citizens will not be prepared to give and hence fails to provide citizens with a sense of ownership over collective decisions.

Our arguments crucially depend on the contrast between the visibility of electoral politics and the black box nature of sortition. We believe that this lack of visibility is an inherent feature of the lottocratic model that can be analyzed at the level of ideal theory and not an empirical shortcoming that could be fixed by institutional fine-tuning. In this context, it is important to emphasize once again that visibility is much more demanding than transparency.Footnote 11 Making minipublics transparent is not hard at all: we could, for instance, livestream the meetings or publish the minutes. The problem is, however, that very few citizens ‘read the transcripts of parliamentary and committee debates’ (Mansbridge et al. 2012, p. 20). Indeed, if we want to generate public awareness and understanding of what goes on inside such venues, we are bound by the laws of dramaturgy. These laws tell us that, if we want to reach a large audience, we need to capture its attention by staging a play with recognizable actors engaged in some minimally agonistic story.Footnote 12 As Parkinson (2005) has shown, this is exactly what minipublics, with their focus on internal deliberativeness, cannot offer.

In the course of our paper, we have dealt with several replies raised by advocates of sortition who believe that the problem of connecting the minipublic with the wider public is not insurmountable. Although we recognize that the matter deserves a more comprehensive treatment, our arguments at least show that the burden of proof rests on the lottocratic side. Think again, for instance, of the proposal to appoint spokespersons within the sortition assembly who have to go out in the media and explain the decisions made (Mansbridge 2019, pp. 197–203). As argued, this would open up a channel of external influence possibly affecting the force of the better argument and, thus, the internal deliberativeness of the process. At the same time, the prospect of becoming a spokesperson would be highly intimidating for most ordinary citizens (Malleson 2018, p. 408; Mansbridge 2019, p. 200) and would, therefore, lead to serious problems of self-selection, further undermining the claim that the minipublic accurately ‘mirrors’ the population at large (Landa and Pevnick 2021, p. 56). The proposal to appoint spokespersons thus provides a particular example of a much more general challenge. Lottocrats face what we have called a sortition dilemma: to the extent that sortition chambers aim to improve their connections with the wider public, they unavoidably undermine the very features—such as descriptive representation or internal deliberativeness—that constitute the core of the sortition ideal, and they, paradoxically, recreate features—visibility, interaction, accountability—that are much more readily realizable through traditional electoral mechanisms.