On blind deference in Open Democracy

Abstract

In this article, I critically assess Hélène Landemore's new model of Open Democracy, asking whether it requires of citizens to blindly defer to the decisions of the mini-public. To address this question, I, first, discuss three institutional mechanisms in Open Democracy, all of which can be read to grant citizens democratic control. I argue that neither the capacity to authorize the selection mechanism (random sortition), nor the lottocratic conception of political equality, nor the self-selection mechanisms of Landemore's model deliver the form of control that would insulate Open Democracy from the charge of blind deference. I then discuss the direct democracy mechanisms that Landemore incorporates into her model. Although these devices grant citizens control and, by extension, offer the resources for repudiating the charge of blind deference, they also, I argue, subvert the logic of Landemore's lottocratic model, rendering Open Democracy predictably unstable. Given the deliberative asymmetries between the mini-public and the citizenry, a high frequency of bottom-up challenges is to be expected that would throw the whole system into jeopardy. Thus, the challenge for lottocrats is to show how democratic control can be achieved in a lottocratic system without undermining the benefits of the legislature by lot.

Introduction

Hélène Landemore's new conception of Open Democracy invites us to fundamentally revisit our notion of democracy. It is one of the few large-scale interventions in the democratic theory of recent times that develops a comprehensive and systemic vision of institutions of “Popular Rule for the Twenty-First Century” (Landemore, 2017, 2020, 2021). Impelled by the urge to combat the oligarchic bias of electoral representation, Landemore envisions a form of democracy where ordinary citizens (rather than elites) would occupy the center of political power and take turns in representing one another. Her most ambitious proposal to achieve this desideratum is to complement or even replace electoral parliaments with open mini-publics composed through lotteries.

Anyone who harbors such aspirations cannot, however, fail to confront Cristina Lafont's recent charge against lottocratic conceptions of democracy. In Lafont's view, lottocratic decision-making requires of citizens to blindly defer to the decisions of randomly selected citizens (Lafont, 2015, 2019). Avoidance of blind deference is a cogent criterion for critically assessing different conceptions of democracy because it does not beg the question in favor of any particular conception of democracy (Lafont, 2021: 13–14). Rather, a requirement of blind deference—implying that citizens would lack the capacity to control and shape policies—is so obviously undemocratic that democratic theorists of very different persuasions should commit to avoiding it (Lafont, 2019: 8). If Lafont is right that democracy necessarily requires the avoidance of blind deference (as seems plausible), then we might usefully employ this criterion as a litmus test on Landemore's Open Democracy to judge whether it (potentially) deserves our endorsement.

To address whether citizens would be required to defer blindly to the decisions of the mini-publics in Open Democracy, I initially arrange my discussion around three institutional mechanisms, all three of which can be read to grant citizens democratic control. First, I discuss Landemore's claim that we can split the selection of legislators apart from their authorization with little to be lost (Landemore, 2020: 108). In Open Democracy, citizens would be authorizing the selection mechanism (random sortition) rather than individuals (as in elections). The problem with this proposal, I argue, is that it conflates two forms of control: control over the decision of whether to adopt a procedure and control as the capacity to impose a direction on a procedure. Avoidance of blind deference requires the latter form of control and the granted capacity to authorize random sortition, therefore, does not release Open Democracy from the charge of blind deference. Second, I turn to Landemore's lottocratic conception of political equality according to which citizens are politically equal because they have an equal chance of access to control and because actual access to control would be equalized over time as all citizens would have a meaningful chance of being drawn once in their lifetimes (Landemore, 2020: 90–91). The first argument offers no grounds for defeating the problem of blind deference simply because it does not realize control is shared equally among citizens but only an equal chance of access to it. The rationale behind the second one also fails to convince, implausibly asking citizens to give up their capacity for control for most of their lives (blind deference) in return for temporarily obtaining it at one unspecified point in the future (avoidance of blind deference). Third, I discuss whether citizens would acquire democratic control through the various self-selection mechanisms in Open Democracy (Landemore, 2020: 93–97). The answer is “no” because citizens—lacking decisional power—would have no means of ensuring that members of the mini-publics would be amenable to their inputs. Moreover, there are reasons to be concerned about the extent to which the influence that self-selection mechanisms make possible would materialize in a democratically representative fashion.

That the aforementioned mechanisms fail to deliver democratic control, may explain why Landemore suddenly changes tack and incorporates quite heterogeneous institutional devices (citizens’ initiatives, right of referral) from the toolbox of direct democracy (Landemore, 2020: 134–38). This marks a significant change in direction from the focus on legislative mini-publics and self-selection mechanisms. Although these mechanisms grant citizens democratic control and therefore arguably insulate Open Democracy from the charge of blind deference, they also, it turns out, subvert the logic of Landemore's lottocratic model. The problem arises from the deliberative disconnect between the mini-publics and the larger public which follows by design in Open Democracy. One would therefore expect frequent scenarios in which decisions of the mini-publics will have a majority sentiment among the broader public against them. Why wouldn’t we then, I ask, expect bottom-up challenges to be extremely frequent in Open Democracy to the point where they would undermine the stability of the political system? Landemore briefly addresses this problem, but, as I argue, her answer leaves much to be desired. The model, therefore, becomes predictably unstable. By introducing direct democracy devices, Open Democracy might steer clear of the problem of blind deference but at the expense of internal coherency.

My article is structured as follows. I begin by unpacking the notions of non-blind and blind deference (1). I then introduce Landemore's new paradigm, Open Democracy (2). Against this backdrop, I apply the criterion of blind deference to the three above-mentioned mechanisms (3). In the final section, I discuss the problems that arise once we combine lottocratic decision-making with mechanisms of direct democracy (4).

Blind and non-blind deference

According to Lafont, a helpful way to evaluate different conceptions of democracy is to examine the extent to which they require or expect citizens to blindly defer to the decisions of others. Such a requirement implies a failure to offer citizens the institutional control mechanisms that would enable them to shape the policies to which they are subject. Lafont identifies two features of blind deference that sets it apart from the democratically acceptable category of non-blind deference. First, she writes: “In representative democracies citizens are expected to delegate decisions to their representatives. However, to the extent that citizens maintain some capacity for control over these actors, they are not doing so blindly.” Call “maintaining some capacity for control” the procedural requirement of non-blind deference. Second, Lafont asserts that in “the first case [non-blind deference], one has some (defeasible) reason to assume that the political decisions endorsed by the agent to whom one is deferring are those that one would have endorsed if one had thought through the issue with access to the relevant information [alignment assumption]. By contrast, in the second case [blind deference], one has no reason to make this assumption” (Lafont, 2019: 8, my italics). In a more recent article, she adds that from the perspective of a blindly deferring person, “the agent's decisions could go either way” (Lafont, 2020b: 98). Call “having some reason to assume alignment” the substantive requirement of non-blind deference. To escape blind deference, both requirements must be met. Following Lafont, I will now illustrate the difference between non-blind and blind deference by contrasting electoral and lottocratic conceptions of democracy.

Electoral and lottocratic conceptions of democracy

The most basic way to understand deference is to think of it as a voluntarist concept delineating the act whereby one actor (e.g. a voter) freely defers decisions to another (e.g. a representative) (Warren, 2020: 86). By voting for a party candidate whose political interests I share, I non-blindly defer to the decisions of that agent. In this sense, electoral conceptions imply an expectation of non-blind deference, because periodic elections enable citizens to transfer decision-making power to like-minded representatives over whom they maintain some capacity for control (procedural requirement) insofar as they can hold them to account once the next election comes around. I emphasize “enable” to illustrate that elections only eliminate an expectation of blind deference. Whether non-blind deference will ensure ultimately depends on which grounds citizens defer to their representatives, which brings us to the substantive requirement. Imagine I am clueless about the political programs of the candidates on the ballot but decide to vote for the candidate whose physical attributes I find the most attractive. There is no expectation of blind deference within the system in this scenario, given the conduction of periodic elections that facilitate control. Yet, I can have no reason to assume that the decisions of the agent to whom I deferred coincide with the decisions I would have endorsed upon reflection, based on my political interests. In this scenario, I have merely given my consent to blind deference, which is to say that the procedural (having capacity for control—satisfied) and substantive (having reason to assume alignment—not satisfied) requirements have split apart. ¹

Relations of non-blind deference between represented and representative according to Lafont serve the purpose of generating alignment between the beliefs and preferences of the citizenry and the laws and policies with which they are expected to comply. Periodic elections enable each citizen of a polity to participate in the non-blind constitution of a group of representatives, who then (under the relevant constraints) get to shape policies in concordance with the political interests of their constituencies. By summating all individual instances of non-blind deference, electoral systems can, in principle and under suitable social circumstances, generate some degree of alignment (which, of course, is not to say that electoral systems automatically will achieve this, let alone that most contemporary electoral systems succeed on this count). This point can be appreciated further if we consider what Lafont refers to as lottocratic conceptions of democracy.

Defenders of lottocratic conceptions often base their arguments on a particular tension between participation and deliberation. Well-ordered deliberation, so the argument goes, can only be achieved in small numbers. Now, if we follow Fishkin in adopting an agnostic stance on participation (Fishkin, 2011: 191), the attraction of descriptively representative mini-publics should be clear: Mini-publics mirror the larger population as a kind of microcosm, and their small size makes it possible to facilitate high-quality deliberation between participants that serves the purpose of filtering out good arguments from bad ones. This combination of the mirror claim (democratic dimension) and the deliberative filter (epistemic dimension) underwrites many standard defenses of mini-publics (Fishkin, 2011, 2018; Goodin and Dryzek, 2006). As Lafont and others have argued, however, mini-publics mirror the population only at the pre-deliberative stage, that is, before the deliberative filter has exerted itself upon the political beliefs of the participants (Lafont, 2019: 118; Parkinson, 2006: 82; Pettit, 2012: 205). As a result of their deliberations, members of the mini-publics will often undergo significant changes in their political opinions. These transformations, however, will transpire only among members, with the larger population remaining largely untouched. Given this asymmetry, it is difficult to see how a lottocratic system, in which mini-publics would be given the legislative competencies that we currently grant electoral parliaments, could generate alignment. Quite to the contrary, an alienating misalignment appears to follow by design (Lafont, 2019: 97). This implication can be traced back to the problem of blind deference. Here is why:

In an electoral system, I am expected to non-blindly defer to the decisions of ideologically like-minded representatives. In a lottocratic system, in contrast, I would be required to blindly defer to the decisions of the mini-publics. This requirement of blind deference follows, according to Lafont, because citizens would lack the capacity for control over political decisions (violation of procedural requirement). Moreover, citizens could have no particular reason to assume that decisions endorsed by the majority within the mini-publics align with the decisions they would have endorsed upon reflection because the opposite might equally be the case, namely that they would have endorsed the minority position upon reflection (violation of substantive requirement) (Lafont, 2019: 116).

On this latter point, it could, perhaps, be held against Lafont's arguments that although electoral systems enable citizens to non-blindly defer to representatives, they do not provide any assurance that they will come to form part of the ruling majority. In both systems, lottocratic and electoral, it can reasonably be expected that an agent within the mini-publics or parliament will be advocating my position. The fact that I will have non-blindly deferred to the agent advocating it in the latter but not in the former case has no bearing on the outcome. Put differently, under a lottocratic as well as an electoral system, it is a procedure (majority rule) beyond the control of the agent defending my position that determines whether decisions to which I will be subject are the ones I would have endorsed upon reflection. And in both cases, “it could go either way,” to stick with Lafont's expression.

To appreciate the difference, we need to adopt a diachronic and aggregative perspective. When I non-blindly defer to a representative, I expect her to remain faithful to the policy agenda upon which my deference is based. The aggregate effect of such relations is the formation of a parliament, whose members will be expected to behave similarly, or else risk being voted out of office once the next election comes around. This implies that elected officials are constrained in terms of the policies they can plausibly pursue. If some representative has received a significant mandate (i.e., has been non-blindly deferred to by many) based on a political commitment, say, to defund the police, then there is a sense in which she will have a difficult time supporting a bill that would add significant resources to police departments across the country. ² In contrast, members of mini-publics would be operating under no such constraints. They would be free to defend whatever position they believe is right and equally free to change it as they deliberate along (Lafont, 2021: 11)—and this freedom can be traced back to the fact that they have not been authorized to pursue certain policy agendas (not been non-blindly deferred to) by citizens. Against this backdrop and given the deliberative asymmetries between the mini-publics and the broader population, it is, therefore, to be expected that actual public opinion over time would become increasingly detached from decision-making (Lafont, 2019: 121–22). Similarly, we should expect to see repeated scenarios where the decisions supported by a majority within the mini-publics would be opposed by a majority of the broader population. I will return to this issue in the final discussion.

What type of control is required to escape the charge of blind deference?

In a more recent article, Lafont specifies what kind of control is required for a conception of democracy to steer clear of blind deference: “Thus, for the purposes of my argument what matters is not how direct or indirect the control one has is, but rather how it bears on the substantive content of the decisions that one is subject to” (Lafont, 2020b: 99 my italics; see also Lafont 2020a: 206). The important issue, as I understand it, is not whether citizens are directly involved in legislation. What matters is rather that they have some capacity for control to ensure that policies will be aligned with their political values. Once indirect control is involved, it is essential that those directly tasked with legislation are suitably sensitive to the concerns of the citizens they claim to represent, not coincidently but because of the control citizens have. Here, Lafont's understanding of control seems to resemble that offered by Philip Pettit in “On the People's Terms” according to which control implies having the capacity to impose a direction, one that reflects the political values of the controlling agent (Pettit, 2012: 153). ³ The cited statement also seems to suggest, however, that we can conceive of control without the capacity to give direction, that is, control that does not bear on the substantive content of policies. As an example, we can imagine citizens being endowed with control over the decision of whether to adopt a particular decision-making procedure. Concerning blind deference, however, the decisive question is whether citizens have control understood as the capacity to impose a direction on that procedure. I return to this issue once I consider whether citizens would acquire democratic control through their capacity to authorize the selection mechanism (random sortition) and the concomitant capacity to revoke that authorization. Before I get to this, I need to provide a brief account of Landemore's Open Democracy.

Landemore's Open Democracy

According to Landemore, the current crisis of democracy can be traced back to a fundamental design flaw of representative democracy, namely electoral representation. Elections, she claims, inherently favor extraordinary individuals (wealthy, well-educated, charismatic, etc.) whose chances of being selected into parliament are disproportionately large. This oligarchic bias of elections implies that ordinary citizens are excluded from participation in law-making, the result being a “closure of power” (Landemore, 2020: 76). Not only do these discriminatory effects bring representative democracy into tension with fundamental democratic values such as inclusiveness and equality, but they also come at a high epistemic cost since electoral parliaments will be inadequately cognitively diverse, thus failing to tap into the collective intelligence of the people (Landemore, 2020: 220). Add to this that electoral democracy structured along party lines is detrimental to deliberation because members of the respective parties, whose main concern is to win elections, will not display the kind of open-mindedness to the claims of others constitutive of good deliberation (Landemore, 2020: 26; see also Mutz, 2006). Thus, there are intrinsically democratic and instrumental (epistemic) reasons for being skeptical of the value of elections.

For Landemore, these allegedly inherent shortcomings of the electoral paradigm provide grounds for fundamentally rethinking democratic institutions. Her new model, Open Democracy, is animated by this quest. The most ambitious proposal, in this context, is to complement or even replace electoral parliaments with descriptively representative mini-publics drawn through lot (Landemore, 2020: 13). In contrast to her previous book, where she offered a defense of random sortition on epistemic grounds (Landemore, 2013), the argument in Open Democracy centers around a claim concerning the improved “democraticity” of mini-publics—a criterion that measures the extent to which access to an assembly is inclusive and equal (Landemore, 2020: 81–82). Random sortition satisfies strict impartiality and inclusiveness because it does not recognize differences between citizens in terms of competence, education, and so forth. Equality would be satisfied because all citizens would enjoy the same chance of being drawn. Full satisfaction of the “democratic principles of inclusiveness and equality” Landemore claims would, therefore, imply “a representative body that is statistically identical with the demos” which explains the obvious democratic advantages of lottocratic over electoral representation: only the former can constitute an assembly whose members approximately mirror the entire demos. In response to the obvious objection that random sortition is not inclusive because by implication all those who are not drawn (the great majority) are excluded, Landemore suggests we adopt a diachronic perspective. Through frequent rotations and a decentralized polity (with numerous local mini-publics), “actual access to power” would be “strictly equalized over the long term” since all citizens would have “a meaningful chance […] of being chosen over the course of their lifetimes” (Landemore, 2020: 90–91). Citizens presumably should be swayed over—and accept being ruled by non-elected officials—by this equality argument not only because they all receive a lottery ticket prior to each rotation but also because of the confidence they can have in knowing that it is likely to yield a seat in a mini-public somewhere down the road.

The mini-publics in Open Democracy would be permanently open to the broader public—an openness facilitated through various self-selection mechanisms (crowdsourcing, participatory budgeting, and self-selected assemblies) enabling citizens to link up with the mini-publics. Such mechanisms are distinct from direct democracy mechanisms, first, in that they typically involve only a fraction of the population and, second, because they do not realize decisional power for those making use of them (Warren, 2013: 74–75). They provide citizens with opportunities to set the agenda, to provide substantial inputs for legislation, but the power to decide resides with members of the mini-publics. With a nod to Mark Warren and the notion of “citizen representation” (Warren, 2013), Landemore claims that these mechanisms can be seen as enabling a form of “self-selected representation” where those participating would be making claims in the name of other (non-participating) citizens with no actual authorization from those on behalf of whom they claim to speak. The democratic value of self-selected representation can be traced back to two sources. First, it derives from “the fact that representatives tend to be ordinary citizens as opposed to professional politicians” whose crowdsourced inputs would acquire legitimacy only to the extent that they are other-regarding (Landemore, 2020: 75). Second, the democratic value of self-selection mechanisms stems from the fact that they would be “spatially open” to everyone, independently of wealth, competence, and so forth (Landemore, 2020: 95).

Apart from these proposals, Landemore introduces five institutional principles of Open Democracy (Landemore, 2020: 135–145), of which I will focus only on participation rights and deliberation. ⁴ Participation rights are fundamental for Landemore because in a true democracy, everyone is entitled to participate in legislation. We have already seen how this normative commitment is anchored in Open Democracy in relation to mini-publics where access to participation would be equalized over time and in relation to self-selection mechanisms where access would be permanently open to everyone. In addition, Landemore proposes to incorporate direct democracy mechanisms that can be activated on the initiative of citizens, qua participatory rights, subject to no constraints other than meeting a certain signature threshold. Among these mechanisms, we can distinguish between constructive and contestatory ones. Constructive mechanisms comprise “citizens’ initiatives” that would provide citizens with the opportunity to trigger a constituency-wide referendum on a self-made proposal for legislation. Contestatory mechanisms comprise “a right of referral” that would enable citizens to force a referendum on a policy enacted by the mini-publics, with the hope of overturning it by having a majority vote against it (Landemore, 2020: 134–138). Unlike the self-selection mechanisms discussed above, these mechanisms entail decisional power and are supposed to involve all citizens (or at least a majority of them).

Deliberation makes up the second institutional principle. Landemore sides with the Habermasian conception of legitimacy, according to which decisions acquire legitimacy only if they result from actual deliberations (Landemore, 2020: 138). In the Habermasian version of deliberation, participants engage in the give-and-take of arguments along several validity dimensions (truth, normative rightness, and truthfulness), and in doing so, they commit to several counterfactual presuppositions (that everyone is the given opportunity to raise and respond to arguments independently of social status, that participants must mean what they say, etc.) (Habermas, 1998: 44). In contrast to the Habermasian model, however, Open Democracy is explicitly committed to “facilitated deliberation”—the conscious effort to approximate the idealized presuppositions that underwrite Habermasian deliberation. Lottocratic representation is crucial to this end because of its non-partisan and inclusive credentials, but careful facilitation is still required to alleviate the power asymmetries that might otherwise disproportionally affect the deliberations in the mini-publics. Among the benefits of deliberation, Landemore points out that deliberation has been shown to induce agreement among deliberators and that it contributes to the structuring of preferences towards single-peakedness, thus making decision-making procedures less vulnerable to majority cycles (Landemore, 2020: 139).

It is important to stress that Landemore is thoroughly committed to a micro-conception of deliberation. She agrees with the assumption that deliberation is possible only in small numbers, and she explicitly distances her model from the more capacious conceptions of deliberation that she associates with authors such as Habermas and Lafont, classifying them as “deliberation in the wild” (Landemore, 2020: 114). According to the latter, deliberation is not limited to facilitated face-to-face interaction that occurs in assemblies but also refers to the anarchic and decentralized flows of communication that move between different agents and arenas in society (Benhabib, 1996: 73; Habermas, 1996: 287–387). Landemore rejects these macro-deliberative approaches, claiming that they fail to do justice to the ideal of deliberation as a “process involving free and equal individuals.” In a telling footnote, she states that “Deliberation […] requires either a small enough scale or a division of labor allowing some to deliberate on behalf of others” (Landemore, 2020: n. 17, 116). This presumably would imply that the cited benefits of deliberation (inducement of agreement, single-peaked preferences, etc.) would transpire only among members in the mini-publics. As I argue in the last section, this restrictive conception of deliberation matters decisively for the overall coherency of Open Democracy, especially when we hold it up against the direct democracy devices of the model.

Does Open Democracy expect citizens to blindly defer?

With this brief account of Landemore's Open Democracy out of the way, it seems that we can now turn to the main question concerning whether Open Democracy expects citizens to blindly defer to the decisions of the mini-public. Before proceeding, however, I need to confront a question that hangs over this critical inquiry: does Landemore even consider some form of control to be necessary for democracy? Some readers of Landemore appear to believe that Open Democracy entails no commitment to democratic control. Daniela Cammack, for instance, writes that Open Democracy “neglects […] the kratic” part of democracy, by which she refers “to the one real bit of political leverage ordinary citizens have at their disposal: their kratos, that is, their overwhelming power to prevail when they come together collectively” (Cammack, 2022: 5). In a similar but more polemical vein, Peter Stone claims that “[i]n Open Democracy, Landemore out-Madisons Madison” because it “reserves no important space for the people to act in its collective capacity” (Stone, 2022: 5). Let's call this the non-control-based reading of Open Democracy. If this reading is correct, then arguably it would be misplaced to critically address whether Open Democracy steers clear of blind deference because, on that reading, Landemore does not share a commitment to its avoidance. The inquiry should then rather be of a more fundamental nature, addressing whether democratic control is a principle that democratic theorists reasonably can dispense with. In my view, however, the non-control-based reading is mistaken, failing to acknowledge the control-related properties that Landemore builds into her model. To show why, I must first consider the features of Open Democracy which may be offered in support of this reading, and then, in a second step, show why they nonetheless fail to vindicate it.

One of the main aims of Open Democracy, as we have seen, is to realize democraticity, a property that is supposed to secure that access to a representative assembly is inclusive and equal (Landemore, 2020: 81). Sortition is crucial for delivering on this aim because it includes all citizens on an equal basis, entailing no discrimination between them on the basis of competence, connectedness, wealth, and so forth. At the same time, it should be clear that democraticity so construed fails to realize control for all those citizens who are not drawn to the mini-public. The strong emphasis on democraticity thus may explain why some authors take Open Democracy to entail no commitment to democratic control. Notice, however, that the non-control-based interpretation would follow from this emphasis only if Landemore conceived of democracy as being reducible to democraticity. While I acknowledge that Open Democracy contains arguments and considerations that could be cited in support of this reductionist reading, I actually think they fail to render that reading plausible. Let me elaborate by addressing these arguments and considerations head-on.

In Open Democracy, Landemore distinguishes democraticity from democratic legitimacy, claiming the latter but not the former to derive from majoritarian authorization (Landemore, 2020: 106–108). If we conceive of majoritarian authorization as a form of control (more on this below) and democraticity as being unrelated to control, we might be led to infer that Landemore considers control to be necessary for democratic legitimacy but not for democracy per se. ⁵ The inference is ultimately misplaced because it suggests that democracy simply means democraticity for Landemore and thus, a fortiori, that Open Democracy is exhausted by its commitment to democraticity. While it is correct that Landemore conceives of democraticity as a property that obtains independently of majoritarian authorization, she makes it clear in this context that a mini-public, to earn the capacity to claim obedience to its decisions, would have to have its selection mechanism (random sortition) authorized by at least a majority of citizens (Landemore, 2020: 116). Now, since democraticity doesn’t confer that authority on mini-publics (Landemore, 2020: 107), it cannot be the case that Open Democracy is reducible to democraticity. Thus, the circumstance that Landemore sees democraticity and democratic legitimacy as distinct properties shouldn’t be read to imply that she thinks democracy requires the former but not the latter—because without democratic legitimacy mini-publics could not claim obedience to their decisions and so could not perform the role that Landemore assigns to them in Open Democracy. If we relate these considerations back to the main issue concerning the plausibility of the non-control-based reading of Open democracy, the following picture emerges. Whereas democraticity hardly can be understood as aiming to grant control to citizens at large, democratic legitimacy can plausibly be understood to have this aim to the extent that it depends on majoritarian authorization of the selection mechanism as a necessary condition (Landemore, 2020: 105, 116). Thus, if I am right that Open Democracy reflects a dual commitment to democraticity and democratic legitimacy, and if we conceive of majoritarian authorization upon which the latter depends as a feature that is supposed to realize control, then it seems to me that we must reject the non-control-based reading on this basis.

Landemore, however, makes an additional claim which may be cited in support of the non-control-based reading. She conceives of responsiveness and accountability as secondary and extrinsic democratic properties, whereas democraticity, in her opinion, is an intrinsic and more fundamental democratic property (Landemore, 2020: 106–108). Since responsiveness and accountability arguably are closely related to democratic control, wouldn’t it be reasonable to interpret this claim to substantiate the suspicion that control, for Landemore, is not a democratic requirement? ⁶ Notice, however, that it would substantiate this suspicion only if we understand secondary and extrinsic to mean something along the lines of optional or dispensable. In other words, the claim would support the non-control-based reading only if it were intended to convey that democracy ultimately could do without responsiveness and accountability. By contrast, I believe that the claim should be understood as follows. First, they are secondary or extrinsic in the sense that they are not distinctively or uniquely democratic because, as Landemore makes clear, they could, in principle, be pursued and realized by non-democratic regimes (Landemore, 2020: n. 8, 89). It would, however, be a clear category mistake to conclude from this point that they are therefore not necessary for democracy. Second and relatedly, they are secondary and extrinsic according to Landemore in the sense of being instrumental in correcting “certain inegalitarian features and exclusion” that inevitably will arise because of the distance between represented and representatives. In fact, Landemore claims “that responsiveness and accountability are required as correctives” to these concerns (Landemore, 2020: 88). ⁷ Thus, the claim that democraticity is an intrinsic democratic property, whereas accountability and responsiveness are extrinsic and secondary ones, should not, as I read it, be interpreted to entail that the former is required in Open Democracy, whereas the latter fall outside the requirement of the model. For this reason, it also does not offer a sound basis for substantiating the non-control-based reading of Open Democracy.

There is a final feature that speaks against the non-control-based reading. In support of his claim that Open Democracy “reserves no important space for the people to act in its collective capacity,” Stone claims that “[r]eferenda play no major role in it …” (Stone, 2022: 5). I am puzzled by this claim. Open Democracy, as we have seen, contains a number of participation rights such as citizens’ initiatives and a right of referral, both of which citizens can activate on their own accord (Landemore, 2020: 134–138). Thus, pace Stone, it seems to be an open matter whether referenda will play a “major role in” Open Democracy because this will depend on the extent to which citizens decide to exercise their participation rights in the form of referenda. It is worth mentioning, in this context, that Landemore claims that these rights, as opposed to traditional liberal political rights, are supposed to facilitate “control of the government by the citizens” (Landemore, 2020: 136). More recently, she has added that “moments of mass authorization via mass referenda are required for the legitimacy of the whole system” (Landemore, 2022: 9). If moments of mass authorization via mass referenda are required in Open Democracy and seen as facilitating control of government by citizens (as seems plausible), then it seems to me that we must conceive of Open Democracy as a control-based conception of democracy (which, of course, doesn’t mean that control is the only aim of the model, keeping in mind the strong emphasis on democraticity).

Taken together, these considerations, I believe, support my reading: with Open Democracy, Landemore seeks to defend a lottocratic model based on the value of democraticity, but with the clear aim of steering clear of Lafont's control- and blind deference-based objections. ⁸ If I am right about this, then it seems to me reasonable to ask whether Open Democracy achieves this through its institutional features and mechanisms and if yes, in a way that is plausible and coherent. We begin with Landemore's proposal to replace the direct authorization of representatives with the authorization of the mechanism (random sortition) through which they are chosen.

Deferring to mechanisms versus deferring to individual participants

Consider the following extract from Open Democracy:

In elections, elected representatives get both selected and authorized in one move […]. But there is no reason why one couldn’t separate the selection and authorization moments. In fact, the individual members of lottocratic assemblies need not be directly individually authorized as long as the selection mechanism picking them (a lottery, an algorithm) is explicitly authorized. (Landemore, 2022: 118, my italics)

The problem with this suggestion is that it conflates two distinct forms of control that should be kept apart. When citizens authorize (or deauthorize) random sortition and its associated decision-making procedure, they exercise a crude, but no less important, form of control: control over whether to adopt a particular procedure or not. When citizens authorize or non-blindly defer to elected representatives, they exercise a form of control that is meant to steer a presumably already agreed upon procedure in a particular direction, one that reflects their political values and preferences. The difference can be further appreciated by considering the following example:

Imagine that I owe a radio-controlled (RC) car that seems to have taken on a life of its own, moving in a wild and unrestrained fashion, not following the instructions that I give through my remote. I can remove its batteries to prevent it from making any further movements, but I cannot give it direction.

In one sense, I control the RC car to the extent that I can put in and take out its batteries at will. As the example illustrates, this opaque form of control, however, does not automatically equip me with the form of control that would enable me to give it direction once the batteries have been inserted. Analogously, control over the decision whether to adopt a particular selection mechanism and decision-making procedure does not entail control in the sense of being able to give the procedure direction once it is operative. Merely consenting to a procedure, in other words, is insufficient for establishing democratic control (Pettit, 2012: 157–60).

This scenario in which citizens would maintain control in one but not in the other dimension appears to be implied by lottocratic conceptions of democracy. In this sense, the relationship between me and my RC-car resembles the relationship between citizens and lottocratic decision-procedures: they get to decide whether to adopt the procedures or not, but once they have been adopted, they cease to have any control over their direction. If citizens authorize random sortition in the way Landemore envisages, they actually give their consent to blind deference, that is, they agree to adopt a procedure upon which they will be unable to impose a direction (Lafont, 2020b: 99). The question that needs to be asked, however, is whether consent or rather the threat of revoking consent could not come to assume a form of control that would involve the capacity to give direction. I could make my RC car obedient to my demands by threatening to remove its batteries. Unfortunately, a similar strategy of threatening to revoke authorization of the selection mechanism to give the decisions of the mini-publics direction would most likely fall on deaf ears among mini-public participants who would legislate only for one short period. Why would they even care? It would be like threatening to withdraw your support for an American president who is well into her second term.

The suggestion that “there is no reason” not to believe that authorization and selection could not split apart appears in the given context to be based on a category mistake that masks an important difference between two levels at which political autonomy can be operationalized. Citizens have political autonomy if they can regulate their collective affairs at both levels:

At the level of normal politics, this requires a political system with control over the relevant issues (e.g. economic, social, and environmental) that provides citizens with democratic rights and institutions. […] At [the] level of constitutional politics, political autonomy requires that citizens are able to determine the conditions of collectively binding decision-making. (Patberg, 2020: 145)

This useful distinction suggests that we should conceive of authorization of the selection mechanism, random sortition, as an expression of political autonomy at the constitutional level, especially if we envisage it as a replacement for elections. Authorization of individual elected representatives, in contrast, happens within the boundaries of a constitution and counts as an exercise of political autonomy at the level of ordinary politics. The requirement of explicit authorization of the selection mechanism thus at most shows that Open Democracy is committed to political autonomy at the level of constitutional politics, but it does not thereby deliver political autonomy (democratic control) in normal politics. So, to sum up this sub-section, the capacity to authorize the selection mechanism and the concomitant capacity to revoke that authorization do not grant citizens the form of control that would insulate Open Democracy from the charge of blind deference.

Political equality in Open Democracy

Recall that political equality is satisfied in Open Democracy not only because citizens stand an equal chance of being drawn into the mini-publics, but also due to the promise that “access to power” would be “strictly equalized over the long term” because everyone would have “a meaningful chance […] of being chosen over the course of their lifetimes” (Landemore, 2020: 90–91). The first aspect has already been effectively rebutted by Thomas Cristiano:

We can see the failure of lottery systems as systems of political equality if we think of the analogous case of substituting equal distribution of chances at material resources for equal distribution of resources. […] For example, suppose that we could choose between paying three equally deserving workers the same wage or paying them with equal chances at all the wages. And suppose that we chose the latter alternative and the consequence was that one of the workers got all the income […] would this be as fair as paying each the same wage? […] the lottery system […] is a very thin form of equality because it contracts the reach of equality to just the initial lottery. (Christiano, 2008: 109).

Substituting control for the wages in Christiano's example, the picture emerges that the lottocratic conception of political equality implies an equal chance of access to control (as in one person, one lottery ticket) but no equal distribution of control (as in one person, one vote). The implications are, as we have seen, that members of the mini-publics would be under no pressure to make decisions aligned with the views of the broader population because they do not earn their powers through processes in which citizens exercise control on an equal basis. Translating Christiano's considerations into the language of deference, we find that lotteries would give citizens an equal chance of being placed in a situation that involves no deference (because they would be legislators and have a great deal of control) but should their lottery tickets turn out to be losers then they would be left in blind deference (due to a lack of control). As Jan-Werner Müller puts it: “Thus lottocracy […] shrink[s] the demos; not all citizens will be the decision-making public” (Müller, 2021: 87, my italics).

Against these considerations, it could be argued that Landemore's conception of political equality really isn’t about the equal distribution of control at all. Thus, scrutinizing it through the lenses of that category appears to miss the point because its justification lies elsewhere: it equalizes the opportunity to be a representative. It is worth recalling, in this context, the additional feature that Landemore builds into her model in response to the abovementioned objection that this conception of political equality entails the exclusion of all those who are not drawn. She believes, as we have seen, that this objection can be met through frequent rotations and a decentralized polity (with numerous local mini-publics)—features that are supposed to ensure that “actual access to power” would be “strictly equalized over the long term” because all citizens would have “a meaningful chance […] of being chosen over the course of their lifetimes” (2020: 90–91). This diachronic solution, I believe, can be read as an attempt to steer clear of blind deference. Let me explain. In the current context, we can distinguish between two types of control which would entail that citizens or a subset of them would not be required to blindly defer. There will be no such requirement if they can exercise a direct form of control over decisions (as participants in a legislature) or an indirect form of control (through electoral authorization and accountability). Against this background, Landemore's diachronic solution can be read as aiming to equalize the absence of blind deference over time: all citizens can reasonably expect to get to exercise a direct form of control in a mini-public in some period of their lives, during which they will therefore not be required to blindly defer.

The remaining question is whether this feature offers a reasonable solution to the problem of blind deference. I submit that it does not. The rationale behind this suggestion implies that citizens would trade no control for most of their lives in return for temporarily obtaining it at some unspecified point in the future. In other words, citizens are promised control in one period (avoidance of blind deference) at the cost of being acquiesced into blind deference for the remainder of their lives. But this excessively diachronic perspective doesn’t capture the democratic significance of control. Citizens do not value control because they harbor a secret desire to become legislators one day but rather because of a sustained concern they have with the laws and policies to which they are subject (Lafont, 2019: 18–25; Rostbøll, 2008: 178; Pettit, 1999: 184). Suppose a mother who flies into a rage when a new school reform is announced by the mini-public, being convinced that it would negatively affect the future of her kids. The assurance that she will get to legislate at some unspecified point in the future (perhaps when her kids have already left school) offers little solace. It is unreasonable to ask citizens to give up their capacity for control for the greater part of their lives in return for the (near) assurance of temporarily obtaining it at one point in the future. At least, if we accept—as we should—that the value of control derives from the continuous concern that citizens have with the content of policies and laws.

Self-selection mechanisms

The mother could turn to the permanently open self-selection mechanisms at her disposal in Open Democracy. A crowdsourcing platform, for instance, would enable the mother to try to change the hearts and minds of the members of the mini-publics and steer them—through the use of well-crafted arguments—toward adopting the school reform that she deems appropriate. Essentially these mechanisms of public consultation would make it possible for citizens to exert discursive pressure on the members of the mini-publics and in so doing come to have a real influence on legislation. Is this then the institutional device that grants citizens democratic control in an Open Democracy, thus releasing it from the charge of blind deference? I am afraid not.

Although such mechanisms might carry the potential to improve the democratic credentials of a political system, they do not seem to qualify as platforms for democratic control because citizens—lacking decisional power—would have no means of ensuring that members of the mini-publics would be amenable to their inputs. The problem consists not so much in the fact that these self-selection mechanisms themselves are forums without decisional power. Rather it derives from the absence of elections (or some equivalent mechanism) that do entail such power and which are typically considered to be the decisive mechanism that provides legislators with incentives to pay attention to public opinion (Gerhards, 1993: 26; Chambers, 2012: 64; Habermas, 2006: 148, 2009: 167; Lafont, 2020a, 212). Without voting rights (strong power), rights to public consultation (weak discursive power) are compatible with being totally ignored—and with no obvious consequences for the members of the mini-publics.

Even if we assume that members of the mini-publics would be naturally disposed to listen to these inputs (I return to this question in session four) and willingly incorporate them into legislation, these self-selection mechanisms would run up against certain structural limitations that should give us pause in thinking of them as platforms for democratic control. It is not possible to enroll the entire citizenry in these public consultation processes, and even if it were, members of the mini-publics could hardly judge more than a fraction of the inputs they receive from citizens on their substantive merits (Malleson, 2018: 407). As soon as participation in these mechanisms rises above a certain (relatively low) threshold, members of the mini-publics will therefore have no choice but to ignore many of the suggestions addressed to them by their fellow citizens. One would be hard-pressed to argue that democratic control would be realized in such scenarios in which many citizens will be left with a sense of not having been heard.

In order to make the case that these mechanisms facilitate democratic control, it, therefore, seems that one would have to grant Landemore's controversial proposal that non-participants will somehow feel represented by those who successfully manage to influence legislation, with the latter serving as self-selected representatives for non-participants. This strikes me as a dubious proposition for a number of reasons. First, it is unclear how non-participants can exercise control through self-selected participants in the absence of some mechanisms of authorization and accountability. Second, even if this problem can be resolved, the question remains of why we should expect this form of influence to materialize in a democratically representative fashion. For this to be the case, all conflicting political views would have to be made present in the process and carry an influence that is somehow proportional to their prevalence in the broader society. It is difficult to see how this could be achieved under self-selection mechanisms in the absence of voting processes meant to determine the relative strength of the conflicting positions. Landemore is aware of the dangers that come with self-selection, as influence might be skewed in favor of certain social groups more likely to participate in processes of public consultation, but she offers little guidance as to how these anti-democratic implications might be mitigated.

To recapitulate, it seems that the proposed self-selection mechanisms fail to release Open Democracy from the charge of blind deference because they do not realize a form of control that would be equally shared among citizens and enable them to effectively impose a direction on the laws and policies to which they are subject. ⁹

Can the democratic credentials of Lottocracy be saved by introducing elements of direct democracy?

That none of the above mechanisms deliver democratic control, may explain why Landemore suddenly changes tack and incorporates quite heterogeneous institutional devices (citizens’ initiatives, right of referral) from the toolbox of direct democracy. These devices would leave it up to citizens to continuously determine which decisions to cede to lottocratic procedures and a fortiori to decide how often to impose their own will(s) upon the laws and policies to which they will be subject. Citizens’ initiatives, on the one hand, would enable citizens to bypass the mini-public by placing proposed statuses on the ballot. A right of referral, on the other, can be seen as facilitating a strong form of accountability, enabling citizens to express their dissatisfaction with the randomly selected legislators and ultimately overturn the laws and policies that they have passed. These mechanisms thus grant citizens some robust capacity to shape the policies and laws to which they are subject and therefore arguably release Open Democracy from the charge of blind deference.

It seems to me, however, that these devices subvert the logic of Landemore's lottocratic model. The problem is that Landemore's defense of mini-publics is premised on the notion of descriptive representation and a conception of deliberation restricted to cover only facilitated face-to-face interactions in small groups. In this sense her model follows a strict Fishkinian logic according to which the verdicts of mini-publics would track the counterfactual considered opinion of the people rather than the actual, raw one (with all its deficiencies) we would get if we were to consult the entire public through voting mechanisms (Fishkin, 2011, 2018). The result is a mismatch between the two opinions. It is important to reiterate that this cannot be swept aside as an empirical issue. It is an implication that follows by design because the newly acquired insights, change of opinions, and so forth that result from the deliberative exchanges will transpire only among the members of the mini-publics (Lafont, 2019: 97). ¹⁰

In an Open Democracy, it is, therefore, to be expected that decisions of the mini-publics (supported by a majority within the mini-publics) frequently will have a majority sentiment among the broader population against them. In such cases, it would be an easy catch for citizens in an Open Democracy to activate their right of referral and trigger a constituency-wide referendum on some law, knowing well that most of their fellow citizens are on their side. And in cases in which the majority opinions of the mini- and macro-publics would coincide by accident, powerful interest groups could try to sway public opinion towards their hidden agendas and, once successful, gather the number of signatures required for triggering a referendum on the law that goes against their special interests. The members of the mini-publics might be shielded from such influences, ¹¹ but no such protection can be offered to the broader population. The bewildering implication of Landemore's Open Democracy can be summed up as follows: political decisions would be based on considered opinions generated in carefully facilitated micro-deliberative environments. Whereas the frequency of the bottom-up challenges meant to challenge those decisions would be determined by actual public opinion(s), originating in dynamic processes of social mobilization outside the reach of orderly facilitation.

This brings us to the pressing question: Why wouldn’t we, given the disconnect between the two publics, expect bottom-up challenges (facilitated through a right of referral) to be extremely frequent in an Open Democracy to the point where they would paralyze the legislative work of the mini-publics and thus throw the whole system into jeopardy? Landemore very briefly addresses this challenge, but her answer leaves much to be desired:

For one thing, in Open Democracy much of the deliberation would take place upstream of the decision process. To the extent that the whole legislative process itself is open and participatory, from the agenda-setting stage to the final passing of the law, including, crucially, the drafting stage, the likelihood of a bottom-up referendum organized to challenge the resulting law would be low […] (Landemore, 2020: 150).

To begin with, it is not clear what is meant by the statement that “much of the deliberation would take place upstream of the decision process,” keeping Landemore's restrictive understanding of deliberation in mind. In the narrowest sense, it could be read as referring to the deliberation in the mini-publics. This seems like a dead end, however, considering that these discursive exchanges would convince only the members of the mini-publics of the correctness of legislation—and therefore hardly offer a shield against bottom-up challenges that would be initiated by other agents in society.

A more charitable reading would be that Landemore is referring to the self-selection mechanisms on offer in Open Democracy, the suggestion being that because these processes would be arranged in a deliberative fashion, the likelihood of a bottom-up challenge would remain low. This is similarly puzzling, bearing in mind that these mechanisms would involve only a fraction of the citizenry. It is telling that Landemore claims that the “whole legislative process” would be participatory, given that the mechanisms she is alluding to would draw in only a limited number of participants. But why then should we expect these mechanisms to decisively affect the frequency of bottom-up challenges? They might give participants a sense of why some law was chosen, but for the remaining citizenry who did not take part (the great majority), there is no particular reason why they should not mobilize in an attempt to force a challenge on a law.

Let us, however, for the sake of argument assume that these mechanisms of public consultation could make the legislative process so inclusive that all citizens (or at least enough of them) would get a sense of having been heard, either directly as participants or indirectly through self-selected representatives, thus rendering bottom-up challenges unlikely. The plausibility of the claim concerning a low likelihood of bottom-up challenges would now seem to rely on a natural willingness of members of the mini-publics to listen to inputs. But how can we be certain about this amenable disposition?

When citizens are drawn into mini-publics, they embark on a journey during which they will be exposed to a great deal of material and information (provided to them by academics and other experts) that will affect how they perceive the most pressing challenges faced by society. The political views that come out of such edifying processes might not sit well with the more experience-based, sometimes less compromising (and, perhaps, manipulated) inputs they receive from citizens through these self-selection mechanisms: “the messy streets and the lottocratic-technocratic isoloir in which randomly selected citizens meet might remain completely separate worlds” (Müller, 2021: 86). Members of the mini-publics in Open Democracy accordingly would be placed in an awkward limbo, not knowing whether to decide as their conscience dictates (in accordance with their deliberatively filtered beliefs) or to incorporate crowdsourced inputs into their decisions that, from their perspective, would detract from the substantive quality of the law or policy under consideration. The latter option would jeopardize the alleged deliberative or epistemic benefits of decision-making by lot. The former, on the other hand, might end up alienating the broader population, thus significantly increasing the chances of bottom-up challenges.

Given the disconnect between the mini-publics and the larger population—public consultation, notwithstanding—there is no reason to believe that referendums in Open Democracy would not be extremely frequent, and the model, therefore, becomes predictably unstable: controlling lottocracy through direct democracy is a volatile proposition. By way of conclusion, we might say that Open Democracy arguably avoids the problem of blind deference but at the expense of internal coherency.

Conclusion

In this article, I discussed whether citizens in Open Democracy would be required to blindly defer to the decisions of the mini-publics. To address this question, I initially arranged my discussion around three mechanisms that could be read to grant citizens democratic control in Open Democracy. First, I discussed Landemore's proposal to replace the authorization of individuals with the authorization of the mechanism through which they are selected. I argued that this suggestion rests on a category mistake that conflates two different forms of control: control over whether to adopt a procedure and control as the capacity to impose a direction upon a procedure. The capacity to authorize the selection mechanism does not grant citizens the latter form of control and consequently offers no basis for defeating the charge of blind deference. Second, I turned to Landemore's lottocratic conception of political equality according to which equality is satisfied because citizens have an equal chance of access to control and because actual access to control would be equalized over time. The first argument fails to convince because it does not realize control is shared equally among citizens but only an equal chance of access to it. The rationale behind the second one is also questionable, implausibly asking citizens to give up their capacity for control for the greater parts of their lives in return for the (near) assurance of temporally obtaining it at some unspecified point in the future. Third, I argued that the various self-selection mechanisms of public consultation fail to realize democratic control because citizens—lacking decisional power—would have no means of ensuring that members of the mini-publics would be amenable to their inputs. I also raised some concerns about the extent to which the influence, that self-selection-mechanisms make possible, would materialize in a democratically representative fashion. The direct democracy mechanisms, in contrast, do grant citizens a degree of democratic control and, by extension, offer the resources for repudiating the charge of blind deference. Yet, as I argued, these elements of direct democracy subvert the logic of Landemore's lottocratic model, rendering Open Democracy unstable. The deliberative asymmetries between the mini-publics and the larger population predictably would lead to an intolerably high frequency of referendums that would throw the whole system into jeopardy. Thus, the future challenge for lottocrats consists in showing how democratic control can be achieved in a lottocratic system without undermining the alleged benefits of legislature by lot.