Lottocracy or psephocracy? Democracy,elections,and random selection

Abstract

Would randomly selecting legislators be more democratic than electing them? Lottocrats argue (reasonably) that contemporary regimes are not very democratic and (more questionably) that replacing elections with sortition would mitigate elite capture and improve political decisions. I argue that a lottocracy would, in fact, be likely to perform worse on these metrics than a system of representation that appoints at least some legislators using election – a psephocracy (from psēphizein, to vote). Even today's actually existing psephocracies, which are far from ideally democratic, are better suited than a lottocracy would be to meet the demands of democratic citizenship (politics must be legible to ordinary people, who must have low-cost opportunities to participate) and the demands of democratic leadership (powerful representatives should be specialized and constrained by competitions for popular support). Democrats therefore have weighty instrumental reasons to reject lottocracy and work to democratize psephocracy, instead.

Aristotle understood ‘electoral democracy’ as oxymoronic, on one influential reading, because elections are for oligarchs and democracy involves choosing leaders by lot (Manin, 1997: ch. 1). Rousseau maintained (improbably) that ‘the old republics […] never had representatives’ and that the ‘marvellous system’ of electoral representation was ‘“found in the woods” […] of Germania’ (Rousseau, 1997: bk. 3, ch. 15). Early modern Florentines and Venetians used elections without thereby seeing themselves as democrats (Manin, 1997: ch. 2). Even the French and American founders often took pains to deny any relation between democracy and elective government: ‘Democracies have ever been spectacles of turbulence and contention’, they proclaimed (Madison, 2003: 44), and ‘representatives are not democrats’ (Sieyès, 2003: 147).

In contemporary political theory it is lottocrats – advocates for randomly selecting all legislators – who view elections as undemocratic, arguing inter alia that a lottocratic legislature would make wiser decisions than an elected chamber because of its greater diversity and its unpartisan open-mindedness (Landemore, 2020); and that it would be less vulnerable to capture, where an ‘official is captured if he or she uses his or her position to advance the interests of the powerful, rather than to create policy that is […] good’, because sortition would block powerful interests from influencing the selection of legislators and regular rotation in office would stop them forming long-term quid-pro-quo relations with representatives (Guerrero, 2014: 142). ¹

This article defends the democratic credentials of elections for top legislative office, or what I will call psephocracy. The common names for such a system – electoral or representative democracy – imply a pre-judgement about whether elections are democratic that is absent from psephocracy, which derives from the Ancient Greek psēphizein ‘to vote’ and to the practice of voting by dropping psēphos – pebbles – into urns (Boegehold, 1963). Such a system can be genuinely democratic, I will argue, insofar as it renders decisions legible to ordinary people; mobilizes the public into broad coalitions; encourages specialization by leaders; and induces competition for popular support. A well-ordered system of elections for top legislative offices could plausibly help fulfil some of the most important demands of democratic citizenship (legibility and mobilization) and demands of democratic leadership (specialization and competition).

A lottocracy would be far less likely to fulfil these demands – less likely even than today's liberal capitalist psephocracies, which are only partly or intermittently democratic. As Dimitri Landa and Ryan Pevnick have argued, lottocratic legislators would lack the disciplining prospect of re-election and the special knowledge, skills, and networks that could help them assert themselves relative to actors from powerful economic and bureaucratic systems (Landa and Pevnick, 2021). A lottocracy would pose a serious threat to healthy civil-military relations, for instance. ² Eliminating elections would also destroy an opportunity to participate in large coalitions at a low cost that is especially valuable to the less powerful. Finally, a legislature free from party politics and its associated attempts at public justification and storytelling would be harder for ordinary people to engage with and understand. This would bias opportunities for political influence through public discourse and action in favour of elites (people who are socially powerful because connected to networks that control substantial resources; see Tilly, 2007) and away from ordinary people (who are not). A lottocracy, therefore, would likely be highly vulnerable to capture by elites and liable to make policy that is biased and unwise.

This argument contributes to the growing debate about the instrumental vices and virtues of lottocracy (Bech-Pedersen, 2023; Guerrero, 2014; Lafont, 2015, 2020; Landa and Pevnick, 2021; Landemore, 2020; Umbers, 2021). ³ Whilst lottocracy has occasionally been defended as intrinsically more democratic than psephocracy (Guerrero, 2014: 168–170), both advocates and sceptics tend to focus instead on its likely consequences, with each critic tending to identify a different bad effect that it could have. ⁴ This article adopts a more systematic approach, grounded in a new theory of democratic benefits of elections for top legislative office in modern bureaucratic states – benefits captured in the idea that a psephocracy could fulfil the demands of democratic citizenship and leadership and that a lottocracy could not.

In conclusion, I suggest that other proposals for incorporating randomly selected representatives into the political process might also usefully be evaluated in the lights of these demands. The present argument therefore also speaks indirectly to a much wider set of debates about the democratic potentials of sortition assemblies (for which see, among many others: Bagg, 2022; Gastil and Wright, 2019; Malleson, 2023; Owen and Smith, 2018; Sintomer, 2018; Stone, 2011; Vandamme et al., 2018).

This defence of psephocracy is compatible with a range of theories of democracy and its value. It assumes that larger groups must inevitably divide the labour involved in making decisions, creating some sort of system of representation, in the minimal sense of a network in which some act on behalf of others (Hutton Ferris, 2022; Rey, 2023) and that those systems necessarily include a legislature: a group that is ordinarily authorized to decide whether a proposal is accepted or rejected as a binding rule. I also assume that democratic institutions must embody popular agency (Booth Chapman, 2022: ch. 6; Stilz, 2015) and political equality (Beitz, 1989) – though my arguments are compatible with multiple ways of cashing out those terms. As will become clear, my defence of psephocracy also rests on two slightly more controversial assumptions: the participatory democratic premise that democratization is impossible without more political engagement and participation by non-elites (Pateman, 1970); and the representative democratic idea that representatives can and should help them do this by triggering interest and participation (Urbinati, 2006). ⁵

Though this article defends the democratic importance of elections, it is not a defence of the electoral status quo. Contemporary electoral politics suffers from a host of well-documented ills, including (in some places) destructive forms of affective partisanship and polarization (Graham and Svolik, 2020) and a media politics of spectacle and diversion with at times worryingly tenuous connection to truth (Hacker and Pierson, 2020; but see Chambers, 2021). Lottocrats are, moreover, right to worry about bad decisions and capture, which occur routinely and can sometimes be deeply at odds with democratic values: party cartelization appears to be relatively widespread (Katz and Mair, 2009; Mair, 2013) and business interests use their capital to exert disproportionate influence over governments (Schlozman et al., 2018). For instance, billionaire-backed organizations like Americans for Prosperity play important informal roles in some US state legislatures, more or less drafting legislation for them (Hertel-Fernandez, 2019; Skocpol and Hertel-Fernandez, 2016).

The empirical evidence for the capture of electoral legislatures that is cited by lottocrats is, however, often not as clear-cut as they suggest. If elected governments are responsive only to socioeconomic elites (Gilens, 2012; as cited by Guerrero, 2014) then why do they sustain welfare states that benefit primarily the middle classes (Elkjær and Iversen, 2020)? And if elections leave representatives ‘basically unaccountable’ to their constituents, (Guerrero, 2014: 165) then why do they tend to partially fulfil most of their manifesto commitments (and 80% or more of them in low veto point systems: Thomson et al., 2007)? The available empirical evidence is sometimes complex and ambiguous, shows variations in outcomes and capture across polities, and does not rule out reforms designed to mitigate capture within an electoral framework. ⁶

In any case, the question asked here is not ‘do bad decisions and capture occur?’ but: ‘would things be better in a lottocracy?’ I argue that, even compared to contemporary psephocracies with their sizeable democratic deficits, things would probably be much worse.

Why psephocracy?

Competition

The most familiar way elections serve democratic ends is by encouraging leaders to compete for popular support. The basic argument is simple and plausible across a wide range of contexts: officials will tend to use their position to serve their own personal or group interests, instead of distributing policy benefits and burdens more fairly, unless they face the prospect of being removed from office by those they claim to serve (Mill, 1865; Shapiro, 1999). This might be true even for those who sincerely want to promote the common good and act impartially but nonetheless understand and sympathize with members of some social groups better than others, for instance, because their friends are all middle class. This simple truth has grown into something of a dogma since the middle of the last century and some have seen elite competition as the hallmark of democratic politics, or perhaps even its sufficient condition (following Schumpeter, 1942).

Lottocrats push back against the simplistic conflation of democracy with the election of officeholders by pointing to practices and regimes whose democratic credentials seem to depend on other institutional features, such as the town-square voting on policy that still occurs in some Swiss Landsgemeinde. Athens is an exemplar of a polity that looks robustly democratic, at least for older free male citizens, but that seems to rely on non-electoral mechanisms to realize popular agency and political equality. By the fourth century BCE, Athens used lotteries to select participants for most administrative offices (hai archai), for the People's Courts (dikasteria) that monitored them, and for the Council of 500 (boule hoi peniakosioi) that acted as a kind of chief executive, receiving foreign dignitaries and setting much of the agenda for the Assembly (ecclesia), which primarily voted on policy. ⁷ The Athenians apparently understood this regular use of sortition as a way of deterring capture of the political system by elites, both by limiting volunteerism and rotating people quickly in and out of office (Hansen, 1991: 235ff), and a tradition going back perhaps as far as Aristotle views Athens’ use of lotteries as the single most democratic feature of the regime (see especially Aristotle, 1998: bk. 4, chs 9 and 15 and bk. 6, ch. 2).

Yet Athens may be an apparent exception that proves the rule that democratic politics benefits when at least some top officeholders compete for popular support. In his funeral oration, Pericles praised Athens not only for its equality but also because ‘preferment for office is determined on merit, not by rank but by personal worth’ (Thucydides, 2013: II37). ⁸ Many of the most influential roles in the polity were unelected, to be sure: most importantly the office of rhetor, or speaker in a formal political assembly, which required little more than talking in public and being responsible for what one said. But Athenians relied on elections to fill positions that demanded more specialized knowledge and skills, such as generalships and top financial officers, and influential rhetors often also occupied formal elective offices and used the formal powers granted by those offices to secure opportunities for influence and achieve their political goals (see, for instance Hansen, 1983). Leadership was provided initially by dominant general orators, such as Pericles, and then later by dominant rhetors such as Demosthenes who also held office as elected treasurers. ⁹ This reliance on elections may look surprising and potentially at odds with the democratic equality Athenians were so proud of.

Why, then, was the use of election viewed as compatible with democratic ideals? Most fundamentally, everyone had the formal opportunity to exert political influence and play leadership roles and it was apparently equality of opportunity that mattered to Athenians, not equality of outcome (Hansen, 1991: 83–84). Elections are at least formally open and inclusive. They also, importantly, facilitate competition between leaders designed to help citizens identify those with specific virtues (Hansen, 1991: 83–84) including, crucially, a clear commitment to democracy itself as a regime (Ober, 1989). Fourth-century Athenians sometimes reasonably worried about elites undermining democracy, as they had done in the oligarchic coup of 411 BCE. Regular elections for office helped ensure that officials who held powerful offices for a long time, such as generals, depended on the continuing goodwill of the people (reaffirmed each year in the case of the strategoi) and on their ability to persuasively present themselves as true partisans of the democratic cause.

Specialization

The other major democratic benefit Athenians derived from their elections was specialization, in the sense of matching people with specific knowledge and skills to offices in which those advantages could be put to good use. If Athenians had selected generals or top treasurers at random from men over thirty, as they did other officials, this would have increased the chances of military defeat, or of an interruption in the flow of silver from the mines. Using elections to harness specific skills to the public interest did not imply disrespect for democratic equality so long as leaders presented themselves and plausibly acted as agents of the wider public, though it did mean that office-seekers had somehow to cast themselves as both democrats in spirit, servants of the people, and as members of a social elite with the special skills and knowledge requisite for effective leadership (Ober, 1989). Specialization is important for deepening democracy too: if Athenian women had started struggling for the franchise their movement would have benefited from skilled and specialized leadership.

The need for highly specialized leadership is greater in modern mass societies with professionalized administrations and market economies. To make them compatible with democratic values, however, legislators must be able to assert themselves relative to skilled and knowledgeable administrators and powerful corporate actors, which demands specific knowledge (e.g. of economics, law, and policy) and skills (e.g. forcefulness and eloquence). ¹⁰ Political self-assertion was simply not as difficult in Athens, where regulatory needs were simpler and most bureaucrats were amateurs. A contemporary finance minister, for instance, needs more statistical training and a better grasp of economic theory than her Athenian counterparts. ¹¹

Legibility

Political legibility is crucial for democracy. Politics is legible to the extent that citizens generally can see the relations between groups, processes, and interests relevant to collective decision-making and also have a lively sense of how they affect their interests. ¹² Making political life legible can serve the interests of the powerful by facilitating manipulation and control (Scott, 1998). But legibility also advantages ordinary people by improving their attentiveness to and understanding of political decisions and public life more generally. For instance, North American socialists won power on the prairies, not the cities, because the former were unusually legible to the farmers gathered around the grain elevator at harvest time, who couldn’t help but recognize their class interest in fighting the exploitative fees charged by its owners (Lipset, 1950). In fact, legibility is intrinsically democratic in one important sense: it reduces information asymmetries between social elites and ordinary people because the former can invest resources such as time and money in analysing and interpreting masses of data about a very complex political situation, whereas the latter are more reliant on quick and easy consumption of accessible information (Gallego, 2015; Hutton Ferris, 2022). ¹³

Elections create democratic legibility by sustaining a party politics that engages ordinary people and helps them to interpret and understand decision-making processes. Athenians could rely on non-partisan sources of legibility because they and their acquaintances regularly participated in political trials and administrative decisions, provoking engaging and informative everyday political talk. Citizens in today's large polities with mediated public spheres rely more on the manufacturing of political legibility by dedicated specialists, and parties have a special role to play in this information environment. Political parties sometimes really have been as detractors claim: manipulative factions that serve special interests, sow division, spread misinformation, and threaten democracy by encouraging loyalty to the party over fidelity to constitutional norms (Rosenblum, 2010: chs 1–2 and 4–6). Yet partisan political discourse can and often does engage and orient the mass public by theatrically staging claims and justifications, imbuing public debates with focus and urgency (Rummens, 2012). Parties help citizens interpret political life by telling a relatively comprehensive story about the state of the political community that is intended to speak to diverse groups and connect their experiences to larger trends and events (Rosenblum, 2010: chs 7–8). Finally, the adversarial process associated with party politics can also play an important epistemic role for citizens by contributing to normative public reasoning about policy, articulating and contesting justifications for past decisions and future goals (White and Ypi, 2016: ch. 3).

Mobilization

The engagement and understanding produced by legibility will not facilitate popular empowerment unless people can do something with it. Elites can relatively easily afford to undertake individualized forms of participation, such as contacting officials, whereas those with fewer individual resources to invest in participation must rely more on joining big and broad coalitions, making up for individual powerlessness by combining what they have (Hutton Ferris, 2022). Those big coalitions depend on organization (Klein, 2022), which is why unions were once such a powerful tool for the working class. And while elites can invest their resources in higher intensity political activity, for instance leading a campaign to reform local services, non-elites depend more on participation that is low-cost, in the sense of requiring little investment of resources such as money and time.

Elections help mobilize by creating low-cost opportunities for ordinary people to join very large political coalitions. In Ancient Athens, elections did not create parties and they were, in any case, less important because citizens were also empowered by participation in bodies such as the People's Courts, which was low-cost because remunerated. But no more than a fraction of the much larger citizenries of modern states can practically wield power at the highest levels of government, as the Athenian ordinary citizens did. Leadership elections therefore serve as a weapon of the weak that helps to mitigate the unfair gaps in the value of formally equal politically liberties. ¹⁴

*

The claims made so far do not amount to a defence of elections or psephocracy in general, and the potential virtues identified are, quite clearly, not well-realized in every electoral regime. Electoral institutions in a ‘no party democracy’ like Peru, for instance, aid legibility or mobilization very little (Levitsky and Cameron, 2003). However, elections for top legislative office can and often to some extent do help promote popular agency and political equality by helping realize the demands of democratic citizenship (by creating legibility and mobilizing ordinary people) and demands of democratic leadership (by incentivizing representatives to specialize and constraining them with competitions for popular support). The remainder of this article will argue that a lottocracy would probably face much more serious problems realizing these goods and that democrats should, therefore, prefer psephocracy. It proceeds by considering the two most systematic and apparently persuasive cases yet made for preferring randomly selected to elected legislatures.

Lottocracy and capture

Alexander Guerrero argues that psephocratic legislatures are more vulnerable to capture than a possible lottocratic alternative. Ordinary people are politically ignorant and therefore incapable of eliciting meaningful accountability from elected officials, which frees the latter to use their position to advance the interests of the powerful, especially the financially powerful. Contemporary politics is simply too information intensive for ordinary people to be well-informed and cognitive shortcuts (e.g. party labels) cannot help them solve the cognitive challenges associated with electoral participation because they are either too broad-brush to be helpful or too fine-grained to be readily utilizable. Guerrero compares this situation to a system of lottocratic governance composed of many single-issue lottery-selected legislatures (SILLs). He suggests that participants could sit for one-year terms, with perhaps three hundred people per SILL. Powerful interests would not be able to influence who participates (because selection relies on sortition) and the frequent rotation of members would make it costly to bribe them. Lottocracy would therefore be able to resist capture more easily than can psephocracy.

There are various concerns we might have about this argument. The use of multiple single-issue legislatures is likely to lead to incoherent policy, in which the externalities of each policy on others are not considered sufficiently attentively by legislators. Proposals for lottocracy that avoid this consequence, for instance those that advocate for a single lottocratic chamber, are more promising. We might also take issue with Guerrero's argument about the problems with cognitive shortcuts (that they are either too fine- or too coarse-grained), which has a suspect form. This is like saying that consumers can only buy cheap shoddy clothes or well-made ones that are extortionately expensive. Presumably, they can find ways to view and solve this as an optimisation problem. We need to hear a specific reason why the virtue between two vices is unobtainable, which Guerrero fails to provide (see Guerrero, 2014: 146). ¹⁵

The central problem with Guerrero's argument is its complacency about the prospects of capture of a lottocratic legislature, which is grounded in a failure to appreciation the demands of democratic leadership and the way psephocracy can help fulfil them. Modern legislatures need to be capable of asserting themselves in relation to a skilled and knowledgeable bureaucracy, on the one hand, and wealthy and assertive assemblage of lobbyists and pressure groups, on the other hand. Their willingness to do this is strengthened by their need to compete for popular support and their ability to do so depends on legislative specialization (possession of knowledge and skills that help representatives make effective policy). ¹⁶ A lottocratic legislature would, by comparison with many existing psephocratic ones, be dangerously lacking in its ability to realize the demands of democratic leadership.

Dimitri Landa and Ryan Pevnick have framed this insight in a useful way by focusing on the question of what the term limits of lottocratic legislators should be in Guerrero's system. They point out that lottocrats who lack the prospect of re-election will have a freer hand to sell their services to outside groups willing to pay them for their support after their term in office (call this societal capture). ¹⁷ Moreover, rapid turnover obstructs the development by lottocrats of the kinds of knowledge and skills that would be necessary to push back against bureaucratic attempts to assert their own agenda and resist unwelcome ideas, for example, with foot-dragging (call this capture by officials). Lottocrats face a dilemma when it comes to term limits: either they are short (e.g. a year), which will facilitate bureaucratic capture, or they could be made very long to resist this (e.g. a decade), in which case the risk of societal capture becomes extremely high (Landa and Pevnick, 2021).

Whilst Landa and Pevnick are right to recognize the danger of lottocratic capture, their positive analysis about why psephocracy is likely to avoid capture is less compelling. They argue that elected legislators are more likely to choose beneficial rather than biased policies for two reasons: they are potentially pivotal voters on bills, which gives them a strong incentive to gain political information and develop policy expertise; and they face electoral accountability, which concentrates them on using that knowledge in ways that benefit ordinary people (Landa and Pevnick, 2020). This defence of psephocracy focuses on the dynamics of democratic leadership and completely ignores the way elections help fulfil the demands of democratic citizenship. In its analysis of inter-elite dynamics, moreover, it adopts an atomistic approach that fails to appreciate the democratic value of coalition and organization. All existing legislative elections involve some relatively safe seats, such that only some psephocrats need to worry about direct electoral accountability. ¹⁸ Furthermore, few legislators are, in fact, routinely pivotal. In the current US Senate, perhaps only one or two representatives face the kinds of incentives to develop expertise celebrated by Landa and Pevnick.

Their defence of psephocracy underrates the role parties play in producing a policy that is good, unbiased, and uncaptured. Whilst only a few legislators are regularly pivotal in a way that incentivizes the development of expertise, partisan divisions of labour (within committee systems, for instance) encourage leadership and deference within specific policy domains amongst co-partisan legislators. Furthermore, whilst many elected representatives are in safe seats, parties help to amplify and transmit the effects of whatever electoral vulnerability there is. The need to fight elections as a team gives every legislator, not only the most vulnerable, important incentives to respond to polity-wide shifts in preferences. Without parties, in short, the electoral vulnerability and expertise of some legislators would not translate into incentives for knowledge and responsiveness across the chamber as a whole – hence also for good unbiased policy. ¹⁹

How could a lottocrat respond to the claim, on which Landa and Pevnick and I agree, that a lottocratic legislature would be much more vulnerable to capture than an elected one? Hélène Landemore points out that the People's Courts in Athens, whose members were selected using sortition from a pool of volunteers, were not captured by the elites (see also Cammack, 2022). It was instead the Assembly, whose members were self-selected by simply turning up in the morning, that was dominated by ‘the gifted, connected, and often rich orators’ (Landemore, 2020: 67). The relevance of this evidence is questionable given the much greater size and professionalism of contemporary state bureaucracies, relative to Athens, and the greater power exerted by commercial interests under a capitalist economic system. More obviously relevant, however, is her claim about our contemporary experience with consultative deliberative assemblies: ‘if we look at empirical evidence from mini-publics and citizen assemblies around the world’, she argues, ‘there is, to my knowledge, not a single case of demonstrated capture by elected representatives, experts, bureaucrats, or lobbyists’ (Landemore, 2020: 196–197).

The evidence here is not, unfortunately, as clear-cut as Landemore suggests. Although deliberative assemblies have had some important benefits (for instance Farrell et al., 2019; Knobloch et al., 2020), they have also occasionally been used strategically by politicians and administrators to provide a democratic veneer for policies they prefer. In Landemore's native France, for instance, there was a very high-profile attempt by the central government to use a minipublic to simulate public support for genetic modification (Curato and Böker, 2016; see also Freschi and Mete, 2009). Some research from Canada suggests their recommendations have often been ignored when they did not align with officials pre-existing preferences (Johnson, 2015) – and officials may also be more tempted to relate strategically to a lottocratic legislature than they presently do towards consultative assemblies, given the much greater opportunities this would create for power and influence. ²⁰

It could be responded that greater influence over the legislative process by unelected officials would be no bad thing. Perhaps we should trust in bureaucratic professionalism and create a lottocratic legislature with shorter term limits on the basis that wealthy lobbyists seem much more dangerous than public-spirited technocrats. One influential vision of a lottocratic legislature adopts exactly this approach by giving administrators control over its agenda (Dryzek and Niemeyer, 2008).

We should avoid this kind of complacency about the risks of capture by officials. Even when they conscientiously strive to serve the public, their institutional roles and loyalties shape their perspectives, which can then lose touch with popular preferences and interests. Given their substantial political power, this can have serious and unwelcome consequences.

Consider, for instance, the likely state of civil-military relations in a lottocracy. Bureaucrats are not the only powerful agents of state and standing armies have been famously worrying to republicans throughout history – and for good reason. A conventional view is that civilian control of the military is achieved by widespread normative support for the idea of a professional and apolitical military. However, there is reason to doubt the pervasiveness of these norms amongst either the soldiers (Brooks et al., 2022) or the citizenry (Krebs et al., 2023), at least in some countries, and it is now normal to model relations between civilian leaders and the military as processes of bargaining in which both sides play strategic games for influence, such as leaking information or making public appeals. ²¹ One of the central dilemmas faced by legislators in overseeing the military is their lack of expertise (even when legislatures have specialized defence committees with professional staff). Oversight of the military is improved when legislators spend longer in office and on defence committees, which allows them to gain knowledge, as well as to plug into relevant networks (Giraldo, 2009). Oversight is also improved when legislators feel efficacious and competent to intervene in military affairs, ²² which is likely bolstered by allegiance to party programs that express more-or-less coherent visions that relate military priorities to other policy areas. ²³ Although elected representatives are probably too susceptible to military pressure, a lottocratic legislature may be far more vulnerable to capture because ordinary citizens are sometimes very deferential to military elites (Golby et al., 2018) and would be much less likely to feel motivated and knowledgeable enough to push back against undue pressure from the generals.

Lottocratic wisdom?

In Open Democracy, Hélène Landemore argues that lottocracy would be preferable to psephocracy because it would tend to make better decisions. ²⁴ Let us set aside the question of capture and assume that we have found some way to make lottocracy no less vulnerable than psephocracy in this respect (for instance by setting up an ‘oversight council’ of the kind discussed by Bouricius, 2013). This will allow us to consider, in detail and on their own terms, the two additional kinds of argument Landemore makes for favouring lottocracy:

Elections introduce systematic discriminatory effects in terms of who has access to power, specifically agenda-setting power. By so doing, elections skew the type of perspectives and input that shape law-making, likely resulting in suboptimal results. Second, elections entail a type of party politics that is itself not all that conducive to deliberation or its prerequisite virtues, such as open-mindedness, rather than partisanship. (Landemore, 2020: 26)

The first argument focuses on the value of diversity. Landemore associates elections with a ‘filter’ model of representation, according to which representatives should ideally be unusually competent and virtuous people: a ‘natural aristocracy’ (Landemore, 2020: 40). She opposes this to a ‘mirror’ model, according to which a group of representatives should perfectly resemble or reflect the group they represent, in the sense that personal qualities and characteristics should be distributed in the same way within both groups. The two models are intrinsically at odds because elections tend to be won more easily by people who are unusually rich or well-connected and, even in more propitious circumstances (in which money played less of a role in political life), voters would select candidates with rare but valued characteristics including not only competence and virtue but also, for instance, eloquence or good looks. The result is a legislative chamber that looks different to the wider population and is more homogenous in terms of the distribution and range of different types of people and personalities it contains. Landemore claims that there is something intrinsically undemocratic about the fact that different kinds of people do not have equal chances of winning elections and – from her epistemic-democratic perspective – this is ultimately lamentable because of its effects on the quality of legislation. The basic problem with the filter theory of representation (and the central virtue of a mirror model) is that contemporary social science has revealed that diversity can be actually much more important for high-quality decision-making than individual virtue and competence (Landemore, 2013a). Landemore's argument in favour of this proposition is complex and involves extrapolating from the findings of an agent-based model simulating a collective problem-solving process. The basic intuition is that individually competent political decision-makers tend to think alike and that injecting some more cognitive diversity into a decision-making group – provided by less competent agents – can productively reveal new solutions to a problem that everyone can recognize as superior once they see them. ‘Diversity trumps ability’ in the sense that (under certain conditions) ‘a randomly selected collection of problem solvers outperforms a collection of the best individual problem solvers’ (Hong and Page, 2004: 16,388; see Landemore, 2013a: 4). Elections either choose incompetent leaders, in which case they fail to secure good representation as it is conceived by the filter theory, or they do choose competent leaders, who then go on to make worse decisions than a more diverse group would – and it is this second and less familiar point that Landemore focuses on in Open Democracy: electoral institutions do not incorporate recognition of the value of diversity for good decisions.

Landemore's second argument against elections appeals to the dangers of partisanship and, though it is not stated as straightforwardly as readers might wish, it may be even more central to her overall defence of lottocracy. Its starting point is the idea that elections create parties and parties create partisanship, understood as ‘a willingness to commit uncritically (at least up to a point) to a set of values and principles’ (Landemore, 2020: 133). Partisans tend to avoid political talk with those they disagree with because it creates unwelcome conflict, which makes it harder for them to cultivate open-mindedness than it would be for citizens of a post-electoral democracy. ²⁵ Though Landemore does not spell out as clearly as she could have what exactly is wrong with the lack of open-mindedness, the most likely answer is that it leads to bad policy. The simulations that underwrite Landemore's argument assume away preference- or value-diversity amongst the problem solvers, because this might stop them from recognizing the same ‘objectively’ superior solution to problems and their diverse cognitive heuristics would not have the right kind of educative effect on each other (Page, 2007: ch. 9). Partisanship is bad because it makes legislators resistant to changing their minds on the basis of good reasons from colleagues in other parties, which undermines policy deliberation and decision quality.

In light of these problems with psephocracy – problems of legislative homogeneity and of partisan rigidity – Landemore proposes policy-making by an ‘open mini-public’: ‘a large, all-purpose, randomly selected assembly of between 150 and a thousand people or so, gathered for an extended period of time (from at least a few days to a few years) for the purpose of agenda-setting and law-making of some kind, and connected via crowdsourcing platforms and deliberative forums (including other mini-publics) to the larger population’ (Landemore, 2020: 13, 219). This minipublic would sit at the centre of a representative system structured as ‘a web of connected mini-publics all staffed with randomly selected citizens’, ‘some of them single-issue, others generalist, operating at various sub-levels of the polity’ (Landemore, 2020: 219). It would be connected to the wider citizenry by lower-level minipublics and crowdsourcing platforms that would help set its agenda and propose policies, as well as by initiatives and referendums that could be used to set its agenda or overturn its policies.

These arguments have some important merits. Landemore's proposed design for lottocracy is superior to Guerrero's, since the single-chamber open-mini-public would avoid the policy externalities and incoherences that would plague a system of single-issue legislatures. Her epistemic argument also contains an important kernel of truth: some failures of descriptive representation are very hard to justify from an epistemic democratic perspective. The better descriptive representation of women and marginalized ethnic and racial minorities, in particular, would improve policy because of the gains to diversity of legislative knowledge and experience, whether or not these are best understood in light of their relationship to ‘cognitive heuristics’ (Mansbridge, 1999).

We do not need to abolish psephocracy to improve descriptive representation, however. Institutional reforms such as party gender quotas or reserved seating could do the job, especially if accompanied by a supportive shift in relevant social norms. They might also help improve attempts to ensure the descriptive representation of class (assuming this would be desirable, but see Phillips, 1995: ch. 7). Landemore might insist that elections will nonetheless always tend to disadvantage candidates unusually lacking in specific skills (e.g. articulacy or sociability) and characteristics (e.g. confidence or bravery), that these qualities are correlated with the kinds of cognitive heuristics people employ (for instance confident people may be more risk-tolerant when considering possible solutions to a social or political problem), and that people with serious social anxiety, say, might make a distinctive contribution to legislative debate and decision. It seems possible, however, for the cognitive diversity of an elected legislature to be improved through, for instance, a system of reserved seating for cognitively or neurologically diverse candidates. ²⁶

Landemore probably cannot be satisfied with such measures because they solve the problem of cognitive homogeneity but not the problem of partisan rigidity. The greater diversity they achieve will not trump ability unless legislators are open-minded and willing to recognize the ‘objectively’ better solutions to political problems suggested by members of other parties.

There is, however, reason to be wary of the idea that diverse deliberating groups will only decide well when members are fully open-minded, in Landemore's sense, to each other's arguments and ideas. In the simulations that ground her argument it is not only deep conflicts of interests that get in the way of cognitively diverse groups making good decisions: ‘tastes, or what we call diverse preferences’ also ‘create problems – huge problems’ (Page, 2007: 239; for a fuller discussion see Ancell, 2017). Lottocratic legislators are therefore apparently assumed not to have diverse interests or values, even of a relatively superficial sort. This kind of assumption is likely a dangerous guide to politics in the real world, which involves diverse preferences, opinions, interests, substantial ‘reasonable disagreement’, and the ‘burdens of judgement’ (Rawls, 1993: 54–58). ²⁷

Even granting Landemore's strongly cognitivist starting point for the sake of argument, we might question whether partisan rigidity is worth abolishing all things considered. After all, this would require abolishing the legislative role for parties, which would stretch already existing gaps in knowledge, attention, and participation between elite and non-elite citizens. The latter are much more reliant on the legibility provided by partisan storytelling and public justification and on the cheap opportunities to participate in very large political coalitions that partisan politics provides. A post-electoral lottocracy would therefore deplete crucial sources of information and attentiveness for ordinary people. It would also be likely to widen participation gaps: by eliminating the lowest cost formally empowered political practice; by drying up partisan sources of interest and information; and by reducing the participation-promoting civic virtues that legislative elections appear to help generate, such as political interest (Zeglovits and Zandonella, 2013) and a sense of participatory competence (Valentino et al., 2009). Eliminating legislative elections threatens to instigate a vicious cycle of political de-skilling and disengagement that would cause greater problems for those who rely more on partisan legibility and mobilization. ²⁸

These threats to citizenship are undesirable impediments to political equality and may also be harmful from Landemore's narrowly epistemic perspective. To see why, we need to consider the position of a lottocratic legislature within wider systems of deliberation, pressure, and representation that link it to civil society. Unequal social power is sometimes translated into partial political outcomes by subtle biases in public discourse that find their way into legislative debate when lawmakers act as conduits for ideas and arguments in civil society. So, for instance, the interpretation of an economic policy aired by business leaders might have a weight in legislative decisions that it not solely related to its epistemic quality. Lottocratic legislators would, like psephocrats, inevitably also serve as conduits for ideas and narratives developed elsewhere. In a lottocracy, however, widening attentional and participatory inequalities would have amplified biases in public discourse. Even bracketing conscious and direct attempts at capture, therefore, decisions and outcomes may end up being less impartial and lower quality than they are now.

Landemore might respond (following Guerrero, 2014: 177–178) that there could be a robust associational infrastructure and strong forms of civic action by non-elites even in the absence of elections and parties. But parties can fulfil the demands of democratic citizenship better than other associations because, whilst other civil society organizations tend to have limited concerns, parties have a ‘take’ on every major political issue. They make a distinctive contribution to public justification by bundling issues together and thinking about trade-offs between priorities more rigorously than any other organization. Moreover, successful parties have many more supporters than any other association, so they facilitate the building of distinctively large coalitions and have an unparalleled capacity to generate solidarity and collective agency. ²⁹ Parties are therefore simply irreplaceable in bureaucratic capitalist polities as vehicles for democratic legibility and low-cost participation.

It could be objected that this argument simply misses a key virtue of lottocracy: legibility and low-cost participation opportunities would no longer matter, in this context, because most people would have no formally defined contribution to make to legislative politics. There would be simply no need for people to be politically informed or engaged, it could be argued, because nothing specific would be demanded of them unless they happen to be chosen to become legislators for a year, which is extremely unlikely. This could be viewed as one way of responding to the incompatibility that some deliberative democrats have identified between the thoughtfulness of decision-making processes and their inclusion of large numbers of citizens (Fishkin, 2009). Landemore could simply accept a more passive role for most.

We should be wary of any fantasy of a democratic regime that could survive without an active and informed citizenry. Officeholders of all sorts, including generals and other top executives, often seek to maintain or expand their capacities for action and to resist constraints on their autonomy, including the democratic constraints provided by civil society and public opinion. They are liable to face substantial counterpressure, however, if the polity is home to an active citizenry habituated to public political communication and action. A group whose members are personally attached to free speech and association will defend its right to engage in them and, ‘by defending its own rights against the demands of power’, it ‘saves common liberties’ (Tocqueville, 2010: 1269; see also Habermas, 1996: ch. 8). A disengaged citizenry, by contrast, would less reliably resist usurpations by the politically powerful (and may also be more vulnerable to demagogic declamations of democratic slowness and compromise, whose political value is hard to recognize for those unpractised in popular politics: Elliott, 2023: ch. 2).

One final possible objection to the present argument must be considered: namely, that the epistemic-democratic interpretation of Open Democracy offered here is mistaken. Landemore has previously argued that instrumental defences of democracy are superior to intrinsic ones, as well as to hybrid approaches that mix both kinds of considerations (Landemore, 2013a: 51, 2013b: xviii, 2014: 192–195), and readers tend to interpret the Open Democracy in purely epistemic terms (e.g. Fumagalli, 2023; Tong, 2022). But Palle Bech Pederson has recently argued that Landemore makes three arguments for lottocracy, rather than two: an argument from cognitive diversity, an argument from partisanship, and an altogether more weighty argument from democraticness (or ‘democraticity’: Landemore, 2020: 85–89, 164). The latter supposedly centres on the open and inclusive way lottocracy selects representatives and presents this as intrinsically valuable, independent of its consequences (Bech-Pedersen, 2023: 7; who points readers to suggestive passages at Landemore, 2020: 81–82 and 220).

This textual issue cannot and need not be conclusively settled here. The epistemic interpretation is suggested by some external evidence: Landemore's own retrospective framings of Open Democracy uniformly present it as a contribution to epistemic democratic theory (see, for instance Landemore, 2021: 370, 2022: 147). And while it seems most consistent with the internal textual evidence too, a full defence of this claim would require a close reading of numerous cryptic passages that can be read in very different ways. ³⁰ In any case, resolving this interpretive disagreement is incidental to the project undertaken here, which is specifically focused on our instrumental reasons for preferring different kinds of representative systems.

None of these arguments imply that all sortition-based reforms are undesirable. Lottocrats have lottocratish siblings who recommend adopting sortition assemblies as supplements to elections, for instance an advisory second chamber of parliament (Abizadeh, 2021) or a fourth branch of government (McCormick, 2011). The use of such institutions at the local, municipal, and regional levels is now widespread and there is reason to believe that the movement in this direction will continue to grow (Martin, 2015), perhaps to the point where participation in local deliberative assemblies becomes a normal practice and duty of citizenship, like jury service in some countries. And there are many reasons to welcome such a change. These institutions can act as focus groups that help policymakers gain specific and specialized forms of knowledge (Fishkin, 2009), improving the quality of political decisions. They can also make elite capture harder by promoting democratic legibility and mobilization by boosting the political knowledge, engagement, and civic virtue of participants (Hutton Ferris, 2019), and by drawing attention to previously marginal ideas and claims that could become foci for civil society pressure and voice and loci around which new issue-based campaigns and constituencies could emerge (Warren, 2009).

The ideas and arguments articulated above can help guide the kind of speculative or ‘predictive political theory’ that imagines a wide range of possible democratizing reforms that could involve lotteries or elections (Bagg, 2016). They suggest that we might reflect on whether such reforms would help people fulfil the demands of democratic citizenship and the demands of democratic leadership. Consider, for instance, the difficult question of whether the randomly sampled citizens in a deliberative assembly should engage in debate with established political actors, such as activists or elected officials (Farrell et al., 2019; Polletta, 2015). Deliberative democrats might worry about the cognitive skew this would create in the deliberating group, or the potential for debate to be captured when some participants are political professionals (see, e.g. Fishkin, 2009: chs 4–5). We should also bear in mind, however, that inserting political professionals into a deliberative assembly could help fulfil the demands of democratic citizenship by drawing attention to its operations and outputs, making it legible and connecting it to broader processes of mobilization (see the related arguments made by Hutton Ferris, 2022).

Pursuing this line of enquiry would take us beyond the scope of the present argument, which has sought merely to consider the relative merits of lottocracy and psephocracy. I have argued that lottocrats tend to overlook some central demands of democratic citizenship (that ordinary people see politics as legible and have low-cost opportunities to participate in it) and the demands of democratic leadership (specialization and competition for support). This make them unduly optimistic about the likely wisdom of a lottocratic legislature and its capacity to resist capture by systemically powerful state and economic actors. ³¹ Existing psephocracies are far from perfect and should be reformed in ways that bolster popular agency and increase political equality. Rather than elevating individualized participation to a defining feature of the regime, however, we should find ways to build big broad coalitions of ordinary people around specialists who compete for popular support. We should therefore prefer psephocracy to lottocracy.