One of the sharpest arrows critics of Canada’s Senate nock on their bow is the charge that it is undemocratic: an appointed chamber with the power to initiate and veto legislation. Since the 1980s, proposals for constitutionally reforming the Senate have accordingly sought to reconstitute it as an elected chamber. Yet competitive elections—used to select the Australian and U.S. Senates—stand in considerable tension with the traditional justifications for a bicameral legislature in the first place: that a second chamber serves a legislative-review function, which enhances the quality of legislation, and a balancing function, which checks concentrated power and protects minorities. An elected second chamber also stands in tension with the justification of representative democracy in general. Because competitive elections enable discriminatory voting, favour candidates with social privileges and resources, and are vulnerable to manipulation and domination by powerful partial interests, they neither treat persons as political equals nor treat conflicts impartially. They produce assemblies composed of elites partial to elite interests and values. This is why until the eighteenth century elections were primarily associated with aristocracy. Democracy, by contrast, was primarily associated with sortition—the random selection of officers. Because sortition gives each candidate an equal prospect of being selected to office, it treats all candidates as equals. And because truly random selection is invulnerable to influence by powerful interests, it tends to produce a descriptively representative “mirror” of society rather than an assembly of elites—thereby helping to treat conflicts impartially. This is why I argue that the Canadian Senate should be reconstituted as a randomly selected citizen assembly.
But my underlying purpose is to advance an argument whose significance reaches beyond the Canadian context in three ways. I take the democratic tradition to be defined by two constitutive commitments: to people’s political agency and political equality. My purpose at the most general level is to show, first, that in specifically representative democracy these two constitutive commitments face an inherent trade-off: realizing the one to some degree unavoidably compromises the other.
Second, I argue that elections and sortition instantiate this trade-off institutionally. On the one hand, elections are a mechanism for giving people the opportunity to pursue office and to hold office-holders accountable, which motivates political participation and influence, and incentivizes efforts to mobilize opinion formation and interest articulation in the public sphere. But elections undermine political equality and impartiality. On the other hand, sortition respects political equality and is a mechanism for promoting the responsiveness of political institutions to shared and conflicting interests and viewpoints in an impartial fashion (by protecting against capture by powerful partial interests and promoting a descriptively representative assembly). But sortition is neither a mechanism for holding representatives accountable, nor for the public’s political agency. The upshot is that realizing, at least to some degree, each of democracy’s two constitutive commitments requires some combination of elections and sortition.
Third, these two theoretical claims about representative democracy undergird a third, novel justification for bicameralism. One way to combine elections with sortition is, of course, via a legislature with one elected and one randomly selected chamber. Hybrid bicameralism is not the only imaginable way to combine the two—one could, for example, elect some members and randomly select others to a single chamber—but it is, I argue, one of the best ways: a randomly selected second chamber is not only compatible with the two traditional justifications for bicameralism, it is also especially well suited to specifically parliamentary (and federalist) contexts. Hence my thesis is that, in appropriate contexts, hybrid bicameralism strikes an optimal institutional compromise between the contradictory imperatives facing representative democracy. I turn to Canada as an especially promising candidate for the proposed reform in order to illustrate and focus my argument because it has an existing bicameral legislature the second chamber of which almost everyone agrees should be reformed (and it is a parliamentary federation to boot).
The overarching argument thus operates across three levels of decreasing generality: representative democracy, parliamentary democracy, and the illustrative Canadian case. I take this manner of proceeding—embedding my primary, more general argument for hybrid bicameralism within an argument centred on a specific institution—to exemplify the “political” political theory many have recently called for (e.g., Waldron Reference Waldron2016).
Bicameralism: Legislative Review and Balancing
Politics is a response to two fundamental problems: How should people forge and jointly act on shared interests or a common good? And how should people structure their relations and jointly act in the face of conflicting interests and disagreements? A second legislative chamber has been traditionally defended as a way to address each problem. It has been defended, firstly, as a means for enhancing the epistemic quality of legislative deliberation. Here the second chamber serves a legislative-review function to better discern and forge—and produce legislation that better tracks—shared interests, values, and norms (Tsebelis and Money Reference Tsebelis and Money1997). Repeating a task may enhance the quality of deliberation by performing a cautionary or redundancy function (Uhr Reference Uhr, Rhodes, Binder and Rockman2006)—by checking for errors or even, as Madison suggested, slowing down legislation to guard against temerity (Elliot Reference Elliot1845, 242; Hamilton, Jay, and Madison Reference Hamilton, Jay, Madison, Carey and McClellan2001, 322 (No. 62); 327 (No. 63)). Legislative review has also been justified on the grounds that second-chamber deliberations will be, thanks to the distinct method of membership selection, superior to first-chamber deliberations. The older, patrician version of the argument was that its members would have greater wisdom and expertise—in part because they would inherit their aristocratic position (U.K.), be appointed (Canada), or be indirectly elected (U.S.). The newer, institutional version of the argument is that the chamber’s superior deliberative potential stems from the fact that its selection method renders it more independent—less beholden to electoral imperatives and party discipline, less shackled to strategic considerations, and hence more capable of genuine deliberation (Thomas Reference Thomas and Montreal2003, 202). Given conflict and disagreement, bicameralism has also been defended, secondly, as a protection against the usurpation of power by special or “sinister” interests or sectarian ideologies. Here the second chamber serves a balancing function to prevent concentration of power and protect minorities (Mill Reference Mill and Robson1977, 514). Federalist institutions, often intended to protect regional minorities, furnish another, complementary balancing mechanism; not surprisingly, a second chamber has been most vigorously defended in federal polities—along with a membership selected partly on a regional basis (Tsebelis and Money Reference Tsebelis and Money1997; Watts Reference Watts and Montreal2003).
The Senate was designed by the 1867 fathers of Confederation to perform both legislative-review and balancing functions (Smith Reference Smith2003). Its balancing function was seen as two-fold: within the parliamentary context, to protect dissent and minorities by checking the power of Cabinet and legislative majorities in the House of Commons; within the federal context, to represent and protect regional interests in federal legislation. To serve this second balancing function, representation by population in the House was complemented by representation by regions in the Senate (Ajzenstat Reference Ajzenstat and Montreal2003). (Today the Senate has 105 seats: twenty-four each for the Maritimes, Ontario, Quebec, and Western regions; six for Newfoundland and Labrador; and one each for the three northern territories). To give teeth to its legislative-review and balancing functions, the Senate was equipped with the power to initiate non-financial bills and to veto House bills.
The contemporary assessment of the Senate’s record is mixed. A common view is that it has performed its legislative-review function well, has sometimes acted as a check on Cabinet and House majorities, but has largely failed to represent the regions (Heard Reference Heard and Smith2009). A primary reason for failure lies in party discipline. Because senators have historically been appointed by the prime minister along federal party lines, they have tended to vote—albeit with less discipline than House members—along party rather than regional lines. That senators have been less party-disciplined stems from two features: first, they are more independent because they are appointed once and for all (originally for life, since 1965 until age 75), are not disciplined by the prospect of Cabinet appointment, and do not need party resources for re-election bids; second, given that the Senate is not a confidence chamber whose rejection of government-sponsored bills could instigate the government’s fall, defection by senators affiliated with the ruling party is less costly (Heard Reference Heard and Smith2009, 128-131).
This failure has been a major impetus for calls for Senate reform. Throughout the 1960s and 1970s, reform proposals sought to enhance regional representation by wresting control of membership selection from the federal government. These proposals refrained from recommending elected senators—on the grounds that elections would, far from improving regional representation, increase the domination of party concerns over regional ones—and focussed instead on enhancing the role of provincial governments in appointing senators. By the 1980s, however, another complaint gathered steam: that an unelected legislative chamber is democratically illegitimate. Reform proposals throughout the 1980s and early 1990s consequently combined a call for stronger regional representation with a call for elected senators (Stilborn Reference Stilborn and Montreal2003, 33-40). In 2016, the government established a new procedure for Senate appointments, via a public application process to a non-partisan advisory board; since then, the Senate has become dominated by a non-partisan “Independent Senators Group.” Although many applaud these reforms for improving legislative review and balancing (Thomas Reference Thomas2019), they fail to speak to the Senate’s perceived democratic deficit, which continues to motivate calls for an elected Senate.
Yet as numerous critics have argued, elections, intended to shore up the Senate’s democratic credentials, would undermine its capacity to fulfill the functions justifying bicameralism in the first place. First, periodic elections would diminish its ability to fulfill its legislative-review function, either because (on the patrician argument) its membership would no longer possess the special wisdom resulting from the appointment process, or because (on the institutional argument) senators would face electoral imperatives compromising their relative independence from partisan, strategic considerations (Smith Reference Smith2003, 158-167). Second, periodic elections would, by strengthening party discipline and weakening senatorial independence, compromise the Senate’s ability to check Cabinet or represent regional concerns—an ability that, in light of the rise of party discipline, has become all the more vital (Smith Reference Smith2003, 179; Joyal Reference Joyal and Montreal2003; Waldron Reference Waldron2016). With respect to legislative review, debate in the House, far from being genuinely deliberative, has been reduced to a strategic performance driven by partisan electoral considerations. With respect to balance, the parliamentary theory of responsible government traditionally assumed that the House could hold the executive accountable because, while the Cabinet depended on the confidence of the House, the latter was itself relatively independent of the government. But party discipline means that any House in which the governing party enjoys a majority has lost the independence necessary to hold the government accountable: the House is controlled by leaders of the governing party (Murray Reference Murray and Montreal2003; Joyal Reference Joyal and Montreal2003, 281-284). These shortcomings generalize beyond Canada: those concerning legislative review apply to directly elected Senates in representative democracies in general (e.g., Australia, United States), while those concerning balancing apply to parliamentary democracies in particular (e.g., Australia).
The theory of bicameral representative democracy therefore faces an apparent dilemma. On the one hand, the democratic legitimacy of legislation appears to rest on the people’s “indirect” consent through elected representatives. On the other hand, bicameralism’s justification appeals to the second chamber’s capacity to fulfill two functions that, in a representative system inhabited by political parties, would be compromised by electing it. I argue that the dilemma’s first horn is chimerical: elections, far from being the exclusive basis of democratic legitimacy, are themselves insufficient.
Representative Democracy: Peace, Responsiveness, Agency, Equality, Impartiality
Recall the two fundamental problems of politics: how to forge and jointly act on shared interests, and how to structure relations in the face of conflicting interests and disagreements. Representative democracy has been defended as the best response to these problems in two ways: in terms of the outcomes it promotes and in terms of the political process itself.
First, with respect to outcomes, some argue that, thanks to the political agency they enable, representative-democratic institutions furnish the best means for securing social interests: they foster the deliberation necessary to discern and forge shared interests; and they respond satisfactorily to those whose interests compose the common good (Mill Reference Mill and Robson1977; Sen Reference Sen1999; Dasgupta and Maskin Reference Dasgupta, Maskin, Shapiro and Hacker-Cordón1999). Footnote 1 Others argue that political equality is the best means for fostering the right social relations—whatever they require, whether distributive justice, non-domination, or individual freedoms—because injustices stemming from non-political sources are best countered by egalitarian political processes giving people agency (Beitz Reference Beitz1989). Footnote 2
Yet political institutions are not merely means to social goods and just social relations: politics constitutes its own, distinct sphere of activity. Partisans of democracy have accordingly defended it not only in terms of social outcomes, but also, second, in terms of the political process itself. They argue that (opportunity for) political agency is constitutive of a valuable human life (Ober Reference Ober2007), Footnote 3 or that democratic institutions are uniquely suited to instantiating or expressing fundamental values—such as the public recognition of all persons as equals (Christiano Reference Christiano2008). Some further argue that, by treating persons as political equals and conflicts impartially, democracy structures specifically political relations (between subjects and rulers and between subjects themselves) in the right—or uniquely just or legitimate—way (Singer Reference Singer1974; Dahl Reference Dahl, Laslett and Fishkin1979; Waldron Reference Waldron1999).
One of representative democracy’s most powerful ideological weapons has consisted in a variant of this last point, namely, the claim that democracy alone can legitimize political power because only it can ground political rule in people’s consent. Many have come to think of (a) consent as the paradigmatic form of democratic agency and the only way to legitimize political power, (b) voting as a way of expressing consent, and (c) democracy as equivalent to popular consent via elected representatives. Footnote 4 These are recent ideological developments. Until the eighteenth century, democracy was primarily associated not with elections, but sortition. (Sortition was a staple of Athenian democracy: citizens had a right to participate in the popular Assembly (ekklēsia), but membership of key offices was determined by lot—including the people’s courts (dikastēria), the Council (boulē) of magistrates, and, in the fourth century BCE, legislative commissions (nomothetai) (Headlam Reference Headlam1891; Manin Reference Manin1997; Dowlen Reference Dowlen2008; Bouricius Reference Bouricius2013). No one can be under the illusion that randomly selecting officers is a way of consenting to their rule: democracy was equated not with consent, but with political agency (participation in the Assembly) and political equality (rotation in office, ruling and being ruled in turn).
Equating elections with consent, however, is one of modern democracy’s great illusory myths (Simmons Reference Simmons1979; Buchanan Reference Buchanan2002). To consent is intentionally and voluntarily to take on obligations (or waive one’s rights) by undertaking (or omitting) actions one intends others to understand as expressing one’s intention to do so (cf. Raz Reference Raz1986, 80-88). Those who vote for losing candidates do not explicitly consent to the winners’ rule. Do they do so tacitly, in virtue of taking part in elections? Not typically. They may merely intend to help a preferred candidate be elected, without any intention of acquiring obligations to the actual winners. Even one who votes for the winners may have no intention to consent. People vote for many reasons. True, some may intend to express their consent to abide by the rule of whoever wins. But others simply intend to affect the outcome, without obligating themselves. Some might find all the candidates unpalatable, and their intention in voting is wholly strategic—to prevent the worst alternative, not to legitimize the elections or outcomes. Perhaps some vote despite deeming the elections fraudulent and illegitimate (Singer Reference Singer1974, 24-26, 45-59). Voting is not intrinsically a way of expressing an intention to take on obligations and of consenting. It is primarily a way of seeking to wield power over the outcome, participate, fulfill a civic duty, or express one’s views.
The justification of representative democracy lies elsewhere. I shall assume that representative democracy is legitimate insofar as it (a) disincentivizes political violence and (b) renders political outcomes satisfactorily responsive to people’s relevant Footnote 5 interests, values, and norms (c) in part by providing the opportunity for political agency—participation and power in decision-making—in a way that (d) treats people as political equals and (e) treats conflicting interests and disagreements impartially. (To call a constitutional order legitimate is to say (1) it is justified, i.e., there are good reasons to install and protect it, (2) officers are presumptively morally permitted coercively to enforce its directives, and (3) subjects have a duty to refrain from overthrowing it. Footnote 6 To say conflicts are treated impartially is to say conflicting interests and views receive equal consideration in collective decision-making irrespective of whose they are, i.e., collective decision-making processes are not captured. Footnote 7 ) I do not seek to vindicate the normativity of peace, responsiveness, agency, equality, or impartiality; I here take them for granted. My question is whether and how representative democracy could live up to these criteria. Footnote 8
The claim that representative democracy disincentivizes political violence hinges on the fact that competitive elections furnish, to forces currently shut out of government, the prospect of taking political power by contesting and winning future elections, without incurring the costs of civil war (Przeworski Reference Przeworski, Shapiro and Hacker-Cordón1999).
The claim that representative democracy is responsive to people’s interests is grounded in two propositions: representative institutions render political rule responsive to people’s policy opinions and preferences, and people’s opinions and preferences furnish an epistemically reliable guide to their interests. Soroka and Wlezien (Reference Soroka and Wlezien2010) provide the most compelling evidence for the first proposition in Canada, alongside the United States and the U.K. For our purposes, their findings—like much of the public-opinion literature (Erikson, MacKuen, and Stimson Reference Erikson, MacKuen and Stimson2002)—suffer from three lacunae: they do not measure for substantively satisfactory responsiveness (they measure dynamic responsiveness, i.e., to changes in public opinion); they are unable to determine the mechanisms of responsiveness; and they do not provide evidence for the second proposition, i.e., they do not show that “public opinion” as operationalized by them, and to which they find policy responsive, is a reliable guide to people’s interests. They therefore cannot rule out the possibility that policy correlates with a public opinion manipulated and distorted by partial interests. Policy and public opinion may vary together not because the former is responsive to the latter, but because office-holders mobilize, manipulate, and distort public opinion (Jacobs and Shapiro Reference Jacobs and Shapiro2000). Footnote 9
The two mechanisms typically supposed to explain the first proposition are grounded in representative democracy’s paradigmatic form of political agency: the power to vote in free, competitive elections. On the one hand, elections enable people to select representatives they believe are disposed to serve their interests—e.g., on the basis of candidates’ promises (prospectively) or past performance (retrospectively). On the other hand, elections enable people retrospectively to hold accountable representatives and hence sanction those they believe have failed to serve their interests (by removing them from office); representatives who anticipate retrospective sanction have a strategic incentive to act in their constituents’ interests (Manin Reference Manin1997; Manin, Przeworski, and Stokes Reference Manin, Przeworski, Stokes, Przeworski, Stokes and Manin1999; Fearon Reference Fearon, Przeworski, Stokes and Manin1999). The effectiveness of either mechanism, however, is compromised by two features of contemporary democracies: voter ignorance and powerful partial interests’ influence over elections and representatives (Guerrero Reference Guerrero2014).
The social-scientific literature abounds with sobering evidence of widespread voter ignorance about basic policy questions (Somin Reference Somin2013). The problem is reinforced by the prospect that pervasive voter ignorance is insurmountable because it is instrumentally rational: because the probability of any individual’s vote being decisive is minuscule, the expected benefit of more informed voting is outweighed by the practical costs to each of gathering information (Downs Reference Downs1957) or psychological costs of giving up false but pleasing beliefs (Caplan Reference Caplan2007). Yet to work, the selection and accountability mechanisms require considerable knowledge.
For a prospective promissory selection mechanism to work, candidates must keep their pre-election promises and voters must be able to discern which promises will serve their interests (Manin, Przeworski, and Stokes Reference Manin, Przeworski, Stokes, Przeworski, Stokes and Manin1999, 30). If voters cannot be sure whether candidates’ announced policies will serve their interests when the time comes, they may want representatives to exercise discretion—indeed, a hallmark of representative institutions is the absence of imperative mandates (Manin Reference Manin1997). But if representatives exercise discretion, prospective promissory selection cannot function (Przeworski Reference Przeworski, Shapiro and Hacker-Cordón1999, 35). Alternatively, the selection mechanism might work on the basis of not fulfilled promises, but judgements about candidates’ policy dispositions—judgements grounded in epistemic shortcuts not requiring resource-intensive information gathering (Lupia and McCubbins Reference Lupia and McCubbins1998). People might, for example, vote for candidates with whom they share descriptive characteristics, on the assumption that representatives like themselves are disposed to share their raw opinions, deliberate and refine opinions as they would have done, and act in their interests—what Mansbridge (Reference Mansbridge2003) dubs gyroscopic representation. Or in an electoral system inhabited by parties, people may rely on partisan identification as an epistemic shortcut, on the assumption that “their” party reflects their own values, will hold policy opinions they would have had if better informed, and is disposed to act in their interests. Achen and Bartels (Reference Achen and Bartels2016) have accordingly argued that ignorant voters do not vote prospectively for parties on the basis of antecedently held policy opinions. Rather, they vote primarily on the basis of prior partisan or group loyalties, and form policy opinions insofar as they are promoted by “their” party. How effective a selection mechanism this is depends on how reliably partisan identities track people’s interests.
For retrospective voting to be an effective mechanism of selection or accountability, voters must be able to discern the extent to which previous outcomes have been due to office-holders’ actions. If they cannot, or if other factors loom too large, then retrospective voting furnishes a rather coarse mechanism (Przeworski Reference Przeworski, Shapiro and Hacker-Cordón1999, 36-38; Achen and Bartels Reference Achen and Bartels2016, 90-104). Achen and Bartels argue that, to the extent voters do vote retrospectively, most do so blindly—a point they illustrate with the 1916 shark attacks in New Jersey prior to the US presidential elections, which, although beyond government control, in beach communities led to a roughly 10% decrease in support for the incumbent. Blind retrospection is no more effective for selecting or sanctioning than voting randomly.
The mechanisms meant to explain the second proposition—that people’s policy opinions or preferences furnish an epistemically reliable guide to their interests—include processes of information gathering, debate, and deliberation. One option is that public debate and discourse lead to an informed public whose actual policy opinions are epistemically reliable. Another is that deliberation amongst representatives filters raw public-opinion inputs to produce a “refined” opinion on which to base policy (Fishkin Reference Fishkin2009). The first option is gainsaid by widespread voter ignorance. This does not imply, however, that public discourse and opinion are epistemically irrelevant. The second option presupposes that, even if people’s raw opinions do not reliably correspond to their interests, discourse in the public sphere serves to articulate interests, values, and opinions that feed into official political processes and furnish representatives with epistemically valuable information on the basis of which they can deliberate (Habermas Reference Habermas1996). Thus some have argued that even if voters on average are individually uninformed, changes in aggregate public opinion, when shaped by public discourse, respond to policies and outcomes in a coherent, orderly way that communicates valuable information (Page and Shapiro Reference Page and Shapiro1992; Erikson, MacKuen, and Stimson Reference Erikson, MacKuen and Stimson2002; Mayhew Reference Mayhew2004). This could be true if changes in aggregate public opinion are driven not by changes amongst the majority of uninformed voters—whose views are either stable or change direction in a randomly distributed way—but by the few informed voters, or are driven by changes in many individuals’ opinions, mostly prompted by errors whose direction is randomly distributed, but also by (even small amounts of) genuine information.
But the second option depends on the epistemic quality of deliberation by representatives. The problem is that competitive elections undermine the mechanism that is supposed to close the gap between public opinion and interests, namely, representatives who, via conscientious deliberation, filter raw public opinion into informed, refined opinion. Competitive elections give rise to partisan politics: winning resource-intensive, modern elections requires the organizational apparatus of a party; in parliamentary systems, moreover, the imperative to control the executive incentivizes legislative party discipline. The result is that parliament is not a forum for conscientious deliberation directed to a common good or to impartial conflict resolution but, rather, a strategic performance geared primarily to communicating the party brand for electoral purposes (Proksch and Slapin Reference Proksch and Slapin2015).
Elections also fail political equality. The notion of political equality at stake in democratic theory concerns being treated as an equal with respect to political decisions, viz., about the laws and policies governing the exercise of political power. One way to instantiate equality might be for all subjects to have equal power over political decisions governing them Footnote 10 —equal power understood as equal capacity, not actually equal influence. Equal influence—even if attainable—cannot be a democratic ideal: democracy relies on political influence through good arguments in deliberation, which supposes that better arguments merit greater influence. Not only would actually equal influence require restricting good arguments in the public sphere, it would also require taking away opportunities for influence from the highly motivated, or somehow forcing the unmotivated to exercise influence (Dworkin Reference Dworkin2000). Where a good is appropriately distributed proportionate to some standard of merit—and hence equality of outcomes is not desirable—equality is instead to be understood as equal opportunity (Williams Reference Williams, Laslett and Runciman1964). This is the basis for construing equal power and hence democratic equality as equal capacity and hence substantive ongoing opportunity for wielding power over political decisions (Brighouse Reference Brighouse1996; Kolodny Reference Kolodny2014).
But in representative democracy, elections do not decide legislation or policy: they select those who decide. The significance of this simple fact has not been adequately appreciated in the literature on political equality. One might think elections treat people as equals in their capacity as selectors of office holders; Footnote 11 and helping to select representatives can provide a mechanism for indirectly influencing political decisions. But without imperative mandates, participating in selecting those who decide is not equivalent to participating in deciding legislation or policy (Green Reference Green2011). Having equal opportunity to wield power in selecting representatives and to influence representatives once selected may be a way of instantiating horizontal equality between non-representatives. But horizontal equality fails to address the formal vertical inequality intrinsic to representative democracy: between representatives empowered to decide legislation and policy and non-representatives who are not. The tendency to parachute a notion of political equality forged with direct democracy tacitly in mind—as equal say in majoritarian decision-making—into a theory of representative democracy (Waldron Reference Waldron1999) fails to take seriously this vertical inequality and the fact that elections select office-holders rather than decide laws. Being treated as an equal qua selector (and having equal opportunity to influence representatives) is therefore insufficient for political equality.
Many think representation could be rendered compatible with political equality if subjects collectively control the selection of representatives who act as delegated agents guided by and answerable to subjects. Kolodny (2014, 317-318), for example, argues that under such circumstances subjects remain representatives’ political equals, just as in non-political contexts principals are not rendered socially inferior by virtue of delegating agents to act for them. But such arguments suffer numerous shortcomings. First, the argument is moot since, as we have seen, the second condition is not met in practice. Second, representatives are not pure delegates even in principle: vertical inequality of power exists precisely to the extent representatives have discretion. Third, the analogy is faulty: equality in non-political contexts does not depend on equal ongoing opportunity for effective power over the decisions delegated to agents; by contrast, political equality depends on equal ongoing opportunity for wielding power over precisely those decisions representatives are empowered to make. Nor does Kolodny’s second analogy fare any better. He argues that just as subsidiary officers (e.g., civil servants) do not acquire a status superior to elected representatives in virtue of being delegated to make subsidiary decisions, representatives are not superior to subjects. But representatives’ power over subsidiary officers includes determining officers’ discretionary scope by establishing the legal framework within which they operate. By contrast, hardly any subjects ever establish the legal framework within which representatives decide. Representatives do not make “subsidiary” decisions.
The question therefore persists: How could vertical inequality be compatible with political equality between representatives and non-representatives, i.e., how could both rulers and subjects have equal ongoing opportunity for power over political decisions? Think of holding office as a good. How could the unequal distribution of this good, which consists in extra opportunities to wield power over political decisions, be compatible with that opportunity’s equal distribution? Because holding office furnishes extra opportunities, equal opportunity for power over political decisions requires treating everyone as an equal with respect to the distribution of the good of holding office, i.e., not qua selectors but qua candidates for office. Treating those who select officers as equals qua selectors is useless because the issue is the distribution of opportunities, not who chooses the distribution (or their equal opportunity to do so).
One way to treat everyone as equals with respect to a good is to give each person an equal share. Representation rules out simultaneously equal shares in office-holding. But people might nevertheless enjoy equal shares over time if they take turns; hence the appeal of the classical idea of rotation of office. Rotation is ruled out, however, by the sheer size of contemporary polities: most subjects will die before their turn.
If a good cannot be distributed equally—if equal outcomes are ruled out—the remaining option is to treat people as equals by granting them either an equal probability (Broome Reference Broome1990) or an equal opportunity (Williams Reference Williams, Laslett and Runciman1964) of acquiring the good. Vertical inequality would be compatible with political equality if everyone had an equal chance or opportunity to hold office.
Elections do not afford candidates equal probability: voters select candidates on the basis of characteristics (such as political skill) distributed unequally amongst the population. Nor do they afford equal opportunity. Formal equal opportunity is compatible with a good’s unequal distribution only if it is distributed on the basis of merit, i.e., criteria appropriate to its distribution. There may indeed be relevant criteria, such as political skill or public spiritedness, for distributing office. But elections are not a meritocratic selection procedure. First, voters can, and typically do, discriminate amongst candidates on the basis of irrelevant characteristics such as race or gender. Formal equal opportunity is incompatible with the arbitrary discrimination that elections intrinsically permit: voters are free to vote as they wish; no voter is accountable to anyone for discrimination. Second, competitive elections advantage candidates with salient, distinguishing features, which often consist in irrelevant social privileges such as good looks, wealth, or fame.
Moreover, even if voters refrained from discrimination, elections would fail to provide substantive equality of opportunity. This requires that people from all relevant sections of society have an equal chance of acquiring the relevant qualifications (Williams Reference Williams, Laslett and Runciman1964; Rawls Reference Rawls1999, 63; Arneson Reference Arneson and Zalta2015). First, many of the qualities voters might deem relevant to qualify will be correlated with social class, giving lower classes fewer prospects. Second, winning elections requires resources to disseminate information, mobilize activists, and shape public opinion. This favours those with power and resources in competitive elections. The effects of resource inequality might be attenuated via public campaign financing (and limits to private financing). But the problem of discrimination means free elections are inherently inegalitarian: elections in which voters are not free to discriminate as they wish would not be free elections. Footnote 12
That electoral processes are susceptible to manipulation and domination—not to say corruption—by powerful partial interests implicates not just political equality, but also the impartial treatment of conflict. An ignorant electorate is especially vulnerable to manipulation because low information renders public opinion prey to incomplete but misleading information, outright misinformation, and framing effects (Fishkin Reference Fishkin2009, 3-4). The result is that the individual “errors” on which aggregate public opinion is based will not be randomly distributed, but systematically biased in favour of influential partial interests; and representative institutions, even if somewhat responsive to public opinion and societal interests, will respond in ways partial to the interests of the powerful—especially those with the resources to shape public opinion and finance campaigns of preferred candidates. Footnote 13 This problem inheres in competitive elections insofar as they facilitate political agency: incentivizing people to participate is to incentivize them to expend time and resources to influence other voters (Dworkin Reference Dworkin2000).
To sum up: elections establish a formal but coarse mechanism by which people may hold their representatives accountable and hence responsive to their interests; have an intrinsically inegalitarian character; and produce chambers partial to the interests of the powerful few (McCormick Reference McCormick2006).
Hybrid Bicameralism: A Sortitioned Second Chamber
Electoral democracy therefore faces a dilemma: the very mechanism meant to secure accountability and hence responsiveness and to facilitate political agency is what undermines political equality and impartiality. Sortition was democracy’s historical response to inequality and partiality. Sortition, unlike elections, treats all candidates as equals: each has an equal prospect of being selected (Manin Reference Manin1997, 56-57). Footnote 14 And it solves the problem of manipulation and domination by partial interests in selecting representatives because lotteries sanitize the process of influence (Dowlen Reference Dowlen2008, Reference Dowlen2009; Stone Reference Stone2011). Sortition also affects its own form of responsiveness independent of any accountability mechanism: sortition selects for a gyroscopic set of representatives collectively disposed to act in constituents’ interests because random selection tends to produce an assembly that, if large enough, is descriptively representative of them. Footnote 15
But sortition has its shortcomings. A purely sortitioned legislature is a potential recipe for civil war: having no prospect for seizing state power by mobilizing and contesting future elections peacefully, social forces opposed to the legislature’s current agenda would have less incentive to refrain from violence to realize their goals. True, they could seek to persuade randomly selected representatives (and the population)—just as subjects can petition monarchs—but they could not mobilize to take office themselves or determine who does. Moreover, because sortition does not render representatives formally accountable to people, it is not a mechanism for facilitating their political agency. The electoral accountability mechanism therefore remains indispensable (Pourtois Reference Pourtois2016): however coarse the mechanism, elections—unlike sortition—promote responsiveness via the public’s political participation and power over representatives. Some attribute intrinsic significance to political agency (Markell Reference Markell2008), but it also has an instrumental significance beyond avoiding civil war. On the one hand, because elected representatives depend on constituents for re-election, they are compelled to mobilize and gain support by justifying their decisions publicly, canvassing, organizing, and building coalitions. On the other hand, without imperative mandates, and given the coarse nature of elections’ accountability mechanism, to induce representatives to be responsive in specific issue areas various publics must seek to influence them (not merely by voting but) via public argumentation, interest articulation, and coalition formation. The gap between the public and its representatives, combined with the electoral accountability mechanism tying them together, incentivizes a two-way circulation of public justification, argumentation, and interest articulation (Urbinati Reference Urbinati2006). This two-way circulation provides inputs indispensable for rendering legislative deliberation and negotiations responsive to societal interests, values, and norms. As a mechanism for responsiveness, elections are not nothing. Even in the Canadian context—with strong party discipline—the more competitive their electoral district, the more responsive are individual MPs to constituents’ opinions and demands (Soroka, Penner, and Bildook 2009; cf. Sullivan and Uslaner Reference Sullivan and Uslaner1978). Elections not only provide opportunities to challenge government from outside official state institutions, and hence foster a culture of contestation necessary for democratic institutions’ long-term viability, they also incentivize the generation of a wide range of policy proposals from multiple sites (Pettit Reference Pettit2012, 201-205)—including from what Habermas (Reference Habermas1996) calls the “wild” public sphere. Again, people could try to persuade even sortitioned legislators. But there would be no formal mechanism incentivizing or forcing legislators to respond (cf. Heller Reference Heller, Thompson and Held1982). (An enlightened monarch, too, may consult subjects at her pleasure.) Footnote 16 A purely sortitioned legislature would fail both to facilitate political agency and adequately to incentivize public discourses crucial to the epistemic quality of deliberation amongst representatives and in the public sphere.
My general thesis is that a hybrid bicameral legislature, with one elected and one sortitioned chamber, would strike the right institutional balance between these contradictory imperatives in representative democracies. The elected chamber would produce rough institutional accountability and facilitate political agency—and thereby disincentivize violence and incentivize the two-way circulation of public justification—while the sortitioned chamber would enhance responsiveness, help secure political equality and impartiality, and foster the kind of conscientious deliberation stymied by strategic partisan and electoral imperatives. This is a compromise because serving democracy’s other normative standards means political equality cannot be fully instantiated for all legislative offices.
My more particular thesis is that federal and parliamentary democracies have even further reasons to adopt hybrid bicameralism: federations have distinct, independent reasons to opt for bicameralism (rather than a hybrid single chamber, for example); and a sortitioned chamber, by acting as a check on the power of cabinet and majorities in the first chamber, provides parliamentary democracies a way to restore responsible government in the face of first-chamber party discipline.
My defence here of these claims must remain preliminary. We do not have enough empirical evidence fully to predict how such a novel institutional arrangement would function. What we can do is establish a research agenda: outline a proposal, articulate its normative motivation, raise central potential problems, and sketch out preliminary responses. To this end, I outline and defend a specific proposal to reconstitute the Canadian Senate as a citizen assembly. Footnote 17
Reconstituting Canada’s Senate as a Randomly Selected Citizen Assembly
I propose a second chamber with the following features. First, its powers would remain consistent with the existing Senate: although the House would remain parliament’s only confidence chamber, the second chamber would retain the power to veto House bills and initiate non-financial bills. It would enjoy a veto in order to ensure that all legislation bear the imprimatur of the chamber instantiating political equality; to prevent an elite first chamber from unilaterally passing legislation serving elite, partial interests; and because even when not exercising a veto, its threat is often indispensable to secure amendments. It would have no right to initiate financial bills because its primary role would not be agenda setting but to respond to the agenda emanating from the government and House (cf. Owen and Smith Reference Owen, Smith, Gastil and Wright2019). (Because it is not a confidence chamber subject to partisan, electoral discipline, its members have less incentive to formulate a coherent policy program reconciling the trade-offs between discrete policies. Footnote 18 ) It would nevertheless retain the right to initiate non-financial bills to shore up its power relative to Cabinet and House—to force their hand if necessary. To prevent legislative deadlock, a supermajoritarian threshold would ultimately enable the House to override the second chamber’s veto. Footnote 19
Second, the chamber’s membership would be selected, like the existing Senate, on a regional rather than strictly population basis. (I propose tripling each regions’ seats to a total of 315, for greater confidence of descriptive representation via random selection.) The selection process in each province would comprise two steps: randomly select a large number from the province’s electoral list; then, from those who agree to proceed to the second round, randomly select officers for staggered four-year terms (preceded by a three-month in-residence preparation and training period).
Selection on a regional/provincial basis and self-selection both depart from random sampling. The first departure is justified by the chamber’s balancing function, to secure adequate representation of regional minorities within the federation. The assembly will not be descriptively representative of Canadians taken as an undifferentiated population, but of the federation comprising distinct confederated units.
The second departure is justified for three reasons. The first is practical: it is difficult to see how those unwilling could be effectively compelled to serve. The second is principled: compelling them would seriously violate personal liberties. The third is also principled: whereas sortition presupposes that everyone in the candidate pool is able to serve, some on the initial list may be unable. Self-selection is a way to introduce an element of judgement into the process consistent with political equality. Consistent, because up to individuals themselves: no one’s opportunity for holding office is diminished by her own power of declining. Self-selection also offers maximal protection against manipulation and domination by partial interests because the element of judgement is maximally dispersed. (The danger is not eliminated: consider a religious leader who forbids female followers from serving.)
But even without manipulation, self-selection can compromise descriptive representativeness and hence impartiality: those willing to serve may tend to have unrepresentative features. The citizen assemblies formed in British Columbia and Ontario to propose electoral reform, for example, which were selected in similar fashion, albeit stratified by gender, comprised a membership on average more educated, more politically engaged, older, and more white than the general population (James Reference James, Warren and Pearse2008; Fournier et al. Reference Fournier, Kolk, Kenneth Carty, Blais and Rose2011).
We must, however, distinguish two kinds of departure from descriptive representation. Strict descriptive representation requires that representatives collectively reflect all population characteristics. But even if a population varied along a limited number of features, no sample could be descriptively representative in this strict sense even in principle: if the representative body does not include the entire population, then one difference is that the former, but not all of the latter, are representatives! This feature is politically relevant: representatives have political powers unavailable to others. But such departures from strict representation are part of the point of representation: they concern features germane to the function of holding office.
Departures from strict descriptive representation therefore divide into two categories: those germane to the ability to serve—such as having the status and powers of a representative, but also political engagement or willingness to serve; and departures concerning features, such as gender or race, that are not. Engagement is constitutive of serving. But no legislator needs to become a man or white to fulfill her legislative duties. We therefore need a more restricted notion of political descriptive representation, compromised solely by non-germane departures from strict representation. Strict descriptive representation is an inappropriately apolitical notion negating the very possibility of political representation; political representation requires that the body collectively mirror only those characteristics not themselves part of the function of serving.
Self-selection poses a significant problem for political representation when—but only when—it introduces non-germane departures. I propose two remedies. First, minimize refusals by offering generous compensation and support (e.g., to caregivers and their dependants) and by contacting and seeking to persuade potential refusers by informing them of these benefits. Second, use stratified sampling: randomly select individuals in proportion to their presence in the population with respect to socially salient features, such as gender, age, visible-minority and First Nations status, primary tongue, and class. The obvious difficulty—beyond data availability—is that stratification introduces discretion and hence potential for controversy and manipulation. But the difficulty can be met via a fair procedure for determining the stratification features (Leydet Reference Leydet2019). I propose a reiterative process: whatever features are chosen in the inaugural lottery should be preliminary; in each subsequent lottery, the incumbent chamber itself should decide the features.
Reconstituting the Senate as a citizen assembly would realize two goals. First, it would secure the second chamber’s democratic legitimacy. Its democratic credentials would be grounded in the political equality instantiated by the selection process and in its claim to impartiality thanks to being more or less descriptively representative of the federation. True, some might even come to see a descriptively representative chamber as legitimate because they see its decisions as representing what they themselves might have collectively decided if they had taken the time to engage, educate themselves, and reflectively and deliberatively consider the matter. This might be the lesson of the British Columbia and Ontario citizen assemblies: the more voters knew about the assembly—how its members were chosen, its freedom from partisan influence, etc.—the more likely they were to vote in the ensuing referendum in favour of its recommendation for electoral reform, even when they did not know much about voting systems (Cutler et al. Reference Cutler, Johnston, Carty, Blais, Fournier, Warren and Pearse2008; Fournier et al. Reference Fournier, Kolk, Kenneth Carty, Blais and Rose2011, 130-132). They did not need to know: they trusted the assembly, perhaps because they assumed its decisions reflected what they themselves would have collectively decided had they invested in studying the matter. But that is not the only potential explanation. The primary value of descriptive representation lies elsewhere: as an instrumental means to securing impartiality. The point is not that the chamber’s decisions somehow represent the “will of the people” but, rather, that the chamber instantiates political equality and is presumptively responsive to people’s interests in an informed, impartial fashion.
Second, random selection would enhance the democratic credentials of Canada’s parliamentary system as a whole. Recall that elections are intrinsically inegalitarian, vulnerable to manipulation and domination, and routinely produce assemblies dominated by elites. A more egalitarian, impartial, and representative second chamber with the power to veto legislation would give its democratic imprimatur to legislation emanating from an elected House—while blocking bills grounded in partial, elite interests. A sortitioned chamber would provide an additional mechanism for holding the elected chamber accountable—to people’s descriptive representatives. Its democratic credentials would grant it the legitimacy and confidence to exercise these powers to a much greater extent than the appointed Senate.
Unlike elections on the Australian or U.S. models, moreover, random selection would be fully compatible with the legislative-review and balancing functions. Partisans of an appointed Senate typically point to senators’ relative independence from electoral and partisan imperatives (Rémillard Reference Rémillard and Montreal2003). But a sortitioned chamber will be just as—if not more—independent than an appointed Senate. Randomly selected members will not have depended on party resources for their selection, will not defer to the government in anticipation of Cabinet posts, and will not need party resources for future re-election bids. Footnote 20 Moreover, unlike House members who must devote considerable time to constituency and other electoral work, randomly selected members would be able, like current Senators, to devote themselves to investigating the merits of legislation. Parliamentary debate would not be reduced, as in the House, to a strategic performance for electoral, partisan gain.
The chamber’s independence would also facilitate its capacity to check the power of Cabinet and House majorities—regardless of which party is in power. Because it would not be a partisan chamber, it would not be controlled, as is the House, by the governing party. And because regional concerns would not be swamped by partisan ones, its capacity to protect regional minorities—based on the partly regional basis of its makeup—would be superior to that of an elected Senate.
Partisans of an appointed Senate often argue that its deliberations are superior to elected chambers also because its members bring wisdom and expertise to the legislative process. Sortition would seem to destroy this advantage: anyone could be selected randomly. What if a racist, misogynistic xenophobe spewing hatred, cheering sexual assault, and inciting violence were randomly selected? This is possible—indeed, given our societies’ makeup, likely. But random selection does not have a monopoly on the Donald Trump problem. Elections regularly install racist, sexist men. So do other selection devices: Canada has appointed its share of racist and sexist judges. The problem is not peculiar to sortition.
The real issue is whether a randomly selected assembly will be competent. Modern legislative and policy questions are complex. High-quality deliberation requires a great deal of information. Yet the typical citizen is woefully ignorant. Wouldn’t a citizen assembly compromise the chamber’s legislative-review function?
The first point is that the basic reason why voters are typically ignorant would be inapplicable. Voter ignorance is instrumentally motivated: given the minuscule probability that one’s vote will be decisive, the costs of gathering information outweigh the expected benefits of more informed voting. But the views of legislators will matter a great deal, giving them every incentive to invest in acquiring the relevant knowledge and competence. (This is also a reason why assembly members will not—as average voters often do—simply form policy opinions on the basis of partisan identification: they will invest in arriving at their own independent judgement.) Fishkin (2009, 119) has argued that deliberative polls, in which randomly selected individuals assemble to learn about and deliberate on a wide range of policy issues, provide considerable evidence that “the public is indeed capable of dealing with complex issues, once it believes its voice matters.” Similarly, the analysis by Fournier et al. (2011, 37-39, 49-49, 79-93) of the British Columbia and Ontario citizen assemblies suggests members were able to gain relevant knowledge leading to informed, consistent, high-quality deliberations and decision-making. They attribute these results to members’ high motivation (resulting from the combination of random selection and self-selection), the importance of the task at hand, members’ perception that they had real power, the assembly’s independence, opportunities for members to develop personal bonds, and the chair’s effective leadership. Other experiments with mini-publics provide further support for these findings (Neblo, Esterling, and Lazer Reference Neblo, Esterling and Lazer2018). True, temporary mini-publics often function with top-down expert guidance in ways inappropriate for democratic legislatures; but legislators would have the benefit of learning that comes with prior training and a longer term, and the incentives that come with having, unlike mini-publics, real power. The point is not that sortitioned members will rival experts in competence, but that we have little reason to expect them to be any worse than the many incompetent representatives who regularly populate existing chambers, whether elected or appointed.
The second point is that collective intelligence is a function of two variables: the individual intelligence and cognitive diversity of group members. It turns out that, above a minimum threshold of individual competence, if individuals can grasp ideal solutions to problems, then cognitive diversity is more important than individual ability for problem solving. In other words, the deliberations and decisions of a cognitively diverse assembly are epistemically superior to those of a lone genius or a less diverse assembly whose members are individually more intelligent or competent (Hong and Page Reference Hong and Page2004; Page Reference Page2007; Landemore Reference Landemore2012). Footnote 21 Random selection is an ideal mechanism for ensuring cognitive diversity.
High-quality deliberation in citizen assemblies, however, depends on institutional design. It requires access to quality information, the representation of diverse views, substantive balance in the pool of pro/con arguments, and equal, conscientious consideration of the arguments’ merits—regardless of who offers them (Fishkin Reference Fishkin2009, 34). There are three basic concerns about citizen assemblies. First, collective deliberations are susceptible to domination by socially privileged members. Even if everyone is treated equally in the random selection process, inequalities can creep in at the deliberation stage, undermining not only equality and impartiality, but also the epistemic benefits of cognitive diversity: not everyone’s views will be considered equally on their merits (Young Reference Young, Benhabib and Cornell1987, Reference Young1990; Mendelberg Reference Mendelberg, Delli Carpini, Huddy and Shapiro2002, 165-167). Second, collective deliberation may suffer from groupthink—where social pressures to conformity discourage giving adequate credence to counterarguments—and group polarization—where given an initially small group bias in one direction, social pressures combine with an initially biased pool of arguments to lead the group to extreme views (Janis Reference Janis1972; Mendelberg Reference Mendelberg, Delli Carpini, Huddy and Shapiro2002; Sunstein Reference Sunstein2006). Third, neophytes may be especially vulnerable to manipulation, capture, and corruption by outsiders on whom they rely for information (Mooney Reference Mooney, Kurtz, Cain and Niemi2007)—whether private lobbies and special interests, Cabinet, parties in the House, or bureaucrats.
Many details beyond the scope of this paper remain to be worked out, but my gambit is that intelligent institutional design can remedy these problems. I propose the following. First, the second chamber would, before deciding on a package of legislative proposals, initiate a three-phase process: learning, public consultation, and decision-making. The learning phase would begin, as in deliberative polls, with common briefing materials on the relevant topics prepared by an advisory committee of experts and stakeholders, designed to ensure quality information and a balance of competing viewpoints. The advisory committee would be appointed jointly by each of the main House parties and the second chambers’ independent research committee. The research committee and support staff would be separate from the executive’s bureaucracy, and appointed via a non-partisan, professionalized process. The briefing materials would be supplemented for each assembly member by whatever materials requested from research support staff. Because meaningful individual participation in deliberation is difficult for large groups, deliberative polls proceed with the learning phase in two stages: small-group deliberations and plenary sessions. I propose the same model. The assembly would begin with small, randomly selected subgroups to discuss the briefing materials and prepare plenary-session questions. Moderators would be trained to facilitate subgroup deliberations, first, to ensure everyone can speak and no one dominates and, second, to avoid the pressures of social conformity—by having participants focus on agreeing on plenary questions, rather than revealing their final view and coming to a substantive decision. (Moderators could be selected by the research committee, but with minimal oversight by House parties to guard against bias.) At the plenary session, a panel of competing spokespersons—experts selected by the advisory committee but also a representative from each party in the House—would advocate for and against proposed legislation, and be subject to questions by assembly members. The learning phase would be followed by public consultations, where stakeholders would be invited to supply briefs for deliberation and to present at plenary sessions. Finally, the assembly would deliberate as a whole and come to a decision.
The evidence from mini-publics is that briefing materials vetted by competing stakeholders; randomly selected and hence diverse subgroups who prepare plenary questions via deliberations facilitated by trained moderators who discourage domination or collective position taking; and plenary sessions with competing spokespersons who articulate pro and con arguments, have successfully avoided domination, groupthink, and polarization in deliberation (Fishkin Reference Fishkin2009, 129-133; cf. Steiner Reference Steiner2012). The evidence from electoral-reform citizen assemblies, already cited, is also promising.
The best evidence we have about manipulation and capture of neophytes comes from the imposition of term limits in some USA state legislatures (which led to less experienced legislators). The clearest effects were the decreased power of legislative-party leaders and increased executive power (Carey, Niemi, and Powell Reference Carey, Niemi and Powell1998; Kousser Reference Kousser2005; Carey et al. Reference Carey, Niemi, Powell and Moncrief2006; Powell Reference Powell, Kurtz, Cain and Niemi2007; Miller et al. Reference Miller, Nicholson-Crotty and Nicholson-Crotty2011). Neither raises concerns here: a citizen assembly is less partisan by design, and informational asymmetries with respect to the Cabinet are offset by the chamber’s independence. The evidence concerning special-interest groups and lobbyists, by contrast, is mixed: neophyte legislators enhanced outsiders’ informational advantage—increasing their influence—but also broke up the long-term networks and relationships of trust between legislators and lobbyists—decreasing the latter’s influence. The net effect seems to depend on interest groups’ prior strength: states with strong interest groups saw a net decrease in their influence over legislators, while states with weak interest groups saw an increase (Carey et al. Reference Carey, Niemi, Powell and Moncrief2006; Mooney Reference Mooney, Kurtz, Cain and Niemi2007; Miller et al. Reference Miller, Nicholson-Crotty and Nicholson-Crotty2011; cf. Kousser Reference Kousser2005, 65-66). But much of the increased influence comes from being more active in the election process—a mechanism stymied by sortition (Richardson, Valentine, and Stokes Reference Richardson, Valentine and Stokes2005).
I propose to reinforce this protection with a legally enforced firewall protecting members of the second chamber (and their research committee and moderators) from lobbyists and from House members and Cabinet (akin to the firewall currently protecting the judiciary): contact would be legally restricted to official and publicly transparent channels. This explains the superiority of a bicameral as opposed to unicameral combination of election and sortition: in a mixed chamber (Pluchino et al. Reference Pluchino, Garofalo, Rapisarda, Spagano and Caserta2011), the sortitioned representatives risk being dominated by their professional, elected counterparts.
The firewall would be supplemented with an annual accountability audit of assembly members (and research committee and moderators) during their term, on departure, and for several years afterwards to ensure strict adherence to anti-corruption codes (which, e.g., outlaw promises of future payoffs). The concern over bureaucratic capture is partly addressed by separating the assembly’s support staff from the executive bureaucracy and protecting the impartiality of their selection process.
These are preliminary suggestions, and without further experimentation and research, we cannot prove to sceptics precisely how effective these design elements will be. But with respect to competence, domination, groupthink or polarization, and corruption, we have no reason to expect sortitioned chambers to be any worse than our rather flawed experience with elected or appointed ones.
Conclusion
Representative democracies face an inherent tension between political agency and political equality. All of them must strike some compromise between these core commitments. Elections and sortition instantiate this trade-off institutionally. Elections are crucial for providing opportunities for political agency and disincentivizing political violence, but are insufficient for political equality, impartiality, and satisfactory responsiveness to interests. Sortition, by contrast, serves political equality and impartiality and enhances responsiveness, but fails adequately to facilitate agency and disincentivize political violence. I have accordingly defended hybrid bicameralism as an institutional compromise between these grounds for representative democracy. Whereas the two traditional justifications for bicameralism initially grew out of anti-egalitarian premises (about the need for elite wisdom and to protect the elite few against the many), the novel justification advanced here stems from egalitarian premises about the need to treat all subjects as political equals and to protect state institutions from capture by the powerful few.
