A Randomly Selected Chamber: Promises and Challenges

This paper explores the idea of a randomly selected chamber of representatives (RSC) through an appreciation of the promises it offers and the challenges it would face. We identify two main promises: a RSC could offset the aristocratic character of elections, thereby increasing the legitimacy of the political system; and it could increase democracy’s epistemic potential, thanks to gains in terms of diversity, deliberations, humility, and long-term perspective. We then discuss four key challenges. First, participation: how can the chamber have diversity without mandatory participation or heavy sanctions? Second, how can we conceive or build legitimacy for this non-elected and somehow unaccountable chamber’s views? Third, independence: how to safeguard randomly selected people from corruption? Finally, there may be a linguistic challenge: if the RSC has a deliberative role, how should it cope with the possible linguistic diversity of its members? We conclude that these challenges are not insurmountable, but reveal some trade-offs that cannot be entirely dissolved. Author Biography Pierre-Etienne Vandamme is completing a Ph.D. in political theory at the Hoover Chair of Economic and Social Ethics, University of Louvain (Belgium). He is working on the relation between democracy and social justice. Antoine Verret-Hamelin is doing his Ph.D. at Université Laval (Canada). He is working in the field of intergenerational justice.


Rejuvenating Representation
Contemporary democratic representation can be considered to be in crisis as indicated by the fact that many people express mistrust towards the political class in opinion surveys (Norris, 1999;Rosanvallon, 2006).As a consequence, voter turnout to elections is decreasing in most established democracies (Mair, 2013) and party affiliation and identification have become marginal (Dalton & Wattenberg, 2000).We interpret this as the result of two factors: 1) people do not believe that their representatives act in their best interests (problem of representation); and 2) democratic states have lost a lot of regulating power in a globalized economy characterized by capital mobility (problem of scale).We believe that the problem of scale partly explains the crisis of representation, but not entirely.This paper, however, will limit itself to addressing the problem of representation.Consequently, we acknowledge that the proposed solution might not be enough to tackle the identified crisis.
In this paper, we will use the term representation in two distinct senses: "statistical" or "descriptive" representation means mirroring the diversity of the people; "active" representation means acting in the best interests of the people (Pitkin, 1967;Przeworski, Stokes, & Manin, 1999, p. 2).Part of the contemporary crisis of representation stems from the fact that elected representatives are perceived as not acting in the best interests of the people, precisely because they are descriptively different, because they belong to a particular social class with interests of its own.Therefore, their decisions are believed to be biased in favor of this class.A different worry is that elections tend to make representatives neglect some minorities or issues that do not directly affect the interests of their constituency, such as environmental justice.
In reaction to these worries, scholars and activists press for revitalizing or improving contemporary democracies through innovative practices giving a more important role to lay citizens.In the last decades, a plethora of minipublic experimentsrandomly selecting participantshave taken place around the world.These democratic experiments are nonetheless marginal in the political landscape: They are usually isolated, temporary, infrequent, brief and depend on elected governments for their organization and macro-political uptake (Goodin, 2008).What is more, because they take place outside the formal sphere of political decisions and limit participation to a happy few, their recommendations lack democratic legitimacy (Lafont, 2015).
Things might be different with a deliberative citizen assembly permanently integrated to our modern democracies, using random selection alongside traditional electoral mechanisms.Here is our concrete proposal.The second chamber of representatives 1 , whose usefulness is now challenged in several countries, should be filled through a random selection among the entire population of the country enjoying political rights.This chamber would exist alongside the elected first chamber, whose prerogatives would remain untouched.The main reason for limiting the use of sortition to the designation of the second (or additional) chamber 2 is that elections have some virtues that sortition lacks, in particular the possibilities of universal participation, consent and contestation (Pourtois, 2016).
One could instead suggest having a single, mixed assembly, composed of both elected and randomly selected representatives.This would be especially interesting if you consider the defects of bicameralismin particular, its potential to block or slow down decisions and thus protect the status quo (Przeworski, 2010;Shapiro, 2016;Tsebelis, 2002).Yet we fear that such an assembly would not reap the main benefits of sortition.Allotted representatives would probably suffer from intellectual domination by professional politicians, and political parties would have a strong incentive to pick them up.The benefits of hearing lay voices and avoiding party discipline would thus be highly threatened.
The randomly selected chamber (RSC), as we see it, would have two main legislative roles.First, it would have a power of initiative, the elected chamber keeping the last word. 3Second, it would have a consultative role: it would examine legislative proposals on issues of wide public interest and then provide a public advice and recommendations of amendments.The selected "representatives" (we will discuss in which sense they can be considered representative of the people) would focus on the most important issues and spend the time necessary to examine them in detail.There would first be a training phase during which participants would get accustomed to their new office.Afterward, they would combine work in public plenary sessionswhere they would audit experts and members of civil society, discuss collectively and make final decisionsand in (randomly defined) sub-group sessions where they would deliberate more deeply, under the auspices of 1 Historically, bicameralism has performed two different functions (Muthoo & Shepsle, 2008).In its aristocratic function, it offered the elite a degree of control over the lower, "common" house.In its federal function, it secured representation of the federated entities.Where bicameralism is only an aristocratic legacy, as in the UK, we believe that the reform makes a lot of sense.Where bicameralism secures the representation of linguistically diverse provinces or states, the randomly selected chamber we propose could use quotas in order to keep some degree of group representation.Where the representation of federated entities is crucial for the political equilibrium of the federation, as might be the case in the US, it could be instituted as an additional, third chamber. 2 Unlike what Bouricius (2013), for example, proposes.
3 Another possibility would be to give the RSC the possibility to submit legislative proposals to referendum, but we do not have the space to explore it here at more length, as it involves taking a stance on the value of referendums, which is another issue.trained moderators.The whole process could be supervised by a steering committee appointed by the first chamber and partly composed of members of major political groups, assisted by specialists of deliberative assemblies.We leave other details (length of mandates, rotation, decision mechanism, etc.) open for discussion.
The reasons for giving the RSC only a secondary role (the main chamber remaining elected) will appear progressively in our discussion.Basically, it is a matter of protecting its independence, its deliberative potential and its legitimacy.It is also a matter of avoiding paralysis.If the RSC had the power to veto legislation, the probable tension between the two chambers would become an additional protection of the status quo.To the contrary, the subordinated role allows for a valuable deliberative input in the legislative process.
Why would such chamber be desirable from a deliberative viewpoint?Because it would compensate for one of the main deliberative shortcomings of electoral representative democracy.Although party competition can be judged to have a valuable deliberative function of organizing publicly visible contradictory debates among competing political projects, party discipline hinders deliberation by impeding representatives from revising their opinions and obeying the law of the better argument (Leydet, 2015).Because selected representatives (SR) would not have electoral promises to hold or party instructions to follow, they would likely endorse a more deliberative attitude.What is more, as SR would enter their function without a predefined political agenda, the chamber's work should be organized so as to help them acquire information and exchange views in a deliberative manner.For this purpose, the various experiments that have been made with citizen assemblies around the world and which have shown deliberative quality (see Fournier et al., 2011;Reuchamps & Suiter, 2016) would be valuable sources of inspiration.And with sufficient media coverage, we could hope to see the RSC having macro effects and elevating the quality of the political debate in civil society (Goodin & Dryzek, 2006).This idea of a randomly selected chamber is not new but has not been fully explored yet. 4Many of its advocates tend to limit the discussion to its 4 The first modern discussion of representation by lotwhich was practiced in ancient Greece and several Italian Republics in the Middle Ages (Manin, 1997) is probably Robert Dahl's (1970, pp. 122-125).Dahl rejects it yet suggests using sortition for "selecting advisory councils to every elected official" (p.123).In reaction to Dahl, Mueller, Tollison and Mullett (1972) have defended the total replacement of general elections by random selection from a public choice perspective.The idea of having one chamber filled through random selection (and the other elected) has been defended by Callenbach and Phillips (1985) for the US Congress; Sutherland (2008) for the House of Commons in the UK; Barnett and Carty (2008) for the House of Lords; Buchstein and Hein (2010) for the EU; Sintomer (2011, pp. 234-235) as a third chamber in France; Van Reybrouck (2014) for the Belgian Senate.Goodwin (1992), promises, neglecting the careful examination of the challenges it would face.We believe it is worth deepening the discussion with a concrete proposal and a realistic appreciation of these promises and challenges.It cannot simply be assumed that "a randomly selected chamber would be independent, representative, and deliberative" (MacKenzie, 2016, p. 283).

Promises The Anti-Aristocratic Argument
In a small-scale society, one could imagine political equality going beyond the formal freedom to run for office and universal suffrage.Every citizen could enjoy the same amount of political power, including legislative power.In ancient Greece, the Athenian model of direct democracy embedded in the ecclesia aimed at this kind of equality: a substantial proportion of the citizenry assembled frequently on the Pnyx to pass laws and appoint magistrates.But in today's mass democracies, there seems to be no way to distribute legislative power equally.
How, then, should legislators be designated?The modern answer to this question has its roots in contractualism: We may give up on legislative equality, but at least those who do get legislative power should be the ones we gave our consent to.As a mechanism that embodies consent, elections satisfy a crucial egalitarian objective: one citizen, one vote, and this vote authorizes a legislative body to exercise power on the entire citizenry.Designation by popular will through formally egalitarian elections seems the obvious mechanism to operate a political division of labor that stays as close as possible to the ideal of political equality.This is the functional explanation of elections.Bernard Manin (1997) has shown how elections also rest on aristocratic justifications.His now famous thesis is that the principleconsentand mechanismelectionthat prevailed in the American and French revolutions and still prevail today were initially seen as opposed to democracy.Elections were put in place to ensure a "principle of distinction," according to which "elected representatives would and should be distinguished citizens socially different from those who elected them" (Manin, 1997, p. 94).Elections certainly constituted a progress towards more equality compared to monarchy and the hereditary transmission of power, but they were not intended to realize an egalitarian objective.They offered a (aristocratic) way to legitimate the elite's power, but at the expense of "the problem of distributive justice in the allocation of political functions" (Manin, 1997, p. 91).
Nowadays we can still see the mark of these aristocratic origins: Even if we ignored the influence of private donations on the outcome of elections and the influence of lobbies during a government's mandate, we would still deplore the fact that governments are dominated by professional politicians who are too often wealthy and highly educated white males.This does not necessarily imply that they act in the interests of their peers.Yet the founding fathers of modern democracies also made sure that representatives would not be too accountable to the people (Przeworski, Stokes & Manin, 1999).Between elections, they enjoy an important freedom to pursue policies that do not necessarily enjoy popular support, and for these deviations from public opinion to be sanctioned at the next election, voters should be sufficiently informed about governments' actions and their consequences, which is seldom the case.So unless the representatives are naturally and consistently publicspirited, this aristocratic character of elections is likely to turn into economic privileges. 5he class bias introduced by elections and the independence of representatives certainly account for a large part of the contemporary distrust in representative democracy.Together, they convey an impression of popular powerlessness.In light of this, sortition has the potential to alleviate this crisis without getting rid of representation in general, whose value goes further than a mere convenient division of labor (Urbinati, 2006).With sortition in the mix, the aristocratic character of elections would be interestingly counterbalanced.Even if a RSC would probably not be fully descriptively representative of the people at large, the increased similarity with lay citizens might strengthen the link between the representative body and public opinion.Although members of the RSC might deviate from public opinion as a result of their deliberations, their initial views are likely to be closer to public opinion than those of elected representatives and their ongoing thought process could be expected to be more transparent, as the transcripts of sub-group sessions would be made available.Furthermore, random selection will likely lessen the impression that politics is a game played by the elite only.Without being a panacea, it would enhance equality of access to political power, since every citizenregardless of social classwould have a real chance of holding a political office.In sum, involving ordinary citizens at the core of legislative processes would probably increase the perceived legitimacy of the resulting decisions.Would that be at the expense of the quality of these decisions?This is the argument examinedand refutedin the next section.

The Epistemic Argument
Epistemic approaches to democracy consider democracy valuable (at least partly) because it has a tendency to engender right decisions (or at least better decisions than its alternatives).The epistemic case for sortition is based on the assumption that the random selection of representatives would increase this epistemic potential of democracy.Given that the fear of lay citizens' political incompetence is generally considered as the knock-down argument against the random attribution of positions of political power, the epistemic argument is crucial.To accept this argument, it is not necessary to agree on what exactly a right democratic decision is, in reference to a specific conception of justice for example.All you need to accept is that some features of a decision process can increase its epistemic quality.We take impartiality and (instrumental) rationality to be two central features of a right political decision (Vandamme, 2016).A decision is right if it does not unjustly favor some citizens over others (impartiality) and if the means selected are appropriate to the pursued ends (rationality).

Diversity
One feature that is favorable to rationality and impartiality is diversity.As convincingly argued by Hélène Landemore (inspired by empirical investigations on collective intelligence by Lu Hong and Scott Page (Hong & Page, 2012;Page, 2007)), diversity matters more than ability for solving problems that are difficult enough to require a collective solving action (Landemore, 2013b).Faced with such problems, diversified groups generally prove more able than groups of experts.Hence the Diversity Trumps Ability Theorem claims that a randomly selected group generally outperforms the best problem-solvers on four conditions: 1) The problem could not be solved by an individual.2) All agents are minimally smart.
3) Participants vary in their reasoning methods.4) People are randomly selected among a wide population (which generally guarantees diversity).
Given the difficulty of seeing these four conditions fully satisfied, Landemore (2013b) endorses a more modest version of the theorem: "It is often better to have a group of cognitively diverse people than a group of very smart people who think alike" (p.103).On the further assumption that democracy is (at least partly) a matter of problem-solving, this is a pro tanto reason to prefer sortition to elections.As already mentioned, elections have the (aristocratic) function to foster the selection of the "bests" among their peers.Although there is no explicit criterion of selection, people tend to vote for a specific kind of person: well-educated smooth talkers who are self-confident.The pivotal role of the televisual media in contemporary societies often gives the edge to the most charismatic candidates (Manin, 1997, pp. 218-234).What is more, elected politicians channel their passion and ambition to their children or relatives.The political vocation is thus partly inherited, partly shaped by elitist school careers.In sum, many elected politicians are similar to each other in some respect and different from the masses.
This might not be a problem by itself if it were not to engender epistemic failures.In short, people following the same kind of careers, going to the same schools, experiencing more or less the same social trajectory, enjoying similar incomes, and so forth tend to think alike. 6The lack of diversity introduces biases, which lessen an assembly's epistemic potential.Elected representatives might fail to consider the problems they are facing from perspectives radically different from theirs; they might lose track of the social reality experienced by having lower income; they might ignore completely what it means to work in some professions (let alone being unemployed).Consequently, some ideas or aspirations are simply left aside.And this accounts for the fact that many people do not feel represented not only descriptively, but also ideologically.
Sortition appears as a natural response to these shortcomings of electoral representation (Landemore, 2013a).By reducing the selection bias 7 and offering the opportunity to access power to people who have not inherited the political vocation, it can bring much more diversity into representative assemblies.Although SR would earn the same income during their mandate, share a common experience and face roughly the same information, they would bring with them the diversity of their past experiences, which would have the effect of widening the collective perspective on the political issues at stake.This is appealing for the desideratum of rationality, because increased cognitive diversity fosters more rational decisions.And it is appealing for the desideratum of impartiality, because positional biasesneglecting some citizens' interests or claimsare reduced.

Deliberation
As Landemore explains, the mere aggregation of diverse views has epistemic value.Yet, what increases the epistemic potential of a diversified assembly is deliberation (Landemore, 2013b), for it helps correcting biases and pushes towards more impartial decisions (Habermas, 1992).
One of the main warnings stemming from recent research on deliberative democracy is that deliberation is not always beneficial and can have, in certain 6 One might take the strong polarization of most political arenas as a denial of this claim.Yet we should not underestimate the electoral incentive to polarization.Politicians face an incentive to publicly distinguish their own position from the others' even if they possibly agree on many fundamental issues. 7As will be discussed in section 3.1, the selection bias will not disappear but might be significantly reduced.
circumstances, perverse effects.For instance, some of our cognitive biases can corrupt the process of deliberation and lead us to reinforce our initial prejudices and overlook some points of view (Smith, 2014).The main conclusion of this body of research is that the quality of deliberation always depends on the context in which it takes place.Consistent with the Diversity Trumps Ability Theorem, one of the factors affecting the quality of deliberation is the heterogeneity of the deliberating group: where groups are homogenouswhen participants share a number of characteristics or hold similar initial viewswe often see the "group polarization phenomenon," meaning that the group will have simply radicalized its initial position at the end of the deliberative process (Sunstein, 2002).But in the right circumstances small and diverse deliberative groups, presence of a moderator, etc.deliberation does make people revise their positions in more impartial and rational ways (Smith, 2014, pp. 281-282).Diversity and deliberation therefore have complementary virtues.
Now, why can we expect an RSC to have a better deliberative potential than existing representative assemblies, whose deliberative record is, to say the least, poor?First, the fact that some representatives would not be selected on the basis of a political program would make them freer to adopt an attitude of intellectual modesty, to listen to experts and to their peers, and to change their minds.Obviously, they would bring with them their prejudices and prior political commitments, but the incentives are nonetheless very different.Second, the fact that they are not attached to a political party also adds to their freedom of judgment.One could imagine coalition-building within the RSC, but at least positions would not be as fixed as they currently are.Third, having only a suspensive (and not absolute) veto power, the RSC would be less vulnerable to external pressure (lobbies) and internal fights for power.Opting for a subordinated role for the RSC, as we recommend, would reduce the fights for power and influence while increasing the incentive to produce reasoned recommendations after high-quality deliberations (Shapiro, 2016).
Hence we must recognize a trade-off between the aim to increase the quality of public deliberations and that of empowering lay citizens to counterbalance class biases in decision-making.
In addition to these spontaneous deliberative effects of sortition, quality deliberations can be fostered through methodological techniques used in deliberative mini-publics such as the presence of a moderator and a devil's advocate in subgroup sessions, the provision of information booklets capturing a variety of viewpoints, or the dialogue with a diversity of experts (Caluwaerts & Ugarriza, 2012).

Humility
Can we nonetheless trust lay citizens in light of the literature on citizens' political ignorance (Caplan, 2007)?One point to make in addition to the value of diversity and deliberation is that the expertise of elected representatives is often overrated.They are generally selected on other grounds than political competence, and what makes them more or less prone to smart decisions is their willingness and ability to listen to experts and advisors.From this perspective, one further epistemic advantage of sortition is humility (Goodwin, 1992): people know that their selection cannot be attributed to their merits.This creates a better incentive to listen to experts.Elected politicians often feel a need for advice, but they were elected for their ideas and abilities (or so they think), which gives them confidence to turn their back to expert advice when they disagree or when this advice contradicts their prejudices or does not suit their interests.
One might judge it desirable to have representatives with strong convictions and confidence in their own expertise because it protects them against the influence of biased experts.This brings us to a further advantage of sortition: Much more attention will be paid, under sortition, to the selection of the experts endowed with the task of informing the SR than is the case with elections.In the latter, it is the responsibility of elected politicians to find information for themselves and to listen to whomever they want.We are less afraid of bad expertise because we know that elected politicians (and their parties) have clear (or at least pre-announced) ideas on each issue.Thus, the fact that the risk of bad expertise increases with sortition actually creates an incentive to publicly scrutinize the mechanisms of expert selection, which potentially increases the likelihood of a more politically neutral (or at least diverse) expertise.

Wider Time Horizon
Another epistemic quality of an RSC stems from its unrestrained temporal perspective, which will be of great help in tackling the environmental and climate crisis and reaching sustainability.By not being subjected to periodical elections, the temporal horizon of deliberants is much more open.It is no longer restrained to the short-termism of electoral cycles, where the temporal horizon is no longer than a few years.With periodical elections, as Philippe Van Parijs (2014) puts it, "even the best democracy remains a dictatorship of the present" (p.49).To be sure, frequent elections are of great importance for electoral democracy: By submitting governments to periodical elections, citizens keep some degree of control over (part of) their representatives (the issue of accountability is further examined in section 3.2).Furthermore, ecologically minded citizens can ask their representatives to account for the long-term effects of their policy choices.But the frequent reiteration of elections also has an unfortunate side-effect: Elected officials have no structural incentive to put forward policies that will bear fruit in the long run, with no visible effects within a few years, because on election day a government is judged on its perceptible accomplishments (Bourg & Whiteside, 2010, p. 70).And this is not exclusive of officials who try to hold office as long as possible.Even a politician who would not run for re-election but would care about his or her political legacy is likely to favor short-term, clearly identifiable accomplishments in order to get the credit.An elected official who would defy this short-termism and try to enact laws and policies looking further down the roadputting forward, for instance, an ambitious and long-term plan to protect the environment and mitigate climate changemight still see her efforts overturned by the following government.For these reasons, electoral democracy is profoundly "myopic," which is an important obstacle regarding the fight against pollution, climate change and biodiversity loss, all problems that are forming slowly and that need long-term solutions (Bourg & Whiteside, 2010, p. 70).Moreover, the "media tempo," always looking for new stories, exacerbates this difficulty to favor a long-term perspective (Sintomer, 2011, p. 26).
Conversely, randomly selected deliberants are not stuck in the short-termism of electoral cycles.Their temporal horizon can be as far as their free and inclusive deliberations bring them.Even if there is no guarantee that SR will be more concerned with the future8 , the hope is that the removal of the disincentives explained above will mitigate the short-termism of electoral democracy and make the overall legislative process more sensitive to the issues that exceed the temporality of an electoral cycle.Once again, this change of perspective is probably an essential step toward sustainability.So it might be true that even the best electoral democracy remains a dictatorship of the present, yet the "bi-representative model" (Van Reybrouck, 2014, p. 179) here defended offers better incentives to take the future into account.

Challenges
Any proposal for democratic reform must try to anticipate its effects and avoid discounting the negative ones.The idea of an RSC faces numerous challenges.
As will be made clear in the coming sections, we are confident that they do not disqualify the project, but they certainly warn against uncritical optimism and inform about desirable (and undesirable) implementation details and side measures.

Participation
The first challenge faced by any use of sortition for the attribution of tasks engages with the issue of acceptance and participation.The rates of acceptance to join deliberative assemblies are almost always very low.9When asked why they refuse to take part, people generally invoke a lack of time, which is probably true but can also hide other reasons for refusing.These reasons, sometimes invoked explicitly by the people, are as various as fear to speak in public, lack of self-confidence, hostility to politics, doubts about the organization, or self-exclusion ("It's not my role; it's for politicians") (Jacquet, 2017).No doubt, similar reasons would ground refusals to serve in an RSC.And the rates of acceptance might be even lower given the national exposure and the public pressure it entails for those who accept the honor/burden.
The only way to overcome this would be to make participation mandatory10 and the obligation binding.After all, participation is generally mandatory in the case of citizen juries.Yet an obligation to serve one or a few years in a political assembly seems much more demanding.Mandatory participation would therefore cause both practical and ethical worries.Practically, it is hard to imagine how we could secure acceptance without important sanctions.The moral obligation would not be enough.Voting is mandatory in some countries, and yet many people do not show up to the polls.11Sometimes it is because they know they will not be sanctioned, sometimes because they prefer to pay a fine.Given the demandingness of the requirement to serve in an assembly, many people might prefer to pay a fine.And if the fine is high, it creates a strong inequality: Some people are simply able to refuse the burden thanks to their good fortune.This is probably not what defenders of the obligation to accept have in mind.What is more, one can wonder about the added value of having unmotivated people serving in the assembly. 12esides, one can express doubts about the supposed obligation to show interest for politics.We surely have a moral duty to take others into account when making political decisions.We might even have a moral duty to vote in cases where we are confident that one option would advance the cause of justice.But it is highly debatable that we should have a moral duty to accept spending one or several years of one's life in an assembly exposed to public pressure and in charge of important and complex decisions.One can have various legitimate reasons for refusing the task: living too far from the capital city, taking care of one's relatives, having no interest in politics or considering oneself incompetent for the mission.One of the virtues of representative democracy is that it secures the freedom to delegate and escape political burdens.The problem is that delegation has gone too far, and the elite enjoy a sort of monopoly over political power.Yet the solution is wider opportunities to participate, and certainly not a general obligation to.
What does this entail for advocates of sortition?First and foremost, it means that proponents of an RSC should abandon any pretension to full descriptive or statistical representation of the people.Self-selection biases are probably unavoidable.Yet we should not be too worried about this if the goal is to increase diversity rather than secure full representativeness.The general epistemic case for sortition is not affected by low acceptance rates, as long as a RSC outperforms an elected chamber in terms of diversity and deliberative quality.
Second, in order to maximize diversity, we should certainly consider using quotas in addition to random selection (Caluwaerts & Reuchamps, 2014), or a stratified sampling method13 (Stone, 2011, p. 134).This will not entirely solve selection biases, because those who accept will probably share some common characteristics (self-confidence, political commitments) even though they are part of targeted minorities.Yet if quotas help increase diversity, it will be enough.Surely, there will be controversies regarding the minority groups that should be entitled to secured representation, but this obstacle is not insuperable.
Third, one should take seriously the possibility of using anonymity to incentivize wider participation.As we mentioned, the RSC could combine public plenary sessions that would be broadcast with sub-group discussions.
The transcripts of the latter could be made public while preserving members' anonymity, and these sub-groups could elect spokespersons that would take the lead in the plenary sessions.This way, one of the biggest disincentives to participatepublic exposure and pressuredisappears.This would have a limited cost in terms of deliberative interactions between the RSC and the wider public, since we could follow the entire thought process of SR.The only thing we would not know is "who thinks what."Moreover, we could keep the participants' names and addresses secret, as was the case with the Irish Convention on the Constitution (2013).
Fourth, we should also think about material incentives to participate.Surely a generous wage will attract more people and increase diversity.Yet, at the same time, there are two negative aspects of a high wage that should be taken into account.First, it might affect the motivations of people choosing to serve in the assembly.It might attract more people who are motivated solely by money and who would remain apathetic in the RSC.Second, given that the elite is overrepresented in elected assemblies, one could see a modest wage as a disincentive for the most well-off to participate, which could be valuable from the perspective of diversity or in order to mitigate the political influence of the wealthy. 14The right wage for SR should be high enough to convince people committed to the public interest to interrupt their career, and some people to overcome their shyness, but low enough to keep away the economic elite (which already enjoys high political influence) and those motivated exclusively by the monetary aspect of the job.
Another incentive which is not directly monetary but might prove crucial for attracting people with low qualifications is the possibility to get one's job back at the end of the mandate. 15Ideally, people with permanent contracts would enjoy a kind of free-of-charge time during their mandate, subsidized by the state.It would probably be impossible to guarantee this to people with temporary contracts, but given their situation, they might be less reluctant to leave their job for a temporary public position. 16

Legitimacy and Accountability
Another source of concern for advocates of sortition should be the perceived legitimacy of the RSC.Elected representatives derive their democratic legitimacy from the fact that everyone has had the chance to vote and thus to influence their selection.Voting entails consenting to be represented and accepting the rules of the game.Elected people have the right to make decisions in the people's name for the duration of their term.Voting does not exclude opposing the government's decisions, but it introduces a distinction between justice and legitimacy.If you have participated to elections, you can both hold the government's decisions legitimatebecause you have had your chance to make your caseand unjustbecause you nonetheless disagree with the decision and thus oppose them on substance, not on process.
14 McCormick (2011) suggests creating a People's Tribunate of 51 randomly selected nonwealthy citizens, with powers reminiscent of the Tribunes of the Plebs in ancient Rome.This institution would squarely exclude the economic and political elite so as to mitigate their disproportionate political power.Excluding the wealthy is not discriminatory, he suggests, because if they want to have access to this institution, all they need to do is getting rid of their wealth surplus (p.13).The risk is that it would increase the wealthy's incentives to exercise hidden influence.An inclusive popular assembly as defended here might counter the power of the elite without legitimizing and entrenching a class conflict. 15We thank Anne Pelsser for pressing this point. 16Especially under the assumption that most temporary contracts are low-paid and most probably lower than what SR would earn.
Sortition does not generate legitimacy in the same way.In a way, it makes accountability unnecessary.If the function of sortition is to select a diversity of interests and worldviews, then SR are legitimate in defending their interests and pursuing their preferred policies without having to account for their decisions (Pourtois, 2016;Sutherland, 2011).But this may not be seen as sufficient, and people would surely need to consent to the use of sortition before the reform is acted.Yet this consent would not be continuously renewed at each (s)election.Thus, we can imagine a situation where people first accept the use of sortition and then come to question the legitimacy of the RSC.This could happen if they do not, after all, feel adequately represented by it.To be sure, this happens with elections too.But there are two significant differences in this respect between elections and sortition.1) Under elections, people who do not feel adequately represented know they have a chance to choose better representatives next time. 17Under sortition, people can only hope they will be better represented next time.2) Under elections, people have had an opportunity to choose representatives with an idea of the policies they would promoteeven though electoral promises are seldom faithfully kept.Under sortition, to accept the legitimacy of a body which does not "actively" represent your political aspirations, the reasoning is much more complex.You need some understanding of probabilities and a great deal of reflexive thinking to accept that the policies promoted by the RSC are most probably in line with the interests of the people.You need to accept the idea that SR are placed in a position of information and deliberation that might legitimately make them see things differently than the people they are representing and that, placed in the same conditions, the latter would probably "come to have those views" (Guerrero, 2014, p. 159). 18Yet this reasoning would work on two conditions only. 1) The rates of acceptance are such that the RSC genuinely mirrors the people in its full diversity.2) People perceive this hypothetical match ("I would think alike in those conditions") and consider themselves represented without having to interact with their representatives and without identifying with their ideas.
The first condition is not plausible, for reasons previously highlighted.For the second condition, evidence that such reasoning is not obvious is provided by various citizen assembly experiments.When policy proposals elaborated by citizen assemblies are submitted to referendums, they are often rejected by the 17 One should nonetheless not be too impressed by the accountability offered by the reiteration of elections.First, for the mechanism to work properly, people should be sufficiently informed, a condition that is almost never met (see Przeworski, Stokes, & Manin, 1999).Second, in the absence of recall mechanisms, rulers still enjoy a lot of freedom between elections.Yet, despite these shortcomings, electoral accountability might already be better than no accountability.If it had no effect, politicians would show themselves much less concerned with opinion polls.We thank Hervé Pourtois for the latter suggestion. 18A divergence between the public opinion and the enlightened opinion of the selected representatives is to be expected if deliberations have the effects often attributed to them.The deliberative macro effects of the RSC might nonetheless reduce this divergence.
people. 19The latter do not seem to trust the descriptive representativeness of the assemblies enough to surrender to their judgment.They seem to vote according to their own judgment or prejudices on the issue at stake.
What is more, one could argue that descriptive representation is not the (only) kind of representation that people expect.They may want to identify more with their representatives' identity, but they probably also want their representatives to "act for them" (Pitkin, 1967), whatever their identity.To secure such active representation, some kind of accountability is therefore necessary, a public perception that the selected officials cannot do whatever they want.But this could still be achieved by the RSC through discursive accountability (Goodin, 2008, pp. 155-185) by encouraging interactions between SR and the larger public.The SR would be held publicly "accountable" for their decisions insofar as they would have to convey numerous public hearings and would have a duty to justify their final decisions to the whole population.This means that they would not only make recommendations of amendments to the first chamber; they would also have a duty to provide to the public an assessment of the legislative proposals they would review or initiate.In return, they might benefit from public opinion's influence through the media, public-spirited lobbying or even demonstrations.Such discursive interactions would be valuable from a deliberative point of view, contributing to a lively public debate, and it would also increase the discursive accountability of elected officials: A more informed and engaged public would keep elected officials on their toes.
The role played by the media will also prove crucial for discursive accountability.As a matter of fact, what partly accounts for the discrepancy between mini-publics and the larger public in citizen assembly experiments is the lack of media coverage (Fournier et al., 2011, pp. 136-139).Quite a few times, only a minority of the people voting in referendums knew that they were voting on proposals stemming from a citizen assembly.But this would certainly be different with an RSC, as it would constitute a central public institution, not an occasional experiment.
This, however, might still not be enough.People could be reluctant to trust the judgment of the SR if nothing makes the latter institutionally accountable as was the case in ancient Greece (Elster, 1999;Sintomer, 2011, p. 204).Because discursive accountability is probably less efficient than electoral accountability (SR do not have the incentive to anticipate the retrospective judgment of those they represent, which is provided by the perspective of re-election), one could imagine securing accountability through a recall mechanism.It could be made possible for citizens to petition against a representative and remove him or her on some conditionsa number of signatures or a (super)majority in a vote, for example.This could be quite harsh for SR, but the mere anticipation of the possibility to be recalled might be enough to incentivize them not to take extreme positions.Now, there are several reasons to resist installing a recall mechanism that would probably sanction unorthodox views.One of them is the deliberative value of diversity previously highlighted.Another is the perception of the RSC as a space for free speech.One of the virtues of an RSC is that it might reconcile with democratic institutions many people who currently see them as embodying the interests and aspirations of the elite or the majority only.The gain in legitimacy might be conditional on the possibility to defend unorthodox views in the new chamber.For these reasons, discursive accountability might still be the best bet, possibly supplemented with a recall mechanism whose conditions would be so stringent that it would seldom be used.Its function would then be to give the wider people a sense of power over the RSC, an opportunity, even if small, to "throw rascals out," in Popper's words. 20Yetas Van Parijs (2014) aptly puts it, "the unpopular need not be rascals" (p.49)they can also be the ones defending foreigners or future generations against the interests of voters.
The main lesson to draw from this discussion is that sortition should never entirely replace elections.Its legitimacy stems from its particular rolea deliberative inputin a broader deliberative system in which electoral accountability also plays a crucial role (Parkinson, 2006, pp. 34-35).Despite all their shortcomings, elections have virtues that are complementary to what sortition offers (Pourtois, 2016).They provide some kind ofalbeit deficient potentially effective accountability; they secure some "consent" legitimacy through universal (opportunity for) participation; and they have some epistemic virtues of their ownin particular the feedback mechanism that they allow between representatives and the people (Anderson, 2006), which is valuable for the circulation of information and the self-correction of the legislative system.

Independence
Related to the issue of accountability is that of independence.Although we want our representatives to be influenced by quality arguments, we do not want them to be at the mercy of influent people or interest groups.In other words, we want deliberation without corruption.Yet the very publicity that increases the discursive accountability of the SR also makes them more vulnerable to corruption.Why should this worry us more than in the case of elections?Because party discipline in voting has this desirable side effect of making corruption more difficult.In order to make sure an MP is going to vote the way you want her to, you have to take into account the position officially defended by the party.And on most issues, because of party discipline, you will have to buy the whole party's allegiance. 21With sortition, if votes are public, it would be easier to identify the pivotal representatives and try to buy their allegiance.
Another comparative disadvantage of sortition is the previously discussed fact that accountability to the represented is lower.As Guerrero (2014) notices, this increases the risks of "capture" by powerful interests."If political officials are free to take a wider range of positionsand, in particular, are free to take positions that are contrary to the interests of the nonpowerfulthis makes the offices held by those officials more valuable, more worth controlling.Thus, as representatives become less accountable, it becomes more worth the effort to control those representatives […]." (p.142) Yet sortition also enjoys some advantages regarding independence.One of them is that, compared to elections, it is very difficult for the wealthy to influence the selection process.According to Oliver Dowlen (2009), "the protection of the public process of selection from manipulation constitutes the primary political potential of sortition."(p.308) Elections make it possible to finance one's favorite candidate's campaign.This not only impairs the principle of equal access to power positions, but also tends to distort democracy at the legislative stage by making MP's mainly accountable to their campaign sponsors.Under sortition, things would be different."While it is always possible for a citizen to become corrupted or to take bribes once in office, the fact that he or she owes their office to no one means there is no one in the potential position to call in the favor at a later date" (Dowlen, quoted in Peonidis, 2013, p. 46).As suggested by Guerrero (2014), it is easier to protect SR against corruption because they do not face the "need to raise funds for reelection" (p.164).Under sortition, there is still the risk of manipulation of the selection process22 , but this might prove difficult in countries with free press and a long-standing democratic tradition.What is more, given that SR will rotate more regularly than elected representatives, corruption would become more costly (Guerrero, 2014, p. 164).
Nevertheless, the risks of corruption are a reality and need to be taken seriously.In order to tackle this issue, secrecy is a potential solution. 23Yet if it is possible to fight corruption without secrecy, it would be desirable given the aim of discursive accountability and our willingness to restore popular trust -which might be difficult without some form of transparency.The formula we suggested earlier, combining plenary public and sub-group anonymous sessions, would make it difficult to corrupt members of the RSC while maintaining an important degree of transparency.
Another possibility would be to provide SR with a very high wage, potentially reducing vulnerability to corruption.But this would come in tension with other goals highlighted in section 3.1 (reduced elite representation and proper motivations).What is more, it might lessen popular identification to their SRwhich is crucial for the RSC's legitimacyby casting doubt on the real motivations of SR (money vs. public interest).Yet the strongest argument against such an idea is that it is a very costly and uncertain way of aiming at reducing corruption.It appears intuitively clear that people with low wages in poor countries are more prone to corruption than people with high wages in rich countries.Yet being wealthy certainly does not protect against corruption.
There is even some evidence that richer people are generally more likely to cheat than poorer people (Piff et al., 2012).
Other incentives against corruption can be imagined, such as a denunciation reward for SR reporting attempts at corruption or an oath taken with high sanctions in case of breach.In any case, even if they need to be taken seriously, the risks of corruption do not constitute a decisive argument against sortition, because compared to elections the prospects are mixed.Those risks nonetheless inform about desirable implementation issues.

Language
A final question that can be raised in multilingual societies concerns the language of deliberation.In what language will the deliberation be conveyed?Will there be any linguistic requirements to participate in the RSC?Many countries around the world are multinational, in that they assemble different historic peoples territorially situatedwhich is different from pluralism due to recent and partly voluntary immigration (Kymlicka, 1995).Proponents of "liberal culturalismˮ contend that in multinational societies, some extent of decentralization should allow national minorities some level of autonomy, so as to preserve their cultural distinctiveness.But even some "civic nationalistsˮ who resist giving privileges to some minoritiesendorse the idea that states can legitimately promote minority languages (Stilz, 2009).Therefore, if there is a wide consensus on the idea that some measures should be taken to preserve minority languages, how can we expect citizens speaking different languages to deliberate in the RSC?If the Canadian Senate were to become an RSC, how could the French Quebecer, the English Ontarian and the Aboriginal Yukoner debate together on complex political issues?And what about an RSC at the EU level?This is the puzzle that leads Will Kymlicka to a conclusion well captured in the title of his book Politics in the Vernacular (Kymlicka, 2001).According to him, if we want the democratic debate to be inclusive, it ought to be conveyed at the level of linguistic groupsit ought to be conveyed in the vernacular.In his words, linguistic groups are the "primary forums for democratic participation in the modern worldˮ (Kymlicka, 2001, p. 213).The main reason for this is the contrast between linguistic faculties of the elitefluent in more than one languageand the average citizen.
So in order to have a public debate that is not dominated by the elite and that allows a truly inclusive and participatory formation of the opinion, the "primaryˮ institutions of democracy should stay within the borders of national-linguistic groups.From this perspective, we should see with skepticism the possibility of an encompassing RSC in multinational states, let alone at the supranational level.To be sure, the linguistic challenge is not peculiar to the RSC.Elected chambers of multinational states face the same problem, and solve the issue with instant translation.But since the promise of the RSC is to foster quality deliberations, one might argue that fluid, spontaneous and unmediated interactions are essential and require a shared language.
On closer look, linguistic diversity may not be the barrier we imagine it to be, especially in light of Nicole Doerr's findings.By analyzing translation practices in the European Social Forums, she concludes that some translation practices can improve the quality of deliberation.This surprising conclusion is explained by the transformative potential of translation in regard to hearing and listening habits: "Multilingual debates, due to what participants perceived as an increased risk of conflict caused by misunderstandings, strongly induce participants to listen attentively to statements madeˮ (Doerr, 2012, p. 13).As one of her interviewees explains, translation requirements slowed down the pace of the debate, and forced everyone to be active listeners: "Right now within this European assembly we need a lot of time at the beginning of every discussion in order to carefully find out what exactly is the position of the others.I think this has to do with all the different languages and backgrounds.ˮ(Doerr, 2012, p. 14) Moreover, debates in multilingual contexts can avoid certain burdens that usually stain deliberation in national contexts, such as stigmas associated with particular accents and ethnicities (Doerr, 2012, pp. 17-18).Finally, Doerr found that translators "made it easier for newcomers and formerly marginalized groups to be heard and included by facilitators,ˮ the latter having otherwise a tendency to "reproduce structural cleavagesˮ (Doerr, 2012, p. 18).So, with face-to-face practices of translation in the sub-group sessions of the RSC, translators could usefully counterbalance the subtle influence of moderators.For all these reasons, multilingual deliberative forums are not only possible, but may conceal attractive features.

Conclusion
It is nowadays a commonplace to say that democracy is in dire need of revitalization.In this paper, we examined the pros and cons of a reform of bicameralism: While elections should remain in place to designate members of the lower house, the high (or additional) chamber should be dedicated to lay citizens, randomly selected, having the prerogative to propose new laws and make recommendations and amendments to bills proposed by the lower house.Such reform can be defended both from the perspective of legitimacyas it could restore popular trust in their representative institutionsand justiceas it would increase democracy's epistemic potential, i.e. its capacity to make right decisions.
An RSC would nonetheless face several challenges.As we have attempted to argue, they are not insurmountable, but they reveal some trade-offs that cannot be entirely dissolved.Some of these trade-offs are certainly in need of further scrutiny.
The most fundamental trade-off at the heart of our proposal concerns the specific mandate and prerogatives of the RSC.The more power it will have, the more significantly it will reap the anti-aristocratic and (most of the) epistemic benefits we highlighted, but the more its members will be vulnerable to corruption, intimidated by public pressure, and questioned in their legitimacy, and this could ultimately backfire on the RSC's promises.The equilibrium we put forward leaves legislative power and primacy to the lower house, while giving the RSC a power of initiative, consultation and amendment.Is this the perfect point of equilibrium between the two chambers?Maybe not.One could argue that the RSC should have the power to initiate referendums and/or to veto legislations.The main goal of the paper was not to defend a specific institutional proposal, but to assess the promises of a bi-representative system in general and the challenges it would face whatever its precise form.The next task will be to compare various detailed institutional frameworks and assess their respective merits.