Electoral Lottocracy - A Critique

> proposes a top-n token system for individual legislators Just a quick note of clarification; the token system is not a top-n system. Each year, each legislator is free to table legislation, and if it reaches 20% support in Parliament, they may use a token to have the legislation be the subject of a legislative trial before a jury. The basic 20% threshold and the token are sufficient to access this trial, and it does not depend upon matching a pre-set agenda. > However, whereas a trial jury seeks to establish the facts of the case (guilty or innocent), the principal task of a nomothetic jury is to reflect the informed preferences of the population that it seeks to ‘describe’. I think the task of the legal jury and that of a legislative jury, in theory, are highly similar. We can look at both facts and values as two elements of the deliberative process, both of which are present in both legal and legislative deliberations. While many tend to emphasize the fact-finding nature of legal juries, there are many instances in which the facts may not present a singular obvious outcome, and as such judgement needs to be applied. The application of this judgement, inevitably, carries with it the values of the participants. Likewise, a legislative jury, in which we examine the values of the participants, will inevitably examine a great number of facts. After all, it is not the case, for example, that those who oppose strong climate change responses are in favour of drought, floods, forest fires and the resulting migration crisis that are predicted by climate advocates, but question the epistemic validity or certainty of their claims. It is similarly not the case that those in favour of dramatic climate action are in favour of significant economic disruptions in a general sense. The balance between climate action and economic challenges isn’t simply a matter of values, but a different sense of the epistemic validity of climate science as a field. Once the matters of fact are settled, differences in values still remain and thus there is still a lot to work out in these areas. > Rose’s reference to ‘bias’ in both variants suggests an epistemic perspective, however, there are no right or wrong answers to political questions. If a person has a particular value, and in pursuit of the realization of that value they encourage policy based upon misinformation, such that it can be shown that the policy would fail to realize, or possibly even harm, their stated goal, I would state that this is clearly an example of a wrong answer. If my desire is to increase economic prosperity within my society, and my stated solution is to declare November as silly walk month, with no reasons provided as to the connection between these things, I feel as though most would agree this is a wrong answer to the question. While there may be multiple answers that can be justified with reasons that demonstrate plausible causality between problem and solution, not all proposed solutions may meet this threshold. Nonetheless, we may be predisposed to accept a solution that fails to meet this threshold when it aligns with our preconceived notions. Our biases do not simply disappear when we are examining political questions. In fact, most, I would argue, would agree the opposite is true. We are most biased in matters of politics. I am not looking to claim one side more guilty than the other, only that the coming together of varying biases is beneficial to the deliberative process and the production of reasons.

>the coming together of varying biases is beneficial to the deliberative process and the production of reasons. Yes, I'm sure that's right. But in both the trial and legislative jury case, what matters is outcomes, not reasons. Trial juries (in the UK) are only required to return an up/down verdict, and the task of the legislative jury is to reflect the informed preferences of the citizens who are not involved. The exchange of reasons may help or hinder this process -- I believe Kai O'Reilly is devising a way of putting this to the test.

> the task of the legislative jury is to reflect the informed preferences of the citizens who are not involved. So, this is not quite how it's positioned inside my own system. How, and why, I deviate from this will be the subject of my next paper, and I don't wish to do my arguments injustice by leaving them unmotivated. However, I agree entirely with Lafont that even if we could determine the sortition body to be fully representative of the larger population, that this would do little to add legitimacy to the process. Citizens that object to the ruling wouldn't see themselves in the outcome, they would see themselves in whatever dissenters were among the jury. While the outcomes are relevant to what laws we have to obey, the reasons that are argued are the reasons we should accept it. A jury that provides it's outcome through a silent vote, who don't need to engage with dissenters amongst them, do not provide any reasons for anyone to accept their judgements. A jury that provides it's outcome through a silent vote has not reason to even consider their reasons for their vote. I would say it would be hard for a general public to accept decisions they disagree with. However, at the very least, I would argue the general public deserves the reasons for the decision to be provided, at minimum.

> I agree entirely with Lafont that even if we could determine the sortition body to be fully representative of the larger population, that this would do little to add legitimacy to the process. Citizens that object to the ruling wouldn't see themselves in the outcome, they would see themselves in whatever dissenters were among the jury. The losers in an election or referendum accept the outcome iff they accept the legitimacy of the procedure ("free and fair" etc). If if could be demonstrated that the "verdict" of a nomothetic jury would be the same, irrespective of which empirical individuals were included in the sample, that's about as good as it gets. No reasons are required in either example, just as nobody knows why trial jurors decided in the way that they did.

I think the paper is compelling overall, and I am sympathetic to the division of labor between legislators and juries. However, I don't necessarily agree with the focus on epistemic considerations and interpersonal deliberation. > I think the task of the legal jury and that of a legislative jury, in theory, are highly similar. We can look at both facts and values as two elements of the deliberative process, both of which are present in both legal and legislative deliberations. While many tend to emphasize the fact-finding nature of legal juries, there are many instances in which the facts may not present a singular obvious outcome, and as such judgement needs to be applied. The application of this judgement, inevitably, carries with it the values of the participants. Likewise, a legislative jury, in which we examine the values of the participants, will inevitably examine a great number of facts. It is true that legal juries involve some consideration of values and legislative juries involve some consideration of facts. However, I would argue that legal juries prioritize epistemic considerations, and legislative juries emphasize value-laden considerations. The goal of a legal jury is fundamentally to determine the truth in a case, and values enter the equation when the truth is not entirely clear, as you argued. The aim of a legislative jury is primarily to represent the interests of the people, and facts allow people to more effectively achieve their interests. In my original paper linked above, I briefly discuss a tradeoff between expertise and responsiveness. In a revised version of this paper that I'm working on, I emphasize this more, framed as a tradeoff between competence (the ability to effectively create and understand policy) and congruence (alignment with the raw will of the people). My justification for the division of labor between elected legislators and randomly selected juries is that legislators have higher competence which is necessary for writing bills, and juries have higher congruence which is important for making the final choice between competing options. Your more epistemic justification for the legitimacy of juries prioritizes competence at the cost of congruence, but because final choices on legislation are highly value-laden, it is unclear whether it is worth sacrificing or de-emphasizing the congruence that sortition provides. Preserving the statistical legitimacy of a jury through a large sample size and confidential voting is arguably critical. > However, I agree entirely with Lafont that even if we could determine the sortition body to be fully representative of the larger population, that this would do little to add legitimacy to the process. Citizens that object to the ruling wouldn't see themselves in the outcome, they would see themselves in whatever dissenters were among the jury. Congruence between the sortition body and the general population may not be sufficient for legitimacy, but I would argue that it is certainly valuable (and necessary to some extent). In my revised paper, I am planning to argue that electoral accountability primarily serves two functions: pushing representatives to be congruent to the public, and incentivizing public justification and debate. My argument is that a hybrid model of legislators and juries can achieve both functions more effectively through the inherent congruence of statistical representation, and the public debate created by having legislators publicly advocate for competing bills in front of juries. This public debate may be more important than seeing the reasoning of the jurors (after all, trial juries don't publicize jury deliberations, but the debate between competing advocates in the trial is key), although if legislative juries were to deliberate, I would certainly support publication of anonymous transcripts. I intend for my proposal of reflective democracy to be neutral on whether deliberation is interpersonal or just internal by weighing reasons after listening to competing arguments, hence the name "reflective" rather than "deliberative" democracy (I make this clearer in the revised version of my paper). I'm working on another paper on the legitimacy of interpersonal deliberation as Keith mentioned, using group-level data from deliberative polls. I can't share results yet, but my initial findings and reasoning cast some doubt on assertions you make in your paper on the effects of deliberation. In particular, you frame deliberation as causing a convergence of opinions. However, deliberation only causes mild homogenization within groups (Luskin et al. 2022, https://doi.org/10.1017/S0007123421000168), does not necessarily cause convergence across groups (the focus of my current analysis), and convergence of opinions is not even necessarily a normatively desirable goal to begin with. My justification for reflective democracy is much less grounded in the epistemic specifics of interpersonal deliberation, and rather in the statistical representation and informed consideration of competing arguments that can be achieved by a large jury, independent of how jurors become more informed, whether by interpersonal deliberation or internal reflection. My paper on the legitimacy of deliberation will thus influence whether there should be interpersonal deliberation in my system, but does not change the overall design and justification.

Kai:> my initial findings and reasoning cast some doubt on assertions you make in your paper on the effects of deliberation. Jim Fishkin would claim that his DP research programme controls for the added benefit of interpersonal deliberation over the provision of balanced information, but I don't believe our diarchic models (advocates debate, juries decide) have ever been put to the test (at least not since the fourth century BC). To my mind the crucial factor is the congruence between mini- and maxi-publics, and this is testable (although it would be a very expensive experiment).

> The losers in an election or referendum accept the outcome iff they accept the legitimacy of the procedure ("free and fair" etc). I argue that legitimacy is not a simple yes/no question, but a spectrum. The "free and fair" certainly can make some decisions more legitimate, but no one is going to accept the legitimacy of a decision in which they feel as though their rights are being dismissed. As soon as a decision of that nature is made law, the decision, in their eyes, is inherently unfair. A decision that affects their livlihood, their self expression, or any number of other issues, will always cause tension with legitimacy. The fact that there exists a majority who believes X can induce a sense of fairness and legitmacy to the minority who believes that X harms them. Especially when the only reason X is law is because a greater number of people believe it. I think we would agree there are a great deal of historical injustices in which the decisions of the majority have produced decisions we believe to be democratically illigitimate and a violation of the rights of the population. While there is little that be done to truly provide a sense of legitimacy in these instances, there are certainly things that can improve things. Free and fair is certainly one of them, yes. However, a population that feels as though they never even had the opportunity to defend their case are going to receive little comfort from the procedure. The three of us agree on this last point, as we share the common goal of including more voices in the decision proccess. Whether it be through superminority, top-N, or the free submission of proposals, we all wish to see more voices included. The adversarial nature of the trial ensures that reasons are provided. The jury, separate from the legislators and without external motivations, provide the population that their views are being properly considered. So we make progress in terms of legitimacy. A population who feels they were meaningfully included in the decision making process will feel the outcome is more legitimate. However, maximum legitimacy won't (and cannot) truly be reached. We can still strive for more. Deliberation ensures that the population feels as though the deciders truly considered their case, and didn't just disregard it. The providing of reasons ensures that the arguments they presented were acknowledged. For a population that loses, this does a lot to provide legitimacy to the process. At the end of the day, true legitimacy can only be achieved through the changing of the publics minds to achieve public consensus. However, the process of getting to that point needs to provide the loses with the most ensurances that their inclusion is not simply for show.

> In particular, you frame deliberation as causing a convergence of opinions. However, deliberation only causes mild homogenization within groups (Luskin et al. 2022, https://doi.org/10.1017/S0007123421000168), does not necessarily cause convergence across groups (the focus of my current analysis), and convergence of opinions is not even necessarily a normatively desirable goal to begin with. I will note that this paper points out exceptions to their results. I have not provided any explanation in this paper as to why convergence of opinions is desierable, though I will be doing so in my next paper. I am aware of the concerns of social pressure, but I do believe there exist democratically legitimate contexts in which social pressure is not only valid, but desirable, but only at the right dosage. Imagine this: A group of people deliberate on an issue X. There are people who are for it and people who are against it. Those who are for it advocate very well, with extremely well reasoned arguments. Those who are against X agree that the arguments provided by the other side are really well constructed defended. They are struggling to argue their side. However, they are not yet, *quite* convinced. I argue this is a pretty common scenario. People recognize the strength of the other side, but are not ready, for various reasons, to give up on their own side. I believe, when it comes to decision making, when an outcome needs to be decided, there is value in those who are in this state of "unconvinced, but reasonably convinced of the merits of the other side" to make the active choice of "deferring" to that other side. I think this is a pretty common outcome in deliberative settings. The goal is not to convince the other side to change their mind, but to convince the other side that it is reasonably safe to defer in this context. In a pluralistic democracy, where there are many competing needs and groups, I believe it is sometimes necessary for groups to defer in this manner. The trick, though, is differentiating between true deference in which one group recognizes the merits of the other side, even if they don't agree, and overwhelming social pressure, which is obvoiusly to be avoided. I think that a decision that is made in which there exists a consensus that everyone can "live with" the decision is far more legitimate than a decision that a number of participants cannot see themselves living with at all, but being overridden through majority rule. Such an outcome can only be reached through deliberation.

Apologies, a slight correction needs to be made: I said > The fact that there exists a majority who believes X can induce a sense of fairness and legitmacy to the minority who believes that X harms them. This should have said "The fact that there exists a majority who believes X *cannot* induce a sense of fairness and legitimacy in the minority that that X will do them harm.

If we have to choose between normative projects and workable systems of governance, I would opt for the latter. Athenian democracy was majoritarian and involved procedures to ensure the agenda was not dominated by partisan elites, but frequently violated the rights of minorities. The latter abuse is best addressed via constitutional safeguards (rather than the exchange of reasons). >At the end of the day, true legitimacy can only be achieved through the changing of the publics minds to achieve public consensus. This utopian (or dystopian) goal is as quixotic as Fishkin's project to transform the full citizen body by the widespread adoption of deliberation. None of this has anything to do with democracy as it's generally conceived.

> Jim Fishkin would claim that his DP research programme controls for the added benefit of interpersonal deliberation over the provision of balanced information, but I don't believe our diarchic models (advocates debate, juries decide) have ever been put to the test (at least not since the fourth century BC). Keith, this is a good point. The information-only treatment in DPs is quite minimal and does not involve debating advocates (although the DP plenary session could be viewed as analogous to this, the two main disaggregation experiments omit this asking of questions to expert panels in the information-only treatment). Goodin and Niemeyer's (2003) study on an Australian citizens' jury with a much more extensive information section found it had a larger impact than deliberation, although deliberation still had an impact and the study is limited by n = 12. Regardless, this study still had more of an emphasis on information than competing advocates. It would theoretically be plausible to concurrently run three treatments in a deliberative poll experiment: one with traditional information and deliberation, one with competing advocates and deliberation, and one with only competing advocates, the results of which would be quite informative. > To my mind the crucial factor is the congruence between mini- and maxi-publics, and this is testable (although it would be a very expensive experiment). I guess I'm not entirely sure what you mean by this. Are you saying that the question is whether the maxi-public would reach the same conclusion as the mini-public after deliberation / listening to competing advocates? The statistical reliability of mandatory non-deliberative minipublics is dictated simply by sampling variance, and I am working on quantifying statistical reliability for deliberative minipublics.

> I will note that this paper points out exceptions to their results. Are you referring to America in One Room? Although that is a compelling example, it is just one example, and my initial findings suggest that true convergence only happens around a quarter of the time. Even when there is convergence, it doesn't imply consensus; opinions still remain quite diverse. > I think that a decision that is made in which there exists a consensus that everyone can "live with" the decision is far more legitimate than a decision that a number of participants cannot see themselves living with at all, but being overridden through majority rule. Such an outcome can only be reached through deliberation. I agree that a decision may be more legitimate if there is a true consensus on it, but such a scenario seems unlikely to arise in general. A political system that requires consensus for legislative decisions is unlikely to be very productive. The only context in which true consensus is plausible is in a small jury, but such a body lacks statistical legitimacy. You suggest parallel juries in your paper, but on charged political issues, I'm not convinced that each jury can come to a consensus, and I think it's unlikely that the different juries would all agree with each other (although this is an empirical question). If the decision is made based on a majority of jury verdicts, this seems to invite the same legitimacy questions associated with majority rule. Can you clarify how the lottocratic body in your system reaches a decision, and whether you think a legitimate decision requires consensus (or just some deliberative version of majority rule)? I agree that legitimacy is a spectrum, but it seems unlikely that we will reach the consensus end of the spectrum in practice. > The fact that there exists a majority who believes X cannot induce a sense of fairness and legitimacy in the minority that that X will do them harm. By this reasoning, a decision is only legitimate if no one believes that it will do them harm. It seems nearly impossible to meet this standard in a large society. I agree that tyranny of the majority is a big problem, but a truly consensus-driven system seems likely to create tyranny of the minority. At some point a decision has to be made, and some people are likely to be dissatisfied with the decision, but that does not necessarily undermine the legitimacy of the system.

Kai:> Are you saying that the question is whether the maxi-public would reach the same conclusion as the mini-public after deliberation / listening to competing advocates? Yes. My proposal for a sortitional alternative to the Brexit referendum presupposed a 6,000+ jury. It this was split into (say) three juries, who all listened to the same debate and the voting outcome was congruent (within an agreed margin of error) then the decision could be deemed to reflect the informed judgment of the maxi-public. Whereas Fishkin's principal interest is opinion change, perceived democratic legitimacy (imo) requires the accurate reflection of the settled will of the citizen body. The outcome could even be 52/48 (as in the Brexit referendum), but it would be an informed decision. We agree that the advocacy element requires election -- under a Superminority threshold of 26% there would likely have been three alternatives: Leave, Remain and Renegotiate. If the latter option prevailed, PM Cameron would have had a very strong mandate to require changes from Brussels in order for the UK to remain part of the EU.

Kai:> It would theoretically be plausible to concurrently run three treatments in a deliberative poll experiment: one with traditional information and deliberation, one with competing advocates and deliberation, and one with only competing advocates, the results of which would be quite informative. Yes indeed. I'm not opposed in principle to intra-group deliberation, I'm just concerned that information cascades and other persuasive imbalances will lead to variance between groups. Citizens assemblies presuppose "impartial" moderators but this would not be possible if the decision outcome were to be anything other than advisory. >The information-only treatment in DPs is quite minimal and does not involve debating advocates. This would be analogous to a courtroom trial in which jurors were just told to read the advocates' briefing notes before deciding the outcome.

>This utopian (or dystopian) goal is as quixotic as Fishkin's project to transform the full citizen body by the widespread adoption of deliberation. None of this has anything to do with democracy as it's generally conceived. The changing of minds of the greater population isn't particularly utopian, it's a matter of reality. The question isn't whether it can be done, it's been done a great number of times. The questions, ultimately, are: 1) are there ways where it can be structurally encouraged (or discouraged) and 2) what is the best way to make the losers of the day feel heard and included, such that they feel they might win tomorrow, while we wait for the issue to be settled in the wider population Neither of these questions are, in my opinion, utopian. >Are you referring to America in One Room? Although that is a compelling example, it is just one example, and my initial findings suggest that true convergence only happens around a quarter of the time. Even when there is convergence, it doesn't imply consensus; opinions still remain quite diverse. I was actually referring to this section: > The task being set. Are the participants asked to reach a conscious, collective decision? To reach a consensus? Or simply to talk, listen, learn, and think about the issues? When the goal is consensus, homogenization is a demand characteristic. It is hardly surprising or informative when a design seeking consensus approaches it (consistent with research on compliance and conformity, as in, for example, Cialdini and Goldstein Reference Cialdini and Goldstein2004; Carlson and Settle Reference Carlson and Settle2016). Striving to reach a conscious, collective decision, too, may create incentives to indulge emerging pluralities. Voilà, homogenization. More subtly, the pressure to agree may also hinder WM and allow SD freer rein, thus facilitating polarization and domination as well. Designs asking the participants only to decide what they individually think entail no such task-based impetus toward homogenization, polarization, or domination. > You suggest parallel juries in your paper, but on charged political issues, I'm not convinced that each jury can come to a consensus, and I think it's unlikely that the different juries would all agree with each other (although this is an empirical question) Certainly. The goal of my next few papers will be to argue the legitimacy, as well as the plausibility, of a consensus-based system. >By this reasoning, a decision is only legitimate if no one believes that it will do them harm. It seems nearly impossible to meet this standard in a large society. Unanamious consent is certainly an impossible achievement, and so that isn't what I call for. I share the concern of tyranny of the minority, and aim to avoid that as well. My hope with my next paper is to reasonably defend both the legitimacy and the plausibility of my model, and, if successful, encourage empirical testing of my theories.