I recently finished a draft paper on sortition, arguing for three hybrid systems that use large, randomly selected juries to choose from among bills and appointees proposed by elected legislators.
The discussions on this forum have helped shape my ideas, and I would greatly appreciate any feedback.
Abstract: Randomly selected representative bodies have the potential to address seemingly intractable problems with electoral systems. However, existing designs are at risk of becoming polarized, insular, and corrupt like elected legislatures. I propose three systems based on short-term, evaluative, conclusive, and multi-choice juries, which are supported by contemporary and historical precedent. These large random juries choose between options proposed by elected legislators, reducing gridlock and polarization. This hybrid approach leverages the differing strengths of politicians and jurors, separating partisan proposers and dispassionate deciders to make both more effective. The three proposed designs address the limitations of both pure sortition and election, achieving the responsiveness and equality of sortition, while retaining the expertise and participation of election. First, deliberative law uses juries to choose agenda items and bills from among those proposed by legislators. Second, deliberative appointment uses juries to choose from among candidates nominated by legislators for judicial and independent executive positions. Finally, a deliberative senate is selected through deliberative appointment in each region. These designs of deliberative government provide a pragmatic pathway for testing and adoption by retaining existing systems while addressing their flaws.
Adam Cronkright
Ali
Ahmed R. Teleb
avanderven
avergne
Alex Kovner
azakaras
Brett Hennig
campbellwallace
Simon Threlkeld
Nick Coccoma
codyhipskind
Common Lot Sortitionist
Conall Boyle
constitutionalism
davidschecter
graytmatter
gtridimas
H.J. Hofkirchner
John Burnheim
jgarry20151
Jochen Krattenmacher
johnedwardtaylor
johnhuang865
Jonathan Crock
keithsutherland
Jacob Richter
kkoreilly
lukewhito
SimonPek
Nathan El
Nicholas Gruen
Owen Shaffer
Patrick Chalmers
Paul Gölz
peterstone
Pierre-Etienne Vandamme
pronfeld
Emil
rcaze
richwbrown
Roger Knights
roslynfuller
Pierre-Etienne Vandamme
shaharavin
tbouricius
Ted Wachtel
Joshua Laferriere
theodorestathis
towardsautonomyblog
Wayne Liebman
Yoram Gat
Oliver Milne
Equality by lot 
How is not making the legislators a choke point, to prevent things the people want done from even reaching the jury?
LikeLike
There is some risk of legislators being a choke point, but it is minimized by the fact that any individual legislator can propose a bill or agenda item. A popular agenda item or policy idea will likely have support from at least one legislator, which is all it takes for it to go to a jury. As discussed in the “Comparing Deliberative Law and the Kovner-Sutherland Legislature” section of the paper, single-legislator proposals give more voice to the nonpartisan will of the people than proposal thresholds. With proposal thresholds, partisan influence on voting is increased, and legislators become a bigger choke point (because a whole party needs to support an idea for it to reach a jury). My proposal gives juries more options and thus more control.
LikeLike
Great paper Kai (but then I would say that, wouldn’t I . . .). I’ll leave Alex Kovner to respond on the details, but just a few quick comments:
When discussing Hansen’s perspective on the nomothetai, I think you need to at least acknowledge that this is a controversial interpretation — Mirko Canevaro claiming that it is based on a mistranslation. And it involved a silent jury without small-group deliberation. I can’t think of a single ancient precedent for this, and am puzzled when deliberative democrats refer to Athens and/or Rosseuau.
I think the role of “neutral” moderators would be contestable on the basis of the Quis Custodet? principle if the jury were to have final lawmaking powers. If (as Jim Fishkin claims) small groups are a vital element in preference change (or moderation), then the overall jury size would need to be closer to Dahl’s original proposal than the wee pretendy parliaments that constitute citizens’ assemblies.
I agree regarding the need for the executive to have emergency lawmaking powers, but there is no reason why that still can’t go through a jury (our proposal separates the (elected/allotted) legislature from the delegated government along Rousseauian lines). I think there is a bigger problem as to how to ensure government stability with rule by ad-hoc juries which Alex and I will be addressing in our book if we ever get round to finishing it. My view is that there will be a need for constitutional safeguards, both for minority rights and fiscal prudence.
When your paper is finished would you like to submit it to the Journal of Sortition?
LikeLike
>With proposal thresholds, partisan influence on voting is increased, and legislators become a bigger choke point (because a whole party needs to support an idea for it to reach a jury)
I guess it depends what you mean by “partisan” — Alex and I use it in the non-pejorative sense of “partial”, as it’s inevitable that large, modern multicultural states will generate a range of different discourses. Our proposal suggests that polical parties under Superminority will be fluid entities and individual legislators will group together on an ad hoc basis. We respect Dahl’s principle that the demos (as a collective noun) must have control of the agenda and don’t see how this is possible without proposal thresholds. It doesn’t require a whole party to support anything — if you want a wider range of proposals then just reduce the Superminority threshold. See: https://www.academia.edu/44790587/Some_Problems_of_Citizens_Assemblies
LikeLike
Hi Kai,
Thank you for contributing to Equality by Lot and for presenting a draft of your paper.
I see that you do not refer to Alex Guerrero’s recent book, Lottocracy, but I think that your general thinking and argumentation is very similar to his (even if the details of your proposed designs are somewhat different). If you are unaware of this book, I think it may be worth your time to read it and refer to it. The book covers the sortition literature quite extensively and, like you, accepts the general terms and frames of thinking that dominate this literature.
My view is that accepting this framework is a fundamental mistake. Please see my review of the book (as published by Sutherland in the first issue of the Journal of Sortition). I think the points I made there largely apply to your work. There is much common ground in your premises and as a result, both of you exert yourselves to design systems in which the allotted are prevented from free interaction and self-management because you claim the allotted cannot be trusted to do so. Another point you share, by the way, is that like Guerrero’s book, your list of references misses the most fundamental work in the sortition literature – Manin’s Principles of representative government.
A couple of key points in my review above are:
LikeLike
Keith, thanks for the comments.
I will look more into Mirko Canevaro’s arguments and add a note about the disagreement on whether the nomothetai were randomly selected. I’m aware that the nomothetai were not deliberative in the sense of small-group interpersonal discussion. When I said that they “would deliberate for a day” in the paper, I was referring to them hearing the competing arguments on the law, but I should certainly word that more clearly. Thanks for pointing that out.
I concede in the paper that “interpersonal deliberation is not normatively necessary,” and I agree that it does not appear to be supported by classical precedent (although it is supported by contemporary precedent in common law trial juries). The reason why I still advocate for interpersonal deliberation is its apparent instrumental usefulness in reducing polarization and increasing knowledge, as demonstrated by deliberative polling. Goodin, who supports deliberation within, still thinks that interpersonal deliberation has a role to play. In his Australia study (Goodin and Niemeyer 2003), the information phase has a larger impact, but the discussion phase still has a significant impact. When I refer to “deliberative democracy,” I am talking generally about deliberation within and/or interpersonal deliberation. In my view, the current evidence seems to support having both, but if new evidence was contrary to that, or if the burden of facilitating interpersonal deliberation empirically exceeded its benefits, I would certainly be fine with focusing on deliberation within. The specific details of how jurors receive information and/or discuss are modifiable.
There is certainly danger of moderator bias, but as I discuss in the section on the Office of Deliberative Democracy and Sortition, I think the risk can be lowered substantially with deliberative appointment of moderators and supervision by the deliberative senate. Juries are ultimately in charge of appointing moderators through deliberative appointment, and of holding them accountable, both indirectly through the appointment of deliberative senators, and also potentially directly by having juries vote on impeachment and/or prosecution attempts. This system of jury-based checks and balances would likely keep moderator bias in check. If it empirically failed to do so, accountability measures could be strengthened, or interpersonal deliberation could be limited or removed.
I propose a jury size of around 500 to 1000 jurors, which should be enough to significantly reduce random variation. Even with variation, if deliberation reliably moves participants in the same direction, would it not be preferable to move a varying amount in the right direction than to not move at all?
To clarify on emergency powers: my argument is that the (elected) legislature should have sufficient emergency powers to, for example, quickly declare war if necessary. If the legislature cannot do that without going through a slow jury process, then more powers will fall on the executive, which is dangerous. Requiring jury input on certain executive actions (which I support, perhaps on decisions where an executive body cannot reach supermajority agreement) helps address executive aggrandizement, but it does not address the power vacuum and lack of stability. I don’t think that the jury process of random selection and invitation, flying people out, giving them information, having them deliberate, etc. can necessarily be effectively executed in a matter of days, especially in an emergency, so most emergency powers cannot require jury approval. If the legislature cannot directly do anything, then the executive and the judiciary are the only standing bodies with direct power, which seems dangerous from a separation of powers standpoint.
For example, if an authoritarian executive prevented any juries from being selected, and the legislature had no direct lawmaking powers, they would be powerless to stop the executive. I think some amount of standing legislative power is important for stability. Can you share more details on how you would address this stability problem while requiring everything to go through juries? Constitutional safeguards are only as robust as the judiciary that enforces them. It is also worth noting that retaining the lawmaking powers of the legislature allows it to handle logistics and less focal topics, while also making the adoption of the system more feasible from a path dependency standpoint.
Once I finish editing the paper, I will consider publishing options. Thanks for the suggestion.
LikeLike
I certainly agree with your overall point, and I make the same argument: there should be a separation between partisan proposers and dispassionate deciders, and legislators will inevitably have some element of partisanship. The job of the legislature is to generate a range of competing (partial) proposals that a jury chooses between in a more dispassionate fashion.
However, in my view, a range of competing proposals should not be generated in a rough correspondence to political parties. I think that political parties are an effective mechanism for electoral organization of a semi-coherent ideological group, but that they are less appropriate for issue-specific policy proposals.
Although you argue that “parties under Superminority will be fluid entities and individual legislators will group together on an ad hoc basis,” in the paper you linked, you state that you “anticipate that the number of political parties will closely approximate the output rule.” In “Isegoria and Isonomia: Election by Lot and the Democratic Diarchy,” you also argue that each party will believe “it can win at least one proposing unit,” and that “parties [will] fight for proposing rights.” Please correct me if I am misunderstanding, but that seems contrary to your argument that proposing rights will be granted to ad hoc groups of legislators.
If parties are in rough correspondence to proposing rights, then proposals will be partisan in the problematic sense of factions. For any given issue, it is unlikely that every legislator of a party will be equally interested in the subject matter. Instead, whoever in the party is focused on the topic (e.g., agriculture) will generate an idea, and the rest of the party will blindly defer to it. If a proposal is backed by all legislators in Party A, and the other proposals are backed by parties B, C, D, etc., then jurors who vote for Party A in elections may just vote for the proposal of Party A, and are likely to vote against the proposals from competing parties, against which they will likely have some partisan animosity (we cannot verify whether this is true without experimentation, but it seems like the risk is greater in this system).
Under my proposal, any given legislator will only propose policy for a limited number of issues that they focus on. Someone on the agriculture and defense committees will generate informed ideas for agriculture and defense without needing to take a stance on other issues. This separation removes blind deference to party positions, increasing the agency of individual legislators and weakening the influence of factions and parties. There will likely be multiple proposals from different legislators within each big party, and there may be compromise proposals from legislators across parties, so there will not just be one option per party. With a deliberative senate, there will be nonpartisan proposals from moderate legislators with expertise, further reducing the influence of partisan heuristics.
I agree that the demos should have control of the agenda, but I disagree that high proposal thresholds are the way to accomplish it. Elections are an imperfect tool, and if the proposal threshold is high, it may be the case that the handful of proposals generated by parties do not include the one that the people when informed would actually want (as in the choke point concern discussed earlier). By reducing the proposal threshold to one legislator (or two) in my design, there is less filtering by the elected legislature and more filtering by the jury, which is a more accurate representation of what the demos wants. There is certainly a tradeoff, but as I argue in my paper, it seems like the risk of too many proposals is less than the risk of too few, although experimentation would be informative. Certainly there could be a compromise supermajority threshold of around three or four legislators, which I would be fine with. What do you think about a low threshold like that?
LikeLike
Yoram, thanks for the feedback and the references. I was aware of Guerrero’s new book and Manin’s work, but I haven’t had the time yet to read them; I’ll take a look.
I agree that electoral systems are flawed at least in part because those in power have different interests than those of the people, and thus will take actions inconsistent with what the people would do in a system of direct or sortition-based democracy. However, I would note that in an (impossible) world in which electoral accountability is perfect, this problem would be much less significant. If the people have perfect knowledge of what their long-term interests are, how to best achieve those interests, and what their representatives are doing, then they can run for office and vote in a way that holds representatives at least roughly accountable to the will of the people. Rational ignorance, echo chambers, and partisanship are problematic because they weaken electoral accountability and thus worsen the gap in interests between the people and the representatives. In some sense these issues are ancillary, but they are still significant parts of the argument. Regardless, I think we generally agree about the flaws of electoral systems, and my paper is not necessarily focused on elaborating that issue.
I agree that the people themselves, represented by an allotted jury, should ideally make the decision on what system of democracy to use. However, for the same reason that I argue for a division between proposing politicians and judging juries in my design, I think that an allotted body need not generate the ideas for what system of democracy to use. As per Dahl (1989), I agree that each person is generally the best judge of their own preferences, wants, and needs, and thus voting equality should be present at the final stage. However, although responsiveness (to the will of the people) is the fundamental ideal of democratic decision-making, expertise takes a greater role for generating coherent competing options. For example, as I argue in my paper, ordinary citizens are perfectly qualified to critique a movie: in fact, what determines whether a movie, restaurant, or product is good is precisely whether the people like it. However, just because ordinary people can decide whether a movie is entertaining or a dish is tasty does not mean that they themselves can make that movie or cook that dish. Sure, almost everyone could learn how to make a movie if they dedicated their lives to it, but then they would no longer be representative of the interests of the general public. There are unavoidable epistemic gaps in generative expertise, and there is a fundamental tradeoff between responsiveness and expertise. The ideal structure of democracy is one that prioritizes expertise when generating options, and responsiveness when choosing between them. As I argue in my paper, this mirrors the economy: e.g., experts need to run the airlines, but the general public gets to choose between them. (Although electoral democracy allows the people to choose between elite-generated options, these options are vague and personality-based candidates and parties, not specific proposals, and the people do not make informed choices between the options due to rational ignorance).
Therefore, although I agree with you that the people (represented by a randomly selected body) should represent their own interests and should make the decision between potential systems of government, I do not agree that they need to come up with those systems of government. Experts still have a role in generating competing options and arguments for and against them. No political philosopher can definitively prove that one system is superior to all others, but they can propose a system and make arguments to convince the people that it is better than the alternatives.
To address your specific proposal, I would argue against a self-regulating plenary sortition body (which is what I presume you are proposing; please correct me if I am wrong) because most people generally do what is best for their own interests. In a short-term statistically representative body, this means their aggregated votes are in the best interests of the general public, since jurors cannot corruptly benefit themselves in a facilitated issue-specific vote between options. In a long-term body without guardrails, the same does not hold: members may interfere with the random selection or compensation processes to unduly benefit themselves, since they can set their own agenda, change the rules for their selection and compensation, and accept bribes. Just as an elected legislature without a constitution, judiciary, or other guardrails would be unwise, an allotted legislature should be limited in its powers, likely even more so than an elected legislature because it lacks accountability. As I argue in my paper, even with guardrails, random selection does not imply ongoing responsiveness to the will of the people in a standing body because members of such an assembly have different interests than the general public by virtue of their long-term position as legislator. A standing, generative sortition body of any design fails to achieve both expertise and responsiveness because it conflates both ideals in the same body, whereas my proposal achieves both by separating them across proposing politicians and judging juries.
LikeLike
Thanks Kai, this is a very fruitful exchange — looking forward to Alex’s comments, as he is the architect of the Superminority principle (my original thesis “Election by Lot and the Democratic Diarchy” was written before I was introduced to his work). From my side, I guess there are three points:
Theory: our work is conceptually grounded in the classical Greek notion that democratic equality requires isegoria (equal speech) and isonomia (equal political right). As 5th/4th century Athens was a direct democracy (all citizens could/should attend the assembly), anyone could speak (although most didn’t) and all votes carried an equal weight. Large modern states require representation, and sortition advocates claim that allotted juries can (under certain conditions) preserve isonomia. But isegoria also requires representation, hence our agreement with Urbinati, Rosenblum etc that political parties play an essential role in ensuring that the demos has kratos — if you want more proposals, then just reduce the threshold. I’ll email you our forthcoming paper that clarifies this point.
Praxis: I agree with Goodin and Niemeyer regarding the primary role of information. Jim Fishkin says his experiments show that discussion is also essential, but I believe his control group only received printed information. In our proposal the jury listens to the deliberative exchange between the advocates, before (silently) deliberating and this is a very close parallel to the nomothetai. I don’t believe preference changing under these conditions has been put to the test.
Perceived legitimacy: Given that we agree that the jury should have final decision power it is essential that the vast majority of citizens who are disenfranchised by the aleatory coup have confidence that it would make no difference whether or not they participated in person, as the outcome is determined by the law of large numbers. This means that Condorcet Theorem conditions (as outlined in Surowiecki’s book) apply and convergence has to be demonstrated by experiments with parallel juries on the same topic (the outcome should be the same, given an agreed margin of error).
>Even with variation, if deliberation reliably moves participants in the same direction, would it not be preferable to move a varying amount in the right direction than to not move at all?
That’s certainly Jim’s view, but I think any talk of “the right direction” will only exacerbate the populist backlash, especially as recent DDL experiments demonstrate preferences moving towards the liberal pole. It’s no coincidence that the biggest fans of citizens’ assemblies are environmental activists and the European Commission.
Yoram>: “both of you exert yourselves to design systems in which the allotted are prevented from free interaction and self-management because you claim the allotted cannot be trusted to do so.”
It’s nothing to do with trust, just the notion that “allotted representatives” is the rare case of a substantive that only exists in the plural case. Pitkin is adamant that all descriptive representatives can do is vote, as individual speech acts cannot be aggregated. Needless to say my political sociology owes more to Dahlian pluralism than Pareto/Mosca/Michels elite theory.
LikeLike
> If the people have perfect knowledge of what their long-term interests are, how to best achieve those interests, and what their representatives are doing, then they can run for office and vote in a way that holds representatives at least roughly accountable to the will of the people
That is indeed the fundamental mistake of the conventional analysis. (“Mistake”, implying good-willed innocence, is not the right term here, but it will do for our purposes.) As Manin clearly points out (and that is why his work is so important), elections put in power, uniformly and without exceptions, people of “distinction”. That is, people which are very different from the average person in important ways. Distinctive people can be expected to have distinctive ideas about what’s right and what’s good – that is, they have ideas that are different from those of the normal person. And since, as you write “people generally do what is best for their own interests”, the distinctive people use their elected power to promote what they believe is right and good – which tends to be different from what normal people think.
Thus, no matter what people know and who they vote for they can expect to get policy that their rulers find right and good and that they find wrong and bad.
LikeLike
> In a short-term statistically representative body, this means their aggregated votes are in the best interests of the general public […] In a long-term body without guardrails, the same does not hold
But if all bodies that are capable of generating proposals are corrupted by their long terms, then all proposals are corrupt. (What a radical claim!). So the source of the proposals is really not of much significance. If the way to launder corrupt proposals is by having a final up-or-down vote by a short term body, then why not apply this miracle device to proposals which originate from a long-term allotted body? Would that allotted body be any more corrupt than an elected body or any other long term body?
By the way, it is important to note that your claim about “long-term” serving bodies (bodies with, say, 4 year terms) being inevitably corrupt expresses a horribly grim view of humanity. According to this view people do not have strongly held values and world views. Rather, you assert that values and world views are no more than a thin veneer over narrow selfish interests, a veneer which is easily and quickly replaced as people’s material conditions change. Do you view yourself and your family and friends as being so horribly cynical? If so, why should anyone take your own proposals as anything other than attempts to further your own personal interests and the interests of your associates?
LikeLike
Keith, thanks for the helpful comments. I read your forthcoming paper and find it very compelling and thorough, and I agree with the general conclusions about separation of proposers and deciders, issue-specific majority rule, etc.
I think that you and Alex make an interesting theoretical argument that viewing a policy debate could be similarly (or more) effective than interpersonal deliberation. However, I think that this is fundamentally an empirical question, and should be resolved through experimentation. In Fishkin’s DPs, there are typically two informational elements beyond printed information: a briefing video, and a plenary session in which (competing) experts answer questions. Neither of these fully constitutes a debate between partisan advocates like a judicial trial as you advocate. The briefing video is typically a sober reading of bullet point pro and con arguments, and the plenary session is typically more nonconfrontational and focused on answering questions rather than making rhetorical arguments. One can argue that these more neutral information sources better achieve the ideals of deliberative democracy, but I can certainly see your argument that more partisan debate could have a greater impact, perhaps removing or reducing the need for interpersonal deliberation.
In the two existing DP disaggregation experiments I’m aware of (Farrar et al. 2010; Sandefur et al. 2022), the design did not clearly address these questions (the plenary session was counted as part of the deliberation, meaning that the information stage was only printed info in the 2010 study, and only a video in the 2022 study). I think a key research idea would be to run a more substantial split experiment, perhaps with four treatments assigned to randomly selected subgroups: (1) viewing partisan debate + engaging in interpersonal deliberation, (2) only viewing partisan debate, (3) reading printed info + engaging in interpersonal deliberation, and (4) only reading printed info. I presume that you would predict (1) and (2) would have the biggest effects, with a negligible difference between them, and then (3) would have a smaller effect, and (4) the smallest. If (1) and (3) had by far the biggest effects, then that would support Fishkin’s argument that deliberation is necessary. It could hypothetically be viable to conduct such an experiment (potentially through the DDL), especially if it was online to lower costs, although that may impact the results.
It is worth noting that in Goodin and Niemeyer (2003), with a comparatively interactive and extended information stage, the information stage had a greater impact, but interpersonal deliberation still had a significant impact. Even if partisan debate has a greater influence than printed info, it may be the case that some interpersonal deliberation is still useful. Regardless, I think that experimentation can provide a much clearer answer to this empirical question than theoretical arguments.
I agree that political parties are necessary in a jury-based system. The main problem with parties from a Madisonian perspective is their factional nature, but if the factions are restricted to submitting competing proposals and arguments, they cannot impose tyranny of the majority, thus removing the problem. However, I do think a potential issue remains if parties and proposals are in direct correspondence. I am still concerned about partisan influences on voting in the jury. If parties are presenting their proposals and making arguments for and against, then they can make rhetorical and partisan identity-based arguments directly to the jurors (e.g., “don’t support their proposal because their party is a bunch of elitists backed by wealthy donors who don’t care about the American people”). I understand that the proposal threshold can be reduced, but I want to clarify whether, in your conception of superminority, you envision there to be a (rough) correspondence between parties and proposals, and if so, whether you think that could cause problematic partisan influences on juror voting.
The issue is that with a low proposal threshold and a partisan debate format, there is potential for too many competing groups. The same problem does not necessarily apply to printed briefing materials and interpersonal deliberation, since there the number of proposals and arguments matters less, partisan balance in legislator speaking time is less relevant, and jury discussions matter more. With a partisan debate format without interpersonal deliberation, it is key that the speaking time for the different sides is balanced, and thus the proposal threshold needs to be higher. Therefore, two of the points of difference we have (proposal threshold and deliberation) are somewhat intertwined. I don’t actually have any personal bias toward either approach, so I think the empirical results on whether interpersonal deliberation is key (relative to viewing partisan debate) should dictate.
I agree that “the right direction” is flawed language, and I think that defining the ideal outcome of a jury is a fraught normative question. Some (like Lafont) would argue that the right outcome is one completely consistent with the beliefs of the general citizenry (without any opinion change based on information or deliberation, which would force “blind deference” by the public to the jury’s outcome). Fishkin argues that the result should be “what the people would think about the issue under good conditions for thinking about it” (Fishkin 2018), but the definition of good conditions is subject to debate. As I argue in my paper, there is a tradeoff between responsiveness and expertise, and the choice of who makes the decisions in any political system sacrifices some amount of responsiveness for expertise and/or vice versa. When I said “the right direction,” I was roughly referring to a position based more on expertise, although there is obviously a limit to that. I do think that in general a jury-based approach is much more responsive than an electoral system, and that even with information and/or deliberation, a properly designed and promoted jury-based system should not suffer from significant populist backlash. Citizens’ assemblies are a different matter due to their low (and biased) response rates, and their frequent lack of balanced competing arguments and anonymous voting. It is worth noting that in my personal experience, the biggest concern (non-expert) people have with my proposal is a lack of expertise/competence in jurors.
LikeLike
Yoram, I agree with your overall points, and I agree that pure election is a flawed method for making democratic decisions.
I agree that elections often result in policy that politicians favor and the general public may not necessarily support. However, I’m not completely sure if it’s empirically the case that ordinary people universally find government policy wrong and bad, since the government doesn’t enjoy a zero percent approval rating. Although electoral systems are flawed, I think they aren’t entirely useless. Regardless, we all agree that elections are flawed and sortition has the potential to improve democracy.
I think that all proposals from a long-term body benefit from review by a short-term body, regardless of how the long-term body is selected. This is not necessarily because the proposals are always corrupt, but because there is a risk of confirmation bias, groupthink, and unfair outside influence in a standing body, and that risk can be reduced by having a short-term jury review. The same principle also underlies checks and balances, peer review, code review, any kind of feedback or editing, etc. Humans are generally good, but they can make mistakes or poor decisions due to cognitive biases, and thus having a second pair of eyes is always useful. This is especially true for influential and controversial political decisions.
To clarify, my proposal is based on multi-choice (approval) voting between multiple (i.e., > 3) options, which prevents oppositional situations where vetos lead to gridlock, and reduces us versus them dynamics. I think that a short-term, evaluative, conclusive, multi-choice jury is a useful tool in many situations, and I would not necessarily be opposed to using it for proposals that originate from a longer-term allotted body. More specifically, for a given agenda item, one could assemble multiple smaller citizens’ assemblies, each of which would generate a proposal, and then a jury would choose between those proposals.
Although I think that could work, I do not necessarily think that it is a better idea than my deliberative law proposal, since the advantage of an elected body is it has more expertise (at least in terms of experience and connections), and can give the general public more opportunities for participation, while being more compatible with existing systems. However, for those who desire more sortition, I think a large random jury choosing between proposals from multiple competing citizens’ assemblies is a reasonable solution.
For example, I like Terry’s multi-body sortition proposal overall (Bouricius 2013). However, it involves up-or-down policy juries, which only have veto powers on legislation originating from a body at risk of confirmation bias and/or corruption. This approach suffers from the oppositional point of the triad of ineffective sortition, as described in my paper. I think a way to improve it would be to have multiple competing citizens’ assemblies (or review panels), with the policy jury choosing between their competing proposals. This would achieve the same benefits of sortition without suffering from oppositional issues.
I think that people are generally good people, I just think that there are risks that some people will exhibit bias and corruption, and it makes sense to take steps to minimize those risks. As Madison argued, “If men were angels, no government would be necessary.” Checks and balances are necessary in many areas of life, and creating a body with unchecked power carries more risks than designing a more balanced system.
I think one should treat my proposal as one competing option of many, just as a bill in deliberative government would be treated as one competing option. I do not think that anyone should blindly follow my proposal, but I think that they should consider it, like a review jury would consider a bill in deliberative government. I have my own experiences and biases that shape my proposal, just as legislators do, which is why the proposers should not be making the decisions, but rather the people (through a representative jury) should decide.
LikeLike
Kai,
> we all agree that elections are flawed and sortition has the potential to improve democracy
I find this conventional phrasing misleading. Elections are not flawed – they work exactly as intended. There is no democracy under and electoralist regime, since, as any casual reading of the writings of the founding fathers would show, elections were never meant to create a democracy. Therefore the accurate phrasing is “sortition has the potential to create democracy”.
> Humans […] can make mistakes or poor decisions
That is a truism. But your argument goes farther – you claim that we can expect short term bodies to make less mistakes than long term bodies. This is quite an unexpected position. It is generally accepted that the more people think about things the better their understanding and decisions become, not the other way around. Here we are suddenly asked to believe that hurried decisions with minimal discussion are better.
But be that as it may, the question of whether we want to have a short term allotted body with an up-or-down decision power is a good idea is a relatively minor one (although not one without importance). The question of where the proposals originate the crucial one.
When you argue that an elected body should be the originator of the proposals you write that the it has the advantage of having “more expertise (at least in terms of experience and connections)”. So now the lifelong political careers of the elected are an advantage, when for allotted citizens 4 year terms are a disqualifying consideration? (Of course, the real expertise is that of academics, professionals and bureaucrats, rather than of elected officials whose true expertise is in getting elected.)
> [Elected bodies have the advantage of] being more compatible with existing systems
That is certainly true. In general, the desire not to rattle the existing power structures, which are seen as imperfect but fundamentally right and good, is the main consideration behind much of the conventional thinking and rhetoric around sortition, including among those who think of themselves as its supporters.
> I think that people are generally good people, I just think that there are risks that some people will exhibit bias and corruption,
Again, this is a truism. But as long as these people are not a majority in the population then they will not be a majority in the allotted body. Why then not trust the allotted body, unless you believe that the majority of the population is prone to corruption?
> Checks and balances are necessary in many areas of life
For “checks and balances” and the Madisonian system I suggest you read Dahl’s Preface to Democratic Theory. https://equalitybylot.com/2014/02/10/early-advocate-of-sortition-in-government-robert-dahl-has-died/
LikeLike
Kai >: I am still concerned about partisan influences on voting in the jury.
Yes, in his paper “debate, not deliberation” Manin insisted that politicians could only participate if their rhetoric was decoupled from electoral considerations. Agree this is a tall order, that’s why Alex’s model is based on shifting alliances of elected legislators, as opposed to party tickets. We also suggest a Rousseauian separation of lawmaking and governance (see below), and this should lessen the red state –vs– blue state shenanigans that we have become accustomed to. The political party under Superminority will be very different from today, but our overall framework is still liberal democracy, rather than agreement (or, preferably, unanimity) as a normative goal.
>: I think the empirical results on whether interpersonal deliberation is key (relative to viewing partisan debate) should dictate.
I agree that the experiments need to be done, until then our perspective in no more than a hunch. The problem is deliberative democrats (including Fishkin) argue that “impartial” moderators and “balanced” briefings are required. If the jury is to have final legislative powers — ‘which would force “blind deference” by the public to the jury’s outcome’ (Lafont) — then quis custodet? is an unavoidable objection. Perhaps that’s why many sortition advocates make the Habermasian case that deliberation is for the public sphere and can only have an influence on (elected) legislators. But, as Bentham pointed out, influences can be sinister, and I’m worried that this will only increase the power of the chattering classes (and the resultant populist backlash). It’s a mistake to think that elites are limited to elected politicians and their financial backers.
I think Alex and my position is best summed up in my last message in the exchange with Jim Fishkin:
“I think the reason we sometimes talk past each other is that our approach to the distinction between the representation of discourses and representation of persons is the polar opposite of most deliberative democrats. We agree with you that sortition can play no role in the representation of discourses and leave this to elected politicians – Superminority just ensures a better (proportional) match to the ideological diversity of the electorate.
Regarding the representation of persons, we take Rousseau’s strictures on popular sovereignty literally and accept his argument that any delegation of political decision making alienates the freedom that is our natural right. My last paper argued that decision making by large randomly-selected juries would not (under certain exacting conditions) undermine popular sovereignty.
In our view “allotted representatives” is a rare case of a substantive that only exists in plural form. We agree with Pitkin’s argument that descriptive representation can only apply to aggregate functions (i.e. voting). If the face-to-face interaction of persons in small groups is essential to reduce affective polarization and directional motivated reasoning, then this will require a very large jury to ensure that Surowiecki’s objections are addressed. My proposal for an alternative to the Brexit referendum (where the decision outcome was 52/48) required a jury of 5,001 and this could be subdivided into (unmoderated) groups of 13 or so. Whether such an arrangement would be downscalable for less momentous decisions is debatable.
Rousseau was opposed to deliberation and sought to insulate the sovereign assembly from corruption by factional demagogues. This is no longer possible in the smart-phone age, hence the need to ensure that the demagoguery is (at minimum) well-balanced. As the final vote requires a simple majority, persuaders will be motivated to adopt a rhetoric that appeals across the broad political spectrum. If the Brexit Leave lobbyists chose to circle the wagons and preach to the converted they would lose the final vote, hence our attraction to Harrington’s analogy of the two girls sharing a cake.
This division of labour – between persuaders and deciders – maps precisely on to the argumentative theory of reasoning (hence the present paper) but puts us at odds with most deliberative democrats (we prefer to use the terminology of the [republican] mixed constitution). I’m puzzled when deliberative democrats claim provenance from Rousseau and/or Athenian democracy, as the only pre-modern example of small group deliberation I can think of is the eighteenth-century coffee house – hence Habermas’s view that the public sphere is the rightful locus for deliberation.”
LikeLike
Yoram,
This is a good point, and an important one to address. It is first worth emphasizing that there are two important related but separate design features underlying my proposal. The first is a separation between proposers and deciders, which I (like Kovner and Sutherland) argue can allow for competing, informed, partisan proposals and more dispassionate, representative, and moderate decisions between them. I think that the separation between proposers and deciders makes sense in an intuitive way, as demonstrated by the “I cut, you choose” analogy and the free market of policy ideas metaphor in my paper.
The second feature is a specific position on the continuum between responsiveness and expertise. As I have argued, there is an inevitable tradeoff between these two important ideals: as one gains more expertise on the policy question at hand, they become less representative of what the general public thinks. In my proposal, I argue for a focus on expertise in the proposing stage, since writing bills requires more expertise than evaluating them, and not all of the proposals have to be perfectly representative of the will of the people: all that matters is a sufficient range of competing options. Therefore, I advocate for election of proposers, as election is relatively weighted toward expertise rather than responsiveness. For the deciding stage, I advocate for a greater focus on responsiveness through a short-term jury, since evaluating bills is less difficult, and accurately representing the will of the people at the decisive stage is foundational to democracy (as Dahl argues). In other words, options can come from elites, but the people need to make the final choice. With multi-choice voting and a sufficient range of options (which should be afforded by legislator proposals with a low threshold), the people can exercise control at the decisive stage.
Thus, the reason why I argue for short-term bodies instead of long-term bodies is because they are more responsive to the will of the people, albeit at the cost of less expertise. I agree with you that people understand more if they think about decisions for longer, but they also become more of a separate “legislator class” that represents the will of the people less and has more vulnerabilities to factionalization and corruption.
Although I advocate for this specific set of proposers and deciders with a certain balance between responsiveness and expertise in my paper, any set of proposers and deciders could work. Proposals could come from elected legislators, citizens’ assemblies, bureaucratic agencies, expert panels, signature petitions, even AI. In fact, there could be multiple sources of proposals, which reduces the lock on power that any one group has. In my paper, I propose that elected legislators and deliberatively appointed senators both be able to propose options. One could also empower a citizens’ assembly and a relevant executive agency to make a proposal for each issue. As long as each source can only make a small number of proposals, this approach distributes proposing power without flooding people with too many options.
Deciders also don’t have to be a jury. A much simpler design more consistent with existing systems would be to have the second elected chamber of a legislator be the deciders. For example, in the US, members of the House could make proposals with some threshold (e.g., five legislators). Then, those members of the House would present their proposals to the Senate and engage in a competitive, lively, policy-focused debate with each other. After hearing the arguments and debate of the house members, the senators would vote on all the proposals with approval voting, with the top option (with at least majority approval) becoming law. This alternative separation of powers would reduce gridlock between the two chambers of congress, giving both a distinct and important role. This approach would emphasize expertise more, as senators have more expertise and less responsiveness.
Alternatively, the deciders could be the entire population, as in a referendum. This option maximizes responsiveness at the cost of expertise. The deciders could also be a standing (long-term) randomly selected body, which would give more expertise at the cost of responsiveness. Everything is a tradeoff between responsiveness and expertise, and it is hard to know what balance is the best without experimentation. Establishing a separation between proposers and deciders seems beneficial regardless.
LikeLike
I agree that this is true in a Bouricius-style up-or-down jury system. However, this does not necessarily apply to my system. To clarify, my proposal does not involve up-or-down juries. My review juries choose between a range of options (ideally around 5-6), which gives them real control over what policy is implemented. Instead of oppositional powers that put the jury in the backseat, the jury is making the main policy decisions through multi-choice approval voting. This means the proposals matter less and the jury decisions matter more. Since the proposals are focused on expertise and the decisions are focused on responsiveness in my design, and decisions have the most impact, the system is more likely to lead to responsive outcomes. One could reasonably object that there should be a slightly greater focus on expertise, in which case the system could be tweaked to a different position on the expertise-responsiveness continuum, but I think the overall design of separate proposers and deciders makes sense.
Everything is a tradeoff between expertise and responsiveness. Lifelong political careers are an advantage in expertise at the cost of responsiveness. Four year terms for allotted citizens are also an advantage in expertise at the cost of responsiveness. Neither is disqualifying for generating proposals, since expertise matters more there. I think both politicians and long-term randomly selected citizens could generate proposals, since expertise matters more than responsiveness for proposals (although I still think politicians have an advantage for this because they have more expertise and are more compatible with existing systems). However, neither politicians nor long-term allotted bodies should be deciding between options in an ideal design, since deciding between options should put more emphasis on responsiveness rather than expertise.
The issue with your design (and those from others such as Guerrero, Gastil and Wright, etc.) is that it confounds proposing and deciding in the same body. I agree with you that a long-term allotted body could potentially play a role in an effective system, but each body should only do proposing or deciding, not both.
I think that humans are social beings who often prioritize their in-group, and thus most people have a risk of being prone to factionalization and other problematic phenomena in contexts conducive to it (Sloman and Fernbach 2018; Kahan 2013; Sunstein 2018). My argument is simply that we should reduce the risk of such problems by separating proposers and deciders and using short-term bodies when possible.
LikeLike
Keith,
I agree that this is a significant problem. Do you think that there is a chance that this issue can be reduced to a manageable extent with a combination of competing proposals from legislators, deliberatively appointed moderators supervised by a deliberative senate, deliberation without moderators (as in trial juries), and/or deliberation with a simple automated moderator (not generative AI)? On a related note, what is your opinion on trial jury deliberations? Do you think that trial juries shouldn’t deliberate, or do you think that the current structure is fine? Regardless, I think that moderator bias is a significant potential issue, and it would certainly be nice if moderators weren’t required for a jury-based governance approach. If further empirical evidence suggested that deliberation is in fact necessary for significant opinion shifts and watching partisan debate has minimal impact, would you still be opposed to some form of deliberation in your design?
I think it’s worth noting that I’m also exploring possibilities for developing an online (social media esque) platform to make deliberation with (effectively random) other people on current news easy and convenient. I think that adoption is fundamentally the biggest obstacle faced by any sortition system. I’m curious what you think about how to increase the chance of adoption of a sortition-based system. Although I evidently think that some form of sortition should be adopted, I do think there is value in additionally pursuing extra-governmental (public sphere) approaches. Given that you think watching debate on a policy issue is more valuable than interpersonal deliberation, do you think that an app that shows users such debate would be more valuable than an app that puts people into small groups and facilitates text-based discussion of the news? Such an app could use A/B testing with different variables (e.g., whether users get to discuss or not, what kind of information they see) to provide more empirical evidence for these questions.
I agree, and I also think that employing approval voting (with a majority requirement) can further encourage moderation. I’m curious what you think about multi-choice voting systems in deciding juries (i.e., do you support approval voting?) I think that having multiple options is a key component of reducing partisan animosity (affective polarization) and allowing people to really think about the policy. If you have a debate between five competing groups, it becomes harder to mudsling at the other groups; instead, you have to make a positive case for your own idea. If you just have two groups, you can demean the opponent.
LikeLike
Kai:> accurately representing the will of the people at the decisive stage is foundational to democracy (as Dahl argues). In other words, options can come from elites, but the people need to make the final choice.
Dahl (1989, pp. 113, 109) insisted that the demos should have kratos at both the proposal and decision stage but suggested (wrongly, IMO) that this could be ensured by two separate minipopuli. Alex and I argue that election is the best way to implement agenda setting, and that Superminority will ensure a better fit to the discursive diversity of the citizen body. Lottocrats like Bouricius argue that if ho boulomenos was good enough for Athens then it should be good enough for us, but ignore the fact that Athens was a direct democracy, characterised by homonoia. Neither of these conditions apply to large modern nation states. We agree with Saffron and Urbinati (2015) that election is the only way to implement representative isegoria.
Trial juries are charged with reaching a consensus, hence the need for deliberation (my own experience showed me how easy it is for one eloquent persuader to swing the jury room debate), whereas with political decisions a simple majority is sufficient. I’ve no principled objection to small-group exchanges, but those of us who advocate legislative juries need to identity and remove any source of variance. Bob Goodin acknowledges the common direction of preference shift in the three Texas power utility DPs, but points out that there were serious differences in degree. If so, then which was the representative sample? Bear in mind the Brexit verdict was 52/48, so any variance is critical, and the impartiality of small-group moderators would (inevitably) be challenged by the losing party. The plenary exchanges could be broadcast live, but not the small groups. If this means face-to-face deliberation has to go, then so be it (as representativeness is the key consideration).
I confess I’m uneasy about using the language of preference change, as it sounds a bit Whiggish to me (take a look at the opinion shifts in Fishkin et al 2021), especially given the enthusiastic uptake of citizens’ assemblies by progressives. I’m more comfortable with the Rousseauian notion of the settled will of the sovereign people.
>: I’m curious what you think about how to increase the chance of adoption of a sortition-based system.
I think two things are necessary: 1) conceptual clarity as to what sortition can and cannot do and 2) experiments that demonstrate whether or not multiple concurrent minipopuli would come to the same decision (within an agreed margin of error).
>: I’m curious what you think about multi-choice voting systems in deciding juries (i.e., do you support approval voting?)
Um, I think so — this is Alex’s area of expertise, but he seems to have gone AWOL.
LikeLike
For an example of the enthusiasm of progressives for sortition as a prophylactic against populism, see George Monbiot in today’s Guardian:
“The shift could begin with a citizens’ constitutional convention, like Iceland’s perhaps, and participatory events all over the country.”
https://www.theguardian.com/commentisfree/2025/oct/23/written-constitution-reform-uk-autocrats
LikeLike
Kai,
> most people have a risk of being prone to factionalization and other problematic phenomena in contexts conducive to it
Unfortunately, this is the crux of it.
As I noted above, your arguments and writing fit well within the conventional academic discourse around sortition, and within a long tradition of elitist, “republican”, thought. The underlying motif is that normal people cannot be trust to control the process and the agenda and must be managed.
The rest is derived from that. People must only choose from a short menu of options created by their betters (what you call “people with expertise”, what Madison et al., a bit more bluntly, called the “wise and virtuous” or “the natural aristocracy”).
Since your arguments and conclusions are so similar to the Madisonian style and theory I suggest again that you have a look at Dahl’s critique of this theory in his Preface to democratic theory.
LikeLike
Keith,
I agree with your overall point. I think it’s important to be clear on terminology: we agree that election is the best way to implement proposing (at least from a path dependency standpoint). Although what you could maybe call “agenda setting” in the sense of generating proposals and arguments (and proposed agenda items) is done by elected legislators in this approach, “agenda setting” in terms of voting on which agenda items to consider should be done by a large random jury, at least in my opinion. In my interpretation, Dahl is arguing that one minipublic could set the agenda (“decide on the agenda of issues”, p. 340) and another could consider a specific issue. Dahl does not necessarily say that one minipublic should be in charge of the proposal stage; proposing and agenda setting are not the same thing. One could reasonably interpret Dahl’s argument as consistent with a separation between proposers and deciders, with legislators proposing agenda items, a minipublic choosing between them, legislators proposing bills for each agenda item, and a different minipublic choosing between them. These four steps are in fact exactly the steps of deliberative law, my proposed system. After all, Dahl says that minipublics would “not [be] a substitute for legislative bodies but [rather] a complement” (p. 340).
I guess my more specific question is whether you think consensus-based (deliberative) trial juries are a flawed design, given your opposition to interpersonal deliberation in legislative juries. Our focus is certainly on legislative (and appointment) juries, but given that trial juries are arguably the most influential modern form of sortition, and we borrow their name, I’m curious whether you think they would be better structured as larger, majority-based bodies without interpersonal deliberation. After all, you make frequent reference to trial juries as an analogy in support of your partisan debate format, even though trial jurors deliberate together in addition to viewing a competitive trial debate; e.g., “Dividing the two forms of deliberation between two institutions (as in the judicial separation of advocates and jury) protects against corruption, and it’s not clear why randomly-selected disposers need to participate in the speech acts that are the defining characteristic of the proposing function” (Kovner and Sutherland 2020, pp. 8-9). I agree that jurors should not generate the proposals and arguments, but if you think that interpersonal deliberation doesn’t make sense for deciders, do you think it is also something that should be changed for trial juries?
LikeLike
Assuming that most people consider trial juries to be (democratically) legitimate, what do you think that legitimacy comes from? I don’t think it comes from statistical representation (since twelve is too small of a sample size, especially with strikes from the prosecution and defense), and I don’t think it comes from repeatability or a lack of variance, since as you noted, one juror can swing the debate.
If I understand correctly, the three deliberative polls in question were conducted in different regions at different times with different samples of around 200, with different experts and information booklets, so a substantial amount of variation was inherent (Luskin, Fishkin, and Plane 1999). A larger jury of 500 to 1000 as in my proposal would be more robust to variation, although not entirely so. Some variation in any sortition system is unavoidable, and what is considered legitimate is somewhat of an empirical question. I would be curious to see if two 1000-person DPs at the same time on the same topic actually have variation substantially larger than two 1000-person control groups without deliberation and thus with just sampling margin of error.
I agree that these two things are useful, at least from an academic standpoint. I think that the bigger issue from a popular standpoint is successfully promoting the ideas with the broader public and people uninvolved in political science, which is difficult, at least in my experience. In my view, more work needs to go into producing compelling articles and videos that effectively promote a moderate form of sortition (like your or my proposal) in rhetorical terms consistent with public opinion. In other words, we should focus on shared discontent with existing systems, the potential for pragmatic reform with legislative juries (not revolutionary, “get rid of elections” rhetoric), the existing real-world applications of DPs, and the transformative potential benefits (putting the informed will of the people in charge, giving the people input on each issue separately, reducing polarization and blind partisan voting, removing the monopoly on policy held by the majority and creating real debate on each issue, even in safe states, reducing gridlock, etc). Once I make some more progress on the academic questions, I hope to spend more time on making things for the general public.
LikeLike
Hi Kai, this is a very fruitful exchange. You are right that I am conflating proposing and agenda setting. Alex is our details man, so I wish he would join in the conversation. I just make the principled distinction between individual speech acts and aggregate judgment and leave it to Alex to operationalise it. Yoram rightly points out that our proposal is republican, in that it reserves a key role for elite proposers. I don’t think there has ever been a “pure” democracy — Athens certainly wasn’t, as most proposals came from the rhetores kai strategoi and (as Moses Finley [1983] insists) ‘the demos never produced spokesmen in the Assembly from their own ranks’. Yoram relies on the Council of 500, but ignores the consensus (?) among classical historians that it was an administrative, not a policymaking body. It’s hard to imagine person-to-person deliberation in such a large group. But I’m happy to be called a republican rather than “deliberative democrat” (whatever that means).
My preferred model for juries is 4th century Athenian law-making. Unlike modern jury trials, the jurors (Cammack prefers to call them judges) just listened in silence (albeit punctuated by dikastic thorybos) to the exchange between advocates (equalized by the use of a water clock) before determining the outcome. Athenian trials blurred the distinction between prosecution (of persons) and nomoi, not so in modern courtroom practice. Consensus is the ideal for modern trials, but there is an old adage in parliaments that consensus leads to bad laws (hence the need for advocatus diabloli).
[The selection of modern trial jurors is another matter, but I think here sortition is primarily a prophylactic against corruption. Helene Landemore’s claim that sortition ensures epistemic diversity (and/or representativity) doesn’t make sense for 12 persons.
Again I’ve got no principled objection to interpersonal deliberation but insist that, for the sake of democratic legitimacy, all potential for variance should be eliminated, that’s why I’m uneasy with Jim Fishkin’s emphasis on preference change (with its Whiggish connotations).]
I haven’t given much though to what form of rhetoric should be employed to spread sortition beyond the dreaming spires and revolutionary vanguard. But I do think that, given the long struggle for universal suffrage, most citizens will want to ensure that they are not disenfranchised in the aleatory coup. And I think an essential prerequisite is for those who are working in this field to come to some sort of agreement as to what sortition can and cannot do (it’s still in the pre-paradigm state). That’s why we’ve launched the Journal of Sortition and would encourage everyone on this forum to subscribe: https://www.imprint.co.uk/product/jos/
LikeLike
P.S. I think the key test for the public acceptance of sortition is when it appeals equally to those of a conservative and progressive disposition. I would be keen to learn the views of Clifford Bates on this: https://uw.academia.edu/CBates
LikeLike
Hi all,
I loved the paper, I think the triad of ineffective sortition is an important rubric, nicely articulated.
I have some catching up to do on the current discussion, so I’ll try to hit the highlights. First, there’s the charge that Superminority retains too much partisanship. I think what’s being left out of the discussion is that the elected body would no longer need to maintain a durable majority, since the agenda is set according to the Superminority rule itself. It is the power associated with durable majorities that toxifies politics, as everyone is trying to climb the greased pole of politics to get to an apex where they can dictate policy unopposed. Without the need for a durable majority, legislators should be much more free to work with others on an issue-by-issue basis. It is durable majorities that give parties power to force their members to vote with leadership, as a rebellious back-bencher has no future.
Parties in the current system don’t actually compete much. They compete in elections, but not in lawmaking itself. I believe the elimination of the durable majority together with the need to compete with the jury will create an environment in which party discipline will give way to looser parties that mostly provide organizational support to like-minded but otherwise independent lawmakers. But I could be wrong about this, and I agree that the ultimate answer to this question is empirical.
Allowing every legislator to submit a proposal on every issue has problems. Many legislators will become conduits for the special interests that back them. Juries will find themselves evaluating 300 bills, all of which were written by 4 special interest groups with minor variations. Gumming up the works is the point here, as many interests just want to maintain the status quo. This is a real problem with, for example, the initiative process in California, where voters are regularly presented with multiple instances of very similar initiatives, where one or more of the proposals has been written by a special interest for no other purpose than to confuse voters. Juries have greater deliberative capacity, of course, but not unlimited capacity.
There are other aggregative solutions to this problem. One is to create a proposing economy by “tokenizing” proposals. Imagine this: every elected legislator gets 100 non-transferable tokens per legislative session. To make a proposal, any group of legislators would have to spend 50 tokens. That way, one person could sponsor two proposals by themselves, or they could team up with like-minded people to conserve their tokens.
One could also follow a “top-n” system. Say we want 5 proposals for a particular agenda item. Elected members could offer up their tokens for specific proposals, with like-minded members supporting a single bill. The five proposals with the most tokens would go to the jury. Those supporting proposals outside the top 5 would be given back their tokens, while those supporting the top 5 would lose theirs.
There is quite a range of potential aggregation schemes for this purpose. Superminority has the advantage of being very simple and easy to understand while also producing a fixed number of proposals, but there are many others. Since none of these have ever been tried, it would be foolish to believe that any one of them is clearly superior to the others.
LikeLike
On the small group deliberation front, I think that small groups are probably necessary for the jury to function. I think that “independence” of the jurors is more about whether or not they will be cajoling each other to vote in a certain way. In a criminal jury, for example, the system requires twelve people to reach a single verdict, which is bad. I think that falls under “oppositional” in the triad. I don’t mind people working together as long as they are free to vote their conscience. I also don’t want them meeting in large groups since this brings a host of problems. 500 people in 50 groups of 10 seems reasonable, or something thereabouts.
I’m more cautious about moderators. The concept of “neutral” is problematic in political systems. It’s all well and good to point to Fishkin and say his team acts fairly, but when you scale up to the political level there won’t be any trust. Politics sets trust on fire even in the best circumstances. These so-called “neutral” moderators will be a prize, just as Supreme Court seats are a prize today.
Phrases like “neutral” and “balanced” sound good, but they require a central authority to determine what they are. Our current political pickle stems precisely from resentment against such elite organs, and the power they have to label some views as acceptable and others as unacceptable. Of course there are good moderators out there, but that’s not the question. The question is whether or not it’s possible to systematically staff an office of jury moderators in a way that is transparently neutral. I don’t think it is, and your observation about the problems with grand juries is appropriate here.
If moderators prove necessary, I think there should be some way to tie them to the elected branch. For example, in a jury with 50 groups of ten, you would need 50 moderators. If you have an elective body of 250, then it would take 5 members to appoint a moderator under Superminority. Similarly, you could use a token system, or every member could choose one and you could randomly select 50 from the 250.
The moderators under this system would not be neutral. But neutrality is in the eye of the beholder anyway. The main danger in politics isn’t volatility, it’s systematic bias. In other words, what we wish to avoid isn’t having moderators with a political perspective, it’s having moderators who collectively have a bias in one direction. Centralized administrative bias tend to create systematic bias in a single direction, and are therefore to be avoided.
Another possibility is to have multiple moderators working with each group in sequence. I imagine juries meeting over a few weeks rather than a few days. Imagine a jury working for three weeks, where they have a different moderator each week. They might get a conservative moderator in week 1, a liberal in week 2, and a centrist in week 3. This helps to smooth out some of the volatility.
LikeLike
Alex, thanks for the comments.
I think that this is a good point, and I agree that the lack of a need for a durable majority would reduce the toxicity of parties. However, I think that this is not the only factor, and the in-group dynamics of parties will be toxic even without the need for a majority. This is certainly an imperfect analogy, but groups from sports teams to ethnic minorities can be in great tension with each other without contestation over a durable majority. Group dynamics are highly influential in social psychology, and I think anything that hints at a correspondence between parties and proposals is somewhat risky, although of course this is eventually an empirical question.
This is a great point, and the creation of a competitive market of policy ideas between elections in which legislators compete for the informed will of the people on specific issues is one of the principal advantages of my and your designs.
This is certainly possible, and the answer is empirical, but I am somewhat skeptical. For example, if the proposal threshold is 40, and each party has roughly 40 legislators, party whips will need every legislator of their party to support their proposal, and thus will likely coerce them into doing so, perhaps with the threat of primarying them. An interpartisan compromise would require members of one party jeopardizing their party’s proposal to support another party, or a substantial number of legislators from multiple parties working together to create a true compromise, which seems unlikely. A Republican and a Democrat working together is not crazy, but 20 Republicans and 20 Democrats working together is much less likely. I could be wrong, but I feel like superminority creates the wrong mindset for compromise.
I think this is a good point, and you have convinced me more that single-legislator proposals are in significant danger of allowing “flooding the zone”.
I think that this is an interesting idea, and it could be effective. I am generally skeptical of tokenized systems, since I think they would at least seem too gamified and complex, but if popular will could be developed for them, I wouldn’t necessarily be opposed.
I think that this is a compelling idea. Given my skepticism of token-based systems, I think that a similar alternative of pure top-n voting could actually be quite effective. For each issue, each elected legislator can propose or cosponsor up to one option. The five options with the most total sponsors go to a jury.
Legislators can work together to propose an option. The legislator(s) who proposed an option work together to develop arguments for their option and against other options. The proposer(s) write and present their arguments, although they can get input from cosponsors. By giving only the proposers argument power, legislators cannot sabotage by cosponsoring an option they dislike and then making bad faith arguments.
This approach would ensure a limited number of proposals (no more than five or some fixed n), without requiring a specific threshold of support that might encourage partisanship. It’s possible that this system would end up with similar problems to superminority, but I think there’s a good chance it would be better. For example, in the earlier scenario with a legislature of 200, a proposal threshold of 40, and five parties with 40 legislators, superminority forces every member of every party to back their proposal. In top-n, legislators don’t need any specific amount of support to get a proposal through, they only need to have one of the five most supported options. In fact, if only one group of legislators from each party generates a proposal, there are only five proposals, and no party whipping is necessary to get a proposal through. If some legislators try to gum up the system by submitting many proposals, then the serious proposers just need to get some cosponsors. Legislators who are uninvolved with an issue can simply not get involved. Top-n reduces party whipping and blind partisan voting, and makes it easier to form multipartisan coalitions.
Top-n is also more compatible with existing two-party systems. If the Democratic leadership and the Republican leadership each generate one proposal, there is space for three other proposals, with an incentive to get more cosponsors on board, potentially through compromises. Instead of needing to meet a high threshold, legislators just need to generate an option with relatively high approval. I need to think through the implications of pure top-n more and generate clearer explanations, but I think it could be an improvement over single-legislator proposals and superminority, while still being simple and easy to understand, so I may change my proposal of deliberative government to use it. What do you think?
LikeLike
Yoram, thanks for the helpful response.
I think there is some truth to this, and there are certainly republican elements of my specific design. However, I think it is important to note that a large part of this republicanism arises because I am trying to design a system that addresses the flaws of existing systems while being maximally consistent with them. Because the American system is republican and Madisonian, my specific proposal retains some of those features through elected legislators generating proposals, and through a deliberatively appointed senate.
However, the principles underlying my approach: separation of proposers and deciders, issue-specific input (ideally from a short-term, evaluative, conclusive, and multi-choice jury), and acknowledging a tradeoff between responsiveness and expertise, could be used in a less republican way. As I mentioned earlier, it would be possible to have citizens’ assemblies generate the proposals and a jury choose between them, which would not directly empower elected representatives or a supposed “natural aristocracy”. Given the lack of compatibility with existing systems and the unproven nature of this approach, I would not necessarily favor it now, but it is not necessarily incompatible with the principles I articulated. My specific design is more compatible with republican principles, and if you want, you could personally view it in more of a transitional role to a more purely democratic approach.
I think that regardless of people’s level of knowledge or status, there is a benefit to dividing the proposing and deciding roles as in the “I cut, you choose” analogy. My advocacy of separate proposing and deciding does not represent skepticism of ordinary citizens, but rather skepticism of the ability of any body, elected, technocratic, or randomly selected, to make and evaluate its own ideas and rules in an unbiased way, especially given the force of group biases and social pressures. I think that Yoram, Alex, and I have some tech background; the widely accepted practice of code review is based on a similar principle. Everyone, even the most experienced programmer, stands to gain from someone else making sure they haven’t made any mistakes or missed an important point. This principle applies even more in a political context, where biases are exacerbated, uncertainty is higher, and the truth is less clear. I think it is not necessarily patronizing to say that humans make mistakes and benefit from feedback and accountability from others. Political systems based on people checking their own work, creating the options and picking them too, are at a greater risk of corruption and authoritarianism. They certainly can sometimes work, but why increase the risk unnecessarily?
I read Dahl’s argument, and I think that it is interesting. I don’t know enough about this yet to give a complete response, but I would say that there is ample historical evidence of what could reasonably be considered tyranny of the majority in a popularly supported way (e.g., slavery, Jim Crow laws, internment camps, etc), and that factions in the sense of groups that attempt to advance their own interests at the cost of other groups exist as highly polarized and extremist political parties today. I don’t think Madison’s solutions have succeeded in preventing these phenomena (e.g., the large republic solution advocated in Federalist 10 has not prevented tyranny of the majority), but that doesn’t mean we shouldn’t try to address these problems in new ways now that may have republican influences. Given the likely moderating benefits of well-structured deliberation, approval voting, and separating proposers and deciders, do you think those features should be pursued? Do you think that reducing polarization and avoiding tyranny of the majority are valuable goals?
LikeLike
Hey Keith,
I agree that minimizing variance is a good goal, but I think that removing it entirely is not really possible in any system involving random selection. It would require a sample size of almost ten thousand to achieve a 1% margin of error, which I think is larger than people are proposing. With a more standard sample size of 1000, the margin of error would be over 3%. Do you think that these margins of error threaten democratic legitimacy? I certainly agree that more variance is worse, but I think that preference change is not inherently negative; if it is relatively consistent, it is arguably beneficial (I don’t agree with the Lafontian criticism that preference changes “usurp” the will of the people).
I certainly agree with this point, and this is part of why I advocate preserving the direct lawmaking powers of the legislature: it makes jury-based democracy more of an addition than a replacement, ensuring that elections retain sovereignty, at least transitionally for path dependency.
I agree that more academic agreement would be ideal, although I think that so many of these questions are fundamentally empirical despite their current state as normative debates, and thus one of the main things that would help resolve these academic debates would be real-world experimentation, which often benefits from more public support. Regardless, I think there is certainly space for more normative debates, and empirical work by organizations like the DDL to answer some questions about the impact of deliberation, as I have argued here. I do think that at some point this should become a more public debate, even without complete academic agreement.
I agree that nonpartisan appeal of sortition is key, and I think that you, Fishkin, and many others in the field do a reasonable job of staying nonpartisan, at least academically. I think for more public rhetorical appeals it is key to emphasize the benefits delivered to everyone, regardless of political affiliation.
LikeLike
Alex,
I agree that small groups with confidential, non-consensus voting are ideal.
I agree that supposedly neutral moderators are problematic, and that is certainly one of the major issues faced by my system. I think that my design for an Office of Deliberative Democracy and Sortition would preserve some amount of neutrality better than some alternatives, but I agree that it would be susceptible at the very least to significant mistrust. There are many moving parts to organizing deliberation (in my experience), and maintaining robust balance and neutrality in a contentious context through a bureaucracy would be difficult, although not necessarily impossible.
Given that trial juries deliberate without a moderator, I would be interested to see a split experiment with a DP sans moderators compared to a standard DP. I think it is often argued that trial juries exhibit problematic dynamics of group polarization, homogenization, and domination because of social pressures for consensus and because of the lack of balanced moderation, but I’m not sure whether there has been empirical work on whether the lack of moderation actually contributes that much. I know that the DDL online platform has shown to be just as effective with an automatic moderator, which primarily only manages a speaking queue, nudges people to speak, and detects toxic language. These tasks are relatively uncontroversial, which makes me wonder whether automated moderation could be used in jury-based democracy, and/or whether moderation is actually that necessary.
The other issue is asking questions: in DPs, each group agrees on one or two questions to ask an expert panel. In deliberative law, the “expert panel” would be the competing legislators. There is a compelling argument that coming up with questions together engages accuracy-based motivated reasoning and reduces polarization by giving deliberators a shared task (for context, see Kunda 1990). In trial juries, I think question asking is typically not allowed. Moderators in DPs facilitate the process of coming up with questions (and the DDL online platform has an automated interface specifically for inputting and voting on potential questions), so whether/how to implement this in a system without moderators is another question.
I agree that competing partisan perspectives can create something more “balanced” than attempting to directly achieve neutrality. The same principle underlies having politicians generating proposals and arguments, not some centralized bureaucracy. I do think that having partisan moderators would take this a step further, and would create a more antagonistic atmosphere in jury deliberations, so I’m not sure whether legislator-appointed moderators would be preferable to deliberatively appointed ones.
I agree after thinking about this more that jury deliberations for a couple weeks would probably be preferable to ones for a few days, since that would help address expertise concerns, and would allow for a fuller consideration of competing proposals and arguments. This represents a small sacrifice in responsiveness for a sizable increase in expertise, which I think is worth it.
If government juries are to be structured as trial juries, the deliberation would come at the end of weeks of proposals and arguments. It is probably an empirical question whether deliberation should be integrated through the jury process or only at the end. If it was integrated throughout, rotating moderators would certainly be helpful, perhaps even daily.
LikeLike
Kai:> With a more standard sample size of 1000, the margin of error would be over 3%. Do you think that these margins of error threaten democratic legitimacy?
It would depend on the issue under consideration. The Brexit referendum verdict was 52/48, hence my specification of a jury of 5,001. If we are proposing a stochatory (Andre Sauzeau’s term) decision making process in deeply divided societies the variance needs to be kept to a minimum.
I agree that public acceptance will require progress on both experimental and theoretical fronts, but I’m no fan of normative political theory and agree with Jeremy Waldron’s sarcastic allusion to 57 varieties of luck egalitarianism. My preference is his notion of political political theory, which owes more to Aristotle and Oakeshott than Rawls. That’s why I titled my paper “what sortition can and cannot do” as opposed to what it ought to do. My claim that “allotted representatives” is a rare case of a substantive that only exists in plural form is just an attempt to clarify the meaning of the concept.
By contrast, the substantive in Superminority is the (individual) elected legislator. Our proposal depends on the Rousseauian separation between the sovereign legislature and the delegated government, which are conflated in all modern democracies, both parliamentary and presidential. Elected legislators (under our proposal) won’t govern, which eliminates the need for durable majorities. This should enable a return to the Madisonian ideal (which descended into partisanship before the ink on the constitution was even dry).
As for the issue of moderators, whether impartial or balanced, this strikes me as a can of worms. I agree that we need to do the experiments to see whether unmoderated deliberation would be viable, but I’d prefer to follow the Athenian/Rousseauian model more literally. “Deliberative democracy” strikes me as a purely normative project and (possibly) even oxymoronic. The problem with face-to-face deliberation is downscaling it for more quotidian decision making (Bloomfield Tracks etc). My hunch is that Goodin is right, but we need to do the experiments.
>: I advocate preserving the direct lawmaking powers of the legislature: it makes jury-based democracy more of an addition than a replacement, ensuring that elections retain sovereignty
This would only serve to increase partisanship. Our proposal for popular sovereignty is a low Superminority threshold combined with a representative jury, which will reduce partisanship. I think there is also a danger that parties will claim that their direct lawmaking power is the democratic one. In our proposal the (delegated) government would have the right to propose legislation, but a jury would still be necessary to turn it into law (and the legislature could appoint opposing advocates, as in Athens). If we do without the deliberative shenanigans it would enable emergency laws to be enacted promptly, but with full democratic oversight.
>: I agree that nonpartisan appeal of sortition is key, and I think that you, Fishkin, and many others in the field do a reasonable job of staying nonpartisan, at least academically.
Hmm, that’s what Rawls thought, until it was pointed out that his universal theory of justice would be better described as the viewpoint of the late 20th century East Coast liberal arts professoriat. We all smuggle in our own preferences, and the unconscious ones are the most pernicious.
LikeLike
Kai,
> I need to think through the implications of pure top-n more and generate clearer explanations, but I think it could be an improvement over single-legislator proposals and superminority, while still being simple and easy to understand, so I may change my proposal of deliberative government to use it. What do you think?
I prefer top-n. I view it as a flavor of Superminority, because under top-n, the Superminority threshold still matters: the Superminority threshold becomes the number of votes you need to guarantee success. Below the Superminority threshold, you can succeed only if votes are split in such a way as to lower the effective threshold. Top-n therefore prevents a type of obstruction from “vote withholding”, where a member simply refuses to vote, or supports an alternative with no chance of success.
LikeLike
Kai,
Yes, and I would like to add the importance of iterative development. If there is one lesson I ever learned from software development, it is that iterative development beats the waterfall approach every time. It is important, therefore, that juries can be called quickly and systematically, so that law can develop by many incremental changes rather than through bloated “omnibus” bills like we have today.
CAs in France, for example, are incredibly difficult to organize, requiring unique preparation for each one based on topic. This is an absolute dealbreaker for me. When we look at variance, an iterative process is just as important as jury size in reducing variance, as successive juries build on previous ones in a convergent manner. It is imperative, therefore, that calling a jury be as systematic as possible, without any choke points where legislators or administrators can obstruct the process.
LikeLike
Keith, thanks for the points.
I agree that separating legislative and executive (government) tasks is valuable. How do you think this can be implemented? An idea I had is to have executive officials be deliberatively appointed, and then have a randomly selected committee of a deliberative senate be charged with supervising and coordinating the executive. The executive would also use juries to resolve controversial decisions.
The general issue is that any executive separate from the partisanship of the legislature may end up as a somewhat unaccountable bureaucracy, which may exhibit “centralized administrative bias” similar to Alex’s worries about “neutral” moderators. Although the idea of a neutral executive faithfully implementing the authoritative will of the legislature is appealing, I’m not sure how it can be implemented in practice, as there will always be controversy in administrative decisions. Using juries to resolve disputes is promising, although juries choosing between competing options seems to resemble the legislature, so I’m not sure if legislative partisanship can be kept out of the executive.
Deliberative democracy is a term assigned different meanings by different people. Although I agree it is purely a normative ideal for many, when I refer to deliberative democracy in my paper, I mean actual political decision-making by an informed microcosm of the citizenry. When Fishkin refers to deliberative democracy, he refers in part to his very empirical work on DPs.
In terms of terminology, I’m not inherently committed to “deliberative democracy.” Jury-based deliberative democracy could be more simply referred to as jury-based democracy, and deliberative government could be jury-based government or jury government etc. The analogy to juries is the more important component of the name, although juries do evoke a sense of jury duty and small, unrepresentative bodies, so any other naming ideas would be welcome. I’m generally skeptical of neologisms like “Jurga”.
I agree that the potential variance created by interpersonal deliberation at a small scale may be problematic, although I think that a single jury of twelve as in Goodin and Niemeyer (2003) has little democratic legitimacy, with or without deliberation. It is an important empirical question whether a smaller jury (e.g., 200 to 300 jurors) for municipal and regional contexts would have sufficient legitimacy, despite its margin of error.
LikeLike
Keith,
Although I agree that there is potential for reduced partisanship, I think there is a danger of intense party whipping and partisan-influenced juror voting in superminority, as I argued previously, and I think that top-n could help reduce this problem. If you could clarify how low of a superminority threshold you are proposing, that would be helpful (e.g., in the U.S. House of 435, what would the proposal threshold be? How do you avoid too many proposals with a low threshold?)
I agree that allowing the legislature to pass laws directly by a majority vote would increase partisanship. My defense against this in my proposal is requiring both the elected house and the deliberative senate to support a bill for it to be passed directly, which prevents a partisan majority in the house from directly setting the law. Regardless, this protection may be insufficient, and the threshold for direct passage of legislation could be set to two-thirds to discourage partisanship.
I agree that removing deliberation would reduce the amount of time it takes to enact emergency laws. However, if I understand correctly, your proposal would still involve days or weeks of trial-style debate between competing advocates, which would prevent immediate emergency action. If the jury is to vote without receiving thorough information, they no longer achieve the informed will of the people. A rushed jury verdict may give false democratic legitimacy to problematic emergency actions.
In emergencies in which there is time to wait a couple days or weeks, the normal process (perhaps slightly rushed) should work fine. In truly urgent emergencies, I would rather suitable emergency powers be granted to the executive and legislature, with the jury voting soon afterward on whether to continue or revoke the emergency after hearing competing arguments. I think a two-thirds supermajority threshold for direct passage in the legislature would be helpful in emergencies without too many negative side effects.
That’s a good point, and part of the reason why I think expanding the discussion from academia to the broader public is so important.
LikeLike
Alex, thanks for the comments.
I agree that the effective superminority threshold still matters in top-n, which is part of my hesitance: if every legislator treats top-n like superminority, it becomes superminority, and inherits the same potential flaws. My hope is that in top-n each proposal would only have a few cosponsors, allowing for compromise options and preventing party domination, and there would be no need for legislators without interest in the subject matter to get involved. However, it is difficult to know whether this would happen, or whether cosponsorship on party lines would become the norm. I think that at its best, top-n is great, and at its worst, it is superminority, so it is at least preferable to pure superminority.
It is also worth noting that top-n would be compatible with a deliberative senate, which would hopefully provide a large pool of potential proposers and cosponsors of compromise options. I’m not sure whether deliberatively appointed senators would be able to stay as moderate as I hope, but if they did, it would help reduce the correspondence between partisanship and proposals. (In this approach, elected legislators and deliberative senators would both be able to propose and cosponsor, with top-n based on the total number of sponsors across both chambers).
I agree that iterative development and legislation are generally beneficial. It is a big shift from the current approach of laws being extremely difficult to pass, and I think one would have to be careful to ensure that the juries do get informed and/or deliberate thoroughly so as to avoid making decisions too rapidly in a way that creates public outrage. Regardless, I think less gridlock and more frequent legislation would be great overall, also in terms of reducing legislatures ceding power to the executive.
I agree that juries should be called systematically. What is your proposal for when they are called? In my paper I propose a periodic, automatically assembled agenda jury which votes on the agenda items for a limited number of other juries, which will be called in the next period of time (e.g., a couple months or a year). I also proposed that either chamber of the legislature be able to add an agenda item (and thus call a jury) by majority vote, which you might object to on the grounds that majority thresholds in the legislature should be avoided. However, if juries are to make most lawmaking decisions, there needs to be some way to populate a continuous agenda. Perhaps frequent agenda juries would be enough, although not for emergencies. I think that giving the executive government proposal power as Keith mentioned may make the executive too partisan, and the agenda unbalanced.
LikeLike
Kai,
> “I cut, you choose”
The cake-cutting analogy of politics is not rather poor, but even it were a good one, I can hardly see how it makes your proposals appealing. Who are the two “sides” here that supposedly get a fair treatment by this division of labor? Both of those sides get significant decision making power. But if one of the sides is the people (or the statistical representative of the people), then who is the other side? Unless that other side is also the people (or a statistical representative), then that other side represents a privileged setion of society who gets preferential treatment, where their ideas and intererests get over-represented in decision making. What justifies an arrangement like that? Thus, the only way to make this arrangement potentially democratic is to have both side (the cutter and the chooser) be statistical representatives of the people.
> Do you think that reducing polarization and avoiding tyranny of the majority are valuable goals?
I don’t see either polarization or “tyranny of the majority” (a term which suffers from an “absence of specific meaning”, as noted by Dahl) as significant problems of our society. The main problem of our society is the perennial political problem of large groups – tyranny of a minority. This simple fact is one which the republican-Madisonian rhetoric aims to obscure.
(If and when we manage to democratize society, i.e., to rid ourselves of the tyranny of minorities, then we can seriously work on improving that society, e.g., by increasing mutual respect and solidarity among the people. Until then, any such efforts are fools’ errands that are doomed to fail.)
LikeLike
Kai:> I agree that separating legislative and executive (government) tasks is valuable. How do you think this can be implemented?
This is currently the work-in-progress part of our joint project. Alex is the implementation man (I just make theoretical generalisations, while waving my hands in the air). The operation of stable and accountable governance may well be the Achilles’ Heel of the ad hoc legislative jury model.
>In truly urgent emergencies, I would rather suitable emergency powers be granted to the executive and legislature, with the jury voting soon afterward on whether to continue or revoke the emergency after hearing competing argument
That sounds like a sensible compromise.
Yoram:> But if one of the sides is the people (or the statistical representative of the people), then who is the other side? Unless that other side is also the people (or a statistical representative), then that other side represents a privileged section of society.
You insist that statistical representation is the only democratically valid form and yet are happy with tiny, voluntary, full-mandate, randomly-selected bodies. Pitkin (along with most political theorists) insists that representation requires a variety of mechanisms. The models that we are discussing presuppose a very low proposal threshold and give the final decision power to a large jury. Elected politicians who chose only to represent the interests of their own peers would be on a hiding to nothing.
LikeLike
Yoram:> that other side represents a privileged section of society.
Sure, but if they are composed of professional politicians, their principal concern will be to secure re-election so, given that Superminority will increase the competition and that the jury has the deciding vote then the demos will have (effective) control of both the proposing/agenda setting and decision stage. It’s hard to imagine any other way of operationalising Dahl’s criteria. As Ledru-Rollin put it “There go the people. I must follow them, for I am their leader”.
LikeLike
Yoram,
Generically, the two sides could theoretically be anything, with the important part being that they are separate entities, one charged with proposing and one with deciding. If you believe that both proposers and deciders should be selected randomly, then I agree that is one way to achieve separation of proposers and deciders, and I think that such an approach might potentially work at some point in the future, as I outlined previously with citizens’ assemblies proposing and a jury deciding. Thus, I don’t necessarily agree with Keith that election is the only potentially effective method for selecting propers.
However, I believe that based on the current context of electoral systems, elected proposers and allotted deciders provide the most pragmatically effective approach, as I argue in my paper. I acknowledge that this is not purely “democratic” in the Athenian sense of democracy. However, I think it is “democratic” and/or republican as in a system that combines various methods to give the people control over the agenda and decisions, not in a perfect way, but in a pragmatic way that facilitates reasonably informed decisions while protecting individual liberties.
I understand that you may desire a more perfectly “democratic” system, and I certainly agree that researching such approaches is valuable. However, it is my (personal) view that elected proposers and allotted deciders would be sufficiently effective (arguably more so than full random selection, due to a potentially more optimal position on the tradeoff between responsiveness and expertise), and significantly more likely to be adopted than allotment of both, so they are the best option to pursue for now. I can certainly agree to disagree on this.
LikeLike
Kai:> I don’t necessarily agree with Keith that election is the only potentially effective method for selecting proposers.
My concern about allotted proposers (CAs or whatever) is how it would be possible to ensure a proportional match between the proposals and the ideological diversity (“discourses”, as Dryzek terms it) of the demos. You guys are the math nerds, but my lay understanding of the LNN is that it requires (a) large numbers (b) Condorcettian independence and (c) units that are open to aggregation. According to Pitkin this rules out deliberative speech acts. Bear in mind that Lafont and Urbinati’s defence of election is based on democratic, not epistemic norms.
Classical-era Athens was a tiny, homogeneous direct democracy, and from a pragmatic perspective, why would modern citizens choose to give up all direct participation in the political process?
LikeLike
Yoram,
I agree that excessive power of a (power, wealth, etc) minority is often a problem.
I think that Fishkin’s definition in Can Deliberation Cure the Ills of Democracy? (2025, pp. 69-76) provides a reasonably specific meaning. He defines tyranny of the majority (or minority) as “the imposition of severe deprivations of human rights when an alternative choice is available that would not impose such severe deprivations on anyone” (p. 76).
I think that there is a reasonable argument to be made that tyranny of the majority is a substantial problem. One could for example potentially argue that the treatment of immigrants in certain countries constitutes tyranny of the majority as per the definition above. From a historical standpoint, there are clear examples of tyranny of the majority in slavery, Jim Crow laws, etc.
I would also argue that polarization is a substantial problem. Perhaps ideological polarization is not inherently problematic, but it can lead to tyranny of the majority and democratic backsliding. If one views the other side as inherently unacceptable, one may be willing to sabotage democracy to keep their side in power. I think it might be difficult to deny that this phenomenon plays a role in some modern contexts.
Affective polarization (animosity toward members of the other party) is problematic from a social standpoint: I think most people would rather not have a world filled with partisan hate and animosity.
Regardless, problems are not mutually exclusive. I think my proposal can address many of the problems you identify with electoral systems, while also addressing problems of polarization and tyranny of the majority, in a relatively pragmatic way. Practical proposals often involve compromises, and jury-based democracy tries to satisfy many competing constraints and tradeoffs.
LikeLike
>my lay understanding of the LNN
Sorry, I meant LLN (Law of Large Numbers), the key characteristic of jury democracy (according to those who believe in statistical representation).
LikeLike
Keith,
I agree that the executive is a challenge in jury-based democracy. I think that, at least in the long run, a combination of increased legislative power (through jury-based law), jury-based appointment of various administrative officials, and jury-based rulings by various important administrative bodies could make the executive reasonably independent, stable, and democratic. Developing a successful jury-based executive will likely require pragmatically acknowledging that administrative government is complicated and will require mixing several approaches, including some existing mechanisms.
I agree, and I think this is important to emphasize: with a reasonable number of competing proposers with some democratic backing, the deciders are really the ones with the bulk of the power. That is the beauty of multi-choice voting: it gives the jury real choice over what policy to implement, without requiring them to learn how to write bills.
LikeLike
Keith,
I haven’t researched the details of this enough to be able to give a conclusive answer, and I think this is a great direction for future research. I would say that if one considers political beliefs (in some abstract multi-dimensional space) to be subject to the LLN, one could make an argument that the policy resulting from a CA would follow a random distribution centered on the average political belief of participants.
There are two problems with this: (1) social pressures and disparities in deliberation may jeopardize this simple average model, and (2) this would not really generate a range of competing proposals capturing the discourses of society. One could try to address (1) with careful structure and/or moderation and (2) with either multiple very small CAs (to increase variation), or with self-selected groups in one CA. I think the latter may be the better approach. In a citizens’ assembly of 100, participants could be instructed to sort themselves into groups of 10-20 based on beliefs, with each group generating a policy idea and arguments to present to a jury.
I’m not sure whether this approach could make rigorous statistical claims to perfect representation, but I think it could fare reasonably well at capturing the discourses of society, barring (serious) concerns about expertise and capture. Again, I don’t think this approach is necessarily preferable to elected proposers, and I think it is much harder to adopt in our current context, but I don’t think it is fundamentally flawed as a (theoretical) approach.
This pragmatic perspective is precisely why I (and Fishkin and others) argue for retaining elections: in our path dependent context, people generally like having the right to vote. Any attempt at reform faces a momentous uphill battle, and I think any initial attempts should certainly retain election in some significant role.
LikeLike
Kai,
> “democratic” and/or republican as in a system that combines various methods to give the people control over the agenda and decisions, not in a perfect way, but in a pragmatic way that facilitates reasonably informed decisions while protecting individual liberties
Again, such appeals to pragmatism and good-enough compromises fit well within the conventional academic discourse around sortition specifically and democracy more generally. This is great fodder for an endless on-the-hand-but-on-other-hand academic discussion that has been going on since the elitarian theories of democracy fell out of favor in the 1970’s. If you feel comfortable in this milieu, that’s fine, but you should not delude yourself that you are promoting anything other than the perpetuation of the status quo.
LikeLike
Kai,
> “the imposition of severe deprivations of human rights when an alternative choice is available that would not impose such severe deprivations on anyone”
The obvious problem with this definition (and other definitions of this kind – this is hardly innovative) is that while it sounds very reasonable and laudable there is no objective standard for its meaning or application. There is no objective standard for what “severe deprevation” is, or for what any other term in this phrase means. Madison et al., for example, when they talked about tyranny of the majority, were thinking about the “severe deprivation” of themselves having being stripped of their wealth and privilege. Others may disagree that this constitutes a severe deprivation. Since there is no objective standard, one has to appeal to some authority to make determinations in specific cases. Now, this authority can either be representative of the people (in which you are back to the “tyranny of the majority”) or non-representative (in which case this is a tyranny of a minority).
The whole notion of “tyranny of the majority” is obviously inherently elitist. The idea is that the majority is more oppressive, less moral than the elites, or at least than some good elites (which those who speak of the tyranny of the majority belong to). In the present case you are telling me that you and I and people like us should design a system that would in one way or another prevent the majority from exercising its evil ways. You are thus assuming that you and I and people like us are better judges of right and wrong than the majority of the people. This is very flattering to us but has no real basis other than self-importance. The very fact that this kind of idea is current in intellectual circles indicates that the elites in our society are very far from actually internalizing the democratic values they profess.
> clear examples of tyranny of the majority in slavery, Jim Crow laws, etc
Actually, all those examples are examples of tyrannies of minorities, since none of those policies were created and implemented by a democratic government. Rather they were created and implemented by oligarchical electoralist governments. It may have been that a democratic government, had one existed, would have implemented the same policies, but obviously even if it did, it would not have been worse than the oligarchical governments that have in fact implemented them. So even if we make the speculative assumption that such oppression by a democratic government would have taken place, it is hardly a good argument for prefering tyrannies of minorities to a “tyranny of the majority”.
LikeLike
Hi Kai, I agree with all of your comments. I think the problem with CA-generated proposals is that it presupposes Byzantine/Venetian levels of complication (for all the reasons that you outline). I also think that voluntarism may well be the most politically significant population parameter (see the open access article by Spada and Peixoto in the first issue of JoS) and that there is no democratic fix for perlocutionary imbalances in speech acts. When you combine this with the inevitable resistance of the vast majority of citizens to the aleatory coup/putsch I think we should acknowledge that the CA movement is on a hiding to nothing (and may well bring sortition into disrepute). I agree that DPs are a far more promising model, as Jim Fishkin is trying to address all these objections, but he told me he’s never really understood the need to focus on discourses.
https://www.ingentaconnect.com/contentone/imp/jos/2025/00000001/00000001/art00009
It does surprise me that many sortition advocates focus on economic elites, when it’s clear that cultural issues are increasingly significant. Proposal-generating CAs may well intensify the populist backlash against the post-Gramscian hegemony.
LikeLike
The focus on the exchange of reasons (as opposed to the representation of discourses) suggests to me that the deliberative tail is wagging the democratic dog.
LikeLike
Yoram,
I agree that my ideas fit within the academic discourse, and I understand that they may not be as transformative as you desire. However, my proposal is still transformative compared to existing systems of government, and I think that most people would not consider jury-based democracy to be the status quo. Regardless of whether our current systems are benign or malicious, we are stuck with them as our starting point, and thus I think we are largely limited to iterative improvements on them. I would be curious to hear what alternative path forward to change you think is feasible.
I agree that academic discussion can seem like an endless back and forth. I think it is important to appreciate that there is nuance and complexity in life, and that black-and-white approaches to politics usually result in tyranny and authoritarianism. A revolutionary approach to sortition would likely create more harm and disillusionment than a more measured approach. Change is difficult and rarely comes overnight; when it does, it is often violent and undemocratic.
LikeLike
Yoram,
I agree that there is no objective definition for this, and I think that attempts at defining “ideal” or “problematic” political outcomes certainly suffer from the lack of an “objective political truth.” I am certainly largely sympathetic to Dahl’s criticisms of guardianship and quasi-guardianship on these grounds.
Notwithstanding these limitations, I think it is still possible to identify that certain injustices involve one group imposing severe deprivations of human rights on another group when it is unnecessary (e.g., the Holocaust, slavery, etc), and that it would be good if such injustices could be avoided, regardless of whether they should be called “tyranny of the majority” or “tyranny of the minority” or something else.
I agree that in some sense these examples qualify as tyranny of the minority, since they were implemented by a minority of citizens (an elected body). However, in the same sense, sortition can result in tyranny of the minority, since a randomly selected minority of citizens is put in power. Sure, these citizens are statistically representative of the people, but after someone spends years as a legislator, I think it would be difficult to argue that they are the same person as before. Through years of service without accountability, sortition legislators develop different interests from the general public, and thus constitute a distinct minority capable of tyranny (without guardrails in your proposal). I am not saying that this is guaranteed to happen, I am just saying that there are substantially greater risks with longer service.
In my view, there are two ways to avoid this “tyranny of the minority”: have the entire population make decisions (direct democracy), which is impractical, or have large, random, short-term juries make decisions. Jurors do not have enough time or unrestricted power to develop minority interests in conflict with the best interests of the general public, thus preventing “tyranny of the minority.”
Juries can also reduce “tyranny of the majority,” since exposure to diverse perspectives in a balanced setting reduces support for injustices against minorities (Fishkin 2025, chap. 3; Kim, Fishkin, and Luskin 2018). My argument is not that a minority elite should be protected from the will of the people (even if that was Madison’s argument), but rather that a short-term microcosm representing the informed will of the people can reduce both types of tyranny.
LikeLike
Keith,
I agree that citizens’ assemblies suffer significantly from nonresponse bias. I do think that this is one of the more addressable problems (with high compensation and quasi-mandatory service), at least for institutionalized approaches. Regardless, short-term bodies can certainly achieve high participation rates more easily.
Moderated small group discussions can fix this imbalance (Luskin et al. 2022; Siu 2017), but that is more difficult to achieve in a standing body, and it’s not clear whether the same holds without moderation (if you consider moderation to be undemocratic).
I certainly agree that DPs are more promising than CAs. I still think that CAs can provide useful information and aren’t entirely negative, but I think that making a clear distinction between DPs and jury-based democracy on the one hand, and citizens’ assemblies and full sortition on the other, is important for boosting support for the former.
One issue is that (poorly structured) citizens’ assemblies are easier to run than (well structured) DPs in terms of cost and effort. If future empirical work shows that a version of DPs with fewer moving parts is similarly effective, then maybe they could be scaled more broadly. In theory, governance juries as in jury-based democracy should be relatively easy to run if legislators provide the proposals and arguments, and if there is either automatically moderated deliberation, unmoderated deliberation, or no deliberation.
I think that the representation of discourses is the job of the proposers, and the exchange of reasons is the job of the deciders if they are to deliberate. I do think that the jury and democracy components are more important than the deliberative component in jury-based deliberative democracy, and I will probably rename my designs accordingly (jury-based democracy, jury-based government, jury-based law, etc).
LikeLike
Kai, pretty much agree, but still think moderated deliberation in full-mandate policymaking bodies would be perceived as illegitimate from a democratic perspective (DPs have always been advisory).
>I think that the representation of discourses is the job of the proposers, and the exchange of reasons is the job of the deciders if they are to deliberate.
Agree on the former, but would prefer the jury to silently adjudicate between the reasons exchanged by the (elected) proposers (i.e. Athenian democracy scaled up for large modern poleis). Do you agree that (apart from the Anglo-American trail jury) there is no historical precedent for the deliberative democracy model? There is little record of the internal workings of the boule, and it’s hard to imagine what deliberation in a body of 500 would be like. So all we are left with is Habermas’s bourgeois coffee house democracy which (according to him) was destroyed by universal suffrage. To my mind deliberative democrats are seeking to create an alternative elite (free from false consciousness), but this will only serve to fuel the populist backlash.
So why the insistence on face-to-face deliberation? The syntax of Fishkin’s “what everyone would think under good conditions” [my emphasis] hints at epistemic considerations — that there is a “right” answer and the job of the jury is to uncover it. This is certainly the case with the judicial trial (hence the need for jury-room deliberation) but it’s a very dangerous idea in politics.
>A revolutionary approach to sortition would likely create more harm and disillusionment than a more measured approach.
Absolutely — all Alex and I are advocating is to change the decision threshold of the elected body (to better represent the discursive diversity of the polis) and to extend the use of juries beyond judicial trials (as an alternative to plebiscites). No need to take to the barricades.
LikeLike
PS, although Jim views the DP as a refinement of polling methodology, it’s one thing to seek to represent the informed preferences of the sample, but measuring transformed preferences is an entirely different kettle of fish/can of worms (especially if the vector is consistently progressive or whatever). To my mind DP organisers should be equally content with zero preference change. If the pre- and post-polling split are both 52/48, then that’s an accurate reflection of the discursive diversity of the polis, not an absence of informed deliberation, so there is no good reason to champion convergence (as in Fishkin et al. 2021).
LikeLike
Keith, I think your points are good. My response is somewhat long, and I would appreciate your thoughts. I may write a paper from this angle at some point.
> Do you agree that (apart from the Anglo-American trail jury) there is no historical precedent for the deliberative democracy model?
I agree that there appears to be no historical precedent for interpersonal deliberation in randomly selected decision-making bodies. Whereas you claim classical (Athenian) and normative (speech acts ought to only be for proposers) backing, my argument for interpersonal deliberation is generally empirical (deliberation increases knowledge and reduces polarization) and contemporary (trial juries deliberate), with emphasis on the empirical.
Of course, my empirical claims are still based on certain normative assumptions, which I agree require justification and should be made more clear so that they can be debated. In my view, there are at least two important normative ideals in a system of government: responsiveness (to the will of the people) and expertise (understanding the subject matter at hand). Responsiveness combines several other ideals (equality, control of the agenda, accountability, etc), and expertise is similar to Dahl’s “enlightened understanding.”
To show the normativity of these values, would someone prefer, all else being equal, for decision-makers to have more knowledge about the issues they are deciding on? Would someone prefer, all else being equal, for decision-makers to be more responsive to the will of the people? The answer to both of these questions is typically yes.
The issue is that responsiveness and expertise cannot be increased with all else being equal. Responsiveness and expertise are generally inversely correlated: as one gains more knowledge about a specific subject, they become less representative of the general public. I would argue that, in order from most responsiveness to most expertise, political systems can be sorted like this:
Madison’s vision was a combination of (5) and (6), and our current system is mostly (5). I think we agree that we should move more in the direction of responsiveness, with me advocating (3), you (2), and Yoram (4).
LikeLike
Keith,
I agree that it is problematic to argue that there is a “right answer” in politics. However, it is the case that there are facts and knowledge that can help shape decisions, and it is arguably normatively desirable to have more of this epistemic good, all else being equal.
Unfortunately, because it is impossible to increase expertise without sacrificing some responsiveness, deciding on a political system requires deciding what balance of responsiveness and expertise is the most desirable.
Analogous to this balancing problem is the economic concept of a production-possibility frontier (PPF). In a PPF, as in this situation, there are two goods to balance, with (productively) efficient combinations on a curve, impossible combinations outside of the curve, and inefficient combinations inside.
For example, I would argue that jury-based (deliberative) democracy is a (productively) efficient combination because it is impossible to increase its responsiveness without sacrificing expertise and vice versa. On the other hand, a corrupt dictatorship achieves neither responsiveness nor expertise, making it decidedly inefficient. A system of direct democracy in which every person is an expert in all policy matters is impossible (outside the production-possibility curve (PPC)).
PPCs are typically concave, since when you are at an extreme of little responsiveness or little expertise, it is relatively easy to increase that value without losing much of the other. For example, adding deliberation to jury-based democracy appears to increase expertise substantially, while still largely preserving the responsiveness to the will of the people.
Most well-designed systems of government achieve “productive efficiency”: their responsiveness or expertise cannot be increased without sacrificing some of the other. However, that does not imply “allocative efficiency”: certain configurations are more desirable than others. For example, by separating proposers and deciders, jury-based democracy achieves relatively high levels of expertise and responsiveness where they matter the most (for proposers and deciders respectively). Lottocracy confounds proposers and deciders, achieving relatively low levels of expertise and responsiveness overall, making it allocatively inefficient (in economic terms).
Jury-based democracy with and without deliberation is productively efficient, but which form is allocatively efficient (and thus preferable for society) depends on to what extent society values expertise and responsiveness. This is a difficult normative question: to what extent should governance jurors become more informed but consequently less reflective of the general public?
LikeLike
Keith,
I guess I’m not sure what distinction you are drawing between “informing” and “transforming.” Can you clarify what your proposal is for the extent to which jurors are informed? If I understand correctly, you are proposing a couple weeks of listening to competing arguments from legislators? If so, especially given the reduction of rational ignorance in a (compensated) jury, that is likely to significantly shift the opinions of jurors, increasing expertise at the cost of responsiveness.
If you consider this sacrifice of some responsiveness for expertise to be legitimate, what makes interpersonal deliberation more problematic? Based on your arguments, I would think that your response would be that (1) moderators are particularly democratically illegitimate/unresponsive and (2) interpersonal deliberation is less subject to the LLN.
(1) could be addressed with unmoderated discussion or automated moderation. (2) means that there might be a somewhat weaker statistical correspondence (less responsiveness), but perfect responsiveness is already jeopardized by jurors becoming more informed. We need more empirical work to figure out the specifics of how much deliberation increases expertise and reduces responsiveness (potentially through greater variance) as I have argued above, but overall, adding interpersonal deliberation to juries is primarily a small shift on the PPC of responsiveness and expertise in my view.
Unless one agrees with Lafont that any sacrifice of responsiveness for expertise is “usurping” the will of the people, I don’t necessarily see how interpersonal deliberation is a fundamental (normatively impermissible) problem.
There are compelling arguments that interpersonal deliberation can increase information gains by encouraging jurors to pay more attention since they know they will have to talk about the issues with others, and that certain forms of deliberation (e.g., coming up with questions together) can engage accuracy-based motivated reasoning.
Given that most people are mostly uninformed about most issues due to rational ignorance etc, the absence of any preference change would seem to suggest to me that either information was not gained or that partisan heuristics overpowered any information gained. I don’t think that either of these scenarios are positive. With or without deliberation, there should ideally be some preference change from information.
LikeLike
kai:> adding deliberation to jury-based democracy appears to increase expertise substantially, while still largely preserving the responsiveness to the will of the people.
What is the evidence for this claim? As far as I’m aware DDL control groups are only provided with (written?) information. What we really need is a comparison between a silent jury and face-to-face deliberation. If it can be demonstrated that the latter is far more “expert” than the former, then I guess face-to-face deliberation might be necessary but, given our shared concerns over group dynamics and the LLN, it would add hugely to the cost of the operation.
> the absence of any preference change would seem to suggest to me that either information was not gained or that partisan heuristics overpowered any information gained.
We’re running the risk of moving into “basket of deplorables” territory. My understanding of politics is that there are competing narratives/ideologies/discourses. We all like to think we’re right and the other guy (and his supporting team of experts) is wrong, but I don’t think that’s necessarily partisan. There are, undoubtedly, policy-related facts, but democratic politics has more to do with interpretation and preferences, so I would agree with Lafont and Urbinati that the epistemic turn (which is embraced by most deliberative theorists) is anti-democratic.
LikeLike
P.S. The empirical foundation of our project is 4th century nomothesia, when large juries determined the outcome of the legislative proposals of competing advocates (half of whom were elected) by listening in silence and then voting. Hansen claims these juries were selected by lot, whereas Canevaro argues it was the whole assembly. But the difference was purely numerical (large vs huge). Deliberative democracy, by contrast, is a modern normative project and the only empirical work has compared outcomes to information-only control groups. Manin argued that the important thing was to be exposed to competing arguments, and this has not been tested since the 4th century BCE.
LikeLike
Keith, I think we agree on most things.
In my view, there are two questions that determine whether interpersonal deliberation makes sense:
(1) Does viewing policy debate between competing advocates have a similar (or greater) impact compared to interpersonal deliberation? This is an empirical question, and we both agree experimentation on this is necessary. If the answer is empirically yes, then interpersonal deliberation is unnecessary.
(2) Is it democratically legitimate for governance jurors to gain information and/or deliberate in a way that changes their opinions relative to the general citizenry? This is a normative question, and Lafont argues that the answer is no. If the answer is no, then viewing policy debate, reading information, and deliberating are all democratically illegitimate regardless of (1).
What is your answer to (2)?
The potential contradiction I see in your argument is that you think viewing policy debate is legitimate but interpersonal deliberation is not, even though both constitute usurpation of the will of the people in Lafont’s view. The LLN does not necessarily apply to “deliberation within” more than it does to interpersonal deliberation: different people will respond differently to arguments, and so viewing policy debate may create the same variance that you object to in DPs.
LikeLike
Kai, massive thanks for introducing some much needed analytical clarity. Agree that (1) is a purely empirical question. As for (2), Lafont’s normative argument is entirely Rousseauian — the general will which is our (collective) natural right cannot be alienated. In my previous paper, I argued that decision making by allotted jury would not involve alienation iff (sic) it could be demonstrated empirically that every sample would return the same verdict (within an agreed margin of error). The LLN would apply to silent deliberation in the sense that the only behavioural output (voting) is open to aggregation and all “deliberators” are independent. In the case of interpersonal deliberation it could only apply if (as Alex suggests), the jury of 1,000 could be broken down into 100 independent (unmoderated) discussion groups. I doubt if such an arrangement would be practical for the large number of policy decisions that characterise modern poleis, whereas silent juries are up/downscalable depending on the importance of the issue (in Athens the range was 501-5,001).
From the perspective of the vast majority of citizens who do not participate directly, all that matters is that the outcome would be consistent across different samples, and our hypothesis is that this would be easier to demonstrate empirically with large, quasi-mandatory silent juries.
Rousseau famously opined that Englishmen were only free on election day: “as soon as the Members are elected, the people is enslaved”. He was sceptical that freedom could be extended beyond tiny nations (such as Corsica), Our claim is that the combination of Superminority and large decision juries will enable freedom by accommodating the competing discourses of large modern states proportionately.
My paper on Rousseauian sovereignty and sortition is here: https://www.academia.edu/126366178/Deliberation_and_Sovereignty_The_Rousseauian_Case_for_Sortition
LikeLike
Keith, thanks for the helpful points.
I think that it is important to distinguish between certain normative conceptions of deliberative democracy that ignore which citizens deliberate (with a purely epistemic focus), and the Fishkinian approach of rigorous random selection combined with confidential preference aggregation. The Fishkinian model is much closer to what you advocate.
I think your Rousseauian argument is interesting. It seems to me that some of the arguments of this kind may overanalyze normative definitions to argue that random selection does not deprive the general public of power. Although I am certainly sympathetic to that argument, I’m not sure that trying to show it normatively completely works.
The most important question for democratic legitimacy is whether the people empirically consider something to be legitimate. I agree that removing unnecessary variance in jury results would help with legitimacy, but that does not necessarily guarantee it. If the public is 60/40 on an issue, and listening to the arguments from competing legislators moves them to 40/60, then 60% of the population will be unhappy with the decision. That’s my simplified interpretation of the Lafontian critique.
In my view, there are three main advantages of not having interpersonal deliberation:
(A) Deliberation within preserves juror independence and thus the LLN. I think this is a good point, and I appreciate you clarifying this.
(B) Deliberation within removes the need for potentially biased moderators.
(C) Deliberation within allows for any interested member of the general public to view the arguments from the legislators, deliberate within, and theoretically experience the same effects.
I think (A) and (B) are valuable, but not enough to ensure legitimacy. I think that (C) could have potential to bring the legitimacy of jury-based democracy across the finish line. The response to anyone disgruntled with the jury decision is “listen to the arguments and see what you think.” There is no magic deliberation behind the scenes: everything the jurors experience is publicly televised for everyone, and generated by electorally accountable legislators.
I still think that a similar effect to (C) could be accomplished with publication of anonymous deliberation transcripts, and that (A) is replicable with enough independent small groups, and that (B) can be accomplished with unmoderated deliberation. Thus, interpersonal deliberation is not unviable, but it is certainly harder. Whether it is worth it depends on the empirical result of (1).
LikeLike
I think this is a good quote, and issue-specific input from the people in the time between elections is one of the main advantages of our proposed systems.
LikeLike
Hi Kai, I’m in full agreement with your first comment.
>issue-specific input from the people in the time between elections is one of the main advantages of our proposed systems.
Superminority would deliver this, especially as Alex is arguing for a system of rolling elections. And the ongoing input of the people is achieved vicariously, as the shifting alliances between elected legislators would reflect their preferences. Legislators who followed their own preferences would lose with the jury, and would be more likely to be ejected at the next election. Whilst Yoram would dismiss this as “electoralist dogma”, Superminority (combined with large juries) would generate a very different kind of legislature.
LikeLike
The exchange between Kai and myself is a good example of the merits of deliberation — we both listened courteously to each other, exchanged reasons, learned from each other and (I think) converged somewhat to a median position. The interesting question is whether or not you need to participate actively in the exchange to gain the benefits of deliberation. This forum has several hundred members (including the facebook and X feeds) but it’s never more than a tiny minority who take an active role. Whilst it may well be the case that most posts go unnoticed, I’d love to know if anyone has been following the exchange and if it has had a similar effect on non-active participants. If the model of advocates (elected or otherwise) and jury were taken up, then all those selected by lot would be obliged to follow the debate as they are acting as proxies for the vast majority of citizens who don’t draw the Golden Ticket. So it would be very different from an online forum, but the same principles would apply. Terry Bouricius has written in a similar vein about his experience as an elected member of the Vermont House of Representatives:
“I had an eye-opening experience while serving on the Commerce Committee. Rather than following the usual procedure of taking testimony from one witness at a time, on one occasion we had a panel of opposing lobbyists essentially conduct an organized argument in front of the committee for a bill on which none of us had prior stances. The lobbyists would call each other out, since they knew when an opponent was intentionally cherry-picking data, or skirting an important issue, which none of us less-informed committee members could detect.” (Bouricius, 2024)
This is the model — debate not deliberation — that Bernard Manin (who helped kick-start the deliberative democracy movement) advocated later in his life.
LikeLike
Keith,
I agree, and I really appreciated this discussion. I would also be curious to know the impact on any non-participants.
I agree that debate plays an important part, and I certainly think a competitive advocates model as in jury trials is important. I agree that viewing debate between legislators is more likely to be effective than reading a briefing booklet (hopefully future empirical work can verify this claim).
However, I think it is worth noting that debate and deliberation are not mutually exclusive. Viewing debate between competing advocates does not preclude one from then discussing, as in jury trials and presumably in Bouricius’ Commerce Committee. As I think we agree, more empirical work is necessary to decide whether interpersonal deliberation is worth it.
LikeLike
Blimey, as we agree about everything for the first time ever I’m lost for words! I would dearly love to hear what anyone else has to say on this topic.
I have a vague memory of a very early British DP with a large silent audience that was televised (on C4 I think). It was plenary only and before and after polls were taken, but I can’t remember the topic and don’t know how the participants were selected. If such a model were followed for actual decision taking, the debate should be televised, so that non-participants could see if they agree with the jury (as you pointed out).
>more empirical work is necessary to decide whether interpersonal deliberation is worth it.
Yes, it would get down to a cost/benefit analysis. It would also be necessary to assess if any divergence from prior preferences was random (in the pejorative sense). If the post-deliberation swing was uniform across different samples, then the benefits might well exceed the financial/organisational costs. I’m still worried by Fishkin’s claim regarding what everyone would think under good conditions — this could just be unfortunate phrasing, but the long tail of Rawlsian deliberative democracy is a possibility. I think Jim originally studied under Dahl at Yale, so let’s hope it’s the former.
LikeLike
Keith, our overall agreement is definitely nice, and I would be curious to hear anyone else’s thoughts.
Fishkin and Luskin (2006) discusses the televising of DPs in the context of a Channel 4 (face-to-face) DP in 1998. Fishkin argues that televising of DPs shouldn’t attempt to replicate the effects on DP participants, and should serve (weaker) roles in encouraging participation, educating the broader public, and providing statistical cues.
Fishkin’s argument is generally that participation in deliberation is necessary to get the experience, whether it be in microcosm or mass deliberation (Deliberation Day etc). This certainly contrasts our jury-based approach with more emphasis on competing advocates, which can be compellingly televised (although I still think interpersonal deliberation is likely also valuable).
I think that members of the general public will shift less from viewing the same debate, since they will lack the context of being selected and having an impactful vote, and will instead be subject to rational ignorance.
However, they will probably shift some, and have the option of paying more attention and seeing all the competing arguments, thus achieving reasonable legitimacy.
I’m not sure which DP without interpersonal deliberation you are referring to. Both the 2006 paper I linked above and Fishkin (1996) seem to involve interpersonal deliberation.
I agree about the need for cost/benefit analysis.
I think Fishkin takes more influence from Dahl (who was his advisor) than Rawls. I don’t interpret his arguments as saying that everyone should think the same (“epistemically correct”) things, but rather that every person (or every randomly selected participant) should have a chance to think more about the issues in question.
Whether those “good conditions” are better achieved through interpersonal deliberation and/or viewing debate is the key question, for which a well-designed split DP experiment would provide very helpful information.
LikeLike
>I think that members of the general public will . . . probably shift some, and have the option of paying more attention and seeing all the competing arguments, thus achieving reasonable legitimacy.
That’s the important point. Anyone who is not included in the sortition (i.e. the vast majority of citizens) can view and assess the deliberative exchange. The exchange of reasons has to be fully in the public domain. That would be very hard to do with small groups, and the participants would know that Big Brother was watching.
LikeLike
I think it’s fair to say that deliberative theorists (and practitioners) are primarily concerned with the internal dynamics of the forum, and the equality of the participants. The deliberative model has been appropriated by advocates of minipublics but the representativity of the group vis-a-vis the target population is assumed to be ensured by stratified sampling (DPs are the principal exception to this rule). If this is not the case then the democratic legitimacy of the minipublic is questionable. The paper by Spada and Peixoto in the first issue of the Journal of Sortition outlines the problems involved: https://www.ingentaconnect.com/contentone/imp/jos/2025/00000001/00000001/art00009
LikeLike
Keith,
I definitely agree. This is also one of my main hesitations with “jury” in terminology. Although trial juries are the most widely accepted and institutionalized implementation of sortition in modern times, they are small and statistically unrepresentative. They are also consensus-driven, poorly compensated, and purely judicial.
The question is whether “governance juries” and “jury-based democracy” can be successfully established as an extension of juries to the legislative and executive domains, building on the legitimacy of juries without making people assume that governance juries would also be small and unrepresentative.
I’m not a fan of neologisms like “jurga” or “demarchy,” and I think that “jury” is still a better word than “sortition” or “lottocracy,” and shorter and better-known than “deliberative democracy.” However, I still wonder if there is a better name than “jury-based democracy.”
LikeLike
>I still wonder if there is a better name than “jury-based democracy.”
Minipublic would be better, had the term not been misappropriated by advocates of citizens’ assemblies. For a long time I’ve been concerned that the sortition movement will shoot itself in the foot on account of a lack of analytic rigour — especially regarding the sloppy use of terms like representative sample. I’m a big admirer of Andre Sauzeau, who coined the term “stochation”. He has a wonderful article in the current issue of JoS: https://www.ingentaconnect.com/contentone/imp/jos/2025/00000001/00000002/art00006
LikeLike
[…] the standard academic sortition mud stirring, one proposal stood out: using sortition to create democratic investor […]
LikeLike