“Limiting who can vote”

Ripples from Van Reybrouck’s book made it across the Atlantic and into the Washington Post where Dutch professors of political science Eric Schliesser and Tom Van Der Meer see fit to discuss his proposals for using sortition together with a proposal to “disenfranchise the ignorant to slant political rule toward experts”. They write:

Both [proposals] limit who can vote and seek to stimulate apolitical and rational decision-making:

1) Representatives by lottery. Belgian author and cultural historian David Van Reybrouck suggests abolishing elections and appointing representatives by lottery instead. Van Reybrouck’s proposal extends the principle of sortition — how juries are appointed — to the legislature: Randomly selected citizens would reach the optimal decision via deliberation, supposedly without a need to be bothered with politicking. When their term is up, they go home.

2) Experts as representatives. Philosopher Jason Brennan at Georgetown University suggests disenfranchising the ignorant to slant political rule toward experts. His proposal recently received favorable discussion in The Washington Post. Inspired by Plato, the rule by properly trained experts, or epistocracy, would prevent politicians from being easily swayed by moneyed interests and demagogues.


The authors are unsympathetic:

Both the election-lottery and the expert-rule plans appear to be democratic solutions to the ills of democracy. They propose better policies and less political intrigue, higher trust, and — in the case of the lotteries — greater equality.

But neither proposal actually provides political equality. While this is rather self-evident for expert rule, it is also a problem of the political lottery. Despite the one-man-one-lot principle, evidence with small-scale political lotteries at the municipal level suggests this approach yields greater political inequality. A study of Dutch political lotteries shows that only 5 to 10 percent of invited citizens participate, creating a small group of mostly educated, motivated, middle-aged, white, male participants not at all representative to the larger population. Dutch political scientists Harmen Binnema and Ank Michels conclude: “Lottery ultimately turns into self-selection, with lack of diversity and legitimacy.”

For those unimpressed with the self-selection argument a back-up argument is offered:

[C]ompulsory involvement could, at least in theory, be [applied] to participation in political lotteries. But even then, simple statistics reveal this plan’s impracticality: A randomly selected body of representatives would likely require thousands of participants to be more or less representative of the U.S. population.

Presumably the authors are hoping to impress the uninformed public they keep worrying about with this meaningless argument, or they are just uninformed themselves.

A couple more facile arguments conclude the article:

Although Van Reybrouck acknowledges that politicians selected by lotteries may have less expertise than politicians elected by voters, he suggests that unelected experts and civil servants would provide them with assistance. Yet that means that legislatures chosen by lottery might be even more influenced by assistants and lobbyists with access to the representatives.

Would lottery-politicians be less susceptible to corruption, as Van Reybrouck claims? That seems unlikely. Precisely because lottery-politicians cannot be held accountable via the ballot box, the institutional safeguards against bribery that curtail our regular politicians would be of little use. Elections, after all, allow voters to throw the rascals out.

Advertisements

131 Responses

  1. All of this is superficial bunk. Random selection of lawmakers has been discussed in great detail with loads of empirical evidence and plain common sense to support it. Yet no one seems to bother to cite these sources. The best RECENT book on this, and perhaps the best is written by two Australian profs–Lyn Carson and Brian Martin–RANDOM SELECTION IN POLITICS (Westprt, CT: Praeger, 1999). Until this becomes the basis of any discussion and critique, everyone involved is just skimming the surface. There is no doubt whatsoever that having at least one branch of any legislature picked by effective random selection techniques would be far more representative of what the citizenry really needs and wants than even an honest election, which is rare to non existent.

    Dr, Ted Becker, Auburn University. USA

    Liked by 3 people

  2. Hi Ted,

    I very much agree. This seems to be a recurring phenomenon – authors appear to be thinking that they can dismiss the idea of sortition by putting together their hastily cobbled-together superficial thoughts. For newspaper pundits this is only to be expected given that this is their profession’s standard procedure. Academics, on the other hand, pretend to have higher standards. It should be expected that a minimum requirement for an academic to write about something would be to have familiarized themselves with the existing literature on the topic. Of course, since Political Science is so dominated by dogma, such an expectation is unrealistic.

    BTW, to me the indispensable academic work about sortition is The Principles of Representative Government by Bernard Manin.

    Liked by 2 people

  3. BTW, both Carson & Martin’s book and Manin’s book are listed in our book list, together with what is, I presume, your book – UN-VOTE for a new America.

    Liked by 1 person

  4. In the above quotes the two Dutch professors show a penchant for especially bad straw man arguments, and seem remarkably clueless.

    Liked by 1 person

  5. However, the two professors do make a good argument against “slant[ing] political rule toward experts.” (Assuming that Dutch political scientists like themselves are to be considered experts on something.)

    Liked by 1 person

  6. HJ Hofkirchner makes a decent comment at the Washington Post page. Following for your possible entertainment are my less than respectful three comments posted there a little while ago:

    Besides showing a knack for especially bad straw man arguments and confused reasoning, Eric Schiesser and Tom Van Der Meer seem remarkably clueless about the citizen jury, lottery, sortition tradition in democracy in Classical Athens, and in recent decades.

    According to the two Dutch professors randomly sampled juries should not be used because in Holland when “using randomly sampled juries only 5 to 10 percent of invited citizens participate, creating a small group of mostly educated, motivated, middle-aged, white, male participants not at all representative to the larger population.” It seems the two professors have managed to remain ignorant of the literature and the real world examples regarding how to avoid this problem. Apparently they have also managed to get their political science jobs without ever learning of the concept of stratified random sampling (as in samples that accurately reflect the population in terms of gender, age, racial identity and other things).

    I’ll have to post a part 2 to comment further on Eric Schiesser and Tom Van Der Meer string of nonsense due to running into a word limit, if more than one post is allowed.

    Part 2.

    According to the two Dutch professors (who wrote the above article) using randomly sampled juries to make decisions lacks democratic legitimacy because “representatives need to be accountable to citizens as a prerequisite for democratic legitimacy.” Really? The professors are confused/mistaken. They are taking a concept that matters in electoral democracy (politicians being accountable in the sense of having to win re-election) and applying it to jury assemblies where legitimacy is on a quite different basis, arguably a much better one. What makes jury assemblies legitimate is that they are random samples of the people, and as such microcosms of the people. This is much better than mere accountability (in the sense of having to win re-election). The reason accountability (having to win re-election) matters for elected legislatures is exactly that they are not a microcosm of the people (but are instead a political elite distinct from the people), and left to their own devices may very well have interests and wishes very different from those of the people.

    Trial juries are widely considered legitimate, even though the jurors are not accountable in the sense of having to win re-election. In fact, part of what makes trial juries legitimate is exactly that they are not punished or rewarded for the verdicts they reach by the general public, nor by the rich interests that often fund the political campaigns of politicians. Eric Schiesser and Tom Van Der Meer just don’t get it.

    It is really not all that complicated, at least for those who’s thinking has not become fossilized, and who bother to look up the answers that have been made to what appear to be the first objections that occurred to them.

    Part 3

    According to the article:
    “Would lottery-politicians be less susceptible to corruption, as Van Reybrouck claims? That seems unlikely. Precisely because lottery-politicians cannot be held accountable via the ballot box, the institutional safeguards against bribery that curtail our regular politicians would be of little use. Elections, after all, allow voters to throw the rascals out.”

    Where does one even begin to parse the flawed reasoning and alternative reality thoughts in this paragraph?

    Shall we abolish the trial jury because of the possibility of jury tampering? If the duo who wrote the article think so, they are in a tiny minority. Instead, jury tampering is a crime, and does not especially worry either lawyers or the public, though of course it needs to be prevented.

    Safeguards against jury tampering are not hard to imagine (or to look up). Make it a felony. Use secret ballot for such things as approving a law or not (secret ballot is what largely ended vote buying in popular election, as the buyer could no longer know they got what they paid for), keep the identities of jurors deciding on a law a secret til they vote on it …

    The need to raise campaign money brings corruption and elite rule into electoral democracy. As President Carter has said, the U.S. is an “oligarchy” in which “unlimited political bribery” has created “a complete subversion of our political system as a payoff to major contributors.” One truly has to live in an alternative reality not to have noticed this critique by Jimmy Carter and many others.

    Far from being the safeguard against bribery that the article apparently suggests, popular elections, at least in America where this article is appearing, are more or less based on political bribery and rich campaign donors.

    Word limit approaching – I’ll leave the rest of the article’s dubious claims to anyone else who wants to bother with them.

    Liked by 1 person

  7. I’m saddened that any critical assessment of the flaws of sortition is immediately dismissed on this forum as facile, dogmatic and nonsensical. Schleisser and van der Meer are right to point out that a sortition with only a 5-10% takeup will be unrepresentative and that stratified sampling will not overcome this. Sortition would have to be quasi-mandatory in order to achieve accurate statistical representation. The figure of several thousand participants is a product of the view that such a body would be an assembly (with active participation) rather than a jury. And the objection regarding ex-post corruption and (lack of accountability) is persuasive. If Kleroterians seriously want to influence debate in political science circles then we need to constructively engage with these criticisms, not just write them off.

    Like

  8. Keith, don’t be sad. This article’s commentary on sortition is quite foolish, and the authors seem primarily interested in doing a bit of a hatchet job rather than providing fair commentary about sortition and David Van Reybrouck’s book.

    Yes of course the general topics they touch on need to be fully and intelligently addressed: sample accuracy, sample size, accountability, democratic legitimacy, jury tampering, possible degeneration of jury assemblies into “politicking” and possible rule by the bureaucracy through influence over randomly sampled citizens. The problem is that these two professors address these topics foolishly and incompetently (when they go beyond a mere mention of them), and seem mostly interested in doing something of a hatchet job.

    You mention stratified random sampling in your defence of the article, but the article does not mention it. Instead the article tries to dismiss sortition on the basis of a straw man case – an apparently very dubious example of sortition in Holland, which (according to the article) is dominated by educated middle-aged white males. The article portrays this straw man as David’s position. An intelligent reasonable discussion on the topic would have pointed out that there are obvious steps to fix this, such as stratification and proportional rep regarding gender, age and race, and such as paying jurors fairly well (so that most would wish to serve). They could also have looked at examples of sortition that are not dominated by educated middle-aged white males. Now yes, the problem may be deeper than simple obvious fixes, but this article does not go there.

    Regarding the article’s reference to several thousand jurors (assembly members), the article is trying to dismiss sortition on the basis of a utopian standard of exact near 100% sample accuracy, which would according to the article require jury assemblies of several thousand citizens. Yes sample accuracy matters and is an issue, however the question is not whether sortition meets a utopian standard of exact accuracy, but whether a well-designed system of sortition is better than the alternatives.

    The Athenian jury courts, boule and legislative juries were not exact microcosms of the demos. Does that mean Athens should instead have been for example an aristocracy, or run entirely by the ekklesia? Of course not. The question is not whether sortition meets a utopian standard, but whether or not it is better than the alternatives. (Yes we want sortition to be as well designed as possible, but the question is not whether it meets a utopian standard, but is instead whether or not it is better than the alternatives.)

    Yes corruption and bribery are a concern, just as they are at trial juries, and steps need to be taken. I find the article’s claim that electoral democracy is good/better at eliminating corruption/bribery daft in the case of the U.S. for the reason I very briefly indicated.

    With regard to accountability, I’m sorry, what the article says is foolish and shows the authors don’t get it. Yes jurors need to be held accountable for things like taking bribes or not showing up to work, but the article is ridiculously suggesting the kind of accountability politicians have (standing for re-election, accountability to public opinion for the decisions they make) is needed for democratic legitimacy. In particular the article says, sortition “fail[s] to acknowledge that representatives need to be accountable to citizens as a prerequisite for democratic legitimacy.” Sortition of course provides a different kind of democratic legitimacy from that of popular election (arguably a much better one). Athenian legislative juries did not lack democratic legitimacy because they were not accountable to, or disagreed with an opinion expressed by, a majority in the ekklesia – on the contrary their whole democratic point was to put laws on a more informed and considered democratic basis than a popular vote (a vote in the ekklesia)! (Similarly, the democratic legitimacy of trial juries is based on their suitability for providing informed fair verdicts from our “peers.” If trial juries were instead somehow accountable to public opinion, somehow catered to the uninformed views of the public as to what the verdict should be, they would lack democratic legitimacy. This is not an obscure point. It is not just that the authors of the article are wrong in thinking that juries lack democratic legitimacy because they are not accountable to the public, but that they also cluelessly fail to see that the democratic legitimacy of juries is in part exactly that they are not accountable to the public. Their thinking is somehow too “fossilized” for them to see what should be obvious to them, or maybe they were just too caught with doing a hatchet job.)

    Also, the view that politicians are meaningfully accountable to the people through popular election is quite dubious. Do the public really exercise anything like informed rule by the people through holding politicians to account in popular elections? Or, for example, is reality more like what is argued by the two political scientists who wrote the recent book Democracy for Realists: “the familiar ideal of thoughtful citizens steering the ship of state from the voting booth is fundamentally misguided.”

    The article’s first comment on David’s book is misleading or at least lacking in nuance: “Belgian author and cultural historian David Van Reybrouck suggests abolishing elections and appointing representatives by lottery instead.” Instead, David speaks of a “bi-represntative” system, with an elected legislature, and a sortition legislature. He for example says (p 160) “In a bi-representative Belgium, the Senate could consist purely of citizens chosen by lot, while the lower house could continue to accommodate elected citizens.” David’s agenda (at least for the near future) appears to be “bi-representative” government, not simply “abolishing elections” and replacing them with sortition.

    Anyway, if you got to the end of this tract, thanks.

    Like

  9. Simon,

    The concerns that the authors raise are shared by the vast majority of political scientists, hence my sadness on account of their airy dismissal on this forum.

    >Instead the article tries to dismiss sortition on the basis of a straw man case – an apparently very dubious example of sortition in Holland.

    On the contrary, it is entirely typical. In the case of the BC constitutional assembly only a tiny minority of those invited responded, and the resulting sortition was based on a sample pool that was heavily skewed in favour of activists and political anoraks. No amount of stratified sampling would solve this, the only solution is quasi-mandatory participation.

    >the article is trying to dismiss sortition on the basis of a utopian standard of exact near 100% sample accuracy, which would according to the article require jury assemblies of several thousand citizens.

    It’s not utopian, it’s merely a concern that the decisions of the sample should reflect those of the target population. As you know this is my prime concern and is the reason why I restrict the mandate of the sample to listening to balanced advocacy followed by the secret vote. Van Reybrouck, by contrast, advocates full-mandate sortition and this would require very large samples in order to ensure representative speech acts (as only a tiny minority would choose to speak, as in the Athenian assembly).

    >The article’s first comment on David’s book is misleading [suggesting] David Van Reybrouck suggests abolishing elections and appointing representatives by lottery instead.

    The misunderstanding is caused by David’s choice of a polemical title (Against Elections) and this folly is replicated on a daily basis on this forum by those arguing to replace elections with sortition and other forms of aleatory fundamentatism.

    Liked by 1 person

  10. Simon,

    I agree with the bulk of your criticism of the authors, except for the suggestion that it was an intentional “hatchet job.” I suspect their ill-informed and unthoughtful arguments for dismissing sortition closely parallel the reason that juries are better than public opinion polls (public judgment rather than public opinion). It seems that the authors dashed off an off-the-cuff reaction to the concept from a mental place so deeply immersed in electoralism that they couldn’t even appreciate the error of certain assumptions (such as their silly accountability argument). Their article would have been more intellectually defensible if they had said “here are our concerns about sortition, and details that need to be carefully thought through in developing a good design. We unfortunately do not have the time or interest to think these things through ourselves, nor consult others who have.” I think Keith is right, however, that this bias is widespread.

    Liked by 3 people

  11. > [“]We unfortunately do not have the time or interest to think these things through ourselves, nor consult others who have.”

    Heheh.

    Liked by 1 person

  12. *** The article the Dutch professors allude to is an interesting essay by Jason Brennan, distinguished ethicist (blog.press.princeton.edu ; June 24, 2016) « Brexit, Democracy, and Epistocracy ». He wants to reduce as possible the political power of the under-educated citizens. « In some sense, republican democracy, with checks and balances, was meant to do just that. And to a significant degree it succeeds. But perhaps a new system, epistocracy, could do even better. » The epistocracy he dreams is actually the current system of our polyarchies, with only a difference : « An epistocracy might grant some people additional voting power, or might restrict the right to vote only to those that could pass a very basic test of political knowledge. »
    *** I doubt Jason Brennan’s proposal will be successful. Specially in its second form : a strength of the polyarchic system is its claim to give to every citizen a say in the political power through the election process. Defranchizing a sizable part of the citizens would be a very dangerous move for the system.
    *** But Jason Brennan’s essay actually develops ideas which, in a less systematic form, were to be found in many comments following the Brexit referendum.
    *** The new technologies get back the specter of dêmokratia. Against it, two strategies are possible. Attacking the democratic principle itself, because it is becoming dangerous – this is Brennan’s option. Add to the polyarchic system some elements akin to dêmokratia, for example sortition, but mixed into the more complex system possible so as to neutralize the democratic venom – this is the other strategy, which has its own risks.
    *** We may note the candor of Jason Brennan: « In some sense, republican democracy, with checks and balances, was meant to do just that ».

    Like

  13. Brennan has a book called “Against Democracy”… http://press.princeton.edu/titles/10843.html. Ironically, the designer of the book cover thought that somehow this idea is best represented by having an x-ed-out ballot box.

    Like

  14. >Add to the polyarchic system some elements akin to dêmokratia, for example sortition, but mixed into the more complex system possible so as to neutralize the democratic venom – this is the other strategy, which has its own risks.

    Yes it does have risks, but likely to be less than siren calls to abolish elections and would be more likely to be acceptable to both political elites and existing voters, especially if the argument can be made for consulting the considered opinion of the demos via minipublics. There is a wonderful opportunity here, so let’s not blow it.

    Like

  15. Keith, I agree. Any experiment of minipublics is a good step. Actually, the risks I mentioned were the risks taken by those who expect to neutralize the democratic venom.

    Like

  16. *** The book of Jason Brennan will be available soon. But considering the essay on the Princeton University Press blog, Jason Brennan can accept mass elections, at least inside a convenient constitution, if the under-educated classes are defranchized. He is a modern Aristotle. We may remember that Aristotle did not care about the progress of sortition in the Second Athenian Democracy (he even omits to mention the legislative juries), because his one concern was to exclude as strongly as possible from power the lower classes.
    *** Jason Brennan in his essay does not propose to reserve the political power to experts, even to « general experts », philosophers-kings. He is a modern Aristotle, not a modern Plato.

    Like

  17. @Keith: “Schleisser and van der Meer are right to point out that a sortition with only a 5-10% takeup will be unrepresentative and that stratified sampling will not overcome this.”

    Can you provide us with evidence on this assertion?

    Isn’t it falsified by the processes run by New Democracy in Australia? For their Melbourne project they describe a rather simple and practical method to ensure a properly stratified sample. We intend to stick to it closely in Vienna.

    http://www.newdemocracy.com.au/ndf-work/183-city-of-melbourne-people-s-panel

    Like

  18. HJ:> Can you provide us with evidence on this assertion?

    The sample will, ex hypothesi, be skewed towards the sort of people who volunteer to participate in deliberative forums. The organisers may decide to stratify this pool of volunteers according to whatever criteria they deem salient* (age, gender, ethnicity, occupation, income, religion, voting history, newspaper subscription etc) but this cannot overcome the obvious fact that the sample will over-represent those who are interested in the topic that the forum will be considering, and that they will not be typical of the target population in terms of their psychological disposition.

    A new book by psephologists Philip Cowley and Robert Ford is of relevance: https://www.amazon.co.uk/More-Sex-Lies-Ballot-Box/dp/1785900900/ref=sr_1_1?s=books&ie=UTF8&qid=1472632330&sr=1-1 Young people are far less likely to vote than retirees and this is generally assumed to be because they are turned off by adversarial politics. Cowley and Ford’s research indicated that this was not the case — young people don’t vote because they simply “can’t be arsed” and the views of the small number of politically-engaged anoraks (who would be far more likely to respond to a voluntary sortition) do not reflect those of their peers. Voluntarism slants the sample towards the left as it’s generally “progressives” who are dissatisfied with the status quo.

    The following is from the book Deliberative Mini-publics. Even in the case of Mansbridge’s “gold standard”, the Deliberative Poll, around 75% of those invited to participate decline the invitation:

    “We have some knowledge of the biases in who turns up. For instance, participants are typically significantly more knowledgeable than the non-participants (Westwood and Sood 2010). That suggests that political interest plays a role in the decision (not to attend deliberative polls. Regularly failing to include those not interested in politics along with the chasmic differences that sometimes appear on socio-demographic variables undercut claims to a reliable process that guarantees [representative] inclusion.” (O’Flynn & Sood, 2014, p. 45)

    And the problem gets even worse once the deliberation begins:

    [t]hose who participate may not be able to participate on an equal footing. The garrulous, those strongly attached to their views, those who think the issue is important, the self-righteously knowledgeable, among many species, all prefer talking to listening, often at the expense of giving the others the chance to air their views. Then there are those who, even when they have the opportunity to talk, talk very little. (O’Flynn & Sood, 2014, p. 46)

    In sum, in order to guarantee the accuracy of statistical representation, three things are necessary. 1) participation has to be quasi-mandatory; all the sample can do is 2) listen to balanced information/advocacy and then 3) decide the outcome via the secret ballot. Deviate from any of these three criteria and the sample loses its only claim to political legitimacy — that it accurately represents it’s target population.

    *Anne Phillips’ Politics of Presence focused on gender, ethnicity and sexual orientation as these were the minority interests favoured by the New Left at the time.

    Like

  19. Thanks Keith, this gives us more to work from.

    Keeping it to the practical:

    1. Why would it be bad that the stratified committee members are more interested and knowledgeable than the average GenPop on a specific topic? We need to keep in mind that a proper demarchy will have thousands of parallel decision committees, and there will be topics for everyone, where they are interested and knowledgeable.

    The opposite is true: being interested ensures intrinsic interest in participating (much better than police turning up to drag you to a mandatory committee) and more pre-existing knowledge will save time and produce better decisions. If stratification from GenPop is otherwise perfect, a complaint that the sample is not “typical” for this one highly desirable difference is grotesque!

    2. Why do we care that “young people are less likely to vote”? Look at the US, why should they be hopping up and down to vote for the privilege to be “represented” by either of two octogenarians? Look at Austria, why should they be excited to vote for a president from either the far left or the far right?

    We do know on the other hand that young people are a huge political force when it matters. The student unrests in Prague 1968, flower power, the fall of the iron curtain, the unification of Germany. It was not octogenarians who climbed the Berlin wall or crossed the barbed wire from Hungary into Austria.

    Morale: we need not fix problems which do not exist. We do not need to force anybody to do anything. We are too long on theory and too short on empirical feedback. Let’s turn to the real priority at hand. Let’s get as many real-life demarchy projects underway as possible.

    Like

  20. 1. I don’t know what a GenPop is but assume it’s a random opinion sample. Democratic norms presuppose that every citizen should have an equal vote in political decision-making (Dahl, 1989, p. 109), irrespective of whether or not they have an active interest in the area. Voluntarism unfairly privileges those who want to change things. Conservatives are generally referred to as the stupid party, as those of a conservative disposition (in Oakeshott’s sense of the term) are not particularly interested in meddling in anything beyond their immediate sphere of operation. But that is not a valid reason for overriding Dahl’s criterion, which is generally regarded as the sine qua non of democratic legitimacy. All votes are of equal weight in a referendum but it is poorly informed; the goal of sortition theorists should be to turn every issue into a referendum that is executed by a well-informed microcosm that represents “what everybody would think under good conditions” and that precludes any element of voluntarism. You argue that demarchy “saves time and makes better decisions”, but ensuring that the trains ran on time was also Mussolini’s justification of fascism.

    2. Regarding student activism, students in the 1980s were a small subset of their age cohort and only a minority of them were involved in the protest movement. While most people would agree that the fall of the iron curtain was a Good Thing, some of the other progressive causes that you list (e.g. flower power) might be greeted in a more equivocal way. Ditto for the unification of Germany, as Bismarck’s original strategy led directly to the First World War (Goethe is very instructive on the merits of the old Germany). But what you or I might think of progressive movements is irrelevant from the perspective of democratic theory — or is that not a normative project that interests you?

    Like

  21. I strongly protest against your Mussolini statement and will not continue a discussion lacking in politeness and mutual respect.

    Like

  22. HJ.

    It was merely a comment on your utilitarian criteria (saving time and producing “better” decisions), it was not intended to be disrespectful at all. Bear in mind that utility is not a claim generally made by democratic theorists (Helene Landemore being the exception that proves the rule). Italian theorists like Nadia Urbinati are particularly insistent about the need to separate democratic norms from utilitarian factors (including epistemic “rightness”), so my comment was entirely in line with academic criteria on robust, but respectful, debate. The convenor of this forum is also keen to separate democratic legitimacy from epistemic considerations, although he has more difficulty in observing the need for politeness and mutual respect when dealing with some of his interlocutors.

    Like

  23. PS I appreciate that the perspective on demarchy of your Vienna Circle is very different from that of John Burnheim, but John has finally acknowledged that demarchy is not a form of democracy, it is just a tool for deriving “better public policy” (a goal shared with both Plato and Mussolini).

    Like

  24. Apology accepted, thanks. I regret if my reference to a better policy process triggered associations with a dictator.

    I am not aware of any generally agreed, stable definition of democracy. Our circle is partial to Karl Popper’s, which I paraphrase as follows: “Democracy is the opposite to dictatorship or tyranny. It gives the people control of public policy and leadership. People can change either without the need for violence.”

    For our New Demarchy concept “change” implies a change of public policy towards the better and in good time, actively and freely controlled by the people. It certainly does not revert to dictatorship to achieve these desirable features, thus firmly fits into Popper’s definition.

    Like

  25. Popper defined everything (including both liberty and the process of scientific enquiry) in negative terms, and he is not a mainstream political theorist. The standard analytic definition of democracy is outlined Dahl’s Democracy and Its Critics (1989), Chapter 8. Political associations have to take binding decisions and “the process for making binding decisions includes at least two analytically distinguishing stages: setting the agenda and deciding the outcome.” (Dahl, 1989, p. 107). For the decision process to be considered a democratic one, this presupposes that in the first stage:

    “The demos must have the exclusive opportunity to decide how matters are to be placed on the agenda of matters that are to be decided by means of the democratic process.” (ibid., p. 113)

    Whereas

    “At the decisive stage of collective decisions, each citizen must be ensured an equal opportunity to express a choice that will be counted as equal in weight to the choice expressed by any other citizen. In determining outcomes at the decisive stage, these choices, and only these choices, must be taken into account.”

    No reference is made to utility, and the theory has no epistemic or consequential entailments. For example a political community could take very bad decisions, including the enslavement of minorities, and still be considered democratic. Although the definition is derived from direct democratic practice (viz. 5th century Athens), it is assumed that in large-scale modern states some form of representation is necessary, although Dahl is prepared to consider statistical representation as democratic (p.340), and most proposals for mini-publics derive their authority from this.

    Like

  26. The snippet which you cite from Dahl only makes sense if such choices are better for the people than tyrannical ones. Not expressly but implicitly.

    Do I understand you correctly, that in our increasingly specialised and complex world:

    You reject decisions by mini-publics (committees) of stratified citizens who are interested and knowledgeable about the choice’s topic?

    You propose choices made by each and every citizen of which the majority will be disinterested and ignorant of the topic to be decided?

    Like

  27. Dahl’s concern (which he shares with other democratic theorists) is with equality as an intrinsic norm, not with making “better decisions” (that’s the province of epistemic theorists, like Plato and Landemore.) Whether tyranny or democracy leads to decisions that are better for the people is a separate consideration. A benevolent dictator might well believe that her decisions are “better for the people” than decisions arrived at by democratic procedures.

    Yes I reject your first proposal (viz. my ongoing dispute with John Burnheim) iff a) the committees are voluntary and b) information and advocacy is endogenous. And I also reject the second because decision-making by nomothetai-style juries will be properly informed by the debate. I certainly don’t advocate referendums (choices made by each and every citizen), my proposal is for decision making by statistically representative minipublics (a procedure rejected by Burnheim-style demarchists).

    (PS I think you mean uninterested rather than disinterested. Disinterest (impartiality) is something to be valued in decision-makers.)

    Like

  28. If you propose mini-publics, we may quite close. Statistical stratification seems to be ok for you, too. I posit that the additional feature of ensuring higher knowledge of committee members will produce better choices for all.

    In Hayek’s sense an polity using this element would flourish more than the one which is not considering the need for better decisions.

    We saw the effect of “everybody is the same” regardless of better choices (by using of the knowledge in society) in the former East Block, just compare East and West Germany.

    Yes, everybody may have had the same in the East … next to nothing.

    Like

  29. My proposal is based on a direct analogy with the law courts (and the Athenian legislative courts). Knowledge, information and arguments are the function of the advocates and expert witnesses, whereas the role of the lay jury is to judge which advocates are more persuasive. This may or may not be optimal from an epistemic perspective, but it’s the only way to ensure that the decision of the minidemos is a reliable proxy for what everybody would think under good conditions. The problem with demarchy (at least in its Australian form) is the assumption that the decisions of randomly-selected “experts” will be accepted on the basis of the provenance of its members. The decisions can’t be judged on the basis of their “rightness” as a) this begs the question and b) in most cases the outcome will not (as Popper and Hayek argued) be known for a very long time, on account of the law of unintended consequences. As Zhou Enlai is reported to have said regarding the impact of the French Revolution — “it’s too soon to say”.

    PS I’m not too comfortable with stratified sampling — my preference would be to make the sample of a sufficient size (and participation mandatory), so that it will naturally include all the relevant strata and in (approximately) the right proportions.

    Like

  30. On stratified sampling:

    There is a tradeoff between deliberative group size and stratification. The number of “relevant” strata is not limited per se, but we do know that there is an optimal group size. It follows logically, that we need to pack the strata within that size, not the other way round.

    On voluntary participation:

    1. We have ample empirical data from prediction markets on this criterion. If we neglect to distill knowledgeable participants or if we force participation, predictions will be measurably far inferior (closer to random) than when allowing self-selection. Refer to Hofkirchner/Puleston: “Exploring the Science of Predictions”, ESOMAR Proceedings 2014.

    2. Mandatory participation in any mini-public where one possesses no knowledge and does not trust one’s own judgement would make a criminal act when abstaining from an activity which some may consider harmful to society.

    3. Citizens wouldn’t trust such a system once they realise how little knowledge is contained in forced committees.

    Like

  31. hjh,

    Promoting increased inclusion in mini-publics of those who are initially better informed is EXTREMELY problematic on many grounds (not least of which is the fact that there can be a difference of opinion about what constitutes valid knowledge). But there are some less intuitive reasons as well. People who have already formed opinions on an issue are LESS able to absorb new information (especially if it conflicts with their previous personal research and opinion) once on a jury, than are people who are “fresh” to the topic. The purpose of the jury is not to become expert, but to listen to the information from a wide range of (conflicting) experts and weigh what they have heard. Good mini-public process will likely deliver better decisions than those delivered by elected elites, NOT because they are better informed than the elites (though they will be much better informed than the typical layperson who is NOT serving on a jury), but because they are relatively unbiased by electoral imperatives (whether serving financial backers, party elites, a desire to cater to ill-informed and easily misled constituents, etc.

    My own notion is that a mini-public of MORE interested citizens may be appropriate for developing and refining proposals, but a much more accurately descriptive representative mini-public (semi-mandatory service) is better suited for making the final decision. A failing of both standard elected legislatures AND all-purpose sortition selected legislatures is that these distinct tasks are given to the same set of people, but optimal decision making requires that agenda setting, proposal development and final decision making should be by DIFFERENT bodies, even though ALL of them can be democratic in some way. The JUDGE of a proposal should not be the AUTHOR of that proposal. The JUDGE (jury) must be MOST like the general population (“gen pop”), While the AUTHOR should have a greater commitment of time and to research than the final jury, and thus may be more volunteer-like.

    Like

  32. Keith:
    “.. on account of the law of unintended consequences…”
    Sure, those and many exogenous factors may impact any outcome. We agree that empirical data itself cannot completely describe the results human action, thus empirical data cannot falsify these decisions. (s. Ludwig von Mises “Epistemological Problems of Economics”).

    A proxy for observation and measurement is to track predictions for the outcome of decisions which produces praxeological insights into the quality of the original decision much earlier, even for long-range problems.

    Like

  33. On stratified sampling, we simply don’t know what strata are appropriate for political decision making, but a jury of 1001 (Athenian juries were as large as 5001) with mandatory participation would likely include most of them on account of the invisible hand of the law of large numbers.

    On voluntary participation, you are ignoring the judicial analogy with its strict separation of function between knowledge providers and lay jurors. Or are you saying that lay people are just too stupid to discriminate between opposing expert arguments? If that’s true then we should give up on any political system that starts with the “dem” prefix. Your proposal for a knowledge-based sortition pool appears more like oligarchy (or aristocracy) than demarchy.

    Like

  34. Keith,
    About the Cowley and Ford book… Do you recommend it to an American? what about their previous one (that mentions UK politics in the subtitle)?

    Like

  35. tboricus:

    We have consensus on the exogenous proposal point. In the New Demarchy process, the person or part submitting a proposal is not allowed as a decision maker in the decision committee. It would constitute a conflict. PS: In Austrian and German law this goes without saying, e.g. it distinguishes sharply between Managing Board and Supervisory Board. I am aware that Anglo-Saxon boards are less strict in this delineation.

    “People who have already formed opinions on an issue are LESS able to absorb new information.”
    Watch out: the knowledge process does not measure somebody’s opinion on the issue at hand, but whether somebody’s opinion on prior issues in the relevant topics turned out to be correct or not. Big difference!

    Empirical results support that people who are more knowledgeable about a general topic absorb new information (about the proposal at hand) faster, have more foresight and decide more correctly (s. Andere Ericsson: “Peak – the Science of Expertise”, 2016). This has been the result for matters like chess, business, science, sports.

    Like

  36. Keith: “Athenian juries were as large as 5001”

    We like and are currently going for the New Democracy way of determining sample size by looking at a required confidence level and interval in relation to the total affected/interested public (hard to express in English “. The resulting numbers for 99% confidence (more between 30 and 50, or below 200 for even the largest countries) also are better group sizes for productive deliberative committees.

    ” …On voluntary participation, you are ignoring the judicial analogy …”

    The analogy does not work.

    While everybody should know and observe The Law and be clever about it, political proposals will touch highly complicated topics in constant flux, endlessly specialised with ever more depth every year, and – to make matters worse, often several of these are relevant for one single proposal. Think tax law, the financial crisis, genetic modification, …

    Do not forget, that there are many million people in GenPop. The headstart by those being layered (having more lots) by virtue of preexisting knowledge in the several topics touched by any one proposal (measured objectively by the empirical capability to predict correctly), is crucial if we want (like John Burnheim) that collective decisions become similarly good and efficient as those in private enterprise in a free market.

    I would not spend a minute of my time on working towards a new political system, if I had reason to doubt that it would produce better decisions than elected representatives. Lack of knowledge was already identified as a problem of unweighted sortition by Socrates in Athens. Today this problem is even bigger by several orders of magnitude.

    Like

  37. hjh,

    I think you are conflating different sorts of knowledge. The expertise discussed in the book “Peak” has relevance to consideration of HOW should we do X, but very little relevance to deciding WHETHER we should do X. We want to over-represent engineers in deciding the details of a bridge’s design, but it is not useful to favor engineers when deciding whether a bridge is worth building across the river at all.

    Like

  38. hjh,

    One of the problems of debating with you is that we don’t really understand your position (GenPop, layered sortition etc). I think you should author a post for this blog outlining who the Vienna Group is and what it proposes. I also have problems with talk of the right answer in politics, and the analogy with chess, sport, business etc. It is easy to operationalise “rightness” in these domains (i.e. winning a competitive struggle); less so in political decision-making beyond the survival of the polis. Operationalising Aristotelian notions like flourishing is very difficult, especially in the light of divergent currents in modern political and social thought.

    I’m attracted to the notion that sample size should be related to confidence level — could you give me a New Democracy URL to follow that up?

    >A proxy for observation and measurement is to track predictions for the outcome of decisions which produces praxeological insights into the quality of the original decision much earlier, even for long-range problems.

    OK, but this is a way of assessing the epistemic reliability of the providers of expert opinion, not decision makers. Such factors will determine the degree to which the views of the proposing advocate should be taken seriously. Terry and I agree 100% about the nature and composition of disposing (isonomic) panels (although we still differ on isegoria — the proposal right).

    Terry,

    “People who have already formed opinions on an issue are LESS able to absorb new information.”

    Yes indeed, and I would go further and say that self-identifying an opinion (via a speech act) makes it psychologically harder to modify it. E.M. Forster’s observation that we cannot not know what we think until we hear what we say is dysfunctional in a jury charged with arriving at a majority decision. That’s why Athenian juries were instructed to listen to the debate in silence before deciding the outcome.

    Cowley and Ford are UK-focused psephologists, not sure how far their observations can be generalised beyond parliamentary systems.

    Like

  39. @ tboricus, why do you assume that I assume that only engineers are required for a committee decision whether to build a bridge? I do not, if you read the thread.

    @ keith, “could you give me a New Democracy URL to follow that up?”
    Here is their design document, the relevant section in p.3-4. http://www.newdemocracy.com.au/docs/activeprojects/City%20of%20Melbourne_newDemocracy%20Foundation_Project%20Outline_July2014%20-%20Revised%20Final.pdf

    GenPop: General population
    Layered sortition: Algorithm to draw first from the last most underrepresented socioeconomic criterion in a group of volunteers.

    Sorry, next to no time to write a blog post in the month of September. You can always look up some of our pages, here’s a starting point: https://www.prediki.com/meta/The-New-Demarchy-Manifesto/

    We are focussing on getting empirical projects going, the field of sortition seems long on theory and short on experiments.

    “This is a way of assessing the epistemic reliability of the providers of expert opinion, not decision makers.”
    If you think it through, it is both, as the mechanism is by its very nature produces more reliable predictions than other methods.

    Gentlemen, this discussion is also mixing up in an unfortunate manner what are (unverifiable) personal opinions on something specific versus (falsifiable) subject matter knowledge in general. You will come to wrong conclusion when lumping them together.

    Like

  40. hjh

    Taken from the New Democracy document that you cite:

    “These statistical labels simply mean that, firstly, we can be 95% sure that the ‘descriptive match’ to the community would be repeated on any random sample.”

    That’s an extremely strong claim for a panel of only 43 members. How can it possibly be justified, given our ignorance of all the demographic factors relevant to the political decision-making process? Few statisticians would seriously consider a sample of less than a few hundred, some insisting on a minimum of 1,000 (along with quasi-mandatory participation). If this were not the case then the surveying of public opinion would be considerably cheaper! I suppose the answer is that the only factors that are considered relevant by the New Democracy folk are (crude) census categories, but this strikes me as a crass form of socio-economic determinism that makes no allowance for ideational differences within each socio-economic category. Marxism certainly has a very long tail.

    Like

  41. @keith – we are quite comfortable with the New Democracy design, for several reasons.

    1. The calculation is correct.

    2. You are right that traditional surveys use those large sample sizes. But how good is their science when it comes to predictions? Look at the recent verifications, with Popper’s eyes: Brexit forecast: wrong, UK election forecasts: 7% off with 48 (!) traditional studies. Hamburg Olympics: a huge 3,000 sample survey one day (!) before claimed 58% “Yes”, the empirical truth was only 48% but 52% “No”.

    3. Other methods, e.g. the prediction market methodology, work with much smaller groups of knowledgeable, self-selected people. Such a distilled crowd of 30 to 50 traders produces more accurate and reliable results than the biggest surveys. (re “cheaper”: sample cost is not the deciding factor.)

    4. Also, they mention a study showing a declining deliberation quality when growing group size beyond that number. You may want to check it our. As a result, 1,000 instead of their 43 would damage deliberation, in practice.

    In summary we consider the New Democracy design as quite plausible, given that they are running a deliberative process before asking for votes. Small difference: we aim for a 99% confidence level, rather than 95%.

    I still do not understand why you say that the decision quality matters not. From our philosophy standpoint, operationalising the intended consequences of a policy decision is actually the key step to render it falsifiable.

    Like

  42. The confidence level is 95%, but the confidence *interval* is 15%. So, +/- 15% over 95%. In other words, if the ideal (post-deliberation) population support for a measure were 50%, the panel would have a 95% chance of returning a result between 35% and 65%. Statistics is not my forte, but for a 43 person sampling this seems perfectly reasonable. If one believes a robust deliberative process should consistently produce the same broadly agreeable consensus position on any particular matter, then this might be a perfectly acceptable trade-off in the interest of improving the quality of deliberation. This is more or less what the paper says. Such a belief is, however, difficult to justify.

    “Small difference: we aim for a 99% confidence level, rather than 95%.”

    What’s the target confidence interval? Any sampling can be said to have a 99% confidence level regardless of its size. You just have to increase the confidence interval enough to make the math work. You pick one of the terms and solve for the other. Or you set both, and solve for the necessary sample size.

    Like

  43. @ noami: Their paper says they require an 80% yes vote from their sample, so we can apply the confidence interval to that, not to the 95% which is just the confidence level. For their parameters it gives a a 95% chance of roughly a 65% yes vote in the GenPop.

    I like this remaining safety margin margin over simple majority.

    First, a 51% deliberated vote (without politicians’ slogans dumbing the masses) is actually not very comforting.

    Second, as Keith said, we should allow for ideational differences within the SOE categories. From what we see, these differences will be noise to some degree which of course cancels out. E.g. assume on Brexit the old category has more Leavers and the young more Remainers. Yes by chance we could get an old Remainer into our sample, but just as well a young Leaver. So the signal, on average, will still be good. That’s why we oversample at the 99% confidence level.

    They went for a 100 sample more recently, so i hope they put out some empirical data on decision quality in larger deliberative groups, soon.

    Like

  44. Hjh

    I confess I’m baffled by all this talk of the “right” answer in political decision making. To take a pertinent example, I voted for Brexit in the full knowledge that this would almost certainly lead to an economic downturn that could last as long as 10 years. But I voted for Brexit because I believe that national self-sufficiency and cultural integrity are more important. So two observations:

    1. Which is the “right” answer, when it comes to pitting economic against cultural factors?

    2. My vote puts me at odds with my census group, but not when it comes to age. However my son (age 27, M.Eng. Oxon, Phd. Tokyo) also voted for Brexit so he is an outlier in every respect. So how can you claim that a random stratified sample of only 43 persons would have a 99% confidence level of representing the informed and considered opinion of the target population? I can’t imagine that my son and I are totally unique.

    Like

  45. >Yes by chance we could get an old Remainer into our sample, but just as well a young Leaver. So the signal, on average, will still be good.

    So you’re claiming that the law of large numbers pertains to a sample of 43? Although my knowledge of statistics is even less than Naomi’s (and I think she is being a little modest), this strikes me as nonsense. If it were true then why do public opinion polls always run to several thousand?

    Like

  46. >>If it were true then why do public opinion polls always run to several thousand?

    Well, this is a good question to ask the traditional opinion poll companies.

    We are focusing on the prediction market methodology because it strictly adheres to Karl Popper’ prescription that every statement of scientific value must be falsifiable, we require this actually for each and every single response.

    A traditional poll asking “How would YOU vote if elections were next Sunday?” is not falsifiable, not on aggregate level, let alone on a respondent level. So we regard it as pseudo-research. No offence to any traditional researchers here, just looking at their terrible empirical track record. Garbage in, garbage out.

    We always ask “How will VOTERS vote on election day?” For our large-scale commercial polling projects accuracy is ca. 60% better than traditional polls, as measured by MAE (mean absolute error).

    >> I’m baffled by all this talk of the “right” answer in political decision making.

    The “right answer” is much simpler to evaluate than you seem to think it is. On your Brexit example, we just need to operationalise all the fantastic claims which each side tries on us. The economy will collapse! Well, predict specifically when and how much and then we will see what really happens. We will have 350 million more a week for NHS! Check it, after Brexit! These predictions we can record on an individual level and on committee level, and in fact for every citizen who cares to make one.

    (I do not think however that the Brexit example is very pertinent for our vision of the future. A demarchy would have never asked citizens such a crude question of huge import and complexity and forced them to a binary Yes-No answer.)

    Decisions following this fundamental principle are perfectly falsifiable. Once you have a collection of these measurements, you will know in the future if the committee decision fulfilled all, some, or none of their predictions. In the “some” case, the overall judgement will be a trade-off, for everybody to judge individually. Still we can easily measure the distribution of individual judgements in the committee or even the GenPop and get a sense of each decision’s quality.

    Like

  47. hjh:

    >it strictly adheres to Karl Popper’ prescription that every statement of scientific value must be falsifiable

    I’m gobsmacked that you think such an approach is applicable to political decision making. How do you operationalize something that is primarily a question of moral judgment (Rousseau would be rolling in his grave) and how can you possibly compare economic and cultural considerations? That would require some sort of common currency, and there is none available. Or are you resorting to the old Vienna Circle (logical positivist) view that any unoperationalized statement is simply meaningless?

    Like

  48. @keith
    “the old Vienna Circle (logical positivist) view that any unoperationalized statement is simply meaningless”
    You are correct, that would be classified “pseudo-statements”.

    “how can you possibly compare economic and cultural considerations?”
    Everybody does this every day when deciding even on ordinary actions.

    We do not think that a representative politician should do this comparison FOR us, simply because they are not representative of us. I guess you do not either, you did not even want to accept voluntary committees. We think that demarchic committees, slightly improved from the New Democracy version for the foresight requirement, are legitimate to make such choices, quasi-representative, measurable, and statistically falsifiable.

    Like

  49. Quasi-representative is not good enough for a descriptively-legislative body, it has to be a reliable proxy for what everybody would think under good conditions. This is testable, in that it would be possible for a number of allotted bodies to deliberate in parallel. If they all came to the same decision then the decision would be held to represent that of the target population. As to what variables would need to be controlled we can only speculate, my best guess being:

    1. An assembly of 501 (the minimum size of an Athenian jury) to 1,001.
    2. Quasi-mandatory participation.
    3. Balanced information and advocacy provided exogenously.

    If different samples came to different conclusions then it would not be possible to say which decision was the representative one, especially when pre-deliberative opinion is evenly split (as in Brexit). That different samples of 43 volunteers with full mandate-deliberation should come up with the same answer is highly unlikely. This is an entirely testable issue, in that sample size and the necessary controlling factors can be determined experimentally. It would be up to statisticians to determine the relationship between sample size and majority threshold.

    Like

  50. This is starting to go in circles and a bit axiomatic to boot, so I guess all was said. It was nice discussing with you. Thank you.

    You do not need a statistician, it’s really trivial. Just use this:
    https://www.mccallum-layton.co.uk/tools/statistic-calculators/confidence-interval-for-mean-calculator/

    Like

  51. hjh

    Thank you. Forgive my total ignorance, but I assume that in the term “true population mean” (from the calculator that you link to) the population referred to is the survey participants, as opposed to the population from which the sample was taken. Statisticians that I have discussed this issue with insist that a proportional sample of 1,000 is the minimum required for a reasonable level of confidence that the sample accurately represents the target population. Most customers for public opinion surveys would be happy with a 95% confidence level, so would (presumably) rather pay a relatively small price for 200 participants if that is all that is necessary to know the state of public opinion to a 95% level of confidence.

    Like

  52. You are a victim of pseudo-science.

    Traditional pollsters try to improve their bad track record by increasing case numbers beyond the mathematical requirement. Stil, it helps not. Their problem is methodology. I explained it above: garbage in, garbage out. They are just doing more of the same, increasing the amount of pseudo-statements from respondents, hoping for a better result.

    These same statisticians took a 3,000 Hamburg sample the day before a referendum and predicted 56% +/- 2% with a 95% confidence level. The true actual empirical result was 48%. You may wish to confront your statisticians with this reality.

    Liked by 1 person

  53. OK, so you are saying that your calculator generates a 95% confidence level that a sample of 200 accurately mirrors the views of the target population. So what is your confidence level for the New Democracy sample of 43 volunteers and what is its confidence interval? Remember that we are not interested in the accuracy of predictions we are seeking merely to understand whether a small sample would reflect what everyone would think under good conditions. The sample (and the target population) might well think that the moon is made of blue cheese — that’s not the issue that we are debating.

    Like

  54. The New Democracy team disclosed their calculations on p.4 (2). We checked them and they are correct.

    Like

  55. “So we regard it as pseudo-research. No offence to any traditional researchers here, just looking at their terrible empirical track record. Garbage in, garbage out.”

    Their track record may not be great, but they do a helluva lot better than +/- 15%. If that’s our standard, then even the most trashy polls are right on the money. Furthermore, the methodology used by polls (that you so criticise) is the same sort of stratification that would be used in these deliberative panels. The issue that polls have is that it’s extremely difficult to predict things like relationships between the various demographic factors, political preferences, poll participation rates, and turnout, making faithful stratification along such lines difficult.

    Like

  56. You misunderstand, Naomi. The methodology problem we criticise is not the stratification or the maths. They are good. The factors you list they can deal with, if you throw enough money at them.

    The problem is the non-falsifiable pseudo-statements which pollsters work with.

    The decision on an exogenous proposal by a member of a demarchic committee is a fundamentally different matter to a respondent’s answer answer to some pollster’s questions.

    Just like a citizen’s real vote on the real election day is a fundamentally different thing to his response to this pollster, when he calls at dinner time, which will often be biased or distorted by some signalling intent. There is a huge amount of literature on the topic so I do not need to expand.

    You also misunderstand the +/-15% which the New Democracy people work with. As they ask for an 80% supermajority, that 80% is the mean (!) expected agreement in the GenPop, and there is even a 95% likelihood that at least 65% would agree. Which, in your words, is a helluva lot better than the measly 51% needed to win a traditional elections.

    Like

  57. Sure. Respondents sometimes lie to pollsters, or change their mind before they vote. There is a fundamental difference between responding to a pollster’s question and actually voting. Not much can be done about that. This has nothing at all to do with the issue at hand.

    Bad sampling methodology will give you results that are generally skewed in the same way from poll to poll, panel to panel, or experiment to experiment. A large statistical margin of error will give you results that are generally skewed in random ways. The two are not related. Both must be managed. It’s the old accuracy vs precision issue (http://cdn.antarcticglaciers.org/wp-content/uploads/2013/11/precision_accuracy.png). Traditional polls have large sample sizes due to the need to minimize the effect of chance on the outcome. In other words, maximize precision. A polling system which showed a politician’s numbers bouncing up and down by five-to-ten points or so on an almost daily basis would be useless. Accuracy is harder to improve. And, yes, there are fundamental limits to what can be done due to the disconnect between polls and voting. But bumping up the sample size does nothing for pollsters who have poor accuracy. Nothing.

    The formal threshold for passage of legislation is not really significant in this context. 50%, 65%, 80%… for results around the threshold it’s always going to be a coin toss. If you have a confidence interval of +/-15% and a threshold of 50%, then a lot of things with an ideal support level of 40% will pass and a lot of things with an ideal support of 60% will fail to pass. Bump it up to 80%. Okay. What’s changed? Many things with an ideal population support level of 70% will pass. Many things with an idea population support level of 90% will fail to pass. Either way, the effect of simple chance on *policy outcomes* is huge. Increase the sample size and you have to get closer to the threshold before you start to see the effect of chance on the results.

    There is a case to be made for making the threshold high enough to keep later samplings from overturing the decisions of previous samplings due to random chance.

    Anyway, as I said before, if one expects broad consensus to be the norm on these panels, then yes, a very large degree of chance should be fine. You just need a threshold high enough to ensure that the desired consensus was found despite any statistical flukes. Keith and I strongly disagree with the notion that consensus would generally be found in these panels. As such we insist on bringing the confidence interval down as low as is practical. This is, I think, the main point of contention here. You say 65% is better than 51%, and I agree, if such a majority can be found. Designing a political system with the idea that such majorities can be found as a general rule is downright dangerous.

    Like

  58. Naomi:
    “Keith and I strongly disagree with the notion that consensus would generally be found in these panels. As such we insist on bringing the confidence interval down as low as is practical. This is, I think, the main point of contention here. You say 65% is better than 51%, and I agree, if such a majority can be found.”

    There seem to be two conflicting theories: One that a 65% consensus is normally doable, the other says, it cannot thus reduce the ambition to 51% only.

    By experimentation, we can see which theory has been falsified. The New Democracy team applies 80% successfully (see their documents in the public domain) and I had a direct report that the typical decision level in their projects is actually close to 95%.

    Ergo, we can be much more optimistic about mankind’s capability to reason and consensus once we establish better democratic processes.

    Like

  59. hjh: >I had a direct report that the typical decision level in their projects is actually close to 95%.

    But will the reasoned consensus still reflect that of the target population that it purports to represent or might different samples come to a different reasoned consensus? What is at issue here is the social psychology of small-group deliberation and arbitrary factors introduced by who gets to speak first and differences in the perlocutionary force of the speech acts involved. Anyone who has first-hand experience of a trial jury will know that much of the “consensus” is occasioned by social rather than epistemic pressure (the forceless force of the best argument). This is why the small-group sessions of deliberative polls require professional moderation, but that would not be possible in a forum with a legislative remit on account of the Quis Custodiet? problem. Fishkin tries to get round this by ending the DP with a secret vote (rather than show of hands in the small groups), but by that time the damage has already been done as the cognitive representiveness of the group will have been skewed.

    Like

  60. 1. Consensus Representation: We can and should design an experiment to measure the typical approval rates of decisions by politicians (e.g. Mrs May in the UK) , and that of a New Demarchy committee. Our theory is that it s will have an enormously stronger approval, simply because it captures much more of the Knowledge in Society and the process is more rational and objective than parties trying to stir the masses.

    2. Group Pressures: I am the first to criticise weaknesses of old-style focus groups for qualitative research. Still, there is a large body of evidence that group DECISIONS are superior to those of individuals, as long as we avoid that some group becomes entrenched. That’s why § 5 of the New Demarchy ruleset specifies: “Decisions are made by committees of members, each appointed strictly only for the time required to arrive at a decision on a proposal or a group of interconnected proposals.”

    3. Quis Custodiet: Secret vote goes without saying but do not suffice. Strictly neutral moderation techniques and rules must be observed too. There are nowadays measurably better moderation techniques which avoid groupthink. (You will not be surprised that they are connected with an implicit, prediction-oriented method.) The side disadvantaged by a rule-violating moderator would be expected object, we could even experiment with a meta-judge who exclusively supervises the moderator’s strict neutrality.

    I re-iterate: the Dutch professors could not be more wrong about the positive potential of demarchy for society. John Burnham’s vision of an informationally efficient collective decision mechanism can be implemented and should be before the next tyrannical period. The Brexit outcome is testament to the threat of the next democratic failure looming above our heads.

    Like

  61. 1. An experiment on approval ratings would be good. My hunch is that democratic norms are so deeply entrenched that people will not give up on the hard-won right to vote unless it can be demonstrated that the decision outcome would have been the same, irrespective of which individuals (including themselves) were included in the sample. Then everyone would be able to have their cake and eat it (i.e. decision making as the considered will of all the people). John Burnheim has acknowledged that his proposal for demarchy gives up on democracy and I trust you guys are prepared to do the same.

    2. The issue is not whether the group decision is “superior” or not, but whether it is representative of the target population. And the decision mechanism that Terry, Naomi and I propose is still a group decision, so long as the group is viewed as an aggregate of individuals rather than a collective.

    3. Quis Custodiet? I guess the model might be the Speaker of the House of Commons. But the individuals that she moderates have all been pre-selected for their rhetorical skills — not so in an assembly comprised of randomly-selected conscripts, where most people would have little or nothing to say.

    >The Brexit outcome is testament to the threat of the next democratic failure looming above our heads.

    That strikes me as a distinctly partisan statement (JB would agree with your verdict), and this is one of the reasons that I am suspicious of demarchy, which seeks to replace popular sovereignty with the rule of the great and the good (or wise in your case).

    Like

  62. On the topic of the relationship between sample size and majority threshold, I’m very grateful to hjh and Naomi for their exchange. As I understand it from https://www.mccallum-layton.co.uk/tools/statistic-calculators/sample-size-calculator/ a supermajority threshold of 60% (common for constitutional changes) from an electorate of 30 million would only require a sample of 271, whereas a simple majority would need a sample nearer to 2.5 million. This is at the highest confidence level (99.9%) which, as I understand it, is an arbitrary figure set by the experimenter, as the crucial variable is the margin of error (confidence interval). Please correct me if I am wrong.

    Presumably Condorcet conditions regarding independence of participants pertain, as there is no way of operationalizing the distortions introduced by group interactions. But I’m encouraged at the small sample size required for a 60% supermajority. If the Brexit referendum had been set at this threshold and a number of samples had reached the same decision having deliberated** in parallel (with the same information and advocacy input), it’s hard to see why anyone would object to that being a binding outcome (or at least as binding as the referendum result). If the outcome had still been 52/48 then Naomi’s argument for deferring the decision for a second round is persuasive (Condorcet would certainly have approved).

    ** in the Latinate sense of silently weighing the competing arguments.

    Like

  63. PS a 55% supermajority only requires a jury of 1,083, and 52% 6,766, so we are now in the ballpark of the Athenian nomothetai.

    Like

  64. Keith “A 55% supermajority only requires a jury of 1,083, and 52% 6,766, so we are now in the ballpark of the Athenian nomothetai.”

    Sure. Your maths mean that a 55% supermajority in the nomothetai corresponds to a 50,1% majority in the population. That is, if the Athenians had stratified correctly, which they did not, e.g. women did not (!) get to vote in that archaic rule-set.

    For a traditional democrat these 50,1% may be enough. Not for us.

    50% decisions — the deplorable holy grail of a party-based representative democracy – is rather a sign of a strong disagreement, a split right in the middle of society, a cause for worry, certainly not indicator of a rational, correct decision.

    Liked by 1 person

  65. hjh

    >Your maths mean that a 55% supermajority in the nomothetai corresponds to a 50,1% majority in the population.

    What does that mean — surely 50.1 is just the minimum, given the chosen confidence interval? That may not be epistemically ideal, nevertheless it’s the holy grail of democracy in general (not just the party-political variant). Whilst you may seek a larger majority it can only be at the expense of representational accuracy because the design for your council breaches Condorcet conditions. Your demarchic council may or may not come up with epistemically ideal conclusions, but it has nothing to do with democracy.

    >women did not (!) get to vote in that archaic rule-set

    For those of us using the nomothetai as a template it’s the representation principle that counts — the criteria for citizenship in the classical world is only of interest to historians.

    Like

  66. Keith: “What does that mean — surely 50.1 is just the minimum, given the chosen confidence interval?”

    For practical purposes, certainly. Strictly speaking, the minimum could be lower, outside your chosen confidence level.

    All of this is of course way better than the deal we get when we must trust in politicians.

    Like

  67. OK, that’s why I went for 99.9% confidence level and a 55% supermajority. This is all very encouraging — thanks for pointing to the calculator. Nobody on this site is arguing in favour of decision making by elected politicians — an intriguing result of the Brexit referendum is that we now have an elected government charged with implementing a headline policy that it disagrees with (and campaigned against).

    Like

  68. Much of the discussion about consensus and super-majorities seems confused. The notion that deliberation can lead to 65% or 80% consensus is only true SOMETIMES. Sometimes a group (even a fully informed one) may be almost evenly split. Not all decisions are prone to consensus. And the super-majority requirement is subject to manipulation by the framing of the question. A 60% threshold of a mini-public to leave the EU? Suppose it got 55%. it fails… no exit. Well, what if the question had simply been framed instead, “Shall we REMAIN in the EU?” It gets 45% so exit occurs. This is why the majority threshold is the only one that consistently works, because a defeat of a proposal is also the agreement to the status quo. NO decision is in fact a decision to maintain the status quo. Super-majorities give those on the minority side more voting strength than those on the majority side…. that is all votes are NOT equal.

    A super-majority scheme CAN work if it is coupled with some other decision-making system (like when a super-majority referendum vote can over-rule an elected legislature), but cannot stand alone.

    Liked by 1 person

  69. Terry,

    In the UK the electoral commission spends a long time agreeing the wording of questions with the relevant parties. The supermajority requirements we have been discussing here are not so much to do with conservative perspectives over constitutional issues as confidence in the reliability of the sample wrt to the target audience. Here’s the UK calculations from the online engine that hjh referenced:

    Margin of error Decision threshold Sample size
    2% 52/48 6,766
    5% 55/45 1,083
    10% 60/40 271

    The sample size for a simple majority is 2,526,148, so you might as well have a referendum.

    Like

  70. Keith,
    You can use a majority threshold, and have a modest sized deliberative body, then, If and ONLY if the outcome is super close, draw a new (perhaps larger -though not necessarily, because the repetition acts as a larger sample itself) mini-public as a confirmation. There is no need to have a massive body as a default (Naomi, you and I discussed this previously).

    Like

  71. Terry

    Obviously we want the body to be as small as possible, for both financial and rational-ignorance reasons but the outcome has to be demonstrable representative of the whole population (hence the above calculations). I imagine the size of the jury would depend a lot on how things worked out in practice, because re-trials are both expensive and time-consuming. In important cases (e.g. Brexit, constitutional amendments etc) larger bodies would be needed, as in the classical example, if only for the sake of public (as opposed to statistical) confidence.

    Like

  72. Roamingspotlight:

    > Perhaps a vote

    That’s an interesting idea — a vote to end voting. Needless to say those who get to vote to deny the vote to the hoi polloi are members of the cognitive elite (followers of EbL and readers of the Washington Post). Reaction comes in many different guises — especially when it comes to the war against “right-wing populism”.

    Like

  73. This just shows how careful one needs to design traditional questionnaires or they do not make much sense. “How are we to decide …” is an independent, technical question. The answer or proposal to this second question is a precondition. People who do not know the answer cannot sensible pick one of the offered options of the first question.

    Like

  74. But the important issue is who gets to answer the question. Democratic norms would suggest a representative sample of the hoi polloi, but then that will involve turkeys voting for Christmas as its a vote against democracy.

    Like

  75. About stratified sortition and mandatory jury service.
    *** Any amount of stratification along some objective dimensions: official sex, age, parameters of skin color, height, blood group, number of children, declared income, etc., will increase the representativity of a mini-public (even, from partial correlations, if the parameter has not clear direct effect on political sensitivity). But we will not know how much for a given kind of political decisions.
    *** And, as Keith Sutherland says, some dimensions will not be taken into account, or indirectly and badly, which may be very important in a kind of decisions. Therefore stratification cannot guarantee statistical representativity, and we may sometimes be very far from it if we establish non-mandatory citizen juries and trust too much stratification.
    *** In a multi-juries system, there is a potential problem of contradiction, which a dêmokratia must try to solve. The problem did exist in the Second Athenian Democracy : political maneuvers aside, the Leiptines’ law could be good for immediate financial reasons, and likewise be deemed bad on long term or constitutional reasons; thus voted by a (legislative) jury, and crushed by another (judicial) jury. But the problem of contradiction becomes much deeper if juries are far from representative, which would be probably the case in a beginning modern dêmokratia if presence is not compulsory.
    *** I specify “in a beginning” because it is dangerous to consider that the contemporary levels of political interest in our polyarchies are intemporal data. In a modern dêmokratia, the sense of political power could be much higher, and political interest could follow. Mandatory political juries could be especially necessary in the beginning of a modern dêmokratia.
    *** But anyway, mandatory jury is better. It was practically impossible in Athens – impossible to make presence compulsory for a peasant of Marathon, far from the city, at the harvest time, or for the sailor of a warship or commercial ship somewhere in the Aegean. Electronic technology has canceled the practical problem for a modern dêmokratia.
    *** Some think that Athenian juries were not mandatory for some “libertarian” reason of individual absolute freedom. It is wrong. There was in the Second Athenian Democracy mandatory military service for the 18 years people, and mandatory service as arbitrators of civil litigations for the 60 years people: the Athenian system included compulsory service, but it was not possible for people in working ages.

    Like

  76. Andre:

    >But the problem of contradiction becomes much deeper if juries are far from representative, which would be probably the case in a beginning modern dêmokratia if presence is not compulsory.

    Agreed, and “mandatory” is just short hand for saying all those who draw the golden ticket should attend. Whether this is by compulsion, bribery (ie salary), or cultural pressure is of no real concern. All that matters is accurate and consistent statistical representation.

    Like

  77. Andre,

    > mandatory jury is better

    The very opposite is true: mandatory participation is a horrible idea. Having people in the jury who do not feel under obligation to do the best job they can, who are only there because they have to, and who still have significant political power, is a sure way to erode confidence in the whole system.

    Like

  78. Andre: > the problem of contradiction becomes much deeper if juries are far from representative.

    Yoram: >mandatory participation is a sure way to erode confidence in the whole system.

    Why would the public be confident in a system that generated contradictory decisions? In the Brexit example, one non-mandatory jury might vote “out”, and another one “in”, and both juries may well have done the best job they could. If so, which decision would be the representative one? Note that this is an empirical issue (and highly likely in a population that voted 52/48), that cannot be resolved by means of a logical syllogism.

    Like

  79. PS delighted to see that this blog now has 500 followers — a significant number for all of us who seek the reincarnation of some form of Athenian democracy.

    Like

  80. *** Yoram Gat describes the jurymen of a modern dêmokratia conscripted without self-will as “people in the jury who do not feel under obligation to do the best job they can, who are only there because they have to”.
    *** This is a dichotomic and overly moralistic description of citizens. A citizen will be eager to participate because of strong sense of civic responsibility, a pleasant feeling of power (lesser if the jury is wide), political militancy or at least strong political consciousness, coming out of ordinary life, kantian ethics (my voice will not have big power, but if everybody reasons like that ..), etc … Other factors will act in the other sense: low involvement with the specific issue, small difficulties with the professional job or private life, idea that one’s voice in a big jury, or a string of small juries, will not really be significant, and, especially in the beginning of an “ortho-democracy”, feeling that political decisions don’t really matter, or unconscious idea that in some classes one is not able, and must leave political decisions to one’s betters.
    *** Non-mandatory political service will have two results. First, a jury will not be representative, it will not be really a mini-dêmos. Second, different juries will have different distributions of political sensitivities, because the mentioned factors for or against participation will act more or less depending of the specific issue; and the problem of contradiction between juries will be so heightened as to be impossible to solve.
    *** The idea that “conscripted” jurymen will act as bad citizens is founded on a simplistic basis. Many soldiers in the two World Wars fought well, who were not pure volunteers, but did accept compulsory equalitarian service.

    Like

  81. Andre:

    >Non-mandatory political service will have two results. First, a jury will not be representative, it will not be really a mini-dêmos. Second, different juries will have different distributions of political sensitivities.

    Exactly. Even in the case of the DP (the ‘gold standard’ of deliberative minipublics), 75% of those invited decline to participate: “We have some knowledge of the biases in who turns up. For instance, participants are typically significantly more knowledgeable than the non-participants (Westwood and Sood, 2010). This suggests that political interest plays a role in the decision (not) to attend deliberative polls. Regularly failing to include those not interested in politics along with the chasmic differences that sometimes appear on socio-demographic variables undercut claims to a reliable process that guarantees [representative] inclusion.” (Iain O’Flynn and Gaurav Sood, ‘What would Dahl say? An appraisal of the democratic credentials of deliberative polls and other mini-publics’, in Deliberative Mini-publics, ECPR Press, 2014).

    Like

  82. Andre: “Many soldiers in the two World Wars fought well, who were not pure volunteers, but did accept compulsory equalitarian service.”

    How can anybody forget that these citizens (most of them “made” soldiers through laws made by elected politicians) were all under threat of immediate execution if denying this (sic!) “equalitarian service”?

    How can anybody think of applying this inhumane principle to an improved democracy? Shoot them if they do not want to serve in a jury?

    It should give us pause if we need to force citizens “for their own good” to do something which should be in their own interest. Is it enlightened to take their freedom to decide for themselves if it is or if it is not? How often have we heard this “for their own good” in history as an argument for the worst crimes?

    Participation must be by free will or we are on the return path to tyranny, which — according to our pragmatic definition — is the very opposite of democracy.

    Like

  83. hjh:

    >How can anybody think of applying this inhumane principle to an improved democracy?

    It was only a loose analogy. Andre, Naomi and myself are merely arguing that political jury service should be a civic obligation on the same principle as the (English) trial jury. The puzzling thing is that Yoram and hjh don’t appear to believe that voluntarism would adversely affect the descriptive representativity of the jury when all the evidence shows that it does (see my earlier quote from O’Flynn and Sood). Descriptive representativity is the only reason for the use of random selection so all conceivable steps need to be taken to ensure that the sample is accurate (stratification will not suffice as we do not know what factors are salient for political decision making).

    >Shoot them if they do not want to serve in a jury?

    Interestingly Harrington did mandate capital punishment for any juror who spoke during the proceedings, presumably as it would have adversely affected the validity of the decision outcome (the context was the violent bloodshed of the English civil war).

    Like

  84. “The puzzling thing is that Yoram and hjh don’t appear to believe that voluntarism would adversely affect the descriptive representativity of the jury when all the evidence shows that it does (see my earlier quote from O’Flynn and Sood). ”

    There is nothing to believe. I know that does not adversely “affect” it because in a proper demarchic process representativity is a factor to be controlled by the sortition algorithm. The evidence cited evidences a bad process, no more.

    Like

  85. >representativity is a factor to be controlled by the sortition algorithm.

    Huh? If you don’t know all the myriad stratification factors involved how can you account for them in the algorithm? DPs have been running for over 20 years and are still subject to these problems (Sood is one of Fishkin’s co-authors) so the folk at the Stanford Center for Deliberative Democracy (which include serious statisticians like Bob Luskin) would dearly love to hear what your magic formula is.

    Like

  86. In the interest of progress, let’s keep this practical. A “myriad” of theoretical factors clearly makes no sense, that’s more than people on the planet. That’s a bad reason to start killing unwilling jury members.

    In reality the problem is quite basic. The typical complaint about the stratification mistakes is that participants were “mostly educated, motivated, middle-aged, white, male participants not representative of the larger population.”

    This shows that your citations did not get in trouble for failing to control “a myriad factors” but for ignoring the most obvious four: age, gender, race and education.

    The solution, in real life: The preparation phase for any specific decision will identify the few relevant citizen strata to an efficient level. No more, no less. The “below-efficiency” rest is up to statistical variation.

    Remember that John Burnheim’s crucial requirement was that in future, collective decisions should be as efficient as in the private economy.

    Like

  87. >collective decisions should be as efficient as in the private economy.

    Although demarchs may be interested in efficiency (along with sundry others who wanted to ensure that the trains run on time), democrats are concerned that the decisions of the minipublic should reflect what everyone would think under good conditions (Fishkin’s aspiration). Given that most people are not interested in participating in deliberative minipublics then a voluntary sortition will underrepresent the majority of the target population. If two minipublics produce two different conclusions then we will cannot say which one is the representative one.

    Like

  88. *** hjhofkirchner writes: “How can anybody forget that these citizens (most of them “made” soldiers through laws made by elected politicians) were all under threat of immediate execution if denying this (sic!) “equalitarian service”?”
    *** The threat explains obedience for an individual, not collective obedience of a big part of a people, and the armed part, at least without efficient secret police or totalitarian network; coercicion is only part of the explanation, it is not the lone explanation.
    *** Compulsory political service (which exists in many polyarchies for judicial juries, at least officially) would be “taking back the freedom of the citizens”; actually it diminishes individual freedom to heighten collective freedom – which has individual sides.
    *** For instance, if some day I am accused of a crime with strong ideological overtones, I would prefer to be judged by a jury of conscripted citizens rather than by a jury of volunteers, which could be packed by militants. And I accept to exchange this safety against some days of compulsory service as member of a jury. I will not consider that as infringement of my freedom, and I will not try to evade it, playing the “clandestine passenger”.
    *** From bad decisions by the political powers I must have to pay high taxes (I am not free to decline !), I may become jobless, I may become victim of a working accident because of bad standards, I may become target of mass killing by a terrorist movement, etc .Many things I am not free to decline. A good working of the political system is worth some political service.
    *** I favour sortition for sovereignty because it is the one process without any selection, and I am afraid of any kind of selection for sovereignty. I favour compulsory jury service because self-selection is a kind of selection.

    Like

  89. >self-selection is a kind of selection

    Absolutely, and this is probably the key stratum to be avoided when it comes to political decision making. In Arthur C. Clarke’s Songs of Distant Earth anyone who volunteered for the presidency was immediately disqualified.

    Like

  90. Andre,

    > *** Non-mandatory political service will have two results. First, a jury will not be representative, it will not be really a mini-dêmos. Second, different juries will have different distributions of political sensitivities, because the mentioned factors for or against participation will act more or less depending of the specific issue; and the problem of contradiction between juries will be so heightened as to be impossible to solve.

    While the objective should be universal voluntary service in allotted bodies, even if that is not achieved forcing unwilling citizens to serve is worse than having them not serve. Both issues you mention are manifested more acutely for unwilling jurists than they are for absent jurists:

    (1) Unwilling jurists will not express a considered and informed opinion and are thus worse than absent jurists,

    (2) Unwilling jurists will be affected by superficial, non-substantive factors and will therefore create a block whose votes are based on uncontrollable circumstances, yielding erratic results.

    In addition, the following problems will be created when forcing jurists to serve:

    (3) Situations in which unwilling jurists will behave in arbitrary, a-social ways will occur regularly and will undermine the credibility of the system,

    (4) The mandatory service will lift from the state the obligation to make sure that citizens are willing to serve and are enthusiastic about serving,

    (5) The mandatory nature of the service will, by itself, reduce the sense of obligation to perform good service.

    In sum – this is not even a matter of having a trade-off. There are zero benefits to mandatory service and there very strong arguments against it.

    > Many soldiers in the two World Wars fought well, who were not pure volunteers, but did accept compulsory equalitarian service.

    This is an inappropriate analogy. Conscripted soldiers have very little power. (They are like voters in a mass election.) One only rises to power in the military after having shown great commitment for the job. Would you put conscripted soldiers in positions of command? I doubt that any military has done so. Certainly not in a position of significant power, as an allotted legislator is in. That would be like putting a conscripted soldier in position of commanding a battalion or a division.

    Like

  91. Yoram,

    In English courts all jurors are unwilling conscripts but there is no evidence that they judge issues in an arbitrary or asocial manner. The main complaint against conscripted jurors is that they are too stupid to understand complex cases.

    >There are zero benefits to mandatory service

    Not so. Mandatory service is the only way of ensuring a descriptively-accurate sample, which would include lazy and unenthusiastic people in proportion to their prevalence in the target population (a virtue applauded by Callenbach and Philips). This is of epistemic value in a law-making body as enthusiasts and idealists are inclined to believe that everyone is like themselves. This is a reason for including convicted felons in the sample as they are in a better position to understand the criminal psyche than civic-minded volunteers.

    >Unwilling jurists will not express a considered and informed opinion.

    They don’t need to — their job is to weigh the evidence and arguments presented to them and then determine the outcome through the secret vote.

    Like

  92. *** I have strong doubts about the idea that stratification using only a few objective parameters– practically age, gender, education, race – will be enough to get statistical representativity with a jury system involving self-selection.
    *** But in a modern dêmokratia this idea would be falsifiable. Let’s convene one jury with compulsory service, and another one with self-selection and stratification along the “few relevant citizen strata”; and let’s see if the results are the same in a given field.
    *** Note that the proposed scheme cannot take into consideration parameters which are not easily measured with trustable results; for instance the level of religiosity (in a given denomination), the level of political militancy, the level of political self-education, the sexual orientation, the level of class consciousness (I am sorry, Keith, maybe the long shadow of marxism), or of race consciousness, the attitudes about social relationships, … If we suppose that any of these parameters is a strong factor, the simple stratification scheme will not work. But, as I said, that will be falsifiable.

    Like

  93. Just in case it wasn’t clear – I am not proposing any stratification. On that I am in full agreement with Andre. The idea that somehow such a technicality can eliminate the inherent bias of self-selection is completely unfounded. On the other hand it opens the door to manipulation and opacity through increased complexity.

    Like

  94. Andre: “Coercicion is only part of the explanation, it is not the lone explanation.”

    Agreed, we have known this reason for “voluntary” soldiers since Gustave Le Bon and Elias Canetti’s times. Tribal man in a mass is an emotional and irrational beast, only by intelligent mechanisms and civilised processes the mob can convert into a wise crowd.

    Andre: “For instance, if some day I am accused of a crime with strong ideological overtones, I would prefer to be judged by a jury of conscripted citizens rather than by a jury of volunteers, which could be packed by militants. “

    This needs empirical data. Intuitively, it should run the other way: mandatory conscripted citizens would statistically be part of the unthinking mob, who’d rather lynch you than give you a fair trial. Whereas volunteers might have more elevated opinions on civil rights, due process etc. Still, Vienna Demarchy certainly neither proposes a case-specific volunteerism nor direct self-selection, so this strand of the discussion misses the point entirely.

    Andre: “And I accept to exchange this safety against some days of compulsory service as member of a jury. I will not consider that as infringement of my freedom, and I will not try to evade it, playing the “clandestine passenger”.”

    Well, this answer works actually against mandatory force, just by logic. If jury service is such a good choice that “you do not try to evade it”, why make it mandatory in first place and shoot citizens if they do not want to serve, as suggested above?

    “A good working of the political system is worth some political service.”

    We differ that it can even be a good political system if it only works by excessive force. “Force” tends to be associated with highly undesirable political systems.

    @Keith: Thanks for stating a very important fact:
    “The main complaint against conscripted jurors is that they are too stupid to understand complex cases.”

    Has this board thought about the implications of this? Consider that political decisions are even more complex and in need of specialist knowledge than criminal juries.

    If we fail to incorporate a high degree of topical knowhow in citizen juries, we have to go back in human progress several millennia. And even for the early civilisation state of Athens the “dumb” sortition process was not adequate after a while, as Socrates noted so aptly.

    Like

  95. I seem to fall between these two views… compulsory vs. voluntary (with or without stratification). I think a little thought experiment is helpful before explaining my current leanings… We are used to thinking of a scenario where those who do not wish to participate are of a “sort,” (either open-minded non-partisans, thoughtless mob-mentality types, low-income over-worked poor people, over-worked self-employed entrepreneurs, or whatever… but I want to pare things back to basics, without these assumptions about what they are like).

    Imagine a small society of 100 adults marooned on an isolated island. They choose to establish a direct democracy. Every week they hold an “all island” democratic gathering to make decisions, All residents can attend, and let’s say nearly all do. But a two choose not to attend… one because he is agreeable and willing to go along with whatever the rest decide, and another because he is philosophically opposed to democracy and thinks the island should have a non-violent battle of wits to select a king. In order to be a true democracy must the rest of the islanders force these two to attend the meeting (where they resentfully sit with arms crossed and refuse to participate)? Are the resulting decisions more democratic or better as a result? Now suppose they decide that only half of the islanders (selected by lot) need to attend each meeting, with the other half attending subsequently and so on…In other words a representative democracy through sortition, but with a few non-willing…

    As long as there are no class, racial, gender or other demographic groups that are handicapped from participating, is it still democratic if those who CHOOSE to defer to others opt out? I think the core of democracy is equal opportunity to participate, rather than mandatory service. However…I ALSO think that service on mini-publics should be quasi-mandatory, because the diffuse knowledge of citizens who might not wish to serve can still be beneficial to the group’s decisions. Also, a citizen who initially THINKS they don’t want to participate, or thinks they are not clever enough, may discover through participation that they enjoy the process and learn a new level of self-respect…(the Republican virtue argument). I think of my own next-door neighbor who is exactly the sort of person I would want to be making decisions for my community, but who would never offer herself (for an election or a jury), but who would be willing to serve if she was told she had to unless she had an excuse. So the lottery for mini-public service should require all those called to serve, or provide adequate reason for a deferment.

    If the rate of those declining is TOO large, I think stratification for any demographic traits that are desired (where there are statistically valid whole-society figures) is acceptable… NOT because it will capture ALL of the relevant traits, but because catching some that people care about (race, income, etc.) can only IMPROVE the descriptive representativity AND improve the legitimacy and public acceptance of the mini-public as “looking like” them.

    My final point is that the level of compulsion can be different for different kinds of mini-publics in order to match the needs of the task (along the lines of my multi-body sortition model).

    Like

  96. Terry,

    As your desert island is a direct democracy, the two people who chose not to participate would only disenfranchise themselves. In a sortition-based democracy they would disenfranchise those who they represent “descriptively” and in all likelihood this would mean unassuming people like your next-door neighbour would be underrepresented compared to activists and loud-mouths. You are more tolerant of such an outcome as your concerns are primarily epistemic, whereas Andre and myself believe that a consistent decision outcome (even if it is consistently bad) is the sine qua non of representative democracy. Otherwise there is no way of knowing which decision is the representative one. I guess our concern is the “wisdom” of crowds whereas you are more concerned with collective wisdom (note the selective use of scare quotes).

    When Andre and I use terms like “mandatory” and “conscription” this is purely to emphasise the need for accurate statistical representation. If very high levels of participation can be ensured by other means — such as bribery or appeals to civic virtue — that’s fine, as the important thing is that the sortition should generate as close a portrait-in-miniature of the target population (warts and all) as is possible (granted the constraints of cost and the rational-ignorance threshold).

    >So the lottery for mini-public service should require all those called to serve, or provide adequate reason for a deferment.

    Agreed, but if the participation level is inadequate, stratification will not solve the problem. It’s not sufficient that the minidemos should “look like America” it needs to generate consistent outcomes, irrespective of which Americans are selected to participate. As Andre has recently argued this can be demonstrated empirically.

    Yoram: >the inherent bias of self-selection

    I’m glad we agree on that, the only outstanding difference would appear to be how best to protect the sortition against self-selection bias. The DP organisers offer a wide range of bribes to ensure participation but 75% of those selected decline the invitation, so those who abjure mandatory participation have a very steep hill to climb.

    hjh: >If we fail to incorporate a high degree of topical knowhow in citizen juries, we have to go back in human progress several millennia. And even for the early civilisation state of Athens the “dumb” sortition process was not adequate

    The fourth-century reforms ensured that the topical knowhow was provided by the competing teams of expert advocates, not by cherry picking the jury. This is the model that most sortinistas advocate as its the informed judgment of the minidemos that we value, but it clearly doesn’t work for demarchs, who require all participants to be well-informed as a precondition for participation. Such an approach is usually referred to as epistocracy, that’s why John Burnheim has confessed that his answer to his original question “Is Democracy Possible?” is No (and neither is it desirable). Most of us on this forum (it’s not a “board”!) don’t want to give up on democracy, and that’s why we’re unsympathetic to demarchy.

    Like

  97. *** I agree with Keith Sutherland comments, and therefore I will not repeat them.
    *** I will nevertheless answer a personal question:” If jury service is such a good choice that “you do not try to evade it”, why make it mandatory in first place and shoot citizens if they do not want to serve, as suggested above? “ I think modestly I am a citizen with civic consciousness not in the low side of the Bell curve. Note I don’t ask to become a philosopher-king. Because I am against any selection procedure, which could give a very good sovereign, as me (in my somewhat subjective opinion), but likewise a very bad one. Democrats (ortho-democrats), as me, do not want to give sovereignty to the supposed (self-assessed) better, but to the ordinary citizens, without selection. And, frankly, I do not fear that a representative mini-dêmos, judging me with good deliberation procedure, will behave in majority as a lynching mob
    *** Note that “lynching mobs” did not follow deliberation procedures, and included self-selection. These two factors separate the “mob” from a sovereign dêmos.
    *** I think this discussion was very enlightening, especially because we see a demarchy proponent criticizing the model of a truly representative jury because it would include “the unthinking mob, who’d rather lynch you than give you a fair trial.” The exclusion of “the unthinking mob” from sovereignty – with relevant reference to Socrates – put the ideology in a side clearly antagonist to dêmokratia, and akin to Aristotle’s views. That is a very useful clarification. I only suggest to choose another word than “demarchy”, which includes dêmos, and could be seen as semantic maneuvering. Why not “epistemarchy”?

    Like

  98. >Why not “epistemarchy”?

    I imagine John Burnheim would agree to that label.

    Like

  99. Keith,
    I feel that you missed the point of my thought experiment… IF it is “okay” if some members of a tiny direct democracy choose to abstain from participation, and still be deemed democratic, then why shouldn’t a mini-public that ACCURATELY reflects the entire population likewise have an equal portion that chooses not to participate. Your analysis of the non-participants that “they would disenfranchise those who they represent ‘descriptively’… ” confuses the members of amini-public with designated representatives of constituents (as in an election). this is ironic because you have consistently pointed out that individual members of a minipublic do NOT represent certain people. It is only the minipublic AS A WHOLE that is representative. Thus, if a segment of a direct demmocracy of 100 people would abstain, then so too should an equal portion of people drawn for a minipublic. Democracy only assures equal opportunity for any citizen to participate. This obviously MUST be WITHOUT society imposed obstacles.

    Now the potential PROBLEM is if lots of citizens refrain (like the 75% example) when called to serve. BUT, we don’t yet have an understanding of how widespread abstainers would be in a mature sortition democracy (with civic expectations, education, good pay, etc.). The examples of a non-binding deliberative poll, or people drawn for a one-off, where none of the jury pool have ever heard of the system before isn’t that indicative.

    Our goal should be to get as high a rate of participation as would occur in a small direct democracy… not 100%

    Liked by 1 person

  100. Terry,

    You have inadvertently made the case for electoral representation. I think we both agree that those likely to choose not to participate in a direct democracy will share a lot of features with those who decline the invitation to participate in a minidemos. We can only speculate on the qualities of such people, but they might well include a degree of modesty about their own abilities, below-average levels of education and self-esteem, lack of self-confidence, fatalism, satisfaction with the status quo, (small-c) conservatism, lazyness, selfishness etc. Such people would be (effectively) disenfranchised in a direct democracy and a non-mandatory lottocracy, but under current arrangements they can choose electoral candidates who are more confident, educated etc., yet purport to share their political views. In your desert island direct democracy they can choose not to participate but in a sortition-based system they would be dependent on the participation of those like themselves who would (if the above catalogue of personal attributes is true) be less likely to participate. That’s why a lottocracy has to be a) quasi-mandatory and b) voting only — so that accurate descriptive representation is not distorted by the disproportionate influence of self-confident and well-educated activists.

    Of course we can’t know the likely numbers of abstainers but my hunch is it would be closer to Fishkin’s 75% (fact) than your 2% (thought experiment), but even if it were somewhere in the middle, the distortion involved would make a mockery of the democratic claims of lottocracy.

    >Democracy only assures equal opportunity for any citizen to participate.

    Yes, and that works in the case of direct democracy and (in theory) with electoral representation, but in a lottocracy, where 99.999% don’t get to participate directly, “equal opportunity” requires that their representation by proxy is guaranteed. It’s alien to the spirit of democracy to claim that those who would have not chosen to participate are “represented” by their non-participating proxies. Or do you claim that the people of Northern Ireland were represented at Westminster by Sinn Fein MPs who refused to take up their seats?

    Like

  101. Keith,

    “Or do you claim that the people of Northern Ireland were represented at Westminster by Sinn Fein MPs who refused to take up their seats?”

    Given that people continue to vote Sinn Fein regardless… quite possibly. Anyway, the analogy breaks down because in a statistical representation system no one is appointing anyone to serve as a delegate. The job of the members of a sampling is to mirror the population. And when participating in a political system is a violation of a person’s principles, as may be the case in the example of a Sinn Fein supporter, forcing them to do so on threat of imprisonment is clearly problematic.

    My stance on mandatory participation has softened considerably over the last year or so. Full participation is strongly desirable, as it is in electoral democracy. But I don’t see how a short-term sampling can have an informed view about both a specific proposal and its context. If a sampling has to persist for one or more years then mandatory participation becomes difficult and we may be better off trying to make the best of opt-out participation with heavy stratification. If the ideological balance ends up being too far off then we could simply put a party preference question on the census and stratify along those lines as well. If the role of elected officials is limited to proposing legislation (with no need to form a governing majority) we can opt for an electoral system which does not necessarily discourage fragmentation. So stratification along ideological lines could be as fine-grained as people deem appropriate in practice.

    In any case, pure opt-out samplings should be just as consistent as quasi-mandatory samplings. The set of people initially drawn by lot is (of course) statistically significant. They are first asked if they want to participate, which is a straight up yes/no question like any other that could be posed posed to a statistical sampling. The subset answering “yes” to the first question is then asked one or more yes/no questions on matters of policy. Statistical significance applies across the board. If a person would have declined the opportunity to participate had they been chosen… then they are not disenfranchised by a similar selectee actually declining to participate. Their representation by proxy is carried out faithfully. If participation is too burdensome, then perhaps the attendance rate will end up being too low. But, then again, there is no rational ignorance at play and the pay should be generous. And electoral and direct democracies often have low turnout rates without too much trouble. If I remember correctly Swiss referenda typically have a 1/3 turnout rate and US Congressional midterms are not much better at this point.

    Hopefully stratification can simulate the effect of full participation to some extent, leaving us somewhere between quasi-mandatory sampling and opt-out sampling. And of course stratification will increase the consistency of samplings across the board, be they opt-out or quasi-mandatory.

    Like

  102. Naomi:

    >I don’t see how a short-term sampling can have an informed view about both a specific proposal and its context.

    I agree, that’s why in my proposal “context” (which I refer to as joined-up government, fiscal responsibility and accountability) is provided by other, non-political, agencies. If you are going to go down the route of stratification by ideological preference then you might as well not bother with all this sortition malarky and focus instead on incremental improvements to the electoral system. And given that most voters (other than in the USA!) are ideologically agnostic, I don’t see how self-stratification would work in practice.

    >In any case, pure opt-out samplings should be just as consistent as quasi-mandatory samplings. . . .The subset [agreeing to participate] is then asked one or more yes/no questions on matters of policy.

    The “policy” dimension presupposes that Average Joe has an opinion on most issues of public policy. And you are also ignoring my principal objection to non-mandatory sampling — that it massively over-represents those with an active interest in political issues and they are (paradoxically) not necessarily the best judges of public policy options. The vast majority of people are not much interested in politics so they should either a) be represented by other people of a similar disposition or b) elect persons who are interested on the basis of their stated policy preferences. I don’t see how you can combine these two desiderata in a single body of persons.

    Like

  103. We’ve already gone a few rounds over the coordination/context function. It’s the most profoundly political function there is. The idea we can take politics out of it reminds me of those deliberative democracy supporters who seek to take politics out of government altogether. And to enforce consistency on the samplings the administration will need have sufficient leverage over them. Any leverage that can be applied for one matter can be applied for any other and if the samplings are tasked with holding the administration accountable then there is plainly a problem. So that leaves us with coordination by either an elected parliament or a longer-term sampling. In the former case we basically have the status quo (as it exists in the UK) with the Lords replaced by a more robust sortiton-based body. This has its advantages as we have discussed before. But there would still an assembled majority, so rational ignorance in the electorate would remain the limiting factor in the quality of the overall system. If we can get away with a longer-term sampling then, in my book, this strategy is clearly the winner. The decision-makers should have the best understanding of the consequences of their decisions possible.

    Incremental improvements in the electoral system cannot overcome rational ignorance or policy bundling, which, if memory serves, were your main reasons for supporting sortition in the first place.

    If most people can be bothered to check a box on a ballot every few years then it’s a safe bet they can be bothered to check a similar box on an already mandatory census. Party membership is a whole other issue that goes beyond making a little mark on a sheet of paper.

    Those with an active interest in political issues are already massively overrepresented. In Switzerland it looks like they may be overrepresented to the tune of 2-to1. No one is up in arms over it. As you have said before, a person is only disenfranchised if their inclusion or exclusion from the sampling would make a difference on the outcome. If a person who would not have participated if given the chance has only themselves to blame. They are deprived of the actual choice, of course, but they are also deprived of the choice to vote a proposal up or down and must live with the statistical equivalent in both cases. As long as there is consistency between samplings there is no problem from a representativity aspect. We must distinguish between representativity, which is necessary, and other factors, which are desirable. It is an open question, I think, whether a sampling which is more interested but less oblivious to context will generally do a better job of voting proposals up or down.

    Also, as a side note, Switzerland is also one of the most politically conservative of the world’s developed states despite being governed by the most interested.

    Like

  104. Naomi,

    What you say may well be true from a democratic perspective, but my preferred option has always been the mixed constitution, in which the monarchical (appointed), aristocratic (elected) and democratic (allotted) elements share power. Doesn’t UK governance post-Brexit — in which the executive is charged to implement a policy that it argued against — demonstrate that the “coordinators” have less leverage than one might have thought? No doubt the government will seek to undermine the popular will (notwithstanding the PM’s claim that “Brexit means Brexit”), but will have to do that in the face of fierce opposition from the other estates (including the media). The role of the executive magistrate/commander-in-chief was never intended to be a political one. According to Mogens Hansen’s 2010 essay the problem was caused by the usurpation of the mixed constitution by the new-fangled (and descriptively inaccurate) doctrine of the separation of powers.

    >if the samplings are tasked with holding the administration accountable then there is plainly a problem.

    Correct. In my (revised) model this is an aristocratic role, achieved via a complex combination of political parties, members of the house of advocates and the media. Most often the media would reveal ministerial incompetence or venality, leading to a censure motion originating in the advocates’ house, with the outcome determined by an allotted jury. In addition to this, political parties will be constantly snapping at ministerial heels. My concern is the opposite of yours — all this accountability might excessively weaken the executive, so that a relatively high threshold might be required for the confidence motion.

    >As you have said before, a person is only disenfranchised if their inclusion or exclusion from the sampling would make a difference on the outcome.

    Yes, and this is a matter that we both agree can be resolved empirically. However if choice (i.e. non-mandatory participation) is allowed, my concern is that this will generate a bias — albeit (perhaps) a consistent one — that would differentiate the sampling(s) from the target population. I guess this is an additional constraint to my earlier concern. Remember that our concern is with what everybody would have thought under good conditions, rather than representing the wishes of volunteers — I believe that this would also fall into your category of requirements rather than optional extras.

    >a person who would not have participated if given the chance has only themselves to blame.

    That’s true. The problem is the disenfranchisement of the overwhelming majority who are not given a chance. This is a bit too close to Calvinist theology, in which the vast majority of people have no hope of salvation (a fate reserved for the tiny ranks of the elect). As this is entirely down to providence, its worth reminding ourselves of the religious origins of the lot.

    Like

  105. “The problem is the disenfranchisement of the overwhelming majority who are not given a chance.”

    But the point is that, by your own reasoning, they are not disenfranchised if their selection/non-selection makes no difference. The subset of the whole population who draw the golden tickets, whether they ultimately go on to participate or not, will be representative of the whole population to some statistical margin of error. The subset of those who were drawn that ultimately goes on to participate will look and vote the same, again with some statistical margin of error, from one sampling to the next. Denying the masses the chance to choose to participate is equivalent to denying them the chance to vote yes or no. The only difference is the question being asked of the sampling. The objections to an opt-out system are then largely limited to the quality of the decisions they make.

    Like

  106. Naomi,

    > we could simply put a party preference question on the census and stratify along those lines as well

    Interesting idea, but it seems that this would motivate people to declare preference for the party which has the highest opt-out rates.

    Like

  107. Yoram,

    Perhaps, but even a huge difference in the opt-out rate would still leave an opportunist with a negligible chance of being drawn. And if there are a large number of opportunists misrepresenting their preferences on the census then the size of the party affiliation-defined pools and their internal ideological breakdowns will shift accordingly. In the most extreme case where everyone simply chooses the party label with the highest opt-out rate then the effect is exactly the same as if there had been no stratification along party lines at all.

    Like

  108. Naomi,

    1. We agree that mandatory participation is appropriate for an ad hoc jury, but not for service of a year or more. Ad hoc is my preference and (I believe), the coordination problem is open to resolution if the jury is just the democratic element in a mixed constitution.

    2. I’m persuaded that in a non-mandatory jury the biases would probably be systematic — i.e. each sample (if it were large enough) should contain them to a similar degree.

    3. The outstanding question is whether the biases can be addressed by the use of stratification. Age, income and education differentials are strongly correlated with participation and can easily be normalised. My concern is more regarding psychological attributes — introversion, modesty, shyness, humility, lazyness, cynicism, indifference etc (and their antonyms) which are likely to affect participation rates. I don’t see how these factors can be corrected by stratification and it’s possible that they may have an influence on decision outcomes. It’s also the case that such a complex stratification could be wide open to manipulation.

    Fortunately it’s not my problem as my model is for ad hoc juries and mandatory participation can easily justified as a civic obligation (like paying income tax). Given that all my jurors get to do is vote, then Yoram’s objections are less significant.

    >Denying the masses the chance to choose to participate is equivalent to denying them the chance to vote yes or no.

    That may be true but it would be quite hard to justify, given that the principal concern here is the behaviour of a proxy. To take an example, if individual x decides not to vote in an election then that’s her choice, but what if her MP decides not to vote in a parliamentary division? In the electoral case the rascal will be thrown out at the next election, but there are no equivalent safeguards with representation by proxy. Far better to have short-term mandatory participation and demonstrate by experiment that it doesn’t make any difference to the outcome which individuals are included. If you allow the proxies to choose whether or not to participate this adds an unnecessary level of complication (if they can choose then why can’t I?) One of the big attractions of sortition is its blind, mechanical nature.

    Like

  109. Naomi,

    Well, no. In such a situation, not only is the stratification not working, it allows the system to pretend that it has some positive effect and use it to legitimize a system that in fact is not representative.

    The only way to make sure that the system is representative is to alleviate all circumstances that make service difficult (e.g., economic hardship, parenthood, illiteracy and disability), so that those who choose not to participate do so for what can really be described as their free will and not because of constraints that effectively bar them from service. In such a situation, uniform sampling probabilities should work fine.

    Like

  110. By the way, it is not difficult to come up with a scenario where adjusting probabilities to ostensibly account for opt-outs will result in a sample that is less representative than a simple random sample with opt-outs.

    Like

  111. >those who choose not to participate do so for what can really be described as their free will

    Now that would present a problem for those seeking to design a stratification system to ensure accurate representation by proxy. My concern, as always, is the will (free or otherwise) of the vast mass of people disenfranchised by the sortition process.

    Like

  112. Yoram,

    If 10% of people define themselves as part of a subgroup on the census then 10% of seats would be reserved for those who defined themselves as part of that subgroup on the same census. If everyone reported the same subgroup then 100% of seats would be reserved for people who defined themselves as being in the subgroup. If everyone picked their subgroup at random you have the same result. Every subgroup would look the same, contain the same number of people, and have the same number of seats. The stratification then has no net effect. As long as there is correlation between politically salient features of the population and an attribute being stratified for then there is a net gain of some sort. If not, then nothing is lost.

    Of course simplicity is a virtue. Stratification makes the overall process much more complicated. If the participation rate is high, say over two-thirds, then maybe leaving it off would be okay, though it would certainly not be my first choice. If it’s low, say less than one-third, we may have no choice. At that point the demographics of the assembly won’t even look right. We don’t know what the participation rate will be. Right now all we can do is guess and plan for different possibilities.

    Keith,

    “but what if her MP decides not to vote in a parliamentary division?”

    There is no line of *delegation* from the people to one or more members of the sampling. No individual member is a proxy for anyone in the general population. The sampling as a whole represents the people as a whole. It can only be considered in aggregate. Non-participation is unrepresentative if and only if you would not have a similar portion of the population declining to participate if they were chosen to serve. This is true for *all* characteristics of the population and has nothing to do with stratification, which is just a way of having our cake and eating it too. Without it the same principle applies.

    It may well be the case that an opt-out sampling would end up being less diverse, cognitively, than would be the case in a mandatory sampling. This would be bad from a wisdom of crowds standpoint. But the wisdom of crowds is also diminished profoundly when a few homogeneous individuals in the executive have significant leverage over the whole legislative process.

    Like

  113. Naomi:

    >No individual member is a proxy for anyone in the general population.

    Yes I know that, I was referring to the principle of proxy representation.

    >Non-participation is unrepresentative if and only if you would not have a similar portion of the population declining to participate if they were chosen to serve.

    That’s true but I can imagine nothing worse than being subject to the unchecked rule of an unaccountable activist subset of the general population.

    >The wisdom of crowds is also diminished profoundly when a few homogeneous individuals in the executive have significant leverage over the whole legislative process.

    I would happily sacrifice a little “wisdom” for the sake of some accountability. The harlot’s prerogative would be equally applicable to your long-term voluntary-sortition-based supreme council. The Athenians had ways of holding their allotted magistrates to account and electoral accountability is the modern surrogate. But there is no way of holding juries to account, so sortition could only ever be part of a mixed constitution. The buck has to stop somewhere, hence my emphasis on the essential coordinating role of (appointed) government ministers.

    Like

  114. Keith,

    We are already being governed by the most activist half of the population. As long as we can do at least as well after stratification we should be in good shape on that front.

    “I would happily sacrifice a little ‘wisdom’ for the sake of some accountability.”

    I still find your argument that accountability can be found by increasing the arbitrariness of the executive (as is the effect of facilitating agency loss) bizarre. As you say, the buck must stop somewhere, but if you set up a transactional relationship between the people and the executive then the buck stops nowhere. It goes around in a circle. The buck should stop with the considered judgement of the people in the sampling.

    Liked by 2 people

  115. Naomi:

    >We are already being governed by the most activist half of the population.

    Not directly. Given that it takes only a small effort to vote in an election, those of us with little interest in politics can still express a preference for a minimal cost. This is particularly true for senior citizens (like myself), many who vote out of a sense of civic obligation, not because they are interested in politics. Most of those in this category would turn down the invitation to participate in a political jury, so rule by non-mandatory jury would mark a significant shift towards governance by activists.

    >if you set up a transactional relationship between the people and the executive then the buck stops nowhere. It goes around in a circle.

    The relationship is not transactional, it’s cybernetic. I agree it goes round in a circle but the trajectory is virtuous, not vicious. Harrington’s example of the two girls dividing a cake (“you divide and I’ll choose”) applies to every aspect of the mixed constitution. The trouble is everyone is so immersed in the (false) theory of the separation of powers that this ancient wisdom has been forgotten.

    I’m surprised that you refer to increased executive agency as introducing an arbitrary element as this would be far from the case. Executive agency is always conservative in nature as the default assumption is to do nothing at all (other than preserving the jobs of the executives), but this would be necessary to counteract the surge of unaccountable activism from all other quarters.

    >The buck should stop with the considered judgement of the people in the sampling.

    Why then did the classical playwrights lampoon jurors as unaccountable tyrants? In Athens the buck stopped with the rhetores and magistrates, not the jurors. There is no reason to think the modern incarnation would be any different as allotted jurors can just walk away and wash their hands of the consequences of their decisions without suffering any penalty whatsoever. In my proposal both elected rhetores and appointed magistrates would be rewarded for success and would pay for any mistakes they make.

    Like

  116. >”Most of those in this category would turn down the invitation to participate in a political jury, so rule by non-mandatory jury would mark a significant shift towards governance by activists.”

    Not necessarily. The burden to participate is greater but the impact of participation is much greater still. It may simply be the case that those who are most likely to participate are those who most strongly feel the weight of the significance of the decision-making capacity they have been given. In other words, it may select in favor of those who are most moved by the absence of rational ignorance and who are most likely to take the task seriously at the expense chiefly of those who would vote based on which advocates look most attractive or whatever. We don’t know what the people who participate will look like in practice at this point.

    >”Executive agency is always conservative in nature as the default assumption is to do nothing at all (other than preserving the jobs of the executives)”

    Power is power whether we stick the label of “executive” on the paperwork or not. Power is neither conservative nor liberal by default. The degree to which an officer is concerned with protecting their job is directly proportional to the ease with which they can be removed. If you make them feel more secure they will feel more free to exercise their powers as they personally see fit and you may or may not like the way in which they use their leverage to direct policy. It depends on their whims. They could easily prove to be more activist than a robust majority in the sampling would like. The personality and preferences of a would-be officer are only ever knowable to a limited extent beforehand. Hence, an increase in agency loss directly translates into a more arbitrary government.

    >”Why then did the classical playwrights lampoon jurors as unaccountable tyrants?”

    The modern media is united in similar critiques over the bureaucracy. They would react quite predictably over the proposed DMVification of the executive (as they may label it). In any case, people are always going to find someone to complain about when they are unhappy about the status quo. The popular media of the day will cater accordingly.

    Like

  117. Naomi,

    My mother voted dutifully all her life but the chance of her agreeing to participate in any form of direct democracy would have been close to zero. But all we are doing is speculating on who would or wouldn’t rise to the challenge under a voluntary scheme — what is beyond dispute is that mandatory participation would remove any potential for systemic bias.

    Executive office holding would involve a compromise between accountability (via a mixture of media, advocate and democratic power) and security of tenure. In my own proposal (the last chapter of my PhD) they have no initiative rights apart from statutory instruments and minor regulations, so there would be little opportunity to implement their whims.

    The classical playwrights lampooned the unaccountable power of jurors because it was a fact, not because they were looking for something to complain about. Any attacks by the modern media on the bureaucracy would be welcomed as this is their role under a mixed constitution.

    Like

  118. >”they have no initiative rights apart from statutory instruments and minor regulations, so there would be little opportunity to implement their whims.”

    The US president also lacks the power to initiate legislation and the Senate lacks the power to introduce money bills. So the president has a like-minded Representative introduce proposals on his behalf and the Senate takes dead money bills and amends them to replace all content with whatever it wants.

    Rules like these have no practical effect.

    Like

  119. >The president has a like-minded Representative introduce proposals on his behalf.

    I have no problem with that because, under my proposal, any citizen can introduce a new bill — directly or indirectly. However the [minister] would have no advocacy rights and the success of the bill would depend on the votes in the allotted parliament. This is a good practical example of the mixed constitution. What your example indicates is the emptiness of the doctrine of the separation of powers.

    Like

  120. *** Hjhofkirchner wrote ( September 11), about compulsory political service: “Force” tends to be associated with highly undesirable political systems.
    *** Well, I have to pay my income tax by force, and I am happy to know that in the desirable system following our epistemarchists, I will be freed from that. Actually, if I have to choose about undesirable acts of political force, I would prefer reasonable political service and no income tax.
    *** Hjhofkirchner implies I suggest to shoot people who would evade the political service. Independently of the death penalty debate, it is clearly against my ideas about sovereignty. Shooting the bad citizens would be a (rather extreme) kind of political selection among the citizens, and I am against any kind of selection among citizens about sovereignty.

    Like

  121. *** Yoram Gat wrote (September 14); “The only way to make sure that the system is representative is to alleviate all circumstances that make service difficult (e.g., economic hardship, parenthood, illiteracy and disability), so that those who choose not to participate do so for what can really be described as their free will and not because of constraints that effectively bar them from service”
    *** Ok about alleviating material parameters which will act against political participation. But, and especially for people in the beginning of a mini-public system, there are other parameters. People who feel powerless in their everyday life, and felt always powerless in the political field, are prone to abstain from political participation. Here a strong factor is social class situation (I am beginning to feel the last Marxist, very strange feeling for me); and those in a “powerless class” who are politically active will be very different from the others, and not representative.
    *** In a deeply established dêmokratia, maybe “political apathy” will be explained through random personal factors (genetics and random events in personal life). But we must not forget the collective mental factors which will be important in the beginnings of a democracy-through-minipublics.
    *** The “desert island” model of Terry Bouricius is not enlightening, because we don’t know the mental legacy of the people who “marooned” here. Did they come from an equalitarian society? Or is this island populated with newcomers with a long history of class division, with powerful ones and powerless ones? It will be important, at least at the beginning of the dêmokratia.

    Like

  122. *** Terry Bouricius wrote (september 12): “So the lottery for mini-public service should require all those called to serve, or provide adequate reason for a deferment. If the rate of those declining is TOO large, I think stratification for any demographic traits that are desired (where there are statistically valid whole-society figures) is acceptable… NOT because it will capture ALL of the relevant traits, but because catching some that people care about (race, income, etc.) can only IMPROVE the descriptive representativity AND improve the legitimacy and public acceptance of the mini-public as “looking like” them.”
    *** “TOO large”? Let’s say no more than 5% of major people (major people = people than society allows, from age and mental state, to make basic choices in everyday life; dêmokratia is founded on the absence of political selection, juries do not have to include children or mentally disabled persons).
    *** Stratification will always heighten representativity. If we use it, we must not restrict to “relevant” parameters, first because we don’t know the relevant parameters in a given field, second because the choice will be enhancing mentally the more conflict-prone parameters. A good stratification would use all the objective parameters (but only the objective ones).
    *** .Stratification would be useful only in the case of small juries. For huge juries, better to use the simpler way.
    *** I disagree strongly with any idea of “public acceptance of the mini-public as “looking like” them”. The mini-public idea is that the mini-dêmos is statistically representative of the dêmos, not that he will “look like” him, in apparent parameters as age, gender or skin color (forgetting especially class, income level, number of children, political sensitivities); quota in an electoral system could result in a parliament “looking like” the dêmos, but deeply different, and it is a temptation in contemporary polyarchies to use this idea against the democratic menace.

    Liked by 1 person

  123. Andre,

    > People who feel powerless in their everyday life, and felt always powerless in the political field, are prone to abstain from political participation.

    It is unlikely that, given a fair opportunity, many such people would turn down the offer of wielding that very power they always felt they lacked – and to be handsomely rewarded – both socially and financially – for doing so.

    But this is an empirical question that would be easily observable once the allotment system is in place. If many people do turn down allotted positions, something would have to be done about it. It would be easy, for example, to determine whether turning down positions is associated with the social factors that you propose be used for stratification. The terms of service would have to be adjusted so as to encourage people to participate, and in particular to address those issues that disproportionately affect people belonging to certain social groups.

    On the other hand, stratification would normalize the situation where a fair opportunity to participate is not given, and as a result many people do turn down appointments, by hiding this situation behind the false claim that stratification annuls its effects.

    Like

  124. Andre,

    >People who feel powerless in their everyday life, and felt always powerless in the political field, are prone to abstain from political participation. Here a strong factor is social class situation (I am beginning to feel the last Marxist, very strange feeling for me); and those in a “powerless class” who are politically active will be very different from the others, and not representative.

    I guess that makes me a Marxist too. Clearly it would not be possible to devise a stratification system to compensate for such persons declining to participate. Would it matter? Yes — if the goal is to empower the powerless, whether for Christian, Marxist or pragmatic reasons (lancing the populist boil).

    >quota in an electoral system could result in a parliament “looking like” the dêmos, but deeply different, and it is a temptation in contemporary polyarchies to use this idea against the democratic menace.

    Exactly. The “looks like America” trope originated with Bill Clinton. It would be good to hear how Yoram intends to reconcile voluntarism, non-stratification, small assemblies and genuine statistical representation (without resorting to logical syllogisms or links to past catechisms), because these desiderata appear to be mutually exclusive.

    Like

  125. Andre,

    I actually agree with all of your points (including those rebutting my posts!)

    To explain…. We can think about early sortition democracy getting started in a non-egalitarian society full of barriers to service, and quite distinctly a utopian future system. In the foreseeable future there WILL be citizens who would decline because they can’t afford the time or effort, or have been beaten down (rather than merely lack interest). Yoram suggests “It is unlikely that, given a fair opportunity, many such people would turn down the offer of wielding that very power they always felt they lacked – and to be handsomely rewarded – both socially and financially – for doing so.” Psychology suggests this is incorrect. We tend to internalize society’s view of ourselves. While I AGREE with Yoram that stratification COULD be used as a ploy to hide the unrepresentativeness of a minipublic (and by the way, I was using the phrase “looks like” in a poetic sense… but of course, I meant “feels, thinks, and reasons like” as well) I also agree with Andre, that when done appropriately it can help.

    the point of my desert island scenario was an idealized society where there are no barriers to service but still SOME decline. My point is that democracy requires that there be genuinely equal opportunity and not that all must take that opportunity. I believe that declining should require an affirmative effort to avoid service (so that service is the default when called).

    Like

  126. Terry,

    >I believe that declining should require an affirmative effort to avoid service (so that service is the default when called).

    That is more or less the current position with UK jury service. It’s probably better to put it this way (abjuring adjectives like mandatory, compulsory, conscripted etc) but the “affirmative effort” should be a very substantial one. I like the idea that jury service should be like paying income tax — in fact at one stage I argued that the sortition franchise should be limited to taxpayers. This used to be uncontroversial up until the turn of the twentieth century. It was called the rate-payer’s franchise and excluded net beneficiaries of the public purse as voting what to do with other people’s money was frowned upon.

    Like

  127. *** I reminded that “people who feel powerless in their everyday life, and felt always powerless in the political field, are prone to abstain from political participation.” Yoram Gat answered (September 16) “It is unlikely that, given a fair opportunity, many such people would turn down the offer of wielding that very power they always felt they lacked – and to be handsomely rewarded – both socially and financially – for doing so.”
    *** I agree with Yoram than we can hope than giving sovereign power to minipublics, and therefore equal participation to sovereignty to any citizen whatever his social weight, will quickly heighten the political consciousness and participation.
    *** But I have nevertheless some fears about the first times – and as the new democracy-through-minipublics will have many enemies, the first times will be especially sensitive to doubts about the representativity of the minipublics.
    *** The problem will be even stronger if during some time we have an hybrid system, as the one some propose in France: elected deputies and allotted senators. An unrepresentative allotted Senate will be used against the minipublics model.
    *** Even in a well established dêmokratia, I am afraid non-mandatory political service will lead to contradictions between minipublics.
    *** Maybe I am especially worried because of the ancient Athens situation. As I said before, political service, either in the assembly or in the juries, could not be made mandatory, for very material reasons. But the result was a disproportionate power for the urban poor, the peasants coming on an irregular way. This fact fueled the anti-democrat strategy of trying to bring into opposition the urban poor and the peasants. And it could only lead to erratic votes, as the peasants would attend more in some cases than others.
    *** Anyway, for me, statistical representativity is an essential point, and only mandatory service can assure it. That does not need strong sanctions against non-participating people; only some kind of moral disapproval and a reasonable tax on it. The aim is not requiring service from people who are not interested – and would be anyway idle in the minipublic debates; but to put a mild public pressure able to compensate for the private pressures of family life and professional life, and some temptations of private life.
    *** Actually there is no much difference between Yoram Gat’s proposal of a handsome social and financial reward for presence and my proposal of a mild social and financial sanction for absence … Maybe some small differences: the financial sanction would be proportional to the income, whereas the “political salary” would be the same for all.
    *** We can combine both.

    Like

  128. Andre

    I think we agree that the carrot is better than the stick — all that matters is that the overwhelming majority of those called end up participating. Even with the DP the participants are effectively volunteers who are then camouflaged by stratification in order to appear representative of the target population.

    Like

  129. Andre,

    > An unrepresentative allotted Senate will be used against the minipublics model.

    As it should be. The allotted body should be representative, but not as a formality – with people sitting there because they have been forced to. A formally representative body which generates a steady stream of examples of disfunctional members would also be an easy target for delegitimizing a sortition-based system.

    An allotted body has to be representative by making sure that there is a real opportunity for anyone who wants to participate to do so. But there must also be an understanding that by taking up your slot in the allotted chamber you are committing to a job well-done. And this is a crucial difference between rewards and sanctions. With sanctions (even when combined with rewards), non-functioning members, or even corrupt, self-serving members, could make the excuse (in their own minds, if not in public) that they only took the job because they had to. Such a stance would be illegitimate if there were no sanctions and the rewards were adequate.

    Like

  130. Yoram,

    >An allotted body has to be representative by making sure that there is a real opportunity for anyone who wants to participate to do so.

    Am I correct in thinking this means that a voluntary allotted body would represent those who choose to participate, and that those who don’t choose to participate don’t merit representation? There are likely to be substantial psychological and sociological differences between these two kinds of citizens, so I don’t see how a voluntary allotted body could ever claim to be a portrait-in-miniature of the target population.

    >dysfunctional members . . . non-functioning members, or even corrupt, self-serving members.

    That would suggest a strong republican ethic underlying your sortition programme which is somewhat at odds with democratic equality (which holds that non-functioning citizens should be just as equal as their activist colleagues).

    PS These is a serious question, so I would be grateful for a reply.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: