Sortition and the need for Internal Deliberation

What should randomly selected citizens actually do when they are deliberating? Should the deliberation occur in their heads or should it be talk-based? This internal-versus-external deliberation question is core to any discussion of the legitimacy of decisions reached by randomly selected citizens. In the attached paper, presented at a conference in Dublin last Friday, we suggest, to put it bluntly, that talking is bad and imagining is good. We further suggest, and operationalise, distinct ways of facilitating imagined deliberation and present tentative experimental findings as to their relative effectiveness. This draft paper is very much a work in progress but we would welcome any thoughts (including robust disagreement).

Imaginative Randomocracy: A General Model of Citizen Decision Making Applied to Northern Ireland (and the UK)

Advertisements

23 Responses

  1. Greetings,

    Here is a recent post of mine where I propose using sortition as a means for selecting candidates for higher office.

    Let me here your thoughts.

    http://dissidentvoice.org/2015/03/do-away-with-elections/

    Like

  2. Also putting it bluntly… Both internal (listening and imagining) deliberation and give and take deliberation (talking) have good and bad characteristics. It absolutely is not valid to dismiss one in favor of the other in all situations. Even in the limited realm of legitimacy (popular acceptance) it is not so stark.

    1. Imaginative deliberation (internal, but based on arguments presented by organizers), is indeed reproducible…any random group (of sufficient size) would likely come to the same conclusions. But so would a completely non-deliberative public opinion poll or one without any information presentations at all. In other words, reproducibility is not the only good….The public might place more value (and legitimacy) on a process that allowed participants to talk and argue and maybe seek common ground, above one that could be easily steered by the organizers (by determining what information passes their filter). Silent (imaginitive) deliberation merely SHIFTS the legitimacy question from the random body to the legitimacy of the ORGANIZERS who structure the information presentations and frame the querstions.

    2. Imaginative deliberation also forfeits one of the prime benefits of the diversity achieved with sortition…. that is the gathering and sharing of knowledge and perspectives. As J. Ober argues in his book “Democracy and Knowledge: Innovation and Learning in Classical Athens,” it is exactly this sharing of diffuse “latent knowledge” (talking) that allowed Athenian democracy to flourish.

    3. On the other hand, talk can lead to go-along-to-get-along group-think, or polarization, and the hardening of positions rather than the tempering and melding of ideas. The devil is in the details.

    4. The challenge is to find the right time and place for each sort of deliberation. In the Athenian Nomothetai (legislative juries), the random sample listened to pro and con arguments and then voted by secret ballot without debate (much as proposed in this paper). However, this was preceded by substantial deliberation in the Council of 500 (the Boule) or the Agora.

    5. In short, active (talk) deliberation is most appropriate at the earlier stages when proposals are being developed, amended, and reviewed in preparation of a “final proposal,” and imaginative (silent) deliberation is most appropriate when merely a yes/no decision is left to be made. Both stages can substantially benefit form the diversity and representativeness provided by sortition, although they would need to be carried out by separate samples of citizens. Neither should be jettisoned.

    Like

  3. John,

    I agree with most of the thrust of your paper, in particular “Talking gets in the way of democracy” and “Decisions made by imagined deliberation are invariant across sample but decisions made by talk-based deliberation are not”. I’ve been arguing this for several years, but it doesn’t go down at all well on this blog and in wider deliberative democracy circles — in fact I presented a paper on this topic (attached) at the PSA conference in Belfast a couple of years ago. John Dryzek was in the audience and I could see the steam coming out of his ears! Have you seen Andy Dobson’s new book Listening for Democracy? It’s very good on the case for silent deliberation within.

    I agree that, if a sortitition-based institution is to play anything more than an advisory role that the decision output of each sample of the population has to be consistent (or invariant, as you put it). Bob Goodin has pointed out that the three Texas utility DP samples came to different conclusions and I’m seeking to understand the reasons for this. Goodin, in his analysis of the Bloomfield Track DP argues that most of the changes in preferences/beliefs comes at the information stage, rather than the deliberation stage, leading me to suggest in an exchange with Jim Fishkin that he should drop the small-group deliberation element of the DP in favour of silent deliberation within. Jim is opposed to this, partly I believe, to defend the DP programme against further attack from the Habermasians, who already claim that his form of deliberation is too thin. But I think this is a necessary price to pay in order to achieve the degree of invariance that would be required by a representative random sample with some sort of statutory function. I was hoping to discuss this with Fishkin’s statistician, Bob Luskin, at his ECPR summer school on statistical methods in deliberative polling but it was unfortunately cancelled.

    In my own institutional proposal the silent deliberators would meet together and hear the information/advocacy in person, rather than just responding to a filmed presentation from home. Although there is no reason in principle to organise things this way it does have two advantages: 1) the descriptive representativity (aka diversity) of the sample would be there for all to see (it would “look like America”, as Bill Clinton put it); 2) there would be a need to ensure that all participants took their civic duty very seriously and this could be encouraged by the privileged decorum of the deliberative assembly (perhaps combined with a modern version of the Heliastic Oath and other dignified flummery). The sortition process of early-modern Venice was an elaborate public ceremonial. The danger with the allotted jurors just watching TV at home and voting secretly is that there is no particular incentive for them to view it as an important civic duty and the whole process would effectively take place in camera — why should the public have confidence in a decision process that they cannot witness with their own eyes? I do believe the perceived legitimacy (referred to as “empirical” legitimacy in the recent Mansbridge post) of sortition is essential if the allotted body is to have anything more than advisory powers.

    Like

  4. Terry,

    >Silent (imaginitive) deliberation merely SHIFTS the legitimacy question from the random body to the legitimacy of the ORGANIZERS who structure the information presentations and frame the questions.

    Yes that’s true, and that’s why I’m increasingly attracted to Naomi’s argument that information advocacy rights (isegoria) should be secured by some kind of elective process. The statistical mandate just doesn’t apply to speech acts, that’s what elections are for.

    >this was preceded by substantial deliberation in the Council of 500 (the Boule).

    Do you have any concrete evidence to support the claim that the council deliberation was “substantial”? I have great difficulty imagining a conversation between 500 people. Of course there was widespread deliberation in the agora (or civil society, to give it its Habermasian name), but in large modern societies, characterised by an absence of homonoia, some sort or representative mechanism is needed to ensure that the information advocacy that gets through to the final policy jury is anything other than random.

    Like

  5. Terry,
    Thanks for your comments. I think I would agree with a lot of what you say in that there’s a time for talking and a time for not talking. The challenge is to identify where the first ends and the second should begin. My concern is that at the decision making stage it would be nice to have, as much as possible, high quality deliberation and reproducibility. If you have both there’s a greater case to be made for legitimacy I think. You are quite right to say that an awful lot of my argument hangs on ‘the organisers’ getting it right; I’m a bit hand wavy on that I accept. At worst, you are right in suggesting that this simply shifts the problem (like leaping on an airbump in your carpet to get rid of it…). One would have to have a robust case that ‘the organisers’ had passed tests of fairness and reasonableness… I accept there are costs to reproducibility and costs worth bearing, as long as the problems are minimised as much as practically possible.

    I don’t think I’m against a multi-stage process whereby real world talking feeds into the identification of the choice that must, at a final stage, be faced by the silent deliberators who produce a reproducible result. .

    Keith,
    Thanks also. John Dryzek was obviously very successfully silently deliberating. I haven’t read Andy Dobson’s book and should do so. I think there’s a lot in your 2 points re actually getting the citizens together. My concern is that they may not remain independent. Laughter and sighing out loud (maybe this is just a Northern Ireland concern…) may influence the other members of the audience. On balance, if the audience were asked to listen in silence then there are good advantages re everyone taking it seriously and it being utterly transparent.

    Also, in those circumstances I would be all in favour – post-voting – of the citizens having a party/ celebration of sorts and engaging in all kinds of talking about what had occurred. The talking and discussion would be great as long as it occurred post-vote. The talking and discussion could aid an interpretation of the result and add rich and nuanced qualitative flesh to an understanding of the result. This may assuage somewhat the concerns of deliberative democrats who see it as very unfortunate that talking and discussion is not a part of the process.

    Like

  6. John,

    >I don’t think I’m against a multi-stage process whereby real world talking feeds into the identification of the choice that must, at a final stage, be faced by the silent deliberators who produce a reproducible result.

    It would be hard to disagree with that, but if there’s a trade-off between the epistemic benefits of talking, and decision-reproducibility then I think we need to opt for the latter. Terry’s argument is that legitimisation will, in part, be a product of the “good” epistemic outcomes resulting from the deliberative exchange, but the nature of the good is the subject matter of that highly contested field that we call politics. Habermasians focus on the benign epistemic results of the deliberative exchange but, if the allotted body is to have a statutory role, then we need to focus on the fact that the vast majority of citizens will be disenfranchised by the aleatory coup d’etat. Universal franchise has been the end result of a long struggle and I can’t see people voluntarily giving it up unless it can be conclusively demonstrated (ideally by experiment with multiple concurrent juries) that it makes no difference at all which citizens are included in the jury, the result will be the same. My own little vote may not count for much, but all voters are equally impotent. In this respect the Mansbridge paper that was referenced yesterday, which focuses on democracy as legitimate coercion, is highly relevant.

    >I think there’s a lot in your 2 points re actually getting the citizens together. My concern is that they may not remain independent. Laughter and sighing out loud (maybe this is just a Northern Ireland concern…) may influence the other members of the audience.

    Yes, but I think this can be reduced to an insignificant level by the adoption of an appropriately dignified decorum (not that far removed from the expectations as to how you should behave in church or a public library). The model should be the House of Lords, not Prime Minister’s Question Time.

    Like

  7. Keith,
    I think I agree on both counts. But there is the obvious point that the 7 billion or so humans on the planet will continue to speak to each other. The politics bits of everyday talk will be part and parcel of what we call political debate broadly among the population and in the media. Whatever issue that is identified (by whatever means at whatever time point) as needing a collectively binding decision will have been affected by talk that has already occurred. That talk cannot be untalked. So, if we stop the clock on talk at this point and let the already occurred talk have an influence (among other influences) on what precisely the choice set is and what the different perspectives may be with respect to that choice set, the silent deliberators can then be let loose on the issue. I think multiple concurrent juries is a good idea, although we can expect that if 20 such juries were instigated 1 of them would get the answer ‘wrong’ (at the 95% confidence level).

    Like

  8. John

    I have no wish at all to put an end to deliberation, my concern is simply not to arbitrarily privilege the speech acts of a tiny number of randomly-selected persons. Public debate should be encouraged as widely as possible, but the information/advocacy input that the silent deliberators receive should be balanced (as in your own movie proposal) and representative (I’m aware that there is a tension between these two desiderata). Ditto with the significance threshold — 19/20 @ 95% sounds fine to me (I would not have been quite so exacting).

    Like

  9. Keith,
    Yes, totally agree.
    Also, on the point about silent deliberators deliberating silently in the their own homes versus in the same large grandiose room… While I agree that the importance of the occasion will be obvious to all (plus transparency advantages) in the same grandiose room there is a potential cost re the number of people involved. 1000 might fit in which would be fine, but if one wanted to minimise as much as possible the margin of error around any estimate (any % distribution regarding a decision) and hence maximise reproducibility then 2000 might be better, but are there any rooms where we can reasonably fit 2000? This isn’t a major deal either way, but it’s a point nonetheless.

    Like

  10. John,

    Your knowledge of statistical significance outstrips my own by several orders of magnitude. I feel that my role as a political theorist is merely to make distinctions between various kinds of representation (descriptive, substantive etc) and suggest how these might correspond to various forms of institution, it’s up to comparative political scientists to work out the details. But apart from cost and practicality issues there is the problem of the rational ignorance threshold — at what point would a juror’s vote be so watered down that she wouldn’t bother to stay attentive to the information/advocacy? I’ve always assumed that Fishkin’s preferred DP size (c. 350) is of adequate statistical significance but appreciate that larger numbers might be required for a body with some kind of statutory role. Anyway, that’s for you institutional polsci guys to decide.

    PS on the marginality issue, I’m assuming a simple majority vote, so how about votes that end up in the 45-55% region being referred to a larger jury? That way the default size could be kept more manageble.

    Like

  11. Keith,
    On the PS: A good suggestion, although for me I’d always want an absolute minimum of 1000 plus I would make participation compulsory to overcome the rational ignorance problem. This sends shivers up many people’s backs but I see attendance as pretty much the same as paying income tax: the system needs it to be non-voluntary, so therefore it’s non voluntary.

    Like

  12. Agree on the need for compulsory attendance, and will bow to your superior knowledge on the minimum number required for statistical representativity. Perhaps we can have a charge hand going round with a whip to make sure that everybody stays awake!

    Like

  13. The whip system is something to avoid

    Like

  14. Another option, besides large sample sizes, might be requiring multiple consecutive passages by more than one sampling. If there’s a delay between votes, then there’s opportunity for large-scale public deliberation (the media, conversations around countless water coolers, etc.) and this will be reflected in the second sampling.

    If the odds that a proposal that should not pass, ideally, passes because of the randomness in the sampling are 10% in a single assembly, they fall to 1% in a two-vote system. If the odds of accidental defeat of a proposal that should pass, ideally, are 10%, they only grow to 19% in a two-vote system. As far as status quo biases go, that’s pretty reasonable. The losers could just try again. Maybe if the first vote clears some supermajority threshold we could skip the second vote. We wouldn’t want to needlessly delay housekeeping proposals, disaster relief aid, etc.

    One other possibility to consider is sample stratification. While it is not possible to stratify every salient detail, as long as there is correlation between factors we stratify for and policy support we will get more consistent results.

    What’s the maximum term length we could have if we also have compulsory attendance?

    Like

  15. Naomi,

    If attendance were mandatory then the term would have to be very short, as in trial jury service. One of the problems with multiple consecutive votes (apart from the cost and administrative delay) is there could be a strong pull away from balanced advocacy (as specified in the original post) as some members of the “public” are a lot more equal than others. This would make it easier to manipulate the result, unlike in the case of a quasi-sequestered jury. I think I could live with John’s suggestion of a minimum of 1,000 jurors, with marginal outcomes being referred on to an even larger sample.

    Like

  16. Right. But just how short? I don’t know much about grand juries, but I believe they often meet weekly for months. I’m thinking the ideal term length is probably on the order of 3 months to a year. A year is probably out of the question, but I’m not 100% sure about mandatory 3 month terms.

    > there could be a strong pull away from balanced advocacy (as specified in the original post) as some members of the “public” are a lot more equal than others.

    You mean the media elite? It’s a fresh sample each time, so I’m not sure there’s a problem. If the media doesn’t like a proposal, and can shift public opinion enough to get it defeated despite the deliberative process, they can surely shift public opinion enough to get the proposal repealed later.

    I’m a bit surprised to hear you endorse such a large group. That’s less than an order of magnitude from the full Athenian ecclesia. The danger of referring marginal results to another body is that the need to pay attention and apply one’s self is diminished in the first round. The members will know that marginal results will go on to a larger body… in terms of rational ignorance this is probably equivalent to just putting everything to a larger body in the first place.

    Like

  17. John,

    Beyond the severe problems of the effects of the constraints you suggest imposing on the allotted chamber (as Terry pointed out), your approach (which is rather common among those advocating the use of sortition in government in one way or another) suffers from a foundational problem: who gets to determine the rules the allotted chamber has to follow? It seems you have appointed yourselves to set those rules. How can that be legitimate? Any attempt to constrain allotted chambers involves subordinating the representative chamber to a non-representative body. How can that be legitimate?

    Like

  18. Naomi,

    >But just how short?

    Most trial juries sit for one case only, and this is the principle that I would go for — exactly how long would depend on the importance and complexity of the particular proposal being considered.

    >You mean the media elite?

    The principal attraction of the DP model (and John’s proposal) is that the aim is to ensure that the final judgment is informed in a balanced way. I’m nervous about anything that departs from this principle. To my mind the role of the media is originating and championing proposals, rather than seeking to manipulate deliberative scrutiny.

    >I’m a bit surprised to hear you endorse such a large group.

    I don’t like to come to a fixed view on anything that’s beyond my pay grade (my knowledge of statistics is very poor). Exactly where the rational ignorance threshold lies is subject to a number of factors and could probably be determined experimentally (by trial and error).

    Like

  19. Yoram,

    >Any attempt to constrain allotted chambers involves subordinating the representative chamber to a non-representative body. How can that be legitimate?

    But that’s exactly what you’ve proposed in your suggestion that the allotted body should allot a rule-making subgroup, as the latter would be unlikely to be of a sufficient size to benefit from the law of large numbers. And that’s before the problem that most randomly-selected conscripts are unlikely to have any experience of (and interest in) constitutional matters, leaving the outcome largely to chance. Why do you consider that to be “representative”?

    Like

  20. Keith,

    Two points…

    1. It isn’t helpful to use name calling (“aleatory coup d’etat”) for a system of sortition with active deliberation. It might be voted in by referendum, and have more popular support than a system of elected legislators elected in many countries by less than half the eligible voters. Rather than DISenfranchising, voters might see it as REenfranchising themselves collectively over the political class.

    2. One of your arguments for preventing randomly selected mini-publics from engaging in active deliberation is that we should not “arbitrarily privilege the speech acts of a tiny number of randomly-selected persons.” But why should we arbitrarily privilege the speech acts of those who are wealthy enough to own television networks, and newspapers, or the ego-driven members of the political class who choose go into electoral politics and are able to master public relations skills and raise enough money to win? Isn’t it fairer to place EVERYBODY on an equal footing in terms of their CHANCES to be part of that minority who get heard by the ultimate decision makers?

    Like

  21. Keith,
    To the extent that there’s a difference between championing a proposal and seeking to undermine the ability of a future statistical sampling to deliberate on a proposal, it would seem that the preferred option for the media would *always* be seek to undermine the ability of a statistical sampling to deliberate. That would seem to be advantageous under any system built around the lot.

    Like

  22. Terry,
    >”But why should we arbitrarily privilege the speech acts of those who are wealthy enough to own television networks, and newspapers, or the ego-driven members of the political class who choose go into electoral politics and are able to master public relations skills and raise enough money to win? Isn’t it fairer to place EVERYBODY on an equal footing in terms of their CHANCES to be part of that minority who get heard by the ultimate decision makers?”

    You don’t need to be rich enough to own a newspaper to win a seat in a nation’s legislature. You don’t have to be ego-driven to win a seat. You don’t have to master public relations to win a seat. Campaign financing is really no business of an individual candidate. I would caution against painting all elections with an American brush. No one here is talking about a candidate-centric approach. We’re talking about privileging *organizations* with a high enough profile to where they can get enough members/supporters to win seats. I don’t have a problem actually acknowledging the movers of public opinion and giving them a related institutional role. Like it or not, they can make or break a system.

    Regarding fairness it seems we are comparing infinitesimals. Is it more fair to give everyone an equal vote or an equal chance? The odds your vote will decide the results are basically zero and the odds you’ll win a chance to speak are basically zero. Completely eliminating mass participation bothers me. It seems there is value in encouraging large-scale political discussion/debate/participation. If only the lottery winners get a say, why should regular Joes give politics any thought at all? That doesn’t seem healthy to me. It also doesn’t seem plausable. Even if the vote is reduced to ceremonial purposes (like monarchy in the UK) it has been cherished too much for too long to just do away with.

    Like

  23. Terry,

    >Rather than DISenfranchising, voters might see it as REenfranchising themselves collectively over the political class.

    Yes, iff it can be demonstrated that it makes no difference which empirical individuals are included in the sample. The second half of your sentence, however, shows the degree to which your perspective reflects a rather old-fashioned notion of class consciousness, the only difference being that your labels are a non-specific (“themselves collectively” — vs — “the political class”, as opposed to proletarians — vs — bourgeoisie). “Aleatory coup d’etat” is a deliberate shock-jock term to indicate the fact that this would be undoing at a stroke centuries of struggle in expanding the franchise. Yoram has ignored my repeated enquiries as to what the kleristocratic vanguard would do if (as it seems likely) his all-powerful small group of allotted rule-makers decide that they don’t want to consign “electoralism” to the trash can of history. This is most likely because he refuses to read any of the drivel emanating from the capitalist lick-spittle “Sutherland”, so perhaps you might respond on his behalf.

    >But why should we arbitrarily privilege the speech acts of those who are wealthy enough to own television networks, and newspapers,

    Because, given a genuinely competitive media market (which we are closer to in the UK), the privileging is anything but arbitrary. People choose whether to subscribe to the Sun or the Mirror, the Telegraph or the Guardian, and those with the most subscribers (rightly) have a stronger claim to isegoria. I believe In the US there are a lot of local newspaper monopolies and in Italy Berlusconi owns everything. Not so in the UK — although Murdoch owns the Sun and the Times, people buy them if they like the views of the newspaper (the Telegraph outsells the Times), not because they like Murdoch. The latter has acute political sensibilities and is brilliant at tracking the changes in public opinion and tacking accordingly.

    >or the ego-driven members of the political class who choose go into electoral politics and are able to master public relations skills and raise enough money to win?

    Because voters can (again) choose the isegoria that appeals to them. We have a general election in five weeks time in the UK and last night there was a two-hour televised debate in which SEVEN party leaders competed against each other on an entirely level playing field. The Green Party (who had exactly the same TV profile as the Conservatives) are extremely poorly funded. If the UK voting system were PR, then both the Greens and UKIP would gain a proportionally strong representation after the election. There were very clear differences between the policies on offer (the Greens, Plaid and SNP are as near to socialist as you can get) and democratic norms would suggest that their policy proposals should be privileged in proportion to how many votes they receive in the election (prior to deliberative scrutiny and an up/down vote by an allotted sample of all citizens).

    Naomi,

    >it would seem that the preferred option for the media would *always* be seek to undermine the ability of a statistical sampling to deliberate.

    That’s why the media’s role should, as far as possible, be quarantined to the earlier stage (initial policy advocacy). DP experiments (mirrored by John’s proposal) suggest that the final deliberation should be guided by balanced information advocacy, rather than being browbeaten by majority opinion. If the arguments of the minority view are best then the deliberative sample should be empowered to overturn the prejudice of the majority, so this would suggest a degree of sequestering while the jury deliberates. In the case of the trial jury, most people accept that the jury, who has considered the arguments for and against, is right, even though their verdict may run counter to expectations. “Trial by media” is synonymous with miscarriage of justice.

    >Regarding fairness it seems we are comparing infinitesimals.

    Yes, in large political systems we should focus on representativity, rather than the chances of a single individual having a casting vote or winning the lottery.

    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: