Sortition is natural to democracy, as elections are to aristocracy

An introductory presentation about sortition for a talk I’ll be giving to a general audience. Comments are welcome.

52 Responses

  1. Hi Yoram,

    Maybe a reference to Montesquieu would be nice in your first slide. Otherwise people would think this is your quote


  2. Hi Klirosi,

    I see this quote as Montesquieu quoting Aristotle. I give quotes of both Aristotle and Montesquieu with attributions on slide #8. Anyway, thanks for the comment – I’ll try to see if there is any confusion when I give the talk.


  3. Hi Yoram,

    Who’s the intended audience?


  4. I am planning to give a talk at my workplace, so the audience are Israeli engineers.

    I would like, however, to have a stock presentation that I could use on any occasion that would call for an introductory talk on sortition, so I am not targeting this audience specifically, except for the last part – the game theory formalization – which is somewhat mathematical.


  5. >representativity – r(d): the government is representative when its efforts are aimed at promoting the general interests (rather than personal or narrow interests)

    Is it the job of allotted members to express an opinion as to what is in the general interest or are you satisfied with aggregating their preferences? If the former then how do you ensure that individuals are doing so and (more importantly) how do you ensure that the policy options presented are in the general interest, given that they have been introduced by a random process? (the volitions of a small number of individuals, most of who would not have previously thought a great deal about political problems). How would you justify privileging the speech acts of these individuals over those of the vast majority of citizens who have been disenfranchised by this process?

    To boil it down to just two questions:

    1. How do you ensure that a body that “looks like America” will also act like America?

    2. Is that sufficient to ensure government in the general interest?


  6. Keith,

    I laid out the argument on slides 4 and 5: The group of delegates promotes its own interests, which are aligned with those of the rest of the population due to the statistical properties of the selection procedure.


  7. I think two specific questions deserve two specific answers, so let me repeat them:

    1. How do you ensure that a body that “looks like America” will also act like America?

    This is the problem that Peter Stone poses in his introduction to Callenbach and Philips: “[The Representative House] just has to exist, and somehow whatever it winds up doing will be right, precisely because it descriptively represents the people as a whole” (2008, p.14). My question is a proxy for Peter’s, and I think it merits a reply.

    2. Is that sufficient to ensure government in the general interest?

    You don’t indicate in any of your slides why individual members would do anything other than calculate their own selfish interests. For example they might decide that it would be in their interests to exterminate all Jews and redistribute their property amongst the rest of the population. So why would government by sortition necessarily lead to the common good? (as opposed to the greatest good of the greatest number). To choose a more plausible example they might decide to continue the present policy of democratic states whereby they run up huge fiscal imbalances by borrowing to fund current expenditure, thereby condemning their grandchildren to a life of penury. Why would a change in ballotting procedure automatically fix this problem? Or perhaps you don’t see it as problematic at all, in that you are opposed to any constitutional checks and balances that constrain the actions of the allotted assembly, so the latter seems distinctly possible (along with the lynching of all paedophiles, even if the odd paediatrician gets strung up by mistake). Government by the random whims of a small number of unaccountable individuals selected ‘by bean’ would certainly not fit very well with Burke’s view of society as a compact between the past, the present and the future, and would be equally problematic for all those of us committed to democratic equality.


  8. Keith,

    I don’t think that the first question you are posing is relevant to arguing for (or against) sortition. I don’t know what “act like America” means. “America” doesn’t legislate, so how would you know if a legislative chamber acts like America?

    As for acting in the interests of the public, the argument, as it appears in the slides, and as it is summarized in my previous comment, is straightforward. If you find it unconvincing, I would be interested to know what are the points in the argument with which you disagree.


  9. An allotted chamber would “look like America” for all the reasons that Callenbach and Philips state (“50% women . . . 12% blacks . . . two accountants . . . and a Buddhist”). But why should the “output” of an allotted assembly with full powers of policy initiation and advocacy “automatically” (C&P) or “spontaneously” (Goodwin) represent the interests of America? Peter argues that this presupposes an “adequate theory of collective rationality”. I would put it rather more prosaically — the speech acts of the Buddhist or the accountant would only represent themselves, and those of the accountant would probably carry more weight. What if the “Buddhist” were a Trappist monk — his speech acts would not carry much persuasive power (on account of his vow of silence), so those who he descriptively represents would be disenfranchised. I agree with you that America doesn’t legislate; introducing and arguing on behalf of policy proposals is the act of individual human beings and this does not benefit from a descriptive-democratic mandate that only applies at the collective (judgment) level — a body that looks like America would certainly judge like America, or else the public opinion industry would long since have disappeared. But the views of individual respondents in a public opinion poll are of no interest to anyone other than themselves.

    On the second point the distinction between discerning the common interest and aggregating preferences is a longstanding one (stretching back at least to Rousseau). Which model do you assume would result from sortition and why? You invited comments, so I think you owe it to us all to answer specific questions, rather than just referring people to your existing text.


  10. Keith,

    Regarding “common interest” vs. “aggregating preferences”: I don’t see how drawing such a distinction is useful for our purposes. Like any group, the group of delegates is expected to promote its own interests as it perceives them. Whether you call that promoting their common interests or aggregating their preferences seems like a matter of labeling without any policy implication.

    Again, I just don’t see how the claims you make are relevant to the matter at hand. I would like to invite you to engage specifically with the argument I present. It seems to me that you would have to argue at least one of two positions:

    1. The delegates would be unable to promote their own interests.


    2. The interests of the delegates would not be aligned with the interests of the population.

    Do any (or both) of these appear valid to you? If so, which one, and why?


  11. >1. The delegates would be unable to promote their own interests.

    It’s interesting that you use the word “delegates”, implying that they have been sent to the assembly by some group and possess a binding mandate. Of course this is not the case in an allotted assembly as there is no formal relationship between allotted members and their “constituents” — they certainly cannot be fired if they are in breach of the delegate mandate (unlike elected representatives). Each individual would promote her own interests, which may or may not correspond with the the interests of the wider political community. However, even if there was an exact proportionate mapping between the interests of each member and the wider community, some individuals would have greater powers of persuasion and a higher perceived social status than others. So some would be better equipped to promote their interests (and those they “represent”) than others and this would be a fundamental source of inequality of political power. Obviously this would not be the case if all allotted members did was vote, as all votes carry equal power. So, in answer to your question, some “delegates” would be better able to promote their interests than others.

    >2. The interests of the delegates would not be aligned with the interests of the population.

    Again I note your unconventional use of the term “delegate”. At the aggregate level there would be a correspondence of interests, the closeness of which would depend on the size of the sample. But the accuracy of the correspondence also depends on each unit of the sample having equal power — that would be true when it comes to voting, not so regarding the other functions that you propose, for the reasons that I’ve just outlined.

    I continue to be puzzled as to why we continue to talk past each other on this issue, and can only conclude that you are assuming a commonality of interest amongst the majority of the “masses”. As the “elite” will, by definition, be in a minority, an allotted assembly would not represent elite interests, it would “automatically” (C&P) or “spontaneously” (Goodwin) reflect the interests of the masses — irrespective of who did most of the talking. In my view modern pluralistic societies are composed of individuals with a very wide range of interests, so any illocutionary imbalance in the allotted assembly will be a serious source of political inequality.

    I’m sure this point would be easy to demonstrate mathematically.
    I’ve probably got the terminology completely wrong (I haven’t done any math for over 40 years), but if you are summing two identical integers and there are a vastly more of one than the other, it matters less if the (Euclydian) vectors associated with each integer are unequal. However this would not be the case if there were a large number of different integers, in which case the associated vectors would play a more important causal role. Votes involve summing simple integers, whereas speech acts involve the interplay of integers and vectors. Apologies again if this is mathematical gobbledegook, I’m aware that I’m trespassing in a domain that I know nothing about. If so, perhaps you’d like to correct the terminology and produce some simple equations that can be understood by a lay audience.

    I would be interested to hear from Peter as to whether or not I have accurately conveyed his criticism of the Callenbach and Philips position, which you have frequently acknowledged as the template for your own proposals.


  12. Keith and Yoram,

    Unlike either of you (perhaps), I think I have understood both of your perspectives on this issue for quite some time…Let me try again to summarize them to see if this helps.

    If we imagine a magical situation in which all members of a society can deliberate and make a decision (I will call “X”) democratically, we can hold X as the standard by which we judge whether a representative body is also democratic.

    Yoram suggests that a sample of sufficient size will not only correspond to the appearance of the population, but also the beliefs and BEHAVIOR (including speech acts) of the larger population. He surmises that this group would also come up with decision X most of the time (with statistical deviations).

    Keith, on the other hand suggests that the unique individuals within many different samples we might draw who are engaging in agenda setting, the initiation of policy, and speech acts would steer the sample in many different directions, perhaps coming to X, but also often coming to Y, Z, and many other decisions.

    If the sample were fully HALF of the population and they could still magically all participate, Keith might accept that they would very likely end at X as we are assuming the full population would have. If the sample were just three people, Yoram would agree they may well NOT arrive at X. If the sample were a few hundred, Keith thinks this is more like a sample of 3 as it relates to X, and Yoram thinks it is more like the 50% sample.

    In short, Yoram believes the individual speech acts sort of all even out in the end to come pretty close to what the full population would end up with if they could all participate.

    Since we can never really know what X actually would be, this is hard to test…but Yoram suggests that a sufficiently large random sample deliberative body with full powers would almost inevitably come closer to X than any elected legislature would.



  13. Thanks Terry, I think that’s a very accurate summary. Unfortunately we’ll never know what X is because the threshold of rational ignorance is remarkably low. The reason I’m doubtful about your last paragraph is because elected politicians are also obliged to take an educated guess as to X. Given the level of uncertainty involved I think we are obliged to take the precautionary principle and limit ourselves to innovations that we can demonstrate (both practically and theoretically) will deliver — even if this produces a pale imitation of what both Yoram and yourself aspire to.

    The other alternative is for Yoram to tap up some rich benefactor and ask her to fund the sort of allotted assembly that he proposes — the trouble is that wouldn’t work either because it would not have any real-world function. So I guess slow incremental progress is the only option.


  14. Terry,

    No – I am not making any claims about a correspondence between the policy outcome of an allotted chamber and the policy outcome of an impossibly large face-to-face forum. The argument I presented does not require resorting to imagining this unlikely hypothetical situation.


    First, I use the term delegation to designate a group of people which are given a specific task (in our case, the task of legislation). The delegates are simply members of the delegation. (If you offer a different term to designate this function, I would be willing to consider adopting it.)

    Substantively, it seems like you base your objection to my argument on the claim that the group of delegates would have inequalities of power within it.

    Of course, in every group some differences in power are inevitable, no matter what the setting is (in a pure voting situation as well, BTW). However, I am not sure how you would tie this claim to an argument about sortition since this objection is really about small group dynamics.

    Are you claiming that a parliamentary procedure is not a good way for a small group to do its decision-making? Are you claiming that it is better for a small group to have its agenda dictated by an external agent rather than be set by the members of the group itself?


  15. Yoram,

    I am confused about your argument then after all. If it doesn’t matter if the decision of the allotted chamber has any correlation with what a hypothetical full-society genuine deliberative process would decide, then why not select one person at random to make decisions for a short time. What is GAINED by having a descriptively representative group if not some hope tor an outcome that would be similar to what the whole society would decide it it could universally participate?


  16. Terry,

    Sortition generates an alignment between the interests of the delegation and those of the population. It therefore creates a situation in which a policy which promotes the interests of the delegation promotes the interests of the population as well.


  17. Yoram,

    I’m disappointed (but nor particularly surprised) that Terry’s wise mediation has fallen on deaf ears!

    The appropriate term for members of an allotted assembly would be “representative” as they have no binding mandate, relying fully on their own judgment. The justification for their presence in the assembly is that they represent other people who would judge in a similar way to themselves and their numbers in the assembly are in proportion to the presence in the community of like-minded people. This is a variant of the Burkean concept of “virtual” representation, both on account of the free mandate, but also in the sense that representatives are bearers of unattached interests (the only difference being that their claim is to be typical, rather than the best). The concept of delegation is wholly alien to Burke’s notion of representative government — it is mostly used nowadays in federations, trade union and partisan congresses, and similar bodies where horse-trading and the wielding of numerical power prevail over deliberative judgment.

    Where are the power inequalities in voting? An individual vote is a pure integer and they are simply added up. It’s been a long time since the abolition of multiple votes. Needless to say the equality of votes presupposes the secret ballot and we are not discussing the influences that persuade people to cast their votes one way or the other, merely the balloting mechanism itself.

    The problem that I’m raising is only partly to do with group dynamics — parliamentary procedures are fine when the participants are elected representatives because they have all been selected (partially) on account of their ability to participate in verbal exchange. You would be unlikely to vote for a candidate who lacked any ability in verbal exchange, even if she had devoted her life to public service and her views were identical to your own. The problem is deeper: most people (unlike parliamentarians) have little a priori knowledge about political issues, would have little to bring to the table (a priori) in terms of policy proposals, and as such would be likely to come under the sway of the tiny number of allotted members with their own agenda and/or high status. As a result there is no particular reason to believe that the deliberations of an allotted chamber would approximate to X — it’s far more likely that a number of different chambers would produce different policy agendas (as they input would be entirely random), so the chamber could not claim to represent the considered view of the whole political community (X). So, although it’s partly a matter of group dynamics, it’s mostly because the members would not have been selected on the basis of relevant criteria.

    Allotted representatives, however, are ideally placed to judge the outcome of the legislative debate. To choose a relevant analogy: few of us (other than art critics) have much knowledge of the history of art, the various schools of painting, or the technical skills involved, but we all know what we would like to see hanging on our sitting-room wall. Art critics have a role to play in suggesting what we might like to consider — they might even force us to challenge some of our prejudices and to look at innovations in the field — but ultimately only we know what we want.

    >Are you claiming that it is better for a small group to have its agenda dictated by an external agent rather than be set by the members of the group itself?

    I am, indeed, suggesting that the agenda should be externally set. The way to avoid “dictatorship by an external agent” is to ensure a plurality of agents alongside democratic innovations that would enable any citizen (not just the few who draw the short straw) to bring forward a policy proposal. Those who insist that such processes are inevitably dominated by imbalances of power, wealth and influence should redouble their efforts to levelling the playing field.

    No doubt Terry will wish to respond himself, but it really doesn’t help the argument to progress if you just repeat slogans. It’s true that Terry was proposing a fanciful thought experiment, but such hypothetical devices (take for example Rawls’s “original position”) are a valuable way of uncovering the norms that we are seeking to pursue in practical politics, even if they have little relevance to empirical reality. If we are democrats then surely we are ALL trying to establish the correspondence between x (the considered will of the microcosm) and X (even if this is entirely unknowable)? If this is not the case then we have to assume that you are pursuing a different agenda from the democratic one. Sortocracy is not a variant of democracy, it is government by random fluctuation.


  18. > I’m disappointed (but nor particularly surprised) that Terry’s wise mediation has fallen on deaf ears!

    I am sorry for your disappointment, but we are not in a situation of a personal conflict which calls for mediation. My objective is to discuss a theoretical analysis in order to achieve a better understanding of the matter being analyzed. This is not something that will be advanced by appealing to a compromise.

    > The appropriate term for members of an allotted assembly would be “representative” as they have no binding mandate, relying fully on their own judgment.

    I see the term “representative” as being too loaded to be useful. It implies that the member of the delegation represents someone or something. This may or may not be the case, depending on various factors, so the term may or may not be appropriate.

    > Where are the power inequalities in voting?

    This is tangential to our discussion as far as I am concerned, but here is a simple example (and many other examples can be imagined): if some of the members of the voting group are not fluent in the language in which the proposed piece of legislation is written then they are clearly at a disadvantage. (Even if a translation is available, there may be ambiguities in the translation which are not present in the original proposition.)

    > I am, indeed, suggesting that the agenda should be externally set.

    So, according to you, if my family considers going on a vacation we should avoid suggesting destinations and discussing the matter. Instead, we should have a travel agent send us materials about a few destinations (of the agent’s choosing) and then vote which of the destinations proposed by the travel agent we prefer?


  19. Yoram,

    We both agree that the prime function of sortition is descriptive representation — to create a portrait in miniature of the whole polis. I also agree that the idea of an individual allotted representative is misleading, as there is no clear and definitive relationship with a finite constituency. Nevertheless the collective function of the group is representation of the interests of the whole polis (we agree on that too) so the word representatives (plural) is preferable to delegates, as they are serving a deliberative function and have no binding mandate (they could not, as they have no finite constituency). Remember that delegation is a (highly restrictive) variant of representation in which the delegate has no freedom of choice, she is just a passive token for the interests that she represents. Assuming, charitably, that isn’t what you are proposing, let’s just call them allotted members that serve a representative function.

    Regarding the travel agent analogy, in the political case the family would be deciding on the destination that everyone should be taking, not just themselves. It’s only right that everyone so affected should have the right to suggest their preferred destination and, if they garner sufficient backing, to have their choice considered by the allotted assembly. This would mean that someone proposing (say) ice-fishing in Siberia would be unlikely to get enough backing to achieve the threshold required to get on to the public votation ticket.

    Assuming the travel agent refers, by analogy, to the professional politician, it certainly used to be the case, in a more deferential age and when few ventured beyond Blackpool or Scarborough, that people would trust the judgment of their local travel agent, whereas now they will be more likely to trust the TripAdvisor rating. If there is a future for political parties it will be in a combination of the two — parties could offer policy options but if they subsequently failed (either at the stage of scrutiny by an allotted chamber or after they were implemented), then the party would suffer the equivalent of a bad rating on TripAdvisor and would lose the trust of the electorate in subsequent elections. They might well switch to a different travel agent. This has the advantage of providing an ongoing mechanism of ex-post accountability. Anyone can set up as a travel agent and, given that the focus would now be on issues rather than parties, it’s likely that many new parties would evolve out of single-issue advocacy groups. TripAdvisor is open to manipulation and it’s always better to shut the stable door ex-ante, rather than wait until the horse has bolted, but a judicious combination of the two principles might well be the best we can hope for.


  20. Again, I see “delegation” as appropriate since this is a group of people given specific task (legislation). I don’t find your arguments in favor of “representatives” particularly convincing, and I don’t think that the word used makes any real difference as long as we understand the meaning, so I suggest we just each stick to the term we prefer instead of carrying on with this pointless quibbling.

    > Regarding the travel agent analogy, in the political case the family would be deciding on the destination that everyone should be taking, not just themselves.

    No – the family simply makes the decision for themselves. My question was simply this:

    Should a small group (trying to make the decisions on their own behalf, promoting their own interests as best they can) rely on an external agent to dictate the agenda to them, or should they set the agenda themselves?


  21. I assumed that the analogy was to an allotted assembly — if not then what?


  22. Again, I am talking about an arbitrary small group of people. They could be an allotted chamber, or a family, or a group of friends, or business partners, etc.

    We agreed, I believe, that within any such group there would be power inequalities to some extent. My point was that – even in the presence of such unavoidable inequalities – it is accepted that the best way to have decisions made in such a group so that the interests of the group members themselves are promoted is by using parliamentary procedures: open discussion with endogenous agenda setting and equal weight voting.

    Do you agree? If not, what is a case in which it is better for the group to rely on a different procedure?


  23. >What is a case in which it is better for the group to rely on a different procedure?

    When the only reason for the existence of the group is that it is a statistically-representative microcosm of a much larger population. The only relevant example in your list is an allotted legislative chamber. Families may choose to rely on their own experience when choosing a holiday vacation (rather than using the services of a professional) but, if they were deciding where everyone else should go on holiday, then relying on their own expertise would lead to accusations of partiality. This would be especially the case if the group were charged with deciding the length of holiday entitlement, the budgetary constraints and the tradeoff between expenditure on holidays and (say) the armed services. In the case of the family, the group of friends and (to a lesser extent) business partners, the decisions they take affect themselves only (leaving out unavoidable externalities). Unlike these examples of small groups, an allotted chamber serves a representative function, thereby necessitating different procedures.

    Although inequalities in speech acts affect all the examples that you provide, the problem is of an entirely different magnitude when compounded with the requirement to represent a body of interests in proportion to the prevalence of those interests in the whole political community. Unfortunately the speech acts would only represent the skills and status of the person involved and the resulting (entirely random) disparities would do irreparable damage to the representativity of the assembly.


  24. Yoram, I’m offline for a few days, back at the weekend


  25. So somehow the fact that a group was chosen as a random sample from the population makes it impossible for them to represent their own interests? Somehow, when it comes to this particular small group, it is better for the group members themselves that their agenda would be dictated against their will by an external agent?

    Seems unreasonable and undemocratic to me.


  26. Individual members of the group may seek to defend their own interests but, in the scenario you propose, there is no particular reason to believe this will reflect the priorities of everyone else. At the Dublin sortition meeting I suggested that statistical representation was rational at the collective level, as the ratio of properties in the sample to the whole is predictable to a reasonable degree of confidence. However it is a-rational (random) at the individual level as there is no way of knowing who will draw the short straw (or the golden ticket). Authorising individual speech acts would result in government by random fluctuation.

    >their agenda would be dictated against their will by an external agent

    In the scenario that I am proposing the agenda will be set by the same public, but manifested via a different mechanism (preference election or direct-democratic initiative rather than sampling). It’s a little eccentric to describe the whole political community as an external agent that is dictating to itself, although that might well be a variant of the dictatorship of the proletariat that liberals would find acceptable.


  27. > Individual members of the group may seek to defend their own interests but, in the scenario you propose, there is no particular reason to believe this will reflect the priorities of everyone else.

    So it appears that you are now abandoning claim 1. (i.e., “the delegates would be unable to promote their own interests.”) and opting instead for claim 2. (i.e., “the interests of the delegates would not be aligned with the interests of the population.”).

    Ok, then, what are the interests of the group of delegates that will not be aligned with those of the population?


  28. There’s no way of knowing. Your example is the proposing of holiday destinations — the product of the speech acts of random individuals, many of whom may not have even travelled oversees. A professional travel agent would be far more likely to know the full variety of destinations available and, as such, would be more likely to represent the knowledge of the whole. But a travel agent always leaves the final choice to the customer and this has two entailments:

    1. The destinations that she proposes will be selected to align with the taste and budget of the client. If she chooses destinations outside this range, she will lose her customers to rival travel agents.

    2. The final judgment (choice) will be the aggregate of individual votes. The members of the small group will have had time to study the travel brochure (and those of competing travel agents) and those who have travelled previously will also use bring their own travel experiences to bear.

    There are very good reasons to believe that other samples of the population would judge in a similar manner, given identical discursive input (from the travel agents), so the verdict of the small group can be extrapolated to the general population. This would not be the case if members of the group made their own proposals as, given modern low-cost aviation, there are a huge number of holiday destinations, so the options proposed by the subset of the individual members of the small group who had holidayed abroad would be likely to vary wildly between different groups. For such a group to have the power to impose the whims of a minority of their members on everyone else would be a case of random dictatorship.

    Note that I would also reject claim 1. Assuming that most of the group had not travelled abroad, they would also be better off if a choice of travel professionals were to suggest to them a range of destinations, rather than relying on the fact that one or two of their members had travelling experience or, worse still, knew someone who had and had vague (and inaccurate) memories of a conversation they had with them some years ago. It would also help if the travel agents had been pre-selected by a process in which all citizens participated, in which the past record (and broken promises) of the travel agents were held up to public scrutiny. This would be particularly apposite for agents who introduced extra charges after the travel contract was signed or who promised ideal-sounding destinations (New Jerusalems) which had not yet been constructed. They might prefer the services of agents who delivered their promises, even if the final destination was closer to home. Better to get some sort of holiday than get on a flashy new airplane that ran out of fuel and crashed.


  29. You seem to be back to arguing for 1., rather than for 2.

    You also seem to be assuming that the options are either (a) having the travel agent dictate a set of destinations or (b) having the family come up with the destinations without any outside help.

    In fact, no one is arguing that (b) is a good option. The alternative to (a) is not (b), but (c) having the family choose the destinations in any way they see fit. They can consult with a travel agent, or multiple travel agents, or use the web, or do whatever they wish.


  30. Family members can do whatever they wish, as they are not seeking to make decisions on behalf of everyone else. When dealing with political issues I think we both agree that most people have very limited information. Although in your proposal members of the “family” could seek information in any way that they see fit, there is no particular reason to believe that they would do this in a balanced and impartial way. As such the information and advocacy gained by each “family” would be likely to be very different, so there would be no way of ensuring the ongoing representativity of the microcosm. This is because the “ratio” element only applies at the collective level, not at the level of individual speech acts. Government by random fluctuations could not be categorised as democratic and would contravene the conditions necessary to ensure the epistemic benefits of cognitive diversity. This would breach 1 and 2, so why should we want it?


  31. > Family members can do whatever they wish

    The question, for our purposes, is not what the family members “can” do, but what would be the best way for them to make their decision. Can they expect to have a better vacation if they let the agent dictate the menu of choices, or would it be better for them to be in control of the process of selection of choices?

    > so there would be no way of ensuring the ongoing representativity of the microcosm

    Again, what is the way in which the interests of the allotted chamber might not be aligned with those of the entire population?


  32. It may or may not be better for them, depending on their level of knowledge, their contacts and, more importantly, their willingness to gain new knowledge in an unprejudiced way. But the crucial difference is that, in your analogy, they are deciding where everybody else should take their holidays.

    >Again, what is the way in which the interests of the allotted chamber might not be aligned with those of the entire population?

    If I had a dollar for every time that I’ve answered this question then I would be a fully paid-up member of the rich and powerful elite. Here’s my answer in a single sentence, capitalised for your convenience:


    Terry explained this at an earlier point in this exchange. To express it in a more general way, representativity is an “emergent” property of the group as opposed to a property that obtains at a lower level of analysis. An example of emergence: most cognitive scientists would claim that consciousness is an emergent property of a complex neuronal system, but few (other than panexperientialists) would argue that consciousness is a property of individual neurons. In the case of an allottted assembly representativity is a property that pertains at the group level, and does not apply at the level of the component individuals.

    It’s THIS argument that you need to refute if you are going to advocate an allotted assembly with the powers to set its own agenda and seek its own information and advocacy. It’s a very simple argument (perhaps that’s the problem you have addressing it — you’re looking for something more profound?), and the principles are widely used in philosophy of science in general.


  33. > But the crucial difference is that, in your analogy, they are deciding where everybody else should take their holidays.

    So, again, you seem to be abandoning your assertion of 1., and reverting to asserting only 2.


    So, it seems you are now abandoning 2. as well.

    Thus, the group is effective at promoting its interests by using small-group democratic procedures, and its interests are aligned with those of the population. It seems to me that this implies that the group is effective at promoting the interests of the population.


    This sentence is self-contradictory. Acts that are in the interests of the allotted group are not random, since the interests of the group are not random – they are aligned with the interests of the population due to the use of sortition.


  34. >Acts that are in the interests of the allotted group are not random, since the interests of the group are not random – they are aligned with the interests of the population due to the use of sortition.

    That’s just a tautology — the speech acts of empirical individuals may or may not be in the interests of the group. All the group can do is to apply its collective judgment to whatever speech acts it is presented with — endogonously generated or otherwise. And I don’t have your faith in small-group democratic procedures, even presupposing that they apply at the level of an assembly comprising several hundred people (which they don’t, the maximum size for a genuinely deliberative group being 12-15). In the case of deliberative assemblies with a tightly-constrained mandate, small-group procedures require careful moderation in order to ensure that they are not dominated by persuasive or high-status individuals. This would be impossible in an assembly with a statutory role and even harder to police if the group had the right of policy initiation.

    Even if the group were persuaded to accept an individual policy proposal as in its interests, do you really think that this would be viewed as legitimate by the general population? (who would not have experienced directy the persuasive powers of the policy advocates). The perceived legitimacy of such an assembly depends on it being comprised of people who “look like America”, but policies are advocated by individuals and would only be legitimate in the eyes of people who resemble those concrete individuals. Or are you suggesting that the assembly should deliberate in camera? If not, then rest assured that the focus of the media and the public would be on the concrete individuals who were seen to advocate the policy in question. Remember that the justification of the allotted assembly is descriptive representation, so only those who self-describe closely to these concrete individuals would consider themselves represented by proxy, everyone else would be effectively disenfranchised. Not so if the assembly was limited to an aggregate judgment role

    This is such a stillborn basket-case — on both the theoretical and practical level — and we haven’t even considered the zero probability of it ever being implemented, given the current distribution of power. By comparison the case for the extension of the jury principle to the trial of legislative proposals is a cake walk. So why don’t we unite behind something that has a faint chance of succeeding? It would certainly save us from having to go over and over the same arguments ad nauseam.


  35. Ok, it seems we are not getting anywhere on the issue of the argument I presented. I feel that you are not able to refute either of the two steps involved. I understand that you think you are, but I don’t really understand your arguments. I’ll leave it at that.

    The issue of public perception is separate. I think that in our age, when democratic ideology is dominant, the argument that the allotted chamber reflects the popular will could easily become conventional wisdom, just as it apparently did in antiquity.


  36. Fair enough regarding the theoretical arguments. Regarding the status of allottment in antiquity, this is the subject of ongoing controversy. Dowlen, for example, claims that there is no evidence to support your view, arguing that it was simply a case of the prevention of factionalism and corruption. Mulgan (1984) offers the received wisdom that the Assembly was the organ designed to express the popular will and that allotted institutions (the courts and the council) were developed to protect it from being dominated by aristocratic influences. Aristotle believed that the role of allotment was to take it in turns governing and being governed (no popular will there either). One can make a plausible argument for the fourth-century innovations (in particular the nomothetai) in epistemic terms — deliberative judgment being more reliable than the rational ignorance that characterised the Assembly. This is my preferred view, and I believe Hansen is also disposed in that direction. So what evidence do you have in support of your claim connecting allotment and the popular will in antiquity? I suppose you might argue for a connection between the popular will and the divine will, but the divinatory status of the lot has long been discredited.

    Or in any other period in history for that matter? Some people (Mueller et al, 1972, for example) connect this idea with Rousseau. But this is a mistake, as Rousseau only considered lot in the context of democracy — one possibility for the executive government — and the legislature is the domain of the popular will.

    So the popular will would appear to be associated in antiquity with direct democracy and in modernity with election. I’m not aware of anyone, either ancient or modern, connecting the popular will with sortition, but would be interested to see any evidence that you may be able to provide in support of your argument.


  37. I am not making any precise assessments about ancient ideology. I don’t think such assessments are possible, or are of much interest for our purposes. I was merely referring to the wide use of sortition in Athens and the matter-of-fact tone which was used by the ancient writers when asserting the connection between democracy and sortition.


  38. There is an important issue at stake here: the scholarly consensus is that the assembly was the instrument of popular will in Athenian democracy, the role of sortive institutions being to protect the assembly from corruption by partisan forces and/or to improve its epistemic output (in the case of the legislative courts). Any attempt to re-establish authentic democracy would need to respect this distinction, the modern analogue of the assembly being direct-democratic initiative. The Aristotle/Montesquieu observation on sortition as the instrument of democratic government was a reference to the appointment of magistrates — an aspect of executive government, not the sovereign legislature, so this is not a reference to the popular will.

    Of course you’re perfectly entitled to say that you couldn’t care less about Athenian democracy, or you may choose to challenge the scholarly consensus — Terry has indicated that he views the sortive aspect of Athenian democracy as more important than normally conceived, but we have yet to hear his evidence. In the absence of any such evidence we are forced to conclude that an allotted chamber as the instrument of popular will (as opposed to deliberative judgment) is entirely without historical precedent.


  39. I would say that I neither accept your various unsubstantiated assertions about the opinions of Aristotle, Montesquieu, and modern scholars nor see this as a point of much importance.

    The evidence that I am aware of – again, statements by Herodotus, Plato and Aristotle, and the institutional arrangements in Athens – indicates that sortition was seen as the hallmark of democracy. It would be very difficult to settle this view with a perception of sortition as having some sort of a passive power-neutralizing role as Headlam, for example, would have it. It may be possible to meet this difficulty, but Headlam is far from doing so and you have not even attempted to do so.


  40. I only mentioned it on account of your connection between sortition and the popular will. The popular will has always been associated with the assembly, with the boule seen as a collegiate magistracy to service its needs. Sortition was needed to protect the assembly from the aristocratic dominance that would have happened under the old council (the Areopagus). Rousseau, parroting Montesquieu’s distillation of ancient wisdom (Herodotus, Plato, Aristotle etc), agreed that sortition was the hallmark of democratic government (and election aristocratic) but this was a reference to the appointment of magistrates. As in Athenian practice, Rousseau allocated the popular will to the assembly of the whole people. The Athenians clearly believed that the courts judged in a way that the whole assembly would have done if they had engaged in the necessary deliberation, but the courts were never given powers to initiate legislation. Although the agenda for the assembly was prepared by a randomly-selected body, all citizens had the rights of isonomia and isegoria.

    So your proposal to restrict these rights to a tiny handful of citizens is without any historical precedent — it would be tantamount to the assembly voting itself out of existence.


  41. PS I should have mentioned that “magistrate” is an archaic word for government officers (including the monarch), in an age before the formal separation of executive and judicial powers.


  42. Two historic and one on vacation destinations…

    1. Keith wrote “Terry has indicated that he views the sortive aspect of Athenian democracy as more important than normally conceived, but we have yet to hear his evidence.”

    The existence of a citizen assembly without a means of democratically establishing an agenda is not a hallmark of democracy. Sparta had a citizen assembly, but as described in Wikipedia (probably trustworthy on points like this) “the Spartan citizen assembly could neither set the agenda of issues to be decided, nor debate them, merely vote on the alternatives presented to them.”

    Here are two pieces of evidence (of many) that the randomly selected bodies, such as the People’s Court and Boule were ESSENTIAL to democracy:

    Observing that the Athenians empowered the randomly selected people’s court to over-rule the Assembly, the Greek biographer and historian Plutarch suggested the randomly selected People’s Court was thus the ultimate sovereign authority, rather than the People’s Assembly.

    Secondly,Classics scholar Josiah Ober has noted evidence that a council selected by lot was the key institution in Greek democracy, and may even have been more central to the Greeks’ concept of democracy than the People’s Assembly. this is based on a edict from the nearby Greek democracy of Eretiria (Ober, 2007 “What the Ancient Greeks Can Tell Us About Democracy.” version 1.0. Stanford. online).
    Here is a quote from that article:

    “a popular deliberative council chosen from the entire citizen body. The Greek recognition of the centrality of a popular council for democracy is underlined by a recently discovered inscription from Eretria. In ca. 340 B.C. the Eretrian democracy promulgated a decree offering rewards to a potential tyrant killer, that is, to anyone who took direct and violent action against those who sought to overthrow the existing democratic government. In a revealing passage, the decree orders all citizens to fight without waiting to receive orders if anyone tries to establish “some constitution other than a Council and a prutaneia (a subset of the Council) appointed by lot from all Eretrians.”

    2. Now, on the matter of vacation decisions…

    I think that it helps if we think of it as a club making a decision about a group vacation (instead of a family). If they select a random committee of sufficient size (that is descriptively representative) to decide and they use “Yoram’s rules” (figuring out whether and who to consult, discussing, arguing, etc.) I think it is likely their decision would reasonably be one that the full group would like. However, if they divided EVERY member of the club into such random committees, and each followed Yoram’s rules, it is unlikely that any two of them would select the same destination. (Likewise any two elected committees from nominated members would likely select different destinations.)

    Keith implies that this indicates they are not making a democratic decision that reliably represents the whole (due to individual speech acts, etc.)

    Yoram might argue that any and all of these destination decisions are democratic, since there is no “right” answer. However, Keith might argue that if each of these random committees was presented with the same menu of options from experts, they might well ALL select the SAME destination, if they followed Keith’s rules, and this is an indication of democracy.


  43. Terry, thanks as always for introducing a little clarity. On the first point we need to draw a clear distinction between the different roles of the nomothetai and the boule. In the former case no-one would disagree with Plutarch that the considered verdict of a deliberative body is better than the snap decision of the assembly. In the latter case, Ober’s evidence would support the consensus view that the role of the allotted council (rather than the old aristocratic version) was to protect democracy from oligarchic or tyrannical influences. Even if you argue that the “centrality” of the role of the boule means pre-eminence over the assembly, it remains the case that any citizen in Athens could make a legislative proposal and speak before the assembly, unlike in Yoram’s proposal where this right is reserved for a tiny group of oligarchs. And the decision process in Athens was the domain of the entire citizen body (until subject to deliberative scrutiny). So it would seem that there are three features of Athenian democracy that require a modern analogue:

    1. The right of anyone to make a legislative proposal
    2. The right of everyone to approve or reject the proposal
    3. The necessity for the decision to receive deliberative scrutiny.

    Only point 3. is covered by sortition; 1 and 2 presuppose some kind of direct democracy.

    On your second point, another difference between Yoram’s thought experiment and an allotted assembly is that in the former, policy proposal is limited to choosing holiday destinations, whereas in the latter the group can do anything it likes, including altering the constitution in any way it sees fit (examples being paying existing members huge salaries, making permanent their tenure of office, appointing a dictator and/or exterminating all Jews), and is not subject to any of the checks and balances that normally apply to civilized societies. But this would not be a problem, as their interests are (by definition) aligned with the larger citizen body.


  44. > I think it is likely their decision would reasonably be one that the full group would like.

    Exactly. And that is really all that matters – the interests of the group would be well represented, meaning the procedure is democratic.

    Beyond being arbitrary, the “all make the same decision” criterion is completely useless as a standard for democracy. “The same decision” assumes a selection from a given set of options, but doesn’t specify how those options are created. If all the options in the set are terrible, who cares which one is chosen, and how would anyone imagine that the selection made from that set, however reached, can be called “democratic”? A necessary condition for democracy is control of the agenda.


  45. Terry,

    Your observation on the difference between the Spartan and Athenian assemblies has persuaded me that a modern analogue for Athenian democracy would require an equivalent of the boule. Was the agenda for the assembly decided endogenously by the council or was it more of a coordinating secretariat? If the former then I remember reading somewhere (I think it was Hansen) that, given the voluntary nature of the lot, the Athenians would have struggled sometimes to find the necessary quorum of 500, hence the decision to limit service to two years, rather than one. If this were the case, then everyone who wanted to put something on the agenda could (given sufficient patience). Alternatively most citizens would know someone on the boule and could introduce a proposal indirectly. This would not be possible in a large modern state, hence the need for the power of initiative by (say) e-petition.

    The modern analogue might look something like this (Athenian equivalents in parentheses):

    1. E-petition (lobbying the boule)
    2. Pre-legislative allotted scrutiny (boule)
    3. Public votation (ekklesia)
    4. Allotted scrutiny (nomothetai)

    In detail:

    1. Lobbying: Isegoria presupposes that any citizen can introduce a proposal. As the modern numbers involved would be huge, they need to be winnowed down, the first requirement being (say) 100,000 signatures to the e-petition (this is the current UK government threshold for a successful e-petition).

    2. Boule: The primary role of the boule being to protect the assembly from elite domination (as in Sparta), the job of the pre-legislative allotted chamber would be to remove proposals that were clearly the result of an organised lobby by a single media source or partisan organisation. This has clearly been the case with a number of UK e-petitions, and it can be demonstrated by correlating the voting pattern with newspaper or internet campaigns. The “boule” would benefit from statistical input from the civil service regarding the sponsor and voting pattern for each e-petition (competing media organisations would also provide discursive input, hence the need for the break up of media monopolies). Like in antiquity the term of allotted service would be a year, with staggered tenure in order to provide the necessary continuity.

    3. Ekklesia: I don’t read Ober as arguing that the assembly was not as important, so only those proposals that received the assent of most citizens would be passed to the next stage.

    4. Nomothetai: The fourth-century innovation of the legislative courts would appear to be largely on epistemic grounds — providing the detailed scrutiny necessary for high-quality legislation. The modern equivalent would also be structured like a court, with competing advocates (for and against the bill) addressing an allotted jury that had the exclusive right to determine the outcome. This would be the same public as in 3, but sampled descriptively. Like in Athens, the jurors would be allotted on a case-by-case basis.

    This would fulfil Yoram’s requirement for the democratic control of the agenda, unlike his own proposal for a single all-powerful allotted oligarchy, which would have been abhorrent to Athenians.


  46. Suerte Yoram. ¿Cómo transmitir la idea y convencer a la audiencia? Me hago esa pregunta muchas veces. Tomas


  47. Keith,

    Your plan above is slowly creeping a bit closer to my own, except I see no benefit from votation with its inherent rational-ignorance, advertizing-manipulation, and elite-domination . Votation would only serve to undercut good agenda-setting and decision-making.

    As to the role of the Council of 500 (Boule) in Athens… it appears that it varied over time. Sometimes it primarily facilitated the functioning and supremacy of the Assembly, and other times it was the real decision-maker, with the Assembly serving only as a last stop rubber stamp, though with a veto possibility.

    As to every citizen having the right to initiate legislation, this could happen at the Assembly, or through the Council. However, the person who initiated a proposal was taking a personal risk. If the Assembly’s decision was determined by the People’s Court to be either unconstitutional, or even simply bad policy, it was not just the decision that was overturned…the procedure was to challenge the initiator for miss-leading the people, and he might be punished or even put to death in some cases. According to Hansen, VERY few citizens were ever willing to initiate things, and those who did REGULARLY had to defend themselves in court for having done so. It was common, though controversial for initiators to find a willing front-man to make the formal motion, as a means of protecting themselves.


  48. Thanks, Tomas.


  49. Terry, that’s all very interesting — especially to learn that the assembly had, on occasion, only a formal role to play. But it’s significant that it was never abolished — if anyone had attempted to I imagine this would have been viewed as an act of high treason.

    Does the votation stage play only a formal role in my proposal? In a sense it does, as it’s partly in order to ensure the democratic legitimacy of our political institutions. (This is why Rousseau insisted that all citizens should rush to the assembly — it made little difference to the epistemic process (establishing the general will) but ensured the democratic legitimacy of the final decision.) But the public votation would play a substantive role as each stage — boule, votation, nomothetai would exercise a powerful constraint on the ability of elite, partisan and special-interest groups to dominate the agenda. Each would “refine and filter” proposals in a different way:

    1. Pre-legislative scrutiny (allotted): Protect the submission process from organised lobbying by media and partisan forces.

    2. Votation (mass democracy): Ensure that everyone has a chance to rank the proposals.

    3. Legislative scrutiny (allotted): Ensure that proposals pass the epistemic test (getting the “right” answer).

    I agree that (2) is somewhat nominal as a) the causal power of each vote is infinitely small and b) rational ignorance is guaranteed. But I think a) is essential for democratic legitimacy (in addition to the isegoria of the e-petition system) and b) will be countermanded by 3 (legislative scrutiny). We need to see the process as a whole rather than focusing on any one stage. Bear in mind also that those initiating proposals will be constrained to design them to pass all three stages; and also that there will be a parallel track for legislative proposals from government ministers (which would go straight to stage 3?). On the issue of advertising manipulation, this really is an American problem — over here we have strict budget limits and advertising plays a minor role in election results. Sensible countries (without constitutions written in stone) have ways of dealing with these practical issues.

    Regarding the ex-post accountability of those who make bad policy proposals, I’m assuming that punishing the individual concerned is not an option (unless they are shown to have broken anti-corruption laws). The modern analogue is the ex-post accountability of the political party. I think this is an important constraint and one which should not be casually rejected — this is why I’m torn between the direct-democratic e-petition (no ex-post accountability) and granting the power of initiative to the manifesto pledges of the party/parties that wins the most votes. Parties that introduce bad policies will lose out at the next election. Perhaps we would see a combination of both — as elections would be over issues rather than personalities, successful single-issue groups will almost certainly extend their remit to additional issues, so there would be no clear distinction between the two. Political parties would be diverse and plural and would ebb, flow and metamorphose in a very different way from current political systems (that fail to differentiate clearly between the legislative and administrative function).

    I’m glad that we are beginning to converge in our views (you’ve certainly convinced me of the need for a modern boule). I hope I will be able to persuade you that the introduction of a rational ignorance element is a small price to pay for democratic legitimacy and that stages 1 and 3 will bring this within acceptable constraints. As the only one of us with experience in the legislative process I’m sure you are all too aware of the need for grubby compromises!


  50. Hola Yoram
    He estado ojeando el documento que has puesto a nuestra disposición y sobre el que vas a apoyar tu charla.
    Me permito hacerte una observación: el discurso es demasiado racional.
    Te sugiero que introduzcas en la charla algún elemento menos formal. A la gente le gusta oír historias, historias que por la noche un anciano de la tribu contaba igual que cientos de años mas tarde otro anciano distinto contaría con las mismas palabras.
    Como sabes, la mayoría de las decisiones que tomamos están fuertemente influidas por mecanismos emocionales. Esto sucede de un modo natural porque estamos diseñados para tomar decisiones emocionales. Las decisiones que entrañan mecanismos mentales deliberativos tienen menos importancia de lo que se suponía.
    Por eso te sugiero que mires en la web del partido azar un apartado que se titula EL CAMBIO SOCIOLÓGICO para que intentes inspirarte para contar alguna bonita historia.
    Algo que a mi me gusta imaginar es cómo sería un mundo sin políticos profesionales: los medios de comunicación dejarían el espacio ahora ocupado por los políticos. ¡Que descanso!



  51. PS It would certainly be the case that stage 2 (public votation) would be prejudicial to innovative and minority proposals. Perhaps then an additional role for pre-legislative scrutiny would be to append recommendations to each proposal explaining why the “boule” has included it on the votation ticket. This would require short presentations from proposers who had passed the 100,000 signature threshold, but would not involve in-depth deliberation. The voting public would be disposed to trust the recommendations of a body composed of people like themselves and this would help counteract the attempts of media organisations to big up some proposals and ignore others. I don’t think we can ignore the case for mass democracy but we need to take steps to ensure that voters are adequately informed, so this might be a sensible compromise.


  52. Hola Tomas,

    You are probably right. However, rationalistic arguments speak to me and I don’t think I would be very effective presenting material that I do not find convincing.

    We can, of course, attack the problem of propagating the idea of sortition using various tools and various styles.



Leave a Reply

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

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

Facebook photo

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

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: