Statistical Sampling

On many occasions I have argued that the representativity of political assemblies constituted ‘descriptively’ (i.e. by statistical sampling) only applies at the collective level, and that this requires members of such an assembly being limited in their function (in contrast to the mandate of elected members). This argument has failed to persuade some participants in this forum, so this post makes the point in a rather stark manner, in the hope that it will challenge my opponents to refute it or else accept it – ‘ to put up or shut up’ – as opposed to merely ignoring it. I’m puzzled as to the continuing necessity to labour this point, as its veracity derives from the meaning of the word ‘statistical’, nevertheless I will seek to hammer the nail in one more time.

Statistical sampling via random selection is widely used for proportionate opinion polling, but the problem with using random selection for relatively complicated issues like political representation (as opposed to preferences over different brands of washing powder) is that such surveys are inevitably of ‘raw’ (unconsidered) opinion. Nevertheless the representativity of the proportional sampling techniques used is hard to deny, hence James Fishkin’s attempt to seek to establish a ‘deliberative’ assembly using random sampling techniques, which combines representativity with informed deliberation in order to represent the ‘considered judgment’ of the whole population. However this requirement leads Fishkin to advocate a very thin form of the deliberative ideal, in which members effectively listen to balanced pro–anti arguments and then decide the outcome via secret ballot, as opposed to the rich active deliberation preferred by Habermasian deliberative theorists. Why should this be?

A proportionate opinion poll on political preferences might provide an outcome indicating that (say) 60% of those polled held a ‘conservative’ perspective and 40% a ‘liberal’ or ‘progressive’ perspective on a particular issue. (For the sake of this simple illustrative example no attempt is made to define these terms or to introduce the additional statistical complexity of the ‘don’t know’ votes.) Such a survey would be held to accurately reflect the raw (unconsidered) preferences (or, if you prefer, prejudices) of the population being sampled.

A Fishkin-style poll would seek to measure these preferences at the beginning and the end of the deliberative process in order to establish to what extent the raw preferences were altered by the information-exchange process. The final vote would be held to represent the considered opinion of the population being sampled (assuming that the information was well balanced and comprehensive).

But what would happen if individual members started behaving as advocates for their own views? Sampling theory would suggest that the deliberative microcosm would be constituted in a broadly similar way (60–40) to the general population, but there is a significant possibility, or even probability, that a small number of participants would have the persuasive and status powers to influence the debate in a manner that did not reflect the underlying preferences of the greater population (note that this is not possible in an opinion poll, where every ‘vote’ [box tick] has the same weight). Opinion polls are only representative at the collective (statistical) level, so allowing individual members of a statistically-representative assembly active roles would be as invalid as suggesting that the views of individual respondents (selected at random from the sample) in an opinion poll reflected the average views of the population. Any such behaviour in the opinion polling industry would result in instant dismissal.

The injustice of the active advocacy model would be even greater if those sampled were empowered to introduce proposals to the forum. Recalling the 60-40 conservative/liberal assumption in the target population, by nature liberals (progressives) are more disposed to introduce political innovations than their conservative opponents, who would, on the whole, prefer to keep things as they are. Hence one might anticipate that the progressives would tend to set the agenda of the assembly, even though their views were in a minority in both the assembly and the target population. It is hard to reconcile such an outcome with the norm of political equality assumed by democratic theory, except via the Whig assumption that deliberation has the automatic effect of turning conservatives into progressives (conservatism being the result of indoctrination by powerful elites, rather than a natural predisposition).

In sum, if we are going to take the word ‘statistical’ seriously then any attempt to assign an active role to individual members of an allotted assembly should be abandoned. The analogy between a proportionate opinion poll and an allotted assembly is an exact one, the only difference being the information factor. This is the reason that the most successful method of establishing deliberative democracy (the Stanford DP) incorporates such a weak form of deliberation – members, in effect, simply ‘weigh’ the arguments and then determine the outcome via the secret ballot.


143 Responses

  1. Must write in haste. Tell me if this captures your thinking. Suppose I randomly select a (reasonably large) sample of the American public. The sample is representative of the entire public, in that we can say that x% of the public holds position y based on the fact that x% of the sample holds position y. Now suppose I get each member of the sample to watch a tv documentary (say, Al Gore’s An Inconvenient Truth). Then if z% of the sample changes their minds as a result, I can infer that z% of the entire public would do the same if they watched the same documentary. But the treatment effect goes to each and every individual in the sample; I could just as easily have mailed a DVD to each sample member so they could watch it in the privacy of their own homes. The sample members never interact with each other; if they do so, then the inference from the sample to the population is broken.

    Is that right? If so, could you say more about the last bit. The inference is broken…why? Because there’s no way to population to do the same thing the sample did?


  2. I think we have had this discussion lots of times and still I am not convinced of your argument.

    If statistical representation only allows for listening and then voting – why would we need any allotted chambers? Polls as they are done today would be sufficient.

    In my view statistical representation as elective representation are means to overcome the scale problem in large democracies. Elective representation has an elitist outcome. Statistical representation provides us with a sample of the population.

    Once the representative chamber is convened, be it through the elective method or the statistical one, the representatives begin to be different of the population at large. In the case of elective representation this leads to self-perpetuation etc. In the case of statistical representation with short mandates this would be impossible.

    But once they convene, the fact is that they have a new role, new information, new functions, which (radically) differentiate them from the population at large – the question is whether we trust more an elective group to take decisions for the whole or whether we trust more a statistical sample to do this or a combination of both.

    Here you will say the elective reps have the “plus” of having a mandate, or an authorisation to put forward their program, their proposed policies – well, that might be true, although we know that this link is very weak in practice.

    In the case of statistical reps what we know or should know is that they bring in the interests, viewpoints and diversity present in the population at large and that -with the right framework of deliberation and decision-making- they would make proposals, deliberate and decide as the population would probably do if the scale problem would not exist.

    Sure, there are more persuasive people and people with more status: but isn’t that the case also (and probably more) in elective chambers? And political professionals are not necessarily elected because of their rhetorical skills – at least not in democracies where 99% of the people who are elected just form part of a closed party list which is decided upon by the respective party leaders.

    Where I still find that further investigation is needed is in how to avoid group-think, polarisation, and other problems in an allotted chamber – although these are problems which also affect elective ones.

    Second I think that the right balance between an allotted chamber and an elective chamber has still to be further elucidated.


  3. The inference is broken for the reason that you give — the sample members would be subject to local effects which would not be replicated throughout the general population. This is particularly problematic in a single forum — ie all members of the allotted chamber are obliged to listen to the views of the opinionated (and will be disposed to defer to high-status members, and affected by consensual group-think). This effect could be partially alleviated by multiple parallel chambers, but then the rational ignorance problem returns when the votes are aggregated. And it woud simply be a case that a wider range of opinionated and high-status members would affect the outcome — it would still act to disempower the average Joe.

    The only way your Al Gore suggestion differs from the DP methodology is that Fishkin would also have members watch the Great Global Warming Swindle in order to obtain balanced advocacy. He would also have a panel of experts available to answer questions, which would have been generated in the small-group sessions. Apart from this it’s hard to see how the value of additional active deliberation could surpass the above-mentioned risks. Why shouldn’t people just listen to the evidence, ask questions and then make up their own minds?

    At root the argument is just an analysis of the meaning of a single word: “statistical”. It’s hard to imagine a thought experiment to circumvent the inconvenient truth that the word applies only in the aggregate sense. There is no such thing as an individual “descriptive representative” in the same way that there are individual elected representative.


  4. My previous response was to Peter.

    Jorge, we are in complete agreement regarding the need for statistical representation. An opinion poll samples uninformed opinion (prejudice) whereas a DP only decides after listening to the evidence and then deliberating (judging) the outcome. The only area I differ from you is in your unsubstantiated claim that the statistical sample would “make proposals, deliberate and decide as the population would probably do if the scale problem would not exist.” Proposals are made by individual people not statistical aggregates and the same is true for arguments — when I stand up and say something it is only me speaking, not a statistical aggregate. The only thing that we can know ex hypothesi is that the statistical sample would decide as the population would probably do, and that only if the advocacy was evenly balanced. Proposing, advocating and deciding need to be kept conceptually separate and only the latter is open to statistical treatment.


  5. Keith: just a thought experiment. Think of a mini-state where we only have say 10’000 people. They take decisions in their assembly. When they decide they do so as the whole population. But proposals, initiatives, deliberation will always be led by some individuals.

    Now take the same mini-state after some years. They have now 20’000 people. They no longer can assemble due to the scale problem. They decide to select half of them by lot to take part at the assmebly and rotate each year. Would you argue that now the members in the assembly can no longer make proposals and lead the deliberation? Would an elective chamber be preferable to make proposals and advocating?

    I don’t think so


  6. Jorge: What you are referring to in your thought experiment is rotation, not statistical representation. This was the rationale behind sortition in the ancient world of the small city state but is of no relevance to large modern states. In Athens proposals and deliberation were led by some individuals but judged by all citizens — the only difference in your example being that you get to judge every other year rather than all the time. In any event enormous assemblies like this offer none of the benefits of considered judgment, hence the need for the introduction of Athenian legislative courts, which I take to be on the numerical scale that most deliberative theorists prefer.

    I’m afraid thought experiments aren’t really going to help as this is a conceptual problem, what philosophers like to call ignoratio elenchi. There simply is no such creature as an (individual) “descriptive representative” as this would contravene the meaning of its synonym (statistic). The latter only applies to aggregates (totals, percentages etc).


  7. jorge,

    I had just constructed an identical thought experiment in my mind while reading Keith’s post.

    I think Keith’s best comeback is about the relatively small number of individuals who will make proposals in any allotted chamber. This small sample of the larger sample is far less likely to be representative of the general population. Thus a 10,000 member assembly can reliably represent a population of 20,000 (or 200,000,000)…but perhaps a 500 member assembly (with only 20 activist speakers) is not so reliable (statistically speaking).

    Imagine two allotted chambers, each descriptively representative, including the number of articulate and charismatic individuals. In chamber A a tall, handsome, articulate, male super-star happens to be a member and advocates policy X. Policy X wins a bare majority vote. But in chambere B, the outspoken advocate for policy X happens to be an ugly, fat, annoying, trans-gendered homeless person. If policy X is narrowly defeated in chamber B, which position is the right democratic decision?


  8. Hi Keith
    It’s not about rotation. It’s about trying to show how limited your argument (which appears to be wholly doctrinal and based on your reading of “statistical”, “descriptive”, etc) is and of getting to the heart of what “a sample” means and may entail: not just an exact depiction of the people at large (which will never exist) but a situation where “people like us” will take decisions on behalf of the population at large for a certain period of time – making present sth which cannot be present.

    And the people at large “taking decisions” would precisely be the voting function I thought you agreed could be performed by allotted chambers – so where would there be the problem?

    So what is the difference between the citizen making a proposal and advocating for something in an open assembly and the same citizen or his/her sample representative in a miniature assembly created by sortition?

    BTW the Boule was the main agenda setting, initiative and control institution in Athens – groupthink and polarisation were avoided with intelligent institutional design while diversity and independence were preserved apparently. Why should we not be able to achieve this nowadays?


  9. Keith: I queried you about this last time you posted on the subject, then re-read Hannah Pitkin’s The Concept of Representation and concluded that you were correct. The best these randomly selected groups can do is decide what comes out of the system on a yes/no basis. Delegates will clearly make decisions based upon their own views but cannot advocate them to others for the reasons you state.

    But that’s not necessarily a problem unless you imagine a single assembly replacing say, the Senate, or the House of Lords. To be anywhere near representative you would need hundreds of small community assemblies making decisions concurrently on national issues with the results aggregated. (Our model now envisages 18,000 randomly-selected people representing 60-70 million at any one time.)

    These community assemblies would be able to make yes/no decisions on community policy with a regional set of assemblies making decisions on regional issues and so on. By the same token, a global set of community assemblies could make yes/no decisions on issues affecting climate change or the global economy.

    Rather than send delegates DVDs, we imagine a large pool of legal ‘advocates’ who are selected at random to present the case for either a ‘yes’ or ‘no’ outcome to an individual community assembly which is allocated to each advocate randomly.


  10. Jorge: How about policy formulation being separate from legislative decision-making in a modern system? How about international, national, regional and community-level policy forums?


  11. Keith, would it be fair to say that what you are looking for from descriptive representation is a sample that yields the same OUTCOME as would have resulted from consulting the people as a whole? In other words, you think that 1) there’s some decision the population would have made under treatment conditions X (watching DVDs, etc.), 2) that’s the decision we want to have enacted, and 3) a random sample, subjected to treatment condition X, will reach exactly the same decision. Is that right? I’m pushing on this because it’s not the only way to think of what a representative body is supposed to do. One could simple want the representative body to follow a certain procedure, without claiming that the procedure would yield the same decision as the population under suitably defined conditions.


  12. Terry: Agreed

    Jorge: Votes in large assemblies are always unconsidered raw preferences, swayed by all sorts of demagoguery. This is equivalent to the elective aspect of modern democracy and it matters little where the proposals come from (petition, initiative, electoral manifesto) so long as:

    a) the proposing rights are open to all (isegoria);
    b) all citizens then have a right to register their raw preferences;
    c) it is followed by a process of informed deliberation, with balanced advocacy and secret votes (as in the legislative court).

    I’m sure what you say about the Boule is correct but the Athenians would not have permitted what was effectively just the secretariat to the Assembly to both set the agenda and determine the outcome. We should not let our obsession with sortition blind us to the fact that the Assembly was the primary institution of Athenian democracy, and this requires a modern analogue in order for a constitutional proposal to claim the mantle of democracy.

    I’m puzzled by your claim that my insistence that we use words correctly is “doctrinal” — words are tokens in a public language and we have to conform to standard usage. Humpty-Dumpty may have claimed that a word means “just what I choose it to mean — neither more nor less”, but that isn’t an option in scholarly debate. Choose what dictionary you wish, here’s one at random: “a statistic refers to a quantity (such as mean or median) calculated from a set of data”. The notion of a “statistical representative” is simply erroneous. You may not like the way this word is used, but please don’t shoot the messenger.


  13. Martin: Thanks for endorsing my reading of Pitkin. Regarding the notion of a multiplicity of sortive bodies (a similar suggestion to Terry’s) how do you determine the outcome without undergoing the rational ignorance effect? If the bodies were to have an active role then there would need to be a lot of them (in order to overcome Pitkin’s caveat) so individual members might well feel as impotent as rational voters.


  14. Peter: Yes and No. Yes from the perspective of the representativity of the sample to the whole; No in that most people would not bother to watch the video (they would switch channel to the X Factor) as they knew that their vote would have little causal value. If this were not the case then electoral democracy would function well and we would not need sortition, because all citizens would take the trouble to educate themselves regarding the available policy options.

    As for proceduralism, I would prefer to leave that to the Schumpeterians. What possible merit might it hold for a sortive solution? Andrew Rehfeld argues that sortition would be a legitimate constitutional procedure if it were approved in a referendum, but I’m more disposed to a substantive notion of democratic legitimacy.


  15. I very much agree with Jorge.

    Keith rehashes here several issues that have been discussed thoroughly on this blog many times before. The new formalistic claim here (that Keith’s points are supported by the meaning of the term ‘statistical’) is both false and irrelevant (as Jorge puts it, it is “doctrinal”). Even if ‘statistical’ meant what Keith thinks it does – and it is not even clear what he thinks it means, that would have nothing to do with the behavior of allotted chambers operating within any set of rules, which is the substantive issue here.

    Repeating what has been written here many times, here is a quick refutation of the substantive claims Keith is making (again):

    > there is a significant possibility, or even probability, that a small number of participants would have the persuasive and status powers to influence the debate in a manner that did not reflect the underlying preferences of the greater population

    Why and how?

    Keith obviously has very little faith in the sense of the average person if he thinks the majority of the allotted group (and thus the majority in the population) can be duped into supporting, after thorough examination and an open discussion, ideas that are against the preferences of the majority of the population (and thus of the majority of the group).

    But even if we accept such a dim view of humanity, as a group of incompetents who can be easily manipulated to act against their best interests, why would the following be true:

    > (note that this is not possible in an opinion poll, where every ‘vote’ [box tick] has the same weight).

    If people are so easily manipulated, why wouldn’t their uninformed and unconsidered opinion (which is what one can expect to be expressed when one is exposed to partial or biased information and deprived of the benefits of an open discussion and inquiry) be even more unrepresentative of the preferences of the majority of the population?

    Going a bit deeper, one can ask “what are the underlying preferences of the population if not the preferences as expressed by a sampled group that has been empowered and motivated to inquire into the issues?” What would be the yardstick against which Keith (or anyone else) be able to measure the expressed opinions in order to determine that they do not reflect the (informed and considered) preferences of the population at large? Obviously the unconsidered answers of an uninformed population cannot serve as such a yardstick.

    I think the only conceivable answer is that the yardstick Keith implicitly uses is the opinion of some privileged group (what Keith likes to call “experts”). These are the people Keith imagines as delimiting the options (and the information) available to the allotted chamber, and this – an oligarchical system – is the only possible alternative to a fully empowered allotted chamber.


  16. Yoram: My usage of statistical is straight from the dictionary (see my last response to Jorge).

    Regarding the potential distorting influence of advocacy from high-status and rhetorically-gifted individuals, this is bog-standard social psychology — see for example Terry’s comment on this thread. If this were not the case then Fishkin would not go to the time and trouble to train moderators to equalise the small-group DP exchanges (and this is without them adopting an advocacy role).

    I agree that partial and biased information will seriously affect the decision process, this is the reason Fishkin goes to such lengths to ensure balance. You choose to scare-quote the word “experts”, but then you have argued that all political appointments (including the highest executive posts) should be filled by sortition and I doubt whether many will share your scepticism regarding the substantive reality of expertise in public policy issues.

    Regarding the yardstick, it’s quite simple — measure attitudes before and after the deliberative exchange. Whilst it may be hard to evaluate preferences, it’s relatively simple to see how well-informed they are, and this is the goal of epistemic democracy.


  17. As usual, Keith, your responses are just a sequence of slogans and non-sequiturs. I have learned after much wasted effort that there is really no point in trying to engage in a discussion with you so I will avoid hopeless attempts to do so.

    As I have written before, if other people think there is some merit in Keith’s points and would like to discuss them, I’ll be happy to try and engage with them.


  18. “I’m sure what you say about the Boule is correct but the Athenians would not have permitted what was effectively just the secretariat to the Assembly to both set the agenda and determine the outcome.”

    I’m not necessarily claiming that the same allotted chamber or the same formation of an AC takes on agenda setting, proposals and decisions. What I’m claiming is that as the Boule (which certainly was much more than a secretariat) we can have an allotted chamber making proposals and setting the agenda, while we can have another allotted chamber or the plenary of an allotted chamber making decisions (as did the Nomothetai). And we can call this “statistical” representation or “descriptive” representation, in the end that is only the starting point, the mechanism with which we create representation, but which does not necessarily define what this representation may entail. E.g. we use nowadays the elective mechanism, but the representation can be manifold and with different remits and powers depending on the constitutional arrangements. In fact I would not call it”statistical” representation if we are to describe the powers of such representatives with such term, but would rather be inclined to call it “citizen by citizen” representation or “homonomous” representation or whatever – what really matters is whether we trust such body or bodies to exercise such powers in contrast to having elective institutions or a combination where the AC only could vote, but not set the agenda or advocate positions.

    The problem I see is the continuity between the people at large and such institutions and the possible relationship between an elective chamber and an AC. As to the first issue I would favor petition rights, citizen initiatives with low thresholds etc as well as the possibility of calling for approval by the citizenship at large if a certain % of parliamentarians think so or if certain decisions are to be taken.
    As to the second I feel that elected reps, when exercising a curtailed and checked power, could act as the modern rhetores and generals, without monopolising that function either.


  19. Jorge,

    It is not clear to me why you think the non-sortitional mechanisms you suggest (elections, referenda, petitions) would improve the representation of the interests of the population in government. As you mentioned in the context of elections, all of these mechanisms are tools for elites. Why would removing power from a representative chamber and moving it to the hands of elites be a democratic design?

    (The elitist nature of referenda is mitigated when the agenda is pre-set so that it is not open for manipulation by elites. For example, a system in which the salaries of the allotted delegates are determined by annual referenda makes some sense in view of the inherent conflict of interest of the allotted delegates have on this issue. Even in such a situation, the risk of uninformed decisions exists, of course, but it could be argued that it is outweighed by the advantages of the system.)


  20. Yoram

    I think that there are several reasons for keeping an elective chamber, starting from such practical reasons as tradition and an evolutionary approach to political innovation.

    However, the principal reasons in my view is 1st to include/integrate in our framework the political potential/power of organised interests -which will always exist in societies based on capitalistic production, race, gender, religious cleavages… (otherwise they will work outside of the institutions and will focus in subverting them)- and 2nd to use their power to formulate policy positions in a democratic setting.

    I see this as analogous to the role elites had in Athens, providing rhetores and generals who were mostly integrated in the system and could develop and advocate policy positions among which the assembly did choose. If we cut political parties off of their monopoly on politics and reduce them to a non-monopolistic agenda-setting, initiative and co-decision function I think they could do a good service also for democratic goals.

    Re Keith: I guess you see sortition as a means which only is able to generate a static miniature picture of the population at large – as a sort of device of depicting the people as they “are”, but this assumption is in contradiction with the fact that a sample will necessarily “be” different from the population at large as soon as the sample assumes its new role, receives new information, develops new functions… so to say, as soon as the sample “moves” the picture is no longer accurate – therefore, if we take the view of what “statistical” representation means in your view to the extreme you seem to advocate, this sample should not even convene, and should not be able to even vote as it is unable to re-enact the decision the population at large (as it “is”) would really take.

    I rather think sortition may serve as a device for not only translating into a workable size the people as they “are” but also as they “would act” and this includes not only the aggregate but also the individuals, as this characteristic of “would act” necessarily entails a dynamic dimension which in turn implies not only aggregate actions but also individual actions.


  21. Representation is partly a matter of convention, as is authority of all sorts. In political representation existing conventions come about as a result of the evolution of earlier conventions.
    Two major influences on changing conventions are dissatisfaction with the results of existing conventions and changing ideas about what the conventions are ought to achieve.
    The argument for sortition is driven partly by dissatisfaction with professional politicians making decisions that are driven by their career strategies rather than the merits of the case, and partly by a move away from the conception of politics as a matter of struggles for power between conflicting interests in a zero-sum frame, towards an attempt to arrive at win-win solutions.
    Keith’s concerns about representation are concerns about power. I am concerned about getting the best solutions to common problems. He seems to think that when one votes for a candidate one empowers her to decide on one’s behalf, giving her a sort of power of attorney. That conception had a certain plausibility when Burke could claim that he was elected to exercise his professional judgement about what is best for the common good, not to represent particular opinions or interests. That view rested on the assumption that an aristocratic political elite possessed an expertise and integrity analogous to that of “the learned professions”. Democracy was a matter of choosing who was to rule. The parliament was to consist of a meeting of sages.
    When opinion moved towards the view that representatives were supposed to represent the opinions of the electors, it became evident that the individual representative was largely powerless to do so effectively. Only an organised body of representatives could do so.
    Voters chose parties rather than individuals as their representatives.
    Under most systems of voting, to vote for an independent was to throw your vote away.
    The parties chose the candidates and the focus of most candidates was to advance their power within the party, and to advocate those policies that would assure the electoral success of the party, while giving as little offence as possible to their constituents.
    The system is not very rational. Its main virtue is that it enables voters to throw out egregiously unsatisfactory governments. The idea that the sort of elections we have authorise representatives as individuals to speak on our behalf is nonsense. Politicians routinely advocate policies that most of those who elected them disagree with. (Often that is a good thing, where they are better informed and more conscious of the consequences of their decisions than most of the electorate). At most they might be said to authorise the government that is formed as the result of the electoral process to pursue its “mandate” to do what it promised.
    What sortition might deliver is a body of people who, in seeking to find suitable compromises between their conflicting opinions and interests, would not be constrained by the requirements of the sort of power politics that one has to engage in to be effective in party politics.
    Unfortunately, where there are very many opinions and interests in play over a wide variety of issues, people will be constrained to do deals. Exchanging their vote on matters about which they do not care much for support on matters dearer to their hearts. The smart operators will use this process systematically to organise parties. It is that rather than the persuasiveness of some advocates that is the danger.
    My contention is that issues must be disentangled from each other before sortition is likely to work well.


  22. Jorge,

    > practical reasons as tradition and an evolutionary approach to political innovation.

    With this I agree.

    > However, the principal reasons in my view is 1st to include/integrate in our framework the political potential/power of organised interests -which will always exist in societies based on capitalistic production, race, gender, religious cleavages… (otherwise they will work outside of the institutions and will focus in subverting them)

    Even if such anti-democratic power must exist, there is no reason to re-inforce and legitimize it by granting it explicit, privileged position in government. In the same way, some criminal activity, and some organized crime, will likely always exist in society – that doesn’t mean such elements should be granted special powers in government.

    > and 2nd to use their power to formulate policy positions in a democratic setting.

    I don’t see this as being democratic, but rather as subverting the democratic underpinnings of the government and de-legitimizing it.

    > I see this as analogous to the role elites had in Athens, providing rhetores and generals who were mostly integrated in the system and could develop and advocate policy positions among which the assembly did choose. If we cut political parties off of their monopoly on politics and reduce them to a non-monopolistic agenda-setting, initiative and co-decision function I think they could do a good service also for democratic goals.

    Once the elites determine, or have large influence over, the agenda, they have gained disproportional power and reduced the allotted delegates into a secondary role as wielding at most veto power. Indeed, as I see things, the power political attained in Athens by the strategoi and later by unelected rhetors was an anti-democratic element that was maintained through the mass procedure in the assembly and that competed with the democratic power of the allotted Boule and magistrates.


  23. Hi Yoram
    In the end we agree possibly on the optimal solution, but probably not on the best possible solution.
    I tend to think that a strongly curtailed elective power (e.g. through the very presence of the AC; with strong internal democracy requirements -including perhaps sortition for control boards-; publicly-only financed elections; open lists, etc, etc) can be useful, as were people like Perikles in Athens, without necessarily transforming themselves back into self-serving elites – it’s perhaps not 100% purely democratic, but in politics I feel that purity is seldom achieved and seldom a good thing in practice – here my model is again Athens were democracy (for citizen males) was a result of a balanced institutional framework.
    In addition, I forgot to add a thought which you probably won’t like: checks and balances. Allotted institutions and elective ones would be able to balance and check each other due to their different kind of legitimacy source.
    Finally, I trust that if ACs work they will overpower such a curtailed elective power – as previously in history the lower chambers have overpowered the higher chambers. This process would need to be evolutionary in any case – with practice comes institutionalisation and legitimacy among the people. In the end perhaps (hopefully) we would have elective chambers as mere relicts of the past as the House of Lords is nowadays in the UK.

    Well you see that I’m not a revolutionary ;-)


  24. Jorge: I’m relieved to see that you do not share Yoram’s visceral opposition to any form of electoral representation. I agree that we need a mixed system, replete with checks and balances, and that the final decision power should be vested in the allotted chamber. Perhaps “secretariat to the Assembly” was a little disparaging for the role of the Boule, Manin’s term is collective magistracy, but he is adamant that the Boule was not a decision-making body, more like a conduit for legislative proposals to the Assembly. We really mustn’t lose site of the fact that the Assembly (with all its flaws) was the primary institution of Athenian democracy and that for any modern system to be described as democratic requires some analogue — sortition alone will not suffice. Options for the modern analogue include election, referenda, initiatives, e-petitions and votation. These are open to elite and demagogic subversion (as was the assembly) but the final decision remains with the allotted chamber — as in the nomothetai amendment to the Athenian constitution.

    I note that you no longer advocate descriptive/statistical representation, but would prefer “person-by-person” or “homonomous” representation. The obvious response to this is which persons? And how do you reconcile this with democratic norms? As far as I understand democracy it has to be either direct or representative, the latter subdividing into categories such as symbolic, ascriptive, representation of interests, and descriptive/statistical. Please tell us more about homonomous representation and how it might attain democratic validity (assuming that is important to you).

    I would agree with you that the informed judgment of an allotted chamber would immediately deviate from the uninformed judgment of their peers. In Fishkin’s words the allotted assembly would “offer a picture of what everyone would think under good conditions. In theory if everyone deliberated, the conclusions would not be much different” (2009, p.194). That strikes me as an acceptable deviation from rational ignorance.

    Yoram: I won’t trouble you with any more slogans or non-sequiturs and will leave it to others to judge the accuracy of your (mis)representation of my arguments.


  25. Burnheim: “[Keith] seems to think that when one votes for a candidate one empowers her to decide on one’s behalf, giving her a sort of power of attorney.”

    John, this is a complete misrepresentation of my position! I’ve consistently argued that:

    1) election is one mechanism (alternatives being referenda, initiatives, e-petitions and votation) for the formulation of policy options;

    2) the judgment role (deciding) being reserved for the sortive body.

    We disagree over the necessity for 1). I argue that a modern constitution that doesn’t include an analogue of the Athenian Assembly cannot be democratic; you have had the good grace to acknowledge that you are not a democrat. Regarding 2) you argue for a volunteer base for the sortition; I argue for random selection from the entire citizen body.

    At no time have I sought to rehabilitate Burke, in my book I doubt whether his trustee theory was even plausible at the time he composed his famous address to the electors of Bristol.


  26. Well, I feel John Burnheim has rightly pointed out that reprsentation is a matter of convention.
    Using sortition as a mechanism to overcome the scale problem and generating an assembly of people of whom the population at large and the not selected individuals could say “they are like me” and “they will act as I would do” is the key to “citizen representation” or whatever you or we want to call it.
    By its very essence and novelty this new flavour of representation cannot fit into (outdated) categories designed by somebody (e.g. Pitkin) working within a paradigm based ultimately on elections.


  27. Jorge,

    > I tend to think that a strongly curtailed elective power (e.g. through the very presence of the AC; with strong internal democracy requirements -including perhaps sortition for control boards-; publicly-only financed elections; open lists, etc, etc) can be useful, as were people like Perikles in Athens, without necessarily transforming themselves back into self-serving elites

    Why would that be any more useful, or any more legitimate, than granting special powers to elite groups in general – aristocrats, the rich, the beautiful, the educated, the church clergy, or, to pick up my previous example, mafia bosses? Would we expect any such group to serve the interests of the public? If not, why would we expect the electoral elite to do any better?

    > In addition, I forgot to add a thought which you probably won’t like: checks and balances. Allotted institutions and elective ones would be able to balance and check each other due to their different kind of legitimacy source.

    It is true I do not think highly of the notion of “checks and balances”. It is nothing but an marketing term for “power sharing”. So what we are discussing here is really sharing power between a democratic mechanism (sortition) and a non-democratic mechanism (elections). Again, if you are going to grant governing powers to some select, non-representative, group, why would you do that through elections? Why would you expect such a group, however selected, to exert its power in favor of the population at large?

    > In the end perhaps (hopefully) we would have elective chambers as mere relicts of the past as the House of Lords is nowadays in the UK.

    It seems, then, that we do agree on the ultimate desired situation as well as on the desired evolutionary approach of getting there.


  28. Jorge: “sortition [will generate] an assembly of people of whom the population at large and the not selected individuals could say “they are like me” and “they will act as I would do” is the key to “citizen representation” or whatever you or we want to call it.

    The key word is “they” (individual acts require a different pronoun). People are different in a variety of ways and only appear similar at a distance (perceptually) or in aggregate (in terms of beliefs, preferences etc.). The latter problem is aggravated by inequalities in personality, ability and status (none of which have any impact on voting, which is why Pitkin singles out voting as the one legitimate task for descriptively-representative assemblies.)

    Jorge: “By its very essence and novelty this new flavour of representation cannot fit into (outdated) categories designed by somebody (e.g. Pitkin) working within a paradigm based ultimately on elections.”

    There is nothing new under the sun and your claim misrepresents Pitkin (who is a philosopher, not a political scientist) almost as badly as John Burnheim has misrepresented me. Let me restate the CONCEPTUAL argument:

    1) Democracy means the rule of the people. There are two ways of achieving it:

    1a) Direct Democracy — where all citizens decide. The problem with this is that decision making is badly informed (as a result of the problem of rational ignorance) and the vote will be swayed by media and other elites.

    1b) Representative democracy. This can be achieved in two ways:

    1bi) Voters choose delegates/trustees to decide issues on their behalf. This made sense at the time of the birth of representative government on account of the combination of widespread illiteracy and the lack of telecommunications (conditions which no longer apply). The process is subject to the same flaws as direct democracy.

    1bii) Sortition — establishing a representative microcosm by random sampling. There are two caveats:

    * If the sortition is based on a volunteer pool, it will not be statistically representative, thus demarchy is not democratic.

    ** The microcosm is only representative as a whole, so any individual acts (other than voting) breach the statistically-representative mandate.

    Conclusion: A fully democratic constitution will involve a combination of 1a (for policy initiation and advocacy) and 1bii (for judgment). If political parties choose to act as policy advocates (as opposed to trustees/delegates) then they would be subsumed under 1a (as a vote for a party is, in effect, a vote for a combination of policies). Political parties will not, however, be able to determine the outcome of parliamentary debates, as this will be in the hands of 1bii.

    As far as I can gauge, the only residual dispute on this forum is over Pitkin’s claim in 1bii**. Martin Davies has had the good grace to acknowledge that she is right about this, so I would recommend everyone to read the book and point to the flaws in the argument (The Concept of Representation is still the seminal text in this area of study — my PhD supervisor has told me to read it at least three times). Philosophical arguments are open to refutation by logic or mathematics; claiming that they are outdated as they do not rely on empirical entities such as elections is another example of ignoratio elenchi. The argument is over the nature of statistical representation (an abstract concept which is not subject to the ravages of time).

    Jorge & Yoram: “In the end perhaps (hopefully) we would have elective chambers as mere relicts of the past as the House of Lords is nowadays in the UK”

    Hmm, I see the Whig mindset is very much alive and kicking. Most analysts would claim that the Lords is the only remaining bastion of freedom in this country. The only serious debate on the 2008 counter-terrorism bill was in the Lords; Labour rebels were only persuaded to vote for it because they were told it would be thrown out in the Lords. In the words of Labour MP John Cruddas (before he passed through the “yea” lobby) “it’s a load of rubbish which will be useless but we might as well vote for it because it won’t work”.


  29. Dear Keith
    As I think I have shown statistical representation -as you define it- in a body which by its very creation, function, information, power etc is made different from the population at large is always a fiction, be it that they propose, they deliberate or they vote. Pitkin notwithstanding.
    And all concepts vary in time – have a look at theories on shifting scientifical “paradigms” – Pitkin’s work does not exist in a vacuum, and it certainly deals with elections as the main mechanism for achieving representation – the same way as all theories before Copernicus saw the earth as the centre of their cosmical parafdigm – you just have to read her book and after all it is inevitable for her to do so ;-)


  30. Jorge

    If statistical representation were a “fiction” then we would need to give up on democracy. Opinion polls, however, are a reliable way of sampling raw public opinion and Fishkin has shown how the principle can be developed to reflect informed public opinion, so I’m not giving up on the possibility of democracy. The Stanford Center for Deliberative Democracy would be a little annoyed to hear that the products of their empirical social science work should be filed on the fiction shelves in the library.

    I’ve spent most of this term studying Kuhn and (more relevantly) the application of the paradigm-shift approach to political languages in Pocock, Skinner, Collingwood etc. and I’m unsure what point you are making with respect to Pitkin’s analysis. She would agree that there are a variety of different (and overlapping) concepts of representation — including symbolic, formal, ascriptive, representation of interests and statistical representation. What other paradigm did you have in mind? Forgive me for putting it like this, but talk of ‘new paradigms” reminds me of the arm-waving you get in New Age circles when they talk about “qualitative” science and other such slippery concepts. The paradigm change required to legitimise speech-acts by individual members of a statistically-representative assembly would be a change in our understanding of mathematical probability. Are you suggesting such a fundamental change is imminent?


  31. institutions of representation create political fictions whereby sth which is not present is made present ;-)

    Politics, Constitutions and law are full of such fictions – it does not make them less real, on condition that people think them legitimate.

    And the concept of representation – if it had to be rewritten today, would be possibly very different as the use and practice of representation has changed in these more than 40 years. I don’t have a definite answer on what those changes would be, but for instance we have new political beings being developed like the EU or we have some tens or hundreds of people asking themselves how to use sortition to represent larger populations as a democratic tool – all thinhs she could not take into account back in the sixties.

    Political conceptions are mutable and representation is a prime example for that – Pitkin’s book is full of such examples – and representation probably did not exist with the Ancients, was developed with Monarchies, used for republics and finally reused for representative democracies… is it so hard to understand that it’s meaning is not written in stone and that our own activity thinking about sortition-representation-democracy is part of a paradigm crisis of fundamental concepts as representation which nowadays still are principally linked to representative government?

    I think we should be aware of that – perhaps it requires some leap of imagination, but precisely that’s what we all sortitionists are doing since Dahls seminal work.


  32. Paradigm change is required when it is no longer possible to explain current phenomena using the existing paradigm. It’s clear that institutional reform is necessary but it’s not at all clear that there are additional conceptual problems in addition to those outlined in my earlier (numbered) commentary. I’m afraid a call for a leap of the imagination is a little too vague for my taste (and sounds suspiciously like a leap into the unknown), so will need to sign off for the time being.


  33. Response to Burnheim’s post: I think John has made an important point about the intertangling of issues being a threat to good decision making. Based on experience of elected chambers it is clear that once multiple issues are to be tackled by the same group of people, vote swapping will quickly become rampant. This is one reason I favor a multitude of single-issue allotted bodies.

    Response to the continued role of an elective chamber (or an elected executive): If any prestige or power remains with political parties or elected officials, there will be an over-powering dynamic to make the work of the allotted chamber serve the elective interests of the elected officials. For example, think of Fishkin and Ackerman’s idea of a “Deliberation Day” prior to elections, where allotted groups discuss the merits of the candidates and make a recommendation. Can all of us agree that such a Deliberation Day would instantly become a venue for partisan gamesmanship, with minimal open-minded “deliberation?”

    Parties will re-purpose themselves to fit the new situation. Parties will assure that their supporters within an allotted chamber sabotage proposals of their political opponents, and promote policies that will put the preferred party in a positive light. In other words, as Yoram fears (I suspect), parties will intentionally poison the work of an AC.

    While I agree that it seems inevitable (short of a revolution) that a transition to a sortition form of democracy will have some overlap of elective and allotted bodies, I don’t think the two can coexist productively for long, without the elective body evolving the means to control the allotted one. The best transition path I can imagine is the adoption of an allotted body that has the function of proposing constitutional amendments to the voters, with the hopes that one day such a body will propose to the voters the elimination of the elected chamber.


  34. Well it seems evident to me that the conceptual elective-representation toolkit we have inherited from previous times is not fully usable for the conceptual needs of sortition-based institutions which in turn respond to a perceived necessity of improving an institutional framework based on elections and party politics which seems to be losing democratic legitimacy.

    If we try to stick to the letter of that toolkit we end with untenable positions like your own where
    1st you are not able to answer the question why a sortition-based institution which is per se different from the population at large -is not truly statistically the same as it starts “moving”- may vote, but not propose or deliberate and
    2nd all the effort spent in creating new sortition based institutions renders them only capable of saying yes/no to a question raised by the very elective institutions we know that are no longer sufficiently legitimate, leaving all agenda-setting, advocacy and deliberation to them.

    So, this is not so vague – it is about designing institutions and weighing the pros and cons from a democratic point of view. Sometimes this thinking and weighing implies the need to update our conceptual toolbox.


  35. Terry

    How would political parties control the votes of a randomly-selected legislature if a secret ballot was in use? Ditto for vote-swapping and other forms of pork-trading. I don’t know how it is in the US, but in the UK very few people have an allegiance to a political party. Parties could only secure votes by offering the most convincing arguments. All the above abuses are caused by the fact that legislative decision making is in the hands of people who need to secure re-election.

    Note: I’m not committed to ongoing political parties if an alternative mechanism for democratic agenda formulation was demonstrated to be better (referendum, initiative, e-petition, votation etc.). The argument for parties is in order to retain some element of ex-post accountability, which would otherwise be completely absent. I don’t think we should entirely dismiss the notion that people should be held to account for their actions as it is a valuable principle in most walks of life.


  36. Jorge: >”1st you are not able to answer the question why a sortition-based institution which is per se different from the population at large -is not truly statistically the same as it starts “moving”- may vote, but not propose or deliberate”

    Because a) if the “movement” is from being ignorant to being informed then we should privilege the latter (see the Fishkin quote in my earlier response regarding representing what the whole population would think under ideal circumstances). b) because policies are proposed by individuals, who have no personal mandate.

    >”2nd all the effort spent in creating new sortition based institutions renders them only capable of saying yes/no to a question raised by the very elective institutions we know that are no longer sufficiently legitimate, leaving all agenda-setting, advocacy and deliberation to them.”

    Legitimacy under elective institutions depends on securing the largest number of votes. The problem with elective democracy is primarily epistemic — most people would agree that government by the majority party is legitimate (otherwise we would be rioting in the streets), we just doubt the quality of the resulting decisions.


  37. You know that the “movement” is much more than acquiring some information, it’s acquiring political functions, a position of power, being part of a collective – which makes the members of the AC different of the population at large: so adieu your “statistical” “descriptive” definition of representation – why should these individuals be able to vote and not to propose/deliberate?
    You have no answer to that, I’m afraid.

    “The problem with elective democracy is primarily epistemic”: that’s really an understatement of the shortcomings of elective representation I fear, if you want to reduce everything to a question of quality of results…


  38. I’m very aware of the problem of institutionalisation (“going native”), that’s why my preference is for an ad hoc timespan for allotted service. Just like in a trial jury, an allotted jury would be summoned for each legislative bill. Of course that would not be possible if the assembly were to have an active function, as training and parliamentary experience would be necessary; not so if all they have to do is deliberate — determine the outcome (in the Latin rather than Germanic derivation of the word). A typical DP session only lasts a few days, jury service only a week or two, and this would be the best span for an AC — there would be no risk of institutionalisation over such a short time span, hence the only “movement” away from the general population would be the acquisition of information. It has the added advantage of including a much larger number of citizens — “It could be you” as the National Lottery slogan goes.

    An AC with a longer time span and active functions would become corrupt in a short period of time, owing to a) the lack of party discipline and b) the lack of ex-post accountability. This is why the Athenians had to go to such elaborate lengths (threefold scrutiny) to protect the process from corruption. I’m not disposed to a political system that requires such an active role for the police.

    As for your last point, my concerns are both epistemic and normative, i.e. getting the best outcome via a process that does not contravene the norm of political equality. There is no need to have to choose between these two ideals.


  39. Keith,

    You wrote: “How would political parties control the votes of a randomly-selected legislature if a secret ballot was in use?…All the above abuses [vote swapping, etc.] are caused by the fact that legislative decision making is in the hands of people who need to secure re-election.”

    I agree that IF their role is limited to secret balloting without debate, the negative influence of continued party elections would be less…but…unless the members of the assembly are isolated (as in some jury trials) their is a risk. Since most of us on this list prefer that allotted bodies be allowed to debate…my concern is more relevant to us.

    It is the same ratchet problem that allows special interest lobbyists to influence legislation, even though they aren’t charged with decision-making. If a concentrated group has high stakes in a decision, then they will devote the needed resources to sway decisions in their favor. Those who control a party (under at least one of your schemes) might not control “decision-making,” but would control agenda-setting, which can be even more important in many cases. Elected officials would still be campaigning and gaining SOME adherents in the general population. When these adherents become part of a random assembly, they would put the re-election interests of their party ahead of their duty to be open-minded.


  40. As to corruption as the new fear/reason for arguing for only voting ACs and very short terms of them: I think there are a lot of institutional devices to provent AC members from being corrupted or corrupting themselves -at least to a much lesser degree than elective reps do- first of all they have no party-professional incentive, they do not strive to be reelected or be appointed to some interesting position, they do not need to return favors to anybody… they just speak for themselves and they will be just ordinary people… if you couple this with a short mandate (one year for example), rotation in functions within the AC, use of sortition to assign functions to AC members (e.g. allotting different bill topics to draft proposals to different groups of AC members each time), collegiality requirements, strong conflict of interest rules, cooling-off standards, and transparency requirements on their income you will fence off corruption much more effectively than with elected officials.

    (Just look a bit at the Boule and how many of such rules they had already in those times to avoid corruption and the concentration of power in the main agenda-setting and proposing – as well as control and administrative day-to-day body in Athens)

    We already talked about accountability – and -as was made clear in that exchange- elections are not at all the only means for holding officials accountable: there are judicial means, checks and balances, transparency controls, responsiveness requirements, supermajorities, performance requirements, referenda… lots of mechanisms which go well beyond the supposed “accountability” of elected officials we now have and which seems to be of such value to you.


  41. Accountability is an interesting concept for an allotted assembly…

    Elections are an after-the-fact throw-the-bums-out procedure, which provides theoretical accountability of individual legislators. The notion is that legislators can be punished if they fail to represent their constituents adequately. In reality, with safe-seat districts in the U.S. or party lists in other countries, individual legislator accountability is essentially just theoretical. So elections really only provide accountability in the coarse sense of replacing one majority party (or coalition) with another.

    With an AC the whole concept of accountabiltiy hardly applies. There is NO accountability concept for individual members, since they are not being asked to represent any constituents, but merely to do what they individually think is best, as others would do if they were in their shoes. And there is still something that serves the same function as accountability on the coarse level, in that every AC will be completely replaced, and any law that was passed by one AC can be reversed (or re-affirmed) by a future AC.


  42. I think we would all agree that interests are ubiquitous, the issue being whether to seek to channel (and thereby contain) them or to attempt to exclude them. Take your pick — Mandeville or Rousseau, private vices/public virtue or the Comité de Salut Public. My preference is the former, via a strict separation of function (advocacy and judgment). I agree that there will always be some “bleed” effect, but attempts to suppress interests never work (or at least that’s the conclusion us world-weary cynics and liberals would draw).

    I would also question why we automatically assume that special interests have to be sinister. Isn’t there a continuum between interests and information? If you take, for example, climate change legislation, why would it be wrong for high-energy-use corporations to lobby an AC, given that there would be serious economic and employment issues for a country if their interests were ignored? Is a Greenpeace lobby information or interests? Surely there’s a continuum between Greenpeace and the corporate lobby as there is a continuum between information and interests (to assume the possibility of disinterested advocacy is more than a little unworldly).

    One of the advantages of political parties is that they bring interests out of the shadows and also require some degree of popular support in order to receive a voice, so it could be argued that parties purify and democratise interests. Deny them and interests simply go underground. The last time America went down the prohibition route, the end result was the mafia. As for the fact that party campaigners will achieve some influence via gaining the support of some members in an AC, the same could be said for religion or any of the myriad influences that human beings are subject to, so I don’t see why you would want to demonise those who seek to influence opinion on political affairs openly, via party campaigning. In this sense the re-election interests of the party only differ from other interests in so far as they are out in the open. As you rightly point out the problem is far worse (fatal, in my opinion) if you assign an active role to an allotted chamber, so I’ll leave it to Yoram, Jorge and yourself to deal with that can of worms!

    I agree that accountability is a crude ex-post phenomenon that is only exercised in an extremely patchy way, but being able to throw the bums out is better than nothing at all (Schumpeter was right in that respect). Otherwise, as Jorge points out, all you are left with is rules and institutional devices. A structural solution, in which advocacy and judgment have their own rightful place, has the benefit of being less reliant on the public prosecutor. As for the argument that bad decisions can be reversed, isn’t it better to close the stable door before the horse bolts? If a politician or government officer knows that a bad decision will put him out of a job she will be far more careful to ensure that she takes the right decisions. If parties were still in charge of the agenda-setting function, then electors would punish those who came up with unsustainable policy options.


  43. Just as a clarification, in case I have not explained myself as to agenda-setting and initiative: in my view these functions would be shared by the elective parliament, the AC and Government (Executive).

    In fact, most initiatives would come from Government (as it happens today) – However, distributing the initiative power prevents any monopolisation of that function by any of these three branches.

    I hope that we may achieve more clarity with the help of David Schechter’s matrix…

    As to accountability: we had a long discussion under
    I think that given the weakness of electoral accountability we should not count out other kinds of “institutional” accountability (which BTW are the rule for holding non-elected officials to account nowadays)

    As to corruption: I did not see any reply to my last posting, especially any reply which may demonstrate that an AC with the rights safeguards would be equally or more prone to corruption than an elective chamber.


  44. Hello Jorge, glad we agree on the need for a multiplicity of agenda-setters. In my first book I restricted this to government ministers (as did Rousseau and Mill) along with members of the house of advocates, but was persuaded that this was undemocratic and therefore reintroduced the elective element extremely reluctantly (“holding my nose”, as I put it in the introduction to the second book). I hope this confession of apostasy indicates that I am not as “dogmatic” as Yoram suggests. In fact I’m happy for another allotted assembly to originate policy (Terry’s suggestion) so long as the policy proposals then pass via a votation or similar democratic filter. This wouldn’t be appropriate for government legislation as it is mostly of a secondary “housekeeping” nature (statutory instruments etc). I’m sure though that we all (with the possible exception of Yoram) agree that it would be entirely wrong for the one AC to both originate and judge policy, as advocates cannot judge the merits of their own arguments dispassionately and may be able to sway the assembly by their own passion and oratory — especially as most ordinary citizens are unlikely to feel strongly one way or the other on many issues. Just in the same way that many people vote for a politician because they like her haircut or pattern of speech (whether folksy or authoritative), similar factors cannot be excluded from a randomly-selected deliberative chamber, notwithstanding Habermasian norms.

    I agree, for the reasons clearly stated by Madison, that elective assemblies are easily corruptible. But this is because they combine two distinct functions — advocacy and judgment (as he forewarned). Separate these two functions and you remove the source of corruption. This is standard separation-of-powers theory — the American founders were very conscious that “placemen” of the executive in the legislature corrupt its proper function and sought to prevent this. Unfortunately they failed to address the corrupting effect of combining advocacy and judgment in the legislature and the corrupting effect of appointing the executive and the legislature by the same mechanism (election). We now know that the executive can corrupt the legislature via party affiliation, hence the need to take the separation of powers seriously: executive, advocacy and judgment powers need to be instantiated by their own unique mechanism (appointment, election and sortition). Get the structure right and there will be less need to depend on band-aid solutions like scrutiny. Isn’t it better to go for a clear and simple solution, rather than piling up institutional fixes, hoping that it will all come out in the wash?


  45. I see lot of common ground. Separation of initiative, advocacy and decision are good ideas, as they promote more neutrality, dispassionate judgements and may be part of requirements for harnessing cognitive diversity, but this can also be done within the same AC. It’s just a question of a good design.


  46. For the same “checks and balances” rationale I’m reluctant to leave initiative only to the government or to the elective chamber. An assembly that only can say yes/no is as powerless as the Roman assemblies or the European Parliament…


  47. Hurrah! At last you guys seem to be getting to somewhere more interesting and you’ve done it through deliberation rather than confrontation! Fatigue setting in maybe?

    Jorge: I believe that accepting Keith’s original strictures about the AC being a yes/no ultimate decision-making body allows a lot more flexibility in the design of the rest of the model. Clearly policy-making requires deliberation and compromise. How about considering a parallel sortition-based policy forum and the various interest channels that feed into it?

    Yoram: The study of juries I’ve made, although by no means exhaustive or scientific, really do tend support Keith’s assertions. Quite often it seems as if a room full of reasonable people will find it easier to tear each other to pieces or simply to acquiesce than reach consensus. Bullying and coercion are commonplace. There was a case of mistrial recently where a juror was afraid to go back in the jury room. Keith: I don’t find Fishkin’s solution to this problem particularly elegant or convincing either. Stick to your premise and it isn’t needed.

    Keith and Terry: I see one possible role for elections: to replace the constituency functions of MPs with an elected ‘community commissioner’ as part of a beefed-up citizen’s advice bureau. I don’t see any case for elections as part of the legislative process for the reasons that the others have already stated.


  48. Martin,

    > Yoram: The study of juries I’ve made, although by no means exhaustive or scientific, really do tend support Keith’s assertions. Quite often it seems as if a room full of reasonable people will find it easier to tear each other to pieces or simply to acquiesce than reach consensus.

    I don’t understand your point. Why would you expect a consensus to be reached?

    As far as I am concerned, open inquiry, discussion and proposal formation is a way for each member to reach a more informed and considered decision, not a way to reach a consensus. Decisions should be made by majority voting within the allotted chamber, not through consensus.


  49. Yoram: Apologies, I misunderstood your position.

    My point is that the outcome of a vote of the AC as you describe it will be influenced by coercion and/or force of personality and the less assertive members will not be heard. However, I agree that the attributes you describe are desirable in policy formulation. I think the conflict can be avoided by policy formulation being carried out by a forum and approval of the resulting legislation being carried out by an AC as Keith describes it. Does that make sense?


  50. Jorge: “I see lot of common ground. Separation of initiative, advocacy and decision are good ideas, as they promote more neutrality, dispassionate judgements and may be part of requirements for harnessing cognitive diversity, but this can also be done within the same AC. It’s just a question of a good design.”

    It’s difficult to see how you can get three outputs from a single constitutive principle. Madison believed that a single chamber constituted by election alone could adequately separate advocacy and judgement. History proved him wrong during his own lifetime. History has also demonstrated the folly of using a single principle (election) for the executive and the legislature, whether that be in a congressional/presidential or parliamentary system. It’s much easier to get three outputs from three inputs, so if we want to separate executive, advocacy and judgment functions we need three distinct constitutive principles (appointment, election and sortition).

    I confess I’m baffled as to why people on this list think that sortition has to do everything — it puts us completely at odds with everybody else who has any interest in public affairs. Do we really want to appear like a bunch of crazy fanatics? The claim that sortition has unique benefits but is not a magic bullet at least has the merit of plausibility. So even if you secretly believe that sortition is all we need wouldn’t it be better to keep quiet about that? Unless that is one has the limited goal of endlessly arguing on internet forums; personally I would prefer to make some difference in the real world. But we won’t get anywhere by appearing like a bunch of fundamentalists.


  51. Martin,

    As I see things there are at least two separate questions here:

    1. Should different parts of the decision making process be separated between different bodies (e.g., should proposal formulation be done by a different group of people than the final approval/rejection vote)?

    2. How should the members of each body be selected?

    The first question is in my mind secondary. There are considerations in favor and against separating functions. It is clear to me that one could easily take the separation of the functions of decision making too far, but I would not insist that having the functions split between two bodies instead of unified in one is necessarily a bad idea.

    The second question is the primary issue and should clearly, in my opinion, be answered by having all political bodies selected through sortition. Any other selection process grants political power to a non-representative body and is therefore anti-democratic.

    How would you justify granting political power to a non-representative body, such as an elected body?

    As for power inequalities within an allotted chamber, that is a question of great interest, but I don’t see any reason to think that such inequalities would be particularly large in a properly designed allotted chamber. Whatever inequalities do exist, they would certainly be much reduced when compared to the enormous political advantages that various elites gain over the average person in an electoral system (or even when compared to the great internal power inequalities in elected bodies). Thus, it is incongruent to be concerned about potential minor power inequalities within ACs while at the same time suggesting building a huge power inequalities into the system by relying on elections.


  52. Perhaps this morning I was too optimistic about the extent of “common ground”. I’ll hope that I’m not being labeled a “crazy fanatic” – otherwise I feel this will be the end of this conversation.

    As to “so if we want to separate executive, advocacy and judgment functions we need three distinct constitutive principles (appointment, election and sortition).” or the newfangled rule that you cannot get different outputs from one constitutive principle (What kind of argument is this? institutions may have many different outputs and BTW initiative, advocacy and judgement are not “outputs” per se but phases in a process) – it seems to me quite otherwordly…

    checks and balances means also that powers are shared to a certain point (at least the legislative and executive functions) – see for instance the power sharing of territorial and popular chambers in federal systems – the problem today is that all powers emanate from the same source (the parties). If an AC only has a yes/no power it will be too weak to check the elective and executive powers.

    In fact, we know that with today’s world we will have most initiatives coming from the administrative apparatus – therefore from the executive.

    This is in my view inevitable and corresponds to the needs of our complex societies. In addition, the executive has a large information advantage over any parliamentary chamber. Hence, it is only advisable not to restrict the initiative to this appointed (elective?) power, but to share that function amongst the three branches (executive, elective and allotted).

    Yes/No power is a very weak form of power, much more so in a framework where the executive has such an overbearing power. Nobody has answered this.

    >”How about considering a parallel sortition-based policy forum and the various interest channels that feed into it?” I can accept that we include institutional safeguards in the functioning of the AC in order to prevent certain members of dominating the different phases of an initiative (which in any case in my view then would also would need to go through the elective chamber), but to do this you don’t need two ACs. You just can separate procedurally initiative from advocacy and from the final vote in the AC – e.g. forbidding initiators of taking an active part in the advocacy phase once they have submitted a proposal.


  53. Yoram,

    I agree with your last point. Within an AC, which has no re-election imperative affecting legislators’ behavior, there is less interest in power hoarding. I also think limiting an AC to one topic, further limits the desire to amass power (and vote swapping).

    Elected legislators in the U.S. routinely collect implied I.O.U.s from each other for future votes, and the chair of a committee, or the Speaker, or party leader can obviously gather the most. It seems to me that all of this corrupting pressure (though not other corrupting influences) would not exist in a single issue AC.


  54. It is off topic, so I won’t belabor it, but I think it would be useful to have a separate string about selection of the executive (whether individual or collective)…election, vs. appointment by an allotted body, or some hybrid as in the Italian city republics.

    I tried to find online any research in the U.S. comparing elected mayors with appointed city managers, and I could find very little, and nothing useful.


  55. Jorge, I use the word fanatic refer only to those who see sortition as the one legitimate appointment method for political office holders — an approach which manages to combine a disdain for history, philosophy and realpolitik. Regarding the difficulty in obtaining multiple outputs from a single input (election) I would refer again to the historical examples that I cited — in practice it just doesn’t work! I agree that Montesquieu’s policy of checks and balances has never been properly instantiated, hence my insistence that each power should be realised by its own unique appointment principle.

    I’m not sure why you consider the power of veto to be a weak one; the key factor is surely who gets to exercise the veto? (in my proposal, the AC). If ministers and elected politicians want to get their policies approved then they have to come up with ones that will gain the considered assent of the representatives of the people. If you look at the history of the English parliament, it was at its most powerful when it was in a position to withhold funding from the executive by vetoing the budget. Bear in mind also that government ministers can be removed by a simple vote of confidence in the AC. This strikes me as very real power — in fact it might well be argued that the power of the AC was such as to undermine the stability of the government, so I’m puzzled why you would seek to grant the AC power of initiative and advocacy as well.


  56. >”government ministers can be removed by a simple vote of confidence in the AC” sounds good. We will need to have a complete picture of AC functions to have a more reasonable and reasoned discussion.

    Veto power is important, especially if the source of legitimacy is different. I agree with that. However, it is a weak power compared to an executive and an elective chamber with initiative and advocacy powers – and codecision (in case of the elective power?)?

    As I tried to explain, nowadays the initiative power (which implies an agenda-setting power) is absolutely key in the real functioning of government as a whole (in fact in Europe at least most initiatives come from the executive and from the EU “ademocratic” institutions) and an AC sharing in that power (opening up the agenda, giving a voice to normally disorganised interests) is key to me. Realpolitik 100%.

    A full matrix could be indeed very helpful to have a full picture.


  57. Terry, you are right in pointing out the need to discuss the method of appointing the executive. May be a topic for a related post as you suggest?
    What I would have in mind in this regard is to start at least with appointments being made by the elective power with the informed consent of the AC. The AC would share in control powers over the appointed officials. In my 2009 article I also suggested that each department of the executive could have a specific advisory allotted citizen board which would compensate more “corporatist” existing advisory boards (for instance the Post department has an advisory board with all interested public agencies, trade unions, trade associations, etc on it with tend to represent only special interests…).


  58. Great idea Terry, do I hear you volunteering to start the new thread?

    Jorge, I certainly don’t advocate co-decision in the electoral (or any other) case, all that politicians get to to do is to propose those policies for which they have received an electoral mandate. All decision powers are reserved exclusively for the AC so, formally speaking, politicians and government ministers have no power at all as they cannot determine the legislative outcome and their job security is at the whim of the AC. In modern parlance they only have “soft” (influence) power as opposed to formal statutory power.

    Political elites would undergo two constraints: the need to please the voters and the need to convince the AC. I admit this would privilege a “populist” agenda, as opposed to radical innovations, but that’s a consequence of majoritarian democracy (as opposed to rule by elites, which come in all sorts of shapes and sizes and straddle the political spectrum). I’m sure that Yoram will view this as another incoherent non-sequitur but it’s no more so than “fighting fire with fire”; the theory of variolation in medical science; Madison’s notion of interests checked by interests; or Mandeville’s argument that private vices can lead to public benefits — however distasteful the latter two may be to civic republicans.

    There’s a matrix at the end of this post:


  59. Jorge, Terry,

    Since both of you (I think) see elective bodies as promoting elite interests, I really do not understand why you insist that it is a good idea to grant political power to such bodies? How does empowering the powerful serve the interests of the population? Wouldn’t the expected result be the use of those powers to maintain and enhance the advantages of the powerful at the expense of the average person?


  60. A quick comment on why one might think veto power to be weak. (I believe I covered this a bit before, but here goes again.) Veto power is politically kinda like a neutron bomb, at least under the right circumstances. Sure, when it comes to small changes to the status quo, veto power can prevent unjust/unpopular/whatever proposals from being realized. But things are different when it comes to big proposals. If there are decisions that absolutely must be made, on pain of terrible consequences for everyone, the proposed decision would have to be insanely bad for a veto to be justified. This is why all kinds of terrible riders get added to must-pass congressional legislation all the time. (Witness the insane fascistic attack on trial by jury included in the latest defense appropriations bill.) And so if you have a bad/corrupt/unjust person in charge of making proposals, the veto might not help very much when the chips are down. If a bill need not be passed, the proposer can simply propose nothing. If there is need to pass something, the proposer will propose his most-preferred option that will not provoke a veto. No rational agent/assembly will veto such an option unless it’s worse than doing nothing–and if the stakes are high, there may be very few things worse than doing nothing.


  61. Terry,

    Vote trading is not a negative phenomenon, in my opinion, but rather a natural way of achieving mutually beneficial solutions. It happens all the time in human interaction. I don’t think there is any reason to discourage vote trading or try to avoid it.


  62. Peter,

    I agree. I would also add that under any circumstances whoever composes the proposals can make them convoluted enough so that it is extremely difficult to decide whether it is better to take them or leave them. As I pointed out before, this is in fact standard practice already for bill coming before Congress. Bills like Obamacare or Medicare part D are very costly giveaways to special interests which also have some desirable components. Weighing the negative parts against the positive parts is very difficult, but coming up with a superior proposal is very easy.


  63. Peter, it’s hard to argue with your point, that’s why my preference is for some form of direct democratic initiative + votation. On the other hand it would be a shame to sacrifice the ex-post accountability that can only be provided by party election. So why not have both at once? Parties can put forward their proposals and so can the public. These would then be ranked in a public votation and those chosen would go to the AC. That way you get electoral accountability, public initiative, democratic choice and considered judgment. In practice a political party would end up as just a regular source of policy initiatives, but those with the most successful track record would be likely to win the public votation the next time round. Similarly an ad hoc body that put forward a successful initiative would benefit from public trust and would eventually morph into a regular political party — this is already happening with the transformation Green politics and other single-issue bodies like UKIP into general political parties. Parties would become very different creatures if final decision power were in the hands of an AC.


  64. Yoram, although I sympathise with your views I feel that a mixed system is the only reasonable thing to advocate nowadays, where the elective power and the parties would need to be strongly curtailed…
    I would not want to repeat what I have already tried to explain:


  65. Keith (Yoram): I am not a contributor, so cannot start a thread about how sortition might play a role in executive selection. One of you might, or i would be happy to become a formal contributor.

    Yoram: Vote swapping is not ALWAYS bad, but generally is. Vote swapping is different than mere “compromise.” It means decisions are made contrary to merits (or even perceived merits). One example…legislator A wants a special amendment that benefits only her town. legislator B wants one that benefits his town…with extrapolation you can see that by agreeing to vote for each other’s amendments, it is possible for many items to pass, each of which a large majority of legislators sincerely believe to be bad policy. This is common in elected legislatures (pork, ear-marks, special interest give-aways, etc.) while there are MORE such items when legislators seek to gain favor from voters, this dynamic would still exist in multi-topic allotted chambers.


  66. Terry, if you email me your wordpress username, I’ll set you up as a contributor. (You will need to create a wordpress account if you haven’t done so yet – go to and click on “sign up”.)


  67. Terry,

    As I see things, the problem with the scenario you described is not the vote swapping itself, but the fact that the two delegates are representing very narrow interests. Like any other political process, vote trading can be used to promote bad policy.


  68. Terry’s observations on vote-swapping would suggest that performing a sortition for each legislative bill is the optimal solution. Obviously this would preclude an active role for the AC, as it would always be populated by novices and would lack the necessary experience. But we must agree with Terry that vote-swapping is a danger with any multi-topic assembly, elective or sortive.

    However the danger could be addressed by the simple expedient of the secret ballot for assembly votes, as vote-swapping is ineffective in a system where votes are not recorded. This is not possible in an elective system as the public needs to monitor the behaviour of their own constituency (or party) representatives, but there is no corresponding reason for recording the votes of allotted representatives.


  69. As to vote-swapping I think much more discussion would be needed before taking it for granted in an AC environment. Some initial thoughts: Vote-swapping happens mainly because party interests, special interests or local interests dictate the beaviour of representatives. Therefore they seek to achieve “deals” when they lack a sufficient majority. They do this because those “interests” are key for their future political career. And this happens in an environment where legislators or their groups know each other well.

    In an AC we would not have such special interests as a main driver for the behaviour of representatives. They have been selected by means that ignore such interests and they lack a prospect for a future political career. Influence by external interest groupings can be disencouraged by conflict of interest, transparency, anticorruption and other rules. Members in the AC (if its duration is say only one year) will not know each other very well and as groupings by district, interest and ideology would not be permitted the deal-making would be severely limited.

    Internally “interest” groupings may be forbidden (at least their explicit forms such as seating together, forming working groups or factions) or strongly discouraged (e.g. in the Boule seating seems to have been allotted).

    Procedural separation of initiative, advocacy and voting would be another tool, e.g. One or more AC members are allotted to a subcommittee on, say, health… They see that a Bill would be necessary and draft an initiative with explanatory documents (with help of the technical services of the AC). The Bill passes to an advocacy/deliberation phase where two teams are allotted to argue in favour and against of it. Then the whole chamber votes on the Bill….


  70. Jorge, would you propose secret voting? If that is the case then vote-swapping ceases to be a problem.


  71. Secret voting may be a tool, but I’m not sure whether it should always be the rule (as it may create certain darkness in the final behaviour of AC members – which in turn may feed corruption) or a possibility which may be invoked by any member of the AC in certain cases.

    What do you think of the possible solutions I have proposed?


  72. As you know, I favour structural solutions rather than procedural rules and don’t believe that ACs should have initiative and advocacy rights. There would be no argument for a non-secret ballot in the case of an AC with judgment functions alone as there are no individual “descriptive representatives”, so the final tally of votes is all that matters. If the AC were to have active functions then all of Terry’s concerns about pork- and vote-trading would apply.


  73. >”all of Terry’s concerns about pork- and vote-trading would apply.” I see that my arguments about the differente situation and incentives in ACs on vote-swapping had no impact on you at all.


  74. Jorge, I’m unsure why you feel that randomly-selected members are more likely to act in a disinterested way than elected members. Allotted members with active powers will be easy targets for rich and powerful lobby groups as they are unconstrained by the very factors that you disparage in elected members (the need to demonstrate integrity and party-political consistency in order to secure re-election from voters who feel that their chosen politicians have represented their interests with an adequate degree of consistency). In addition we all have our own personal interests, preferences and views (eccentric or otherwise) and individual members will be unconstrained by the need to ensure that their own views adequately reflect the incidence in the general population (when voting, this takes place automatically). In this sense the electoral process can be seen to sanitize politics by bringing interests and partiality out into the open — not so with the lot, which only sanitizes ex-ante. I’m afraid that the deeper you go into it, the more you demonstrate that the mandate of allotted assemblies should be limited to aggregate functions (i.e. voting), or a very extensive role for institutional rules, procedures and the police. As Burke put it, albeit rather dramatically, “at the end of every visto, you see nothing but the gallows”.

    I’m reading Carl Schmitt’s Concept of the Political at the moment: his central thesis is the connection between the political and the friend-enemy distinction. He’s certainly right regarding the origin of political parties (in the UK case, the Civil War); Hobbes’s theory of sovereignty involved the abolition of the political. I suppose one might counter Schmitt that the problem is the party — abolish it and you will get universal love and peace. Hobbes and Schmitt would argue that you’re more likely to end up with the war of all against all. Better therefore to contain (and restrain) the political within the formal adversarial joust, leaving the AC to do what it does best, i.e. judging the outcome of the chivalrous tourney.


  75. Well, Schmitt is certainly an interesting read but his solution to parties was the NSDAP!

    As to AC members and possible dangers: lot+ short terms + anticorruption rules + transparency + the lack of career incentives + lack of special interests as their main driver will render them less corruptible. There is a big difference between pork-barrelling as your endogenous driver (elective reps) and being corrupted by external interests (which may be prevented by strict rules). You may believe so or not, but it’s definetly different.

    As to individual interests: well, after all the meaning of ACs is to have a sample of citizens (and their interests – which determine the way they propose, they advocate, they deliberate or they vote) in the institutions. The key is to prevent grupings, factions or individuals to hegemonise the workings of the AC – but this can well be achieved to a large extent by procedural rules.

    An theur proposals, positionings and votes (unless made secret) would be also in the open: subject to scrutiny by courts, by the population at large, the media and the other branches of Government… this leaves small space for corruption.


  76. I agree that pork-trading and external corruption are separate issues; unfortunately a multi-topic AC with initiation and advocacy functions is a potential victim of both. At heart it gets down to whether structures or procedural standards are the best solution. My preference is always for self-regulating structures. At heart the Harringtonian constitution is cybernetic — a self-correcting mechanism that does not require external policing.


  77. Well, Keith, I think that in your framework the AC would be at the expense of proposals made by others (executive, elective chamber?) subject to “leave it or take it” tactics, with an agenda dominated exclusively by those elective powers and proposals full of the pork-barrelling and corruption we see nowadays (product of our elective system).

    I don’t think that such weak veto-power could sanitise the whole institutional framework you propose more than the institutional safeguards and procedural rules I’m thinking of.

    At least it seems that we share the goals ;-)


  78. Yes we agree on the goals. As to the means you are right about the agenda setters, but politicians would need to formulate policies that passed the macro- and micro-democratic filter. Only the most popular policies would be selected in the public votation and these would then need to pass the scrutiny of a descriptive sample of the same public. I can’t see how pork-barrelling could possibly survive such a dual filter and am puzzled as to why you describe it as a weak veto power. I’ve already acknowledged the accuracy Peter’s criticism on this point, and have suggested therefore that agenda-setting should be a combination of party and direct-democratic initiatives, which would be subsequently narrowed down by a public votation.

    By all means describe the setting of the agenda by political elites as corruption if you wish, but the Disraeli-Constant definition of the political party as “organised opinion” is not entirely anachronistic (the obverse of this — agenda setting by an AC — would, of course, be disorganised opinion). You view the political party as a sinister force, whereas I see it as a mechanism to bring interests out into the cold light of day in order to be scrutinised by the public and media alike. No-one is forced into party affiliation and there are good grounds to believe that parties continue to some extent to reflect organised opinion (notwithstanding Manin’s observations on audience democracy). Note that the latter aberration would be much less significant if party leaders were no longer charged with forming a government.


  79. >”combination of party and direct-democratic initiatives, which would be subsequently narrowed down by a public votation” experience teaches us that those devices will be dominated by party politics and special interests – I’m more than skeptical as to the workings of direct democratic devices… In the end structured powers will make the proposals and the AC is left with a yes/no which is weak.

    “Agenda setting by the AC” I don’t claim this. I claim a co-initiative power for the AC which will coexist with initiatives by the elective power, the executive (who will dominate) and possibli citizen petitions with low thresholds.

    Political parties: I am myself advocating for their survival in this forum (as they can provide today’s analogue to the function developed by the elites in Athens in structuring public debate and providing some key political personnel) BUT leaving initiative and advocacy solely to an elective power (and/or direct democratic devices) mantains their monopoly on the key power in politics (agenda-setting, initiative). This would maintain their hegemonic position – which leads to absolute power which absolutely corrupts… Therefore I’m advocating for this monopoly to be broken up and I trust an AC much more than direct democractic devices to be able to open up that key function – and by this means establishing a structural check against established special interest domination.


  80. Jorge: “I claim a co-initiative power for the AC which will coexist with initiatives by the elective power, the executive (who will dominate) and possibli citizen petitions with low thresholds.”

    I’m glad to see that we agree completely on the necessity for a plurality of agenda-setting initiatives. The problem with pluralities is that there will inevitably be conflicts between the proposals from each group (elective, initiative, petitionary, sortive, executive etc.), so who decides? To my mind if we really believe in democracy this has to be a public votation. I would be perfectly happy with sortive initiatives so long as they passed the public-choice test.

    The only exception should be government secondary legislation of a purely housekeeping nature.


  81. Conflicts are inevitable and even healthy in mixed systems I think ;-)

    >”So who decides?” That’s a very good question (typical of Schmiit: who has in the end the decisive power…).

    I guess that once the Bill is underway the question of who decides is between the elective and the allotted chamber and the role the population at large may play.

    I would be in favour of granting both chambers the same rights – so they can block each other if needed – probably with a supermajority of the AC being able to trump the elective chamber, but with the possibility of any of both chambers (or a % of its members – 40%?) calling for a popular vote.


  82. I think the current US fiscal impasse should warn us against gridlock solutions. If you go down the supermajority route then you have the possibility of a single elective chamber making a proposal and then judging it itself. Terry has ruled that out for very good reasons, so I think his alternative suggestion — a different higher-level sortive body to decide between competing proposals — should be seriously considered. I would certainly not rule it out on principle as an alternative to a public votation. But there are a number of potential problems:

    1) The benefit of sortive over votive decision-making is that it is properly informed, so for a higher-level sortive body to select between conflicting proposals each one would need to be subject to balanced deliberation. So it would depend on how many alternatives were being viewed. If there were a lot of alternatives, and the resulting debating time for each was limited, then the sortive solution would offer no benefit over the (uninformed) public votation.

    2) However the principal danger would be if the higher-level sortive body selected an option that originated from a sortive assembly (and then went on for final yea/nay deliberation in a sortive chamber). This (Terry’s model) would mean that the law would at no time have anything other than a sortive mandate, and I seriously doubt if that would be acceptable to the majority of citizens who did not get to show any sort of preference. Once again, we would have a klerotocracy, not a democracy, albeit a pluralistic klerotocracy.

    Needless to say this presupposes your acceptance of the multiple sortive body model as opposed to Yoram’s suggestion to concentrate all powers in a single body (subdivided or not).


  83. Some cases of gridlock should not negate the fact that even with elective chambers the two symmetrical chamber system works remarkably well (see Germany and Switzerland) – gridlocks could also be solved by calling a popular vote


  84. Keith,

    I am confused about how votation could work, unless you are imagining that only a handful of high-visibility issues would ever be tackled by an AC. When I was in the state legislature, nearly 500 separate bills were introduced each year. Some of these dealt with the same topics, but still…that is a lot of raw material for a votation filter to cope with.


  85. Is it the case that the majority of the bills were of an incremental housekeeping nature? I had assumed that votation would apply to the big ticket manifesto and initiative proposals — that’s why I drew a distinction between government-initiated and party/citizen-initiated bills (the former would proceed straight to the AC without the need for an intermediary stage). How would the 500 break down?


  86. Keith,

    As you know, the U.S. Congress and state legislatures operate very differently than the UK (or other parliamentary systems.) Here is a rough run down of Vermont’s legislation.
    Vermont holds elections every other year.
    No party presents a manifesto. (Nobody knows what’s in the “platform.”) Each candidate is free to introduce (alone or with co-sponsors) any bill they like (legislative staff provide drafting assistance). In a two year period, out of around 1,000 bills introduced, about 100 to 200 get enacted. A majority are in some sense “house-keeping” though many of these also have extremely controversial and hotly debated policy elements. The annual budget bill is the most policy-laden “housekeeping” bill. Maybe 300 introduced bills are policy-changing non-house keeping bills. Maybe 30 of these big bills are enacted.

    In other words there is a spectrum from minor to fundamental, with no sharp break points.


  87. Yes, the two systems seem to have very little in common — if anything the VT system sounds a lot closer to an AC with policy-making powers as, without an electoral manifesto, voters’ choice of candidates would appear to be almost random! Although there is no formal role for partisan organisations in your description, these presumably developed incrementally (as they did in the US in general and the UK). The UK constitution has only very recently acknowledged the existence of political parties, although they have for some time been the principal causal factor in both electoral choice and government behaviour. The elephant in the room has only just been noticed.

    As a thought experiment, just replace the sentence “Vermont holds elections every other year” with “Vermont holds sortitions every other year”. What changes might one anticipate as a result? Whilst we might like to imagine that policy proposals would result from the free exchange in the Assembly, is it not equally likely that lobbyists would seek to control the agenda by bribery? One can also imagine the continuation of pork-trading, as members sought to achieve their sponsors’ goals by trading off one policy for another. So what would change?

    Meanwhile the people continue to be excluded from the political process. If so then we might welcome the return of political parties as a way of blocking the sinister influences of lobbyists and providing some voice for the excluded public.

    At heart the difference between us would appear to be one of political anthropology: is the venality in human behaviour a product of structural arrangements (elections, parties, capitalism etc) or tendencies already present in the human heart? Are we born free but everywhere in chains or is man a fallen creature? It’s hard to come to a verdict over such fundamental anthropological issues, but cautious assumptions are less likely to lead to disappointment — assume the worst but hope for the best. Liberals like to see themselves as optimists, but that is a recent aberration — most political anthropology is ultimately pessimistic (as Schmitt rightly argued). I can’t offhand think of a political movement founded on an optimistic anthropology that has led to any arrangements that have sustained longer than a few years — the French Revolution being the obvious case in point — and it would be a tragedy if the fledgling sortition movement had a similar history.


  88. Yoram: “The primary issue [should be] having all political bodies selected through sortition. Any other selection process grants political power to a non-representative body and is therefore anti-democratic.”

    Assuming this is based on remarks by Aristotle, Montesquieu and others that elections are associated with oligarchy, sortition with democracy, we shouldn’t forget that these are references to the ancient world where sortition was a mechanism to enable rotation (rule and be ruled in turn). Aristotle would have viewed sortition in a large political community as a form of oligarchy (rule by the few), the only difference from classical oligarchies being the selection principle. Further argumentation is required to establish that sortition in large states is a form of democracy, anything else, including the use of the word “representative” is just sloganising (I won’t claim that it’s a non-sequitur as that presupposes an antecedent argument). As we are all aware, “representation” is a concept with a variety of different meanings and the devil, as always, is in the detail.


  89. At the beginning of this post, Keith wrote, “On many occasions I have argued that the representativity of political assemblies constituted ‘descriptively’ (i.e. by statistical sampling) only applies at the collective level, and that this requires members of such an assembly being limited in their function (in contrast to the mandate of elected members).”

    I agree that an allotted chamber would only be “descriptively representative” as a whole, not as individual members. However, I’m not sure why that requires the assembly members to be limited in their function, or why the mandate of elected members is better.

    Suppose that we have an allotted chamber, selected in a way that makes it a true statistical sample of the whole population, that is considering whether to approve or reject a particular bill. Let’s say that they will be first hear arguments from proponents and opponents of the bill, and have the chance to ask questions, and that after that, they will deliberate with each other, and then cast secret votes about the bill. And suppose they were polled about their views on the bill three times:
    1. Before hearing pro and con arguments
    2. After hearing pro and con arguments, but before deliberating
    3. After deliberating

    Now, let’s suppose that the entire population could go through a similar process about the same bill — hearing pro and con arguments, asking questions, deliberating, and voting. And suppose that the entire population was surveyed at the same three points. To what extent would the allotted chamber’s views (the distribution of support and opposition) match the views of the entire population?

    Before hearing pro and con arguments, the chamber’s distribution of views would probably have a very strong relationship to the distribution of views in the whole population. After the allotted chamber and the whole population heard pro and con arguments, how strong would be the relationship be between the chamber’s views and the whole population’s views? Probably not as strong (I only took one class in statistics). After the chamber and the whole population had deliberated, I’m guessing that the relationship would be weaker still.

    However, I’m guessing that even after hearing arguments and deliberating, the views of the allotted chamber would be much more “representative” of the whole people than the views of an elected legislature would be, simply because the allotted chamber’s membership would be so much more “descriptively representative” of the whole population than the elected legislature’s membership.


  90. David,

    > after hearing arguments and deliberating, the views of the allotted chamber would be much more “representative” of the whole people than the views of an elected legislature would be, simply because the allotted chamber’s membership would be so much more “descriptively representative” of the whole population than the elected legislature’s membership.

    I completely agree. If you ever read a useful response from Keith to this point, please let me know. In my many discussions with him I have never seen him rebut it.

    As for:

    > After the chamber and the whole population had deliberated, I’m guessing that the relationship would be weaker still.

    I don’t see why. The allotted chamber is representative and is therefore representative of the post-deliberation position of the population as well as of the pre-deliberation position of the population.


  91. David, I’m delighted to agree with you (and, more surprisingly, Yoram as well). The procedure that you describe is very close to the Deliberative Polling methodology, and Fishkin has robust experimental evidence that confirms your prediction of the difference between pre- and post-deliberation views. I’m sceptical about the value of the chattering, but defer to Jim Fishkin’s judgment, who assures me that the small-group deliberations are an essential part of the process. I’m more cautious about the value of the final plenary chattering for the group-psychology reasons that Terry has provided. (Note: I use the word “chattering” rather than deliberation as I adhere to the Latin meaning of the word (weighing the outcome of a debate) and this can as well be done silently.

    I’m relieved that your proposal does not assign an advocacy or policy-initiation role to a statistically-representative assembly as this would be a clear breach of its democratic mandate. So, in sum it would appear that any differences between us are over emphasis rather than functional role, as the prime task of the statistically-representative body would be to weigh the evidence and then vote in secret. The only minor quibble is whether or not the weighing should be silent or vocal. As a matter of interest what value do you think the final vocal deliberation would add, bearing in mind Terry’s strictures on the dangers and that there is no need for trial-jury consensus?

    Needless to say I’m puzzled as to why Yoram thinks I have argued that elected members are more descriptively representative than allotted members, as that would clearly be absurd (I’ve consistently argued that elections are for the active representation of interests, not descriptive judgment). Why would I seek to rebut a claim that I agree with and why would I be doing a PhD which advocates sortition as the best way of implementing descriptive representation in large political communities if I thought that elections would suffice?


  92. Keith and Yoram,

    Stop talking past each other…:-)

    The difference you have is whether a self-selected SUBSET of assertive individuals, if given the right to help shape a bill in a descriptively representative body will distort a final bill in a way that if the whole population was involved, they would not want.

    Is the subset of “talkers” so unrepresentative of the whole that they undermine the democratic process (Keith’s view), or so much MORE representative than elected officials would be, that they enhance the democratic process (Yoram’s view)?


  93. Agreed, but my primary concern is not so much the shaping (fine-tuning?) of legislation as introducing it in the first place. A small minority of people tend to feel very strongly about certain issues and they would naturally exert a disproportionate influence over agenda-setting. Elected representatives can at least make claims about the supposed interests of their constituents, whereas individual allotted members only represent themselves. This is equally true of voting, but the sum total of the votes “represents” the whole population to the extent that the sample is statistically accurate (no similar claim can be made regarding the sum total of the speech acts of individuals).


  94. Terry,

    I think you gloss over the main defect of the “talkers” argument:

    A subset of self-selected assertive individuals would exist in the AC to the same extent that it exists in the population. Why then would the effect of the “talkers” on the allotted delegates be more pronounced than it would be on the population as a whole?

    In fact, there is every reason to assume the opposite would be true. In the AC the delegates have the opportunity and the motivation to gather their facts independently, employ their own judgements and make up their own minds. For the average citizen, on the other hand, making an informed and considered judgement would require an investment that is completely disproportional to their political power. The result is that in mass politics settings the “talkers” have a much better opportunity to manipulate and control the discussion.

    In addition, in the mass political context, the “talkers” can employ superior resources – such access to mass media – that amplify their powers. Such resources are much less effective in a group of hundreds than they are a group of millions.

    The effects are not theoretical: the self-selected “talkers” do control the entire mass-political discourse. The elected are all “talkers” and they are backed by various other “talkers” in various positions of power. Thus, by the time decisions are made by the elected, the “talkers” have already completely overpowered the rest of the population. Whatever advantage the “talkers” may have in an AC, it would never be as total, as complete, as absolute, as it is in the mass political arena.

    The whole point of sortition is to turn politics into a small group activity – and the advantage of that is exactly the reduction in the advantages of the elites over the masses. For the reasons I mentioned (and for other reasons) the dynamics of small groups make them naturally conducive to equality of power, while the large group setting is naturally conducive to power inequality. Sure, some advantages may persist in the AC to some extent, but speculating about potential inequalities in the small group setting of the AC is fine, but doing so and at the same time ignoring the much more severe inequalities that exist in the large group setting of mass politics is just not credible.


  95. Yoram, you are right to point out that the incidence of loudmouths in the AC would be proportionate to their incidence in the whole population. Democracy, however is premised on political equality, hence the need to limit the role of the AC to those activities that do not impair political equality by empowering the domineering. You are, however, conflating “loudmouthing” in an AC with elite power in general, and I would like to see your evidence that the dynamics of small groups make them naturally conducive to equality of power.

    No one would deny the domination of elites in elective democracy, but it’s a further step to claim that there is such a thing as an elite of “talkers” who have already overpowered the rest of the population. Pluralists would argue that there is no such monolithic elite and that a more constructive route is to institute a competitive inter-elite struggle, whilst simultaneously enabling any citizen to attempt to set the agenda themselves. If you look at the (albeit brief) record of UK epetitions there is some evidence that this is working — an obvious example being the epetition over fuel-price increases which has already secured a change in government policy.


  96. I find Yoram’s argument quite compelling.

    for Keith: You believe that democracy=equality=don’t empower the “domineering.”

    But that would suggest that NOBODY should ever be allowed to talk about politics (because some are better at it).

    This brings me back to Jorge’s December 8th thought experiment above. If the ENTIRE population could effectively convene, pay attention and participate in decision making, we would expect some people to participate more and more effectively than others. It is expected that some will be more persuasive than others. If ALL are present and able to participate, we would call this democratic. So why would the rules change if it is a representative sample participating instead?


  97. Everyone should be free to do whatever they like (within the law) but the primary purpose of the institutions of representative democracy is the establishment of political equality. It’s an unfortunate fact that statistical sampling can only secure this benefit in terms of the counting of votes, thereby constraining us to locate the other elements (policy initiation and advocacy) in areas that are subject to the dominance of elites. The way to overcome this is to ensure a plurality of choice, and to enable any citizen to (in effect) join the political elite by launching her own policy proposals.

    I don’t quite understand your reference to Jorge’s thought experiment — convening, paying attention and participating in decision making (via the secret ballot) does not contravene democratic equality. This would be the case in both examples (the AC and the whole population, the only problem with the latter being that it’s impossible in practice). Loudmouthing would be inappropriate in both cases, that’s why polling stations only supply ballot boxes, not loudhailers or soap boxes for voters to stand on to harangue their neighbours. What people do in private is of no concern constitutionally speaking.

    Note that these constraints are dictated by the inherent nature of the two processes (statistical sampling and electoral choice), therefore we have no alternative but to work within those constraints.This is a conceptual problem, so our own political preferences are of little consequence.


  98. To reduce Jorge’s thought experiment to its essence…Suppose ten people crash on an island with no hope of rescue. They have a meeting to decide how to function as a mini-society. They all are allowed to propose ideas, and debate pros and cons. They follow ideal procedures with time limits per speaker, etc. After they have all said their piece, they vote. Do you agree that this is democratic? What if it happens that two of the ten made all the proposals (but all were allowed to make their own) and did most of the debating? Does that violate political equality?


  99. Your example is one of direct-democratic equality, but what is its relevance to large societies where representation is required? In the case of direct democracy all are free to participate actively, not so in the case of indirect (representative) democracy, where the impact of my own views depends on the speech acts of others, and there is no reason to think they will be instantiated proportionately to the prevalence of particular views in the population — the relationship is arbitrary, rather than statistical. Arbitrariness is the core of the blind-break argument for sortition, but the argument for sortition as a system of political representation presupposes a statistical relationship (the opposite of randomness).

    Representation is either formal, symbolic, active, ascriptive or statistical — under which category does the AC fall where members introduce and advocate policies? Or is there some category that scholars of the representative process have overlooked?

    For a more relevant thought experiment, I really would appreciate a response to


  100. Keith, every time I think I understand your position, something goes wrong. As far as I can tell, you agree that it is democratic for a small population to make decisions directly, even though some people talk more than others, know more than others, etc. You also believe that it is democratic for a random sample to make decisions for a population that cannot make decisions directly, because the sample will act in the same way that the population would. And you seem now to believe (although I think it’s a highly debatable inference) that you’d have the same proportion of people knowing more, talking more, etc. in the sample that you’d have if the entire population could do it. That seems to warrant the inference that the sample will, with deliberation, reach the same outcome that the entire population would, with deliberation. So why not let the sample deliberate? Am I missing something here? How can it be a violation of political equality for the sample to have loudmouths and know-it-alls, but not for the entire population to have loudmouths and know-it-alls?

    Perhaps it would help if you could lay out in advance all of the values you think political decision-making needs to take into account. Then we could ask how well the various alternatives advance to those values. Most of this seems to be about democracy, but you occasionally throw in other issues like competence as well, and I’m still hazy on how they interact.


  101. Gotta rush out in a minute, so will reply in full later. It all hangs on the simple distinction between the individual and the collective. Every individual is present in person at a direct-democratic event but their presence in a representative event is only as part of a aggregate (either statistical or via electoral choice). Therefore anything that distorts that statistical average affects that presence. Neo-nazis and anarcho-syndicalists are in a small minority in the whole population but their “presence” could be increased by a single AC member with a charismatic voice, whereas an election under PR rules would accurately match their presence in the population (if only one member out of 600 were returned in an assembly, she would be marginalised as in a minority of one). Although this is an extreme example, the principle applies across the board, hence the illegitimacy of speech acts (neo-nazi or otherwise) by individual members of a statistically-representative body. It really is as simple as that — is the difficulty that people assume I’m making a more sophisticated argument?


  102. From what you just said, it sounds like it all hangs on how you regard the law of large numbers working. That law says you’d have real proportionality within some margin of error. If I understand you now, your argument concerns that margin of error; even with a sample of several hundred, there’s still a chance of some lack of proportionality–one or two extra bigmouths on one side of an issue, and that could distort the entire process. If so, I’d suggest you need to make the argument formally, with some statistics to show that the variance problem is as real as you think it is.

    One more thing. Assuming I’m following your argument now, this isn’t really a concern about political equality, contrary to what you said. It’s an argument about preserving the inference from the sample to the entire population.


  103. > Keith, every time I think I understand your position, something goes wrong.

    I know the feeling. After a while it seems that for Keith the conclusions are the only fixed point. The arguments for the conclusions are completely fluid.


  104. Peter, the argument has nothing to do with the margin of error, it is to do with the distinction between the individual and the aggregate (a gross distinction by any standard). The distortion of aggregate (statistical) representivity has nothing to do with the margin of error that is inherent in any sampling process, it’s to do with failing to view an aggregate for what it is (a statistical result), and focusing on the individual components that make up that statistic. The error involved in authorising members of a statistically-representative assembly to act in an individual manner (that is not open to mathematical representation) is exactly the same as taking an individual response (a completed questionnaire) from an opinion poll and claiming that it is macroscopically representative rather than being just a datum. An individual “descriptive representative” has exactly the same status as an individual completed questionaire, and the latter is of no intrinsic interest to the operators of proportionate opinion polls. If your argument for sortition is to establish descriptive (statistical) representation then you need to adopt the same mindset as the polling industry. Fishkin does this, but most deliberative democrats (including the majority of participants on this list) don’t. This is such a simple and obvious point that I’m surprised that it keeps being misunderstood and misrepresented.

    Regarding your competence point, I always seek to establish a clear distinction between epistemic and equality-based arguments for democracy. As you rightly pointed out in your paper at the first Paris meeting, the epistemic argument for deliberative democracy (diversity as an essential element in sound decision making) does not require proportionality. For this reason most deliberative democrats (Helene Landemore being a good example) aren’t really bothered about the latter. To my mind this is an argument for epistocracy and it may well be that a diverse group can make better decisions than Plato’s guardians. But my concern is for democratic equality and that does require proportionality, in so far as democracy generally cashes out in practice as majoritarianism. It may well be that the sort of fully-mandated assembly proposed by many on this list would come to better decisions than elite groups, but it would have to be subsumed under the category of oligarchy (rule by the few), hence the need for the restrictions that I have proposed in order to ensure it can be subsumed under the category of democracy (the rule of the many).

    >>”Assuming I’m following your argument now, this isn’t really a concern about political equality, contrary to what you said. It’s an argument about preserving the inference from the sample to the entire population.”

    In what sense are the two things not the same? Equality presupposes counting (each vote being equal), so if the sample is a microcosm of the entire population then counting the votes of the microcosm would be a proxy for the votes of the entire population — you just wouldn’t need quite so many vote-tellers or hanging chad machines. Acceptable deviation from the entire population (from a deliberative perspective) would be that voting in the microcosm would be based on balanced information; unacceptable deviation (from the egalitarian perspective) would be any factor that allowed any member of the microcosm more (or less) influence than her peers. My claim is, therefore, that preserving the inference from the sample is identical with the preservation of political equality. If not, then the argument for sortition would have to be on the basis of considerations other than descriptive representation. In your papers and recent book you have expressed scepticism with the latter, so I suppose I need to read your comments in that context. But I would still hope that you can understand my argument, even if you don’t agree with it.


  105. PS. I should restate that this argument is not my own, it is derived from Pitkin. Like most Austinian arguments it revolves around the way we use words — in this case the word “statistical”. I suppose I should be flattered when my detractors attribute the argument to my own eccentricities, but I would actually prefer it if correspondents did the necessary work and studied the book (the leading text on representation) for themselves. Martin Davis did this recently, and had the good grace to acknowledge that my interpretation of Pitkin’s arguments were accurate. Of course if you’re interests in sortition are other than political representation then it’s understandable.


  106. Keith: I do indeed find Pitkin’s arguments on representation compelling and your interpretation of them accurate with regards to statistical sampling. But whereas you employ Pitkin’s thesis as a pillar for your central premise of a mixed constitution within which an AC is a component, I take it as the underpinning of a model in which almost all political institutions are selected by sortition.

    Is it not the case that the writing is on the wall? The world will not go back to the way it was before the global financial crisis. Elected representative governments no longer meet the needs and aspirations of their societies or of mankind as a whole. Political parties are bankrupt institutions in every sense of the word. I don’t wish to sound overly melodramatic but the inability of elected representatives to get to grips with fundamental economic issues or reach any kind of consensus on measures related to climate change are surely early indications of a complete collapse in our systems of organisation. Has no one here noticed that unelected technocrats are now running two European countries on behalf of their creditors, rather than for the benefit of their people?

    I think time is of the essence in finding alternative systems for organising human affairs and it is incumbent upon those of us with an interest in doing so to get on with it.

    Contributors to this site have helped to shape my own ideas and you yourself have fundamentally changed my thinking on some issues, Pitkin’s arguments on representation being a good example. So I was profoundly disappointed by your assertion that the views expressed here should pander to the mainstream in public affairs for fear of being branded ‘fundamentalist’. Why?

    It sometimes seems that you’re prepared to use any argument, no matter how tenuous, to defend a position that doesn’t warrant defending. Is your thesis really that valuable to you? You seem to be so combative that, as Yoram has pointed out more than once, it requires too much effort to engage with you.

    I find joy in good ideas, no matter whether they’re mine, yours or anyone else’s. I’m delighted when I find myself capable of original thought but, fundamentally, I believe ideas are for sharing. They’re not weapons with which to bludgeon others into submission.


  107. Martin, I completely agree with you regarding the crisis that we are facing, but draw different conclusions for the following two reasons:

    1) It’s important not to replace one set of malfunctioning institutions with another set that are potentially even worse. Leaving aside the election/sortitition issue, the functional separation between policy inititiation, advocacy and decision-making strikes me as a valid and sensible one. If so then the problem is that all three powers are currently in the same hands. Given that you are persuaded by the argument on statistical representation, why would you wish to replace one oligarchical system with another? It strikes me as a better option (both epistemically and normatively) to open up the decision-making system to the public as a whole, and that requires a combination of election (or direct democracy) and sortition. One may find that distasteful but it’s nevertheless a natural consequence of acknowledging the nature of statistical representation.

    2) Regarding the need to pander to the mainstream in public affairs, I would like to think that we can come up with proposals that our fellow citizens find plausible and attractive. From my own non-scientific straw polls I have not been able to find a single person (outside this list) who is attracted to a sortition-only solution. Indeed most people initially find sortition to be a crazy idea but come round when you explain the trial court analogy in which the final decision is made by a randomly-selected jury. However they balk at the prospect of appointing the judge, prosecution and defence counsel in the same manner.

    As for my combative approach, I confess to a little frustration when Pitkin’s arguments are misunderstood. The argument is so simple that I wonder if it is a case of wilful misrepresentation. I try to be courteous in my responses, but it’s a challenge sometimes. As for being defensive over my thesis, I’ve changed my position dramatically over the last few years in response to counter-arguments and would be happy to do so again when I hear a convincing one. Until then I am content to remain in a minority of one (at least as far as this list is concerned).


  108. The problem I’m having with the argument, whether it be Pitkin’s or yours, can be stated as follows. You presumably do not feel that the democratic credentials of direct democracy are compromised by the fact that a direct democracy might have several loudmouths in it who exercise disproportionate influence. So why should the democratic credentials of a random sample by compromised if it has exactly the same proportion of loudmouths as the general population.

    At the risk of repeating an argument already made several times here–and I really don’t think you’ve responded to it properly–consider the following. There is a population of 300 people deliberating and making a decision. (You don’t like 300? Name whatever size you like. And structure the deliberation however you like.) There are a few people in the group who know more, and talk more, than others, but presumably not in a way that is obviously bad (i.e., nobody gets threatened physically, etc.). If I say to you, that’s a group engaging in direct democracy, you’d say, oh great, it satisfies political equality, it’s perfectly legitimate, etc. But if I say, that’s a random sample of people who are deliberating on behalf of a larger population, you’d say, oh that’s not legitimate, it doesn’t satisfy political equality, etc. And you’d say this even though you apparently concede that the proportion of people in the sample who say more, know more, etc. is exactly the same as in the general population.

    Sorry, but I still don’t get it. What’s the difference between the two?


  109. The argument on this forum is between two competing models of indirect democracy (elective and sortive), not between sortive and direct democracy. In the case of elective democracy everybody is equally free to choose their representative and can pick a loudmouth, a bozo, or whoever they wish. However in the case of sortive representation you have no freedom to choose your own representative, as your proxy is selected randomly. This contravenes political equality as those who are allotted a competent and persuasive representative are more equal (in Orwell’s sense) than those allotted a lazy bozo. But equality is not contravened if the function of the allotted rep. is only to vote, as all votes (loudmouths and bozos) carry the same weight, but political equality is breached as soon as allotted members are permitted an active function (introducing and arguing for/against legislative proposals).

    I agree that the difference between direct democracy and the AC with full powers is primarily one of scale and apologise if I have given a different impression. The problem only arises with large political communities, where representation is required — this indeed being the focus of Pitkin’s (and Manin’s) work, neither of whom are interested in direct democracy. Manin argues that sortition is impossible to implement as a democratic mechanism in communities where representation is required, whereas I have attempted to show that it is legitimate so long as its role is limited to functions that do not contravene poltiical equality. Given that a sortive model with active functions would (as you rightly point out) replicate the principal flaw of direct democracy (loudmouthing) there are good reasons for taking Madison’s arguments for representative government seriously, while at the same time seeking to counter the corrupting effect of relying solely on electoral representation. Once again I apologise if I have suggested that there was a fundamental difference (other than scale) between direct democracy and a sortive chamber with full powers.

    The topic of direct democracy (referendums, epetitions, initiatives etc) was only raised on account of your own arguments regarding the weakness of veto power — elected reps alone may not provide a suitable range of legislative alternatives. Loudmouths will ex hypothesi dominate the agenda-setting process, hence the need for a public votation (where political equality is reinstated) to select between the proposals put forward for consideration by the allotted chamber.


  110. I think the point is more basic than you suggest. The question is why should we think an AC is at all legitimate. You have given two answers to this question that are closely interlinked. First, the AC is legitimate because descriptively the sample is representative of the whole. Second, the AC is legitimate because it would do the same thing that the whole would do under the same circumstances. You seem to think that the two claims are equivalent. But according to you, if there are inequalities within the AC, then something goes horribly wrong. It cannot be that the AC is no longer descriptively representative of the whole; we both agree that there will be the same percentage of loudmouths, etc. And it appears that you believe that the inference still holds–that the loudmouths will lead the AC under deliberation to make the same decision as the population would if it deliberated with the loudmouths present. And yet you still think that there’s a problem.

    I stress again–nothing in the story I just told mentions political equality. If you introduce that demand, then you’re introducing a separate demand, and that demand is doing the work in torpedoing deliberation in the AC. Finally, note that here political equality in a strong sense works AGAINST descriptive representation. Your understanding of political equality would require an AC in which nobody was more outspoken, influential, smarter, etc. than anyone else. And in that respect, you very much DON’T want the AC to be descriptively representative of the whole. So your argument for political equality is doing very real work, and I cannot see any connection between it and the “AC will make the same decision as the whole would” argument.


  111. Loudmouthing (along with power, wealth and status differentials) contravenes political equality right across the board — macrocosm or microcosm, ancient or modern, direct or representative. This is because equality is a mathematical concept, which requires the counting of discrete (digital) units. You can count votes, but there is no equivalent process for the measurement of (analogue) speech acts or other forms of influence. All votes in the macrocosm are of equal weight and the same would be the case in the microcosm so I’m puzzled by your claim that descriptive representation has nothing to do with political equality (as the microcosm is simply a portrait, in miniature, of the whole electorate).

    But there is no point counting votes unless there is something to vote about, so someone has to provide an agenda and as soon as this happens substantive political inequality is introduced — once again this is the case across the board — macrocosm or microcosm, ancient or modern, direct or representative — on account of the ubiquity of Manin’s principle of distinction, which would apply as much to debate within an AC as it does to the electoral hustings.

    Democratic norms require us to do everything possible to minimise the above inequalities, so elective democracy presupposes free choice and we rightly dismiss single-party elections as fraudulent. If the decision function were in the hands of a lower chamber constituted by sortition then electoral choice needs to be supplemented by direct-democratic initiatives in order not to succumb to the weak form of power that you argued results from the veto. And every step needs to be taken to ensure that inequalities are further minimised by capping funding, redistributing media power and ensuring the right of all citizens to actively participate by proposing initiatives themselves. These measures will never overcome inequality, but it’s hard to see what else one can do.

    Returning to your first question, the AC is legitimate only in so far as it mimics the political equality that legitimates the wider democratic process and, in both cases, the equality is realized by counting votes. We are not interested in descriptive accuracy per se (the number, say, of people with red hair), only the way that people might vote and there are good reasons to believe that the voting behaviour of a random sample would be a good proxy for the voting behaviour of the whole population (the only difference factor being the presence or absence of well-balanced information).

    So, in this respect (counting votes), I see no difference between accurate descriptive representation and the preservation of political equality.


  112. I may have located the crux of Keith’s problem with individual members of an AC being initiators. It may have to do with the concept of individual citizens having representation…
    Keith wrote:
    “In the case of elective democracy everybody is equally free to choose their representative and can pick a loudmouth, a bozo, or whoever they wish. However in the case of sortive representation you have no freedom to choose your own representative, as your proxy is selected randomly.”

    Setting aside, for the moment, that at least 49% (or more, depending of voting and nominating method) of the voters may NOT be able to “choose their representative,” in an electoral system…It is crucial to understand that unlike an elective chamber, NOBODY has a “proxy” representative in an allotted chamber. AC members have no “constituency” to be accountable to…and no citizen can point to “their” representative. Each AC member simply acts as she thinks best. Period. The NET effect (we surmise) is to replicate the decision making that would occur if EVERY citizen could and did participate.

    However, I share the concern about assuring that any citizen who wishes can have input into societies decision making. Elections provide an illusion of this input. I think that a sortition democracy should provide AT LEAST equal opportunity for individual input .But rather than giving input through the doubtful tool of voting (or putting up a sign with a candidate’s name on it in your lawn)… I imagine an ACTIVE role through the institutionalization randomly selected Interest Panels of volunteers, who would be the primary initiators of legislation that an AC or sortition Review Panel would take as their raw material.


  113. Terry, I agree of course that no individual has their own proxy representative, otherwise the allotted assembly would be the same size as the total population. The AC only looks like the population collectively (as if at a distance, where it cannot be differentiated into its component individuals).

    >>”Each AC member simply acts as she thinks best. Period. The NET effect (we surmise) is to replicate the decision making that would occur if EVERY citizen could and did participate”

    Unfortunately your second sentence doesn’t follow from the first. I agree that the decision making (judgment) of the small group would mirror the judgment of the whole (the only differential being the presence/absence of balanced information), as judgments (votes) can be aggregated. But your first sentence is about each member acting as they see fit. There is no reason to believe that the individual speech acts of members of a small group would mirror the speech acts of the whole community, and, unlike votes, speech acts cannot be aggregated. Yet this is the claim that you would need to substantiate if you wish to argue that an AC with an active role performed a democratic function. Such a body would be oligarchic, the only difference from most oligarchies being that it was not selected on the basis of money or power. By all means continue to advocate an AC on that basis (I suspect that this would be Yoram’s preference), but we are moving a long way from democracy (the rule of the many).


  114. Keith: Clearly I wouldn’t want to replace one oligarchical system with another. Briefly and simply so that I don’t bore everyone with detail (I’m happy to go over individual elements at length separately):

    I imagine a network of hundreds of community ACs (equivalent to community councils), each of around 30 members where participation is compulsory. For the reasons we’re already worked over at length, members of the AC would not have an active role but would simply be required to provide yes/no decisions. Community ACs would be organised into sets to make decisions on local matters (equivalent to town councils), regional matters (equivalent to county councils, city councils or unitary authorities) or national matters (equivalent to parliament) where the decisions of the individual community ACs within each set are aggregated.

    Mirroring the network of community ACs, would be a network of community policy forums, each of around 30 members, where selection is by sortition but participation is voluntary. Clearly it is desirable that these institutions be deliberative so participants would have an active role. Community AFs would be organised into sets mirroring those of the ACs.

    A wide range of channels would be able to feed into the AFs enabling individuals and interest groups to make representations without being able to exert undue influence or coercion.

    The size of the sets involved in specific policy-making/decision-making would vary depending upon those affected by the outcome. Policy on local allotments or village halls could be dealt with by one AF/AC, for example, regional transport or police authority business by perhaps a few tens of AFs/ACs, whilst a decision to intervene in an international conflict might need the approval of every AC.

    Acting between the AFs, the ACs and the civil service would the ‘Advocacy’, a network of pools of lawyers who would be allocated at random the task of making the case either for or against a matter at a randomly allocated community AC.

    Executives of technocrats, selected by and answerable to the respective sets of ACs would be responsible for implementing policy on a local, regional, national or international basis, much as already happens. (I imagine a rolling approval rating system.)

    I think this model might resolve Keith’s issues to do with representation, Peter’s issues to do with legitimacy, Terry’s issues to do with accountability and Yoram’s issues to do with elites and participation. I think it is reasonably true to John’s original conception and eminently practicable in implementation.

    No doubt you’ll all let me know what you think.


  115. The discussion evolves a bit in circles, I fear.

    So I won’t repeat myself. I’m only a bit sad that for the sake os some missconceived conceptual purity, all key powers -safe yes/no decision- are left in Keith’s approach in the hands of a truly self-perpetuating oligarchy (which we well know) instead of sharing those powers with a sample of citizens (of whom we could say “those are people like we are”) who -with the right institutional and procedural setting- would very much not only vote but also propose and discuss like the population at large.

    BTW Keith’s model permanently ignores (apart from refering to Fishkin which is only a referral but not a rebuttal) the argument made that the essential difference between the sample and the population at large kicks in as soon as the sample is made and convened and, therefore, if he were consistent with his “pure descriptive” argument, that sample would not be entitled to vote on behalf of the population at large as it is already essentially different from it (in power, information, role, status, etc) – hence, in his view -if consistent- it could only be termed as a “rule of the few”…


  116. OK, let’s take a direct democracy again. And let’s assume for the sake of argument that the agenda and information comes from God almighty (or James Fishkin, if you prefer). Then a direct democracy which engages in NO deliberation before it votes will perfectly reflect political equality in its decisions. Is this right? Once deliberation is introduced, it violates this equality, and so would have to be justified on other grounds. (Perhaps there is a tradeoff between political equality and some other value, like decision quality. But that, of course, presupposes a standard of decision quality independent of the political process itself.)

    If so, then (again, setting aside the question of agenda and information control) what you want is an AC that does not deliberate because it will do the same thing as a direct democracy that does not deliberate. Is that right?

    Now the question is–what about agenda and information? Well, I gather you think it is literally impossible for everyone to have the same influence on the agenda, the information supply, or the deliberation. But if political equality were really all that was at stake, then the solution would be obvious–random selection. Select the agenda-setters randomly, select the information suppliers randomly, select the speechmakers randomly. It’s commonly accepted (and I defend the idea at length in my book) that when people are all supposed to receive something good, but they cannot all do so, then the way to treat them equally is to use a fair lottery. So why not randomly select an AC, then randomly select who in the AC will form the agenda, etc.? Again, IF political equality were all that mattered–and I’m not saying it does for you, or should do for you–then this would seem to be the obvious solution.

    Finally, I don’t understand the sentence, “Democratic norms require us to do everything possible to minimise the above inequalities, so elective democracy presupposes free choice…” I simply don’t see how the second part follows from the first. How does electoral democracy mitigate or minimise these inequalities? The issue of choice seems orthogonal to the issue of political equality to me. (Obviously, there is a sense in which everyone is equal in casting votes for representative, but a) it’s not the choice part that’s doing the work here, and b) it just shifts all the issues of political inequality back one stage.)


  117. In addition, seeing that the “thought experiment” likening the AC with an assembly of the population at large seems not to be sufficiently compellin to Keith, what about the Jury and Jurors analogy?

    The Jury is selected as a fair cross-section of the population, but it does not only vote in aggregate. Its Jurors are allowed to speak, to propose, to discuss specific pieces of evidence more than others – and loudmouths are inevitably present. Isn’t this democratic?

    I concede that some “loudmouths” probably attain excesive inflluence in Juries – more so in a small group of 12 – but people see Juries nonetheless as democratically legitimate. And in a wider AC this can be prevented by rotation in institutional strongholds (presidency of the AC, presidency and speaker status in commissions, time limits for speaking, separation of proposal and advocacy…) – we just have to look at the Athenian Boule – if we couple that with a framework where initiative is shared by three branches (government, AC and elective chamber) and where the people at large can be called upon to decide gridlocks by referendum, I sincerely don’t see why Keith holds on to his proposal where the key power would remain in the hands of the Executive (elective?) and the elective chamber as we all know how manipulable direct-democratic devices are…

    (well, I’m repeating my self, I confess. apologies for that).


  118. Thanks, Peter, that was a really good point!


  119. Response to Martin, Jorge and Peter

    Martin, unfortunately increasing the number of ACs will only overcome the conceptual problem once you move into figures when rotation takes over from representation as the legitimising principle. Even if we discount the practical difficulties involved, this would be a long time after the problem of rational ignorace has been reintroduced.

    Jorge 1, I share your sadness, but don’t want to indulge in wishful thinking — in the case of statistical representation “those are people like we are” is true; “she is like me” is false; and it takes a good deal of conceptual obfuscation to obscure the veracity of this simple disctinction. If the AC was appointed on a case-by-case basis and if it only deliberated in the latinate sense then the only differences between the sample and the whole would be information and power. Although the latter does follow Acton’s dictum, it would take more than a few days of political jury service for the corruption to occur.

    Peter, there is one element missing from your argument, namely representation (this is understandable, as in your book you don’t consider sortition to be a representative mechanism). In a direct democracy everyone is free to deliberate or not as they choose, not so in a representative democracy, hence my earlier point that I was seeeking to contrast sortition and election, not sortition and direct democracy. Your suggestion to select agenda-setters and advocates randomly complies with your own “lottery principle” (fairness, arationality etc) but would not function as a representative mechanism (unless there is some aspect of representation that Pitkin and Manin have overlooked in their studies). Who would your randomly-selected agenda-setter represent apart from herself, and if so how?

    Yves was right to point out in Paris that there are in fact FIVE lottery principles and it’s important not to think that they can all be subsumed under a single principle. In large political communities political equality presupposes representation, so thought experiments about direct democracy or appointing individual office holders by lot have no bearing. If you could rephrase your argument to include the necessity for representation then I can respond, but I don’t see how that’s possible when considering a totally different function of sortition (fairness) — it’s no more relevant than Yves’ other categories (divination, distribution etc.)

    As for the relationship of political equality and free choice, my argument (summarised from earlier) is:

    1) the only political act that is perfectly equal is voting, however
    2) people need something to vote about, so we need agenda-setters
    3) this immediately introduces political equalities, but
    4) exercising choice via the vote reduces this inequality.

    This is not to suggest that attempts to equalise financial, status and power resources do not impact on (3), my claim is simply that all votes count the same, but all voices don’t. Shifting the unequal voices from the hustings to the allotted chamber doesn’t impact on the inequality, it just moves it around.

    Jorge 2, as we’ve discussed before, jury-room deliberation is only on account of the need for unanimity and it’s widely accepted that it distorts the working of the Condorcet theorem.


  120. Keith, I fear that you are using the word “representation” as a kind of talisman here. I don’t think it has quite the power you attribute to it. As you know from reading Pitkin, there is not one form of “representation,” but several. There is the kind that happens when x resembles y, there is the kind that happens when x authorizes y to act on her behalf, etc. And I do indeed discuss one of these types–descriptive representation–in my book. Most of my questions, I think, are poking at the question of just what you want a representative body to do. What makes a representative group representative for you, and why? I may not be using the word “representation” as much as you have, but I like to think I’m trying to unravel just what makes an AC “representative” in the ways that matter. If you think I’m missing a critical element of it, then fine, please tell me what I’m missing. But simply evoking a complex political concept like “representation” is not, I must confess, leaving me any the wiser.


  121. I assume that point #3 in Keith’s last post should have the word “inequalies” instead of “equalities.” If Keith confirms this, Yoram, is there a way to edit previous posts to clean up significant typos like that?

    Martin: Your proposal is almost identical to the one I have been describing. But, the one piece I don’t like is the idea of contracted advocates…I agree with Keith, that advocates should be those with sincere motivations, to avoid intentionally weak presentations, and openness to bribery.


  122. Thanks Peter, my principal concern is with what forms of political representation different institutions can and cannot offer; what we want them to do being something of an additional luxury. So let’s look at the two relevant institutions, elective and statistical (I prefer this word to any of its slippery synonyms like mirror or descriptive).

    Elective representation
    I don’t have Pitkin and Manin to hand (I’m currently nearing the end of a 12 hour shift here at the print works), so will have to do this from memory. Elective representation originated as a form of ascription (the need for king to bind his subjects to accepting his decrees via the assent of their representatives), but over time this developed from a passive into an active remonstrative function (OK, we’ll accept the edict, but only if you settle the following grievances). As the power of parliament increased (along with the extension of the franchise) this developed into a means for expressing a preference — one could choose from a variety of representatives touting their wares and could replace the snake-oil salesmen at the end of their shift. As we all know, it doesn’t work out quite like that in practice, nevertheless with proportional representation, tight constraints on election budgets, media plurality and a leavening of direct-democratic initiatives, elections do provide a reasonable means for most voters in large, distributed polities to express their raw preferences. And it’s difficult to imagine what an alternative method would look like.

    However the intrinsic downside (over and above the empirical problems just acknowledged) is that a) there is no incentive for voters to educate themselves, as their single vote is of no significance and b) politicians will inevitably be a class apart, owing to the principle of distinction inherent in the electoral process, hence the need for:

    Statistical representation
    Sortition will produce a mini-public that is much more like the electorate itself — “people like us”, as Jorge puts it and, assuming it’s only a few hundred strong, members will be disposed to listen to the arguments before casting their votes, so statistical representation complements electoral representation by overcoming the two intrinsic problems noted above. Statistical representation, however, only applies at the aggregate level (“people like us”, rather than “someone like me”) therefore elective representation is required as well if all citizens are not going to be denied an expression of raw preferences. Manin also claims that the ascriptive aspect of representation (consent) cannot be supplied by sortition, although I think the case for electoral consent is, at best, tacit.

    So it’s not a question of what you want from representation, it’s more a case of what each procedure (election or sortition) is capable of offering. If you want active representation of interests and ascriptive representation, then I don’t see what choice there is other than election and/or direct- democratic initiatives; if you want legislators to look more like the rest of us and if you want them to listen to and weigh up the arguments before voting then you need statistical representation. I think democracy requires all of these things, hence the need for a sensible balance of elective and sortive institutions. But given that both systems provide different kinds of representation if you then say, OK let’s create a balance by having (say) an elective senate and a sortive lower house, then the functions of the two bodies need to be differentiated accordingly unlike (as in the proposals by Callenbach and Philips, O’Leary, Gat etc.) each chamber running in parallel, but with similar policy-initiative and judgment powers. Pluralism isn’t sufficient, you also need separation of powers, according to what each institution is capable of offering.

    Note: Pitkin’s other distinction — between trustee and delegate represenatation — is orthogonal as it’s a subcategory of elective representation. Helene has attempted to argue that it’s part of the same axis, but she would need to present that case herself.

    Terry, thanks for pointing out my typo. I’ve been up since 5.00 running three print machines all day and my focus is not too good!


  123. Two points:

    – separation of powers and checks and balances entail an overlap in functions most of the times. Otherwise you run the risk of having separate but very unequal powers, like in your proposal, Keith.

    – >”“people like us”, rather than “someone like me”” – in fact I believe that in a “people like us”-AC there will be always “someone like me” – perhaps my (not so individualistic) belief is that I am (we are) not so original as an individual(s) and, hence, if a sufficient number is there, there will always be somebody or a collection of somebodies in an AC who will reflect my (our) interests, points of views and will act like I would do – much more, in any case, than in an elective chamber where they are definetly not “like me” and have and pursue “raw preferences” which reflect their own needs, but not the population’s.


  124. Obviously members of an AC will, on average, be “more like me” than in an elective assembly — that’s an inevitable consequence of the principle of distinction. But being and doing are two very different things that are constantly being conflated on this forum. C&P are (unintentionally) very good on this point when they argue that beach bum- and dropout AC members would honour their descriptive duties well by not turning up at the assembly. True, but how about if you as a passionate idler, were relying on the no-shows to actively pursue the interests of lazy people like yourself? (Peter points out this anomaly in his introduction to the new edition of A Citizen Legislature.) Beach bums and dropouts will need someone to actively champion their interests (such as Tom Hodgkinson, the editor of the Idler, who Yoram cited the other day). I’m sure Tom is a very active and articulate advocate of the interests of the idle.


  125. I could live with those “anomalies” and would definitely prefer that to an hegemonic elective power.

    “Doing” includes voting I’m afraid – you are consistently inconsistent about that fact which destroys your alleged conceptual purity.

    And, I accept that those “people like me” and that or those “somebody/ies like me” could well represent me not only in being (which is in my view almost impossible), but what’s more important, in doing – not only voting, but discussing and proposing.


  126. > Beach bums and dropouts will need someone to actively champion their interests (such as Tom Hodgkinson, the editor of the Idler, who Yoram cited the other day). I’m sure Tom is a very active and articulate advocate of the interests of the idle.

    This is silly. The possibility that those chronic idlers would get a Tom Hodgkinson elected is laughably remote. In fact, it is oxymoronic. Idlers wouldn’t even show up to vote, much less organize a campaign. (After all, many non-idlers don’t bother showing up to vote because the cost/benefit ratio is so high.)

    On the other hand, if one of those idlers finds herself in the AC, she can simply and effortlessly appoint Hodgkinson as her advisor, have him take care of business in her name, while she idles about and only show up for votes in the chamber.


  127. A quibble over a word is not at all the same thing as a conceptual disagreement. The reason that Pitkin (not me) singles out voting as the *only* action that members of a descriptively-representative assembly can do without compromising democratic equality is because all votes are equal and can be aggregated mathematically. The same cannot be said of speech acts. We really must try and understand each others’ arguments, not just score points. As for your final claim, I’m afraid that’s just a statement of faith, not an argument.

    I do find it ironic that members of a small discursive forum who share similar political prioritiess (and yet fail to listen to each other’s arguments properly), still believe that unstructured discursive exchange is the best way to bring harmony to the bear pit that we call politics. This strikes me as the triumph of hope over experience.


  128. It’s amusing to see how the noble republican ideals soon melt away as the newly enfranchised mimic the ways of their indolent predecessors. Plus ca change. (not intended to be a serious point, it’s just time to go home after a long shift).


  129. Well “quibbles” over words are certainly not my speciality as a non native English speaker and writer – please do not use this disauthorisation tricks with me…

    We already discussed at length on Pikin’s “sacred” book – and despite being it “her words” they are nonetheless inaccurate when applied to an AC at least – “descriptiveness” ends (and therefore YOUR alleged source of legitimacy for an AC) when the sample gains powers, information and roles which makes it essentially different from the people at large. I.e. to follow your argument if you were really consistent: if the sample is made different from the population at large (as it is rendered in an AC), the sample is no longer an exact picture of the population, it is no longer descriptively or statistically representative and, therefore, should not be entitled to DO anything on behalf of the population at large.

    The descriptiveness or “statistical” sort of representation -as described by Pitkin- was I believe tailored to reflect the practice of mere polls, and logically cannot reflect something which was in nobodies’ minds in the 60s (an allotted chamber). I haven’t heard anything rebutting that apart from vague references to Pitkin’s sacred book or to Fishkin’s model where people have by no means the same role and power an AC would have.

    >”The same cannot be said of speech acts” – it never can, be it in an elected chamber (which is oligarchic by definition), an AC or an assembly of the people at large – only institutional and procedural means and rules can prevent that from happen and this applies to all cases alike where democratic equality is at stake. This danger would happen in the same fashion if loudmouths dominate an elected chamber (which mainly reflects special interests and builds a self-perpetuating elite – providing many times people who only have gathered 25% or 30% of the electorate’s vote with the only representation in their district or with an absolute majority -depending on the electoral system at work in each case) and in an AC.

    So, where is the democratic equality of an elective chamber? in the possibility of ‘selecting’ the lesser evil among two pre-established options no ordinary folks can influence and which depend on special interests for their survival and which function as an elite? There is little equality in fact and less if we start to look at all the gerrymandering and all inequalities in vote which in fact exist in electoral systems -not as pathologies, but as a consequence of the electoral principle.

    So we are again at the starting point it seems: Keith trying to construct a new political framework with a toolbox inherited from electoral times (like producing automobiles with coal, horses and wood) and the rest of us trying to make sense of it, I fear.

    >”This strikes me as the triumph of hope over experience.” that would certainly be the case if you trust a yes/no AC to sanitise an electoral-dominated system


  130. Keith: “…unfortunately increasing the number of ACs will only overcome the conceptual problem once you move into figures when rotation takes over from representation as the legitimising principle. Even if we discount the practical difficulties involved, this would be a long time after the problem of rational ignorance has been reintroduced.” > You what? I don’t understand.

    Terry: Cool! It can only be a good thing if we’re both describing a similar system. Therefore, since we’re agreed that the community ACs are inactive in the way that Keith describes, shouldn’t the arguments for or against a particular outcome be presented within a consistant legal framework to ensure legitimacy?

    If the advocates are allocated their briefs at random (either to argue ‘for’ or ‘against’), and the location of the AC at which they are to make their case is also selected at random, how would anyone be able to bribe them since their identities would not be known in advance? On important national issues, multiples of advocates would be making the same case to multiples of community ACs, thus removing the possibility that a single powerful orator could skew the result simply by force of personality (although I accept that probability might conspire to allocate all of the advocates who are powerful orators to the same side in the same case at some stage – hopefully that won’t be on a bill to bring back hanging!)


  131. I and others on this list have frequently compared the extremely faulty existing electoral systems, to some ideal sortition model. In trying to counter Keith’s (and others’) defense of a continued role for elective representation, we should imagine an ideal electoral system (or imagine a faulty sortition system)…Apples to apples.

    A point about beach bums and idlers…
    What about the concept of allowing randomly selected AC members to appoint a proxy replacement officially? This would move away from perfect statistical representation, but may be closer than simply letting them opt out.

    Indeed, there are some public choice theorists who advocate switching all elections to a cascade of delegable proxies…so that every citizen has a chain of appointments to somebody in parliament. This wouldn’t solve our concern about elite rule and the principle of distinction…but might be the best that an electoral system could get.


  132. Jorge, Martin, Terry

    Jorge, the whole point of judgment by a microcosm (as opposed to public referendum) is in order to achieve an informed decision. Of course information makes a difference, but the microcosm is, in Fishkin’s words a proxy for “what everybody would think if everybody deliberated” (this presupposes Fishkin’s minimal conception of deliberation as the weighing of competing arguments). The microcosm will only differ in any other respect if membership is long-term, hence the compelling argument for a fresh sortition for every legislative event. Of course this is not possible if you want the AC to perform an active role, as experience will be necessary. But that’s your problem, not mine. As far as regards the DP model, the only deviation between the microcosm and the whole population is information: that is the reason for appointing it in the first place and there are no accidental by-products. So I’m puzzled as to why you continue to claim that this is a problem (I’ve answered this very issue several times before).

    Given that you feel that the concept of statistical representation is outmoded, what model of representation would you invoke for an AC? Or are you seeking to argue a non-representative role for sortition? (as with the Stone-Dowlen “lottery thesis”). I think we all need to be very upfront with exactly what it is we want to achieve with sortition. Your posts to date have argued for an AC comprised of “folk like us”, and that would naturally presuppose statistical representation. But perhaps you have something else in mind?

    >So, where is the democratic equality of an elective chamber?

    I’ve never argued for an elective chamber. My model is unicameral, with elected (or direct-democratically mandated) advocates introducing policies for judgment by an allotted assembly. Advocacy and judgment are separate functions within a single chamber. As for the democratic equality of the elective element, this is provided via the counting of votes (which are of equal weight); those advocates who get most of the votes get the right to introduce their policy proposals. I have consistently argued that equality in active political functions is instantiated via voting and this is the case in elections, referendums, votations and the judgment of a legislative assembly. Equality in passive (descriptive) functions is via sortition, but the mandate established thereby only applies to aggregate (statistical) functions. If you wish to legitimise the speech acts of AC members (from a democratic perspective) then you will need to propose something other than the “outmoded” concept of statistical representation.

    I’m well aware of the empirical pathologies of elective democracy but that’s no reason for introducing institutions that are known to be illegitimate at the design stage. Although it’s hard to rule out unanticipated consequences (for all the reasons that Popper has outlined) there’s no reason to introduce problems that are an inevitable consequence of wrong-headed institutional design.

    Martin, rational ignorance is the product of one’s own vote being diluted by large numbers of competing votes. As I understand your proposal, there would need to be a large number of sortive bodies, arranged in a hierarchical structure, especially if you are trying to get closer to the rotation case for sortition. This just reintroduces rational ignorance by the back door — why take the trouble to study the issues if the verdict of your AC is simply going to be overturned higher up the food chain?

    Terry, nice point on apples and pairs. To that might we also add the distinction between empirical and conceptual issues. The former category is primarily unanticipated consequences (eg the lack of foresight of the US founders for developments in inequalities and developments in communications technologies); in the latter we have the pre-empirical understanding of what a particular mechanism can and cannot do (you can work that out in before the experiment — on the back of an envelope or in conversation on Equality by Lot). Or perhaps that’s just rephrasing the ideal/actual distinction.

    The trouble with beach-bums choosing their own proxies is that this opens the floodgates for corruption. Lobbyists won’t have to bother with the expensive and time-consuming task of influencing people, they can just buy their seats. It would be like putting the clock back to the time of Old Sarum and the other rotten boroughs. Whilst the beach-bum might choose to give his seat to the editor of The Idler, he’s just as likely to sell it to the highest bidder (the same argument applies if he retains the seat, but just appoints an “advisor”). Once again this will all lead to a greater reliance on the police in order to protect the integrity of the system.


  133. Thanks, Keith

    >”As far as regards the DP model, the only deviation between the microcosm and the whole population is information” – do you claim that your AC would be the same as Fishkin’s DP? If yes, the problem is that Fishkin’s DP is a sample without power – your AC has, and this makes it further different from the population at large.

    >”what model of representation would you invoke for an AC?” –> certainly I’m thinking about representation as “making sth present which isn’t”. Due to scale we need a smaller group – sortition serves as a means to produce a sample which is like the population at large and whose members are like examples of citizens in the population at large. People could say “they are folks like me” and “there are individuals like me in there” and “they act individually/collectively” as I/we would do. Their mandate is to express their views as citizens – providing the population at large with a operational sample of individuals which would be a proxy of how the people at large would think and decide if there were not a scale problem. A fair cross-section of the population. My point is that we can entrust people like us to do certain things, like propose, discuss and decide (within a checks and balances system). How to call it – I don’t know – we already had this discussion before. Representation by one elected person (president) seems to be accetable as representation, by an elective body is too, by judges, jurors and other institutions is also accepted it seems. Hence, I don’t see why people would not accept “representation” by “their likes”. As to democratic equality I think that if elective institutions can be acceptable, allotted will be so so much more.

    My view is that endowing your AC in your model with decision powers breaks the purely ‘statistical’ link/source of legitimacy as you see it – using your words: voting is not “passive”, voting and exercising power is “active” – therefore I think that in the end your AC has not so different conceptual problems than an AC with initiative.

    As to the possible problems of loudmouths -as danger to democratic equality: we have discussed this a lot of times and there are many ways to minimise this danger in an AC.

    >”As for the democratic equality of the elective element, this is provided via the counting of votes”: you know how weak that democratic element is and how it is overcome by an elitist working of this system – and it’s not a pathology it’s part of the essence of election


  134. Jorge, Yes I do claim that the AC would operate in a similar manner to the DP, but I’m proposing that the AC would have final decision power, rather than (as in the DP) only advisory power. It’s true that possessing power would make it different from the population at large, but this does not affect its representativity. The claim is simply that the whole population would (given information) decide in the same way. The only *relevant* difference between the micro and the macro is information — the power would only be possessed for a few days so significant corruption of judgment would be unlikely to occur.

    >Their mandate is to express their views as citizens – providing the population at large with a operational sample of individuals which would be a proxy of how the people at large would think and decide if there were not a scale problem.

    I’m sorry to keep repeating this, but in what way would this be different to allowing an individual participant in a statistical opinion poll a platform to grandstand their views? As you well know the significance of an opinion poll is statistical, an individual respondent having the status of mere datum. If you worked for a polling organisation and pursued this policy you would be sacked. In your comment piece you continually elide the individual and the collective, but this distinction is absolutely crucial to the case for statistical representation.

    And if not statistical, then what? “Making present what is absent” is merely the starting point for Pitkin’s work, she then goes on to split that down into a variety of different types of representation. So if not statistical then what? I really would appreciate an answer to this (that’s why I’m repeating the question). In the case of the elected president, the representation would be a combination of ascription and the active representation of interests and, as in the case of Obama, there is sometimes an added frisson of statistical representation. So that’s how it is with elected politicians, judges are not normally viewed as a representative function.

    As for jurors, their mandate is (weakly) descriptive as their role is to decide the outcome of a legal case, so they would have a similar function to an AC with decision powers. But how would you feel if individual members of the jury decided to act as advocates for either the defence or the prosecution, or even chose to bring their own charges against the defendant? That would appear to be the right that you are claiming for an AC.

    It’s a little pedantic to refer to voting as an “active” function, especially if you only have to lift a finger to press a button. By “active” I’m referring to speech acts in the Austinian sense that pervades Pitkin’s thesis — ie how we do things with words. And the trouble is we can’t measure such analogue processes. I really don’t see how you can substantiate equality without counting things — surely that’s what the word means? I appreciate that sounds a bit positivist, but equality is a mathematical concept, and mathematics requires units to aggregate (fingers, votes, bits, widgets or whatever). OK you can aggregate money, postal codes, university degrees, occupational class, but there’s no *direct* connection between them and political behaviour. You can’t count speech acts (despite the slightly deranged efforts since the empirical turn in deliberative democracy), so all that’s left is votes.


  135. >”It’s true that possessing power would make it different from the population at large, but this does not affect its representativity” – that’s as big a leap as allowing people in the AC proposing things. Power and even more son decision-power would make that bunch of people very different from the population at large. Keep ignoring it – but it goes to the heart of your supposed consistency.

    >”It’s a little pedantic to refer to voting as an “active” function, especially if you only have to lift a finger to press a button.” It’s not pedantic – in fact it’s really a joke to deny that deciding is not acting – especially if political final decisions are at stake

    >”in what way would this be different to allowing an individual participant in a statistical opinion poll a platform to grandstand their views?” An Ac – including yours- is lightyears from a poll. Anyway, in polls people are not only asked yes/no questions, but also are asked to give their own impusive input on issues (e.g. “what are your main political concerns?”)

    And nobodie’s going to “grandstand” anything: please read again all insitutional and procedural means already discussed above.

    >”In the case of the elected president, the representation would be a combination of ascription and the active representation of interests and, as in the case of Obama, there is sometimes an added frisson of statistical representation” –> hodgepodge is allowed, unless you don’t like it, huh?

    >”So if not statistical then what? I really would appreciate an answer to this (that’s why I’m repeating the question).” I already explained – sorry it doesn’t feed your Siecle XX toolbox


  136. “Power and even more son decision-power would make that bunch of people very different from the population at large.”

    The issue is whether or not it would make a difference if one sample was replaced with another. With aggregate functions (ie voting), statistical theory claims that it would make very little difference. Not so if the members of the AC had active powers. For an example of the latter, take two hectoring and highly-opinionated individuals (for example, Yoram and myself). Are you suggesting that the AC would be exactly the same if we were interchanged? If, however, our only role was to vote, then one could safely predict the same outcome in either case, as each of us would only have one vote and that would not make much difference. So I’m not ignoring anything, merely distinguishing between relevant and irrelevant differences (from the outcome perspective).

    As for the distinction between voting and speech acts, the relevant factor is the “illocutionary force” (persuasive power) of the latter. This varies hugely between individuals and cannot be measured or aggregated. So long as votes are secret, they contain no illocutionary force and are therefore not considered as speech acts. As I’ve said before, Austinian philosophy is concerned with speech acts, as opposed to mechanical motion, that’s why I referred to your insistence that pressing a button is an act as “pedantic”.

    Fishkin refers to his process as “deliberative polling”, all I am saying is that the result should be mandatory rather than advisory, so why do you describe my model of an AC as “light years” away from deliberative polling? Of course it’s very different if your AC performs active roles, but that’s your problem, not mine. All I argue for is Fishkin with teeth.

    Yes elective representation is a hodgepodge. For example the tallest candidate usually wins presidential elections (at least in the US). But this can be explained by the principle of distinction — although it’s a complete nonsense that there is any connection between stature and political competence, nevertheless voters choose to make that connection and that is their prerogative. Some voters may choose people who they feel resemble them (a significant factor in the Obama example) and there again some may choose the opposite. Voters may choose a politician because they feel she represents their interests or because they trust her judgment. And citizens are also bound by the outcome of the poll to put up with the policies of the government, so elective representation is a hodgepodge of ascriptive, active, trustee, delegate and descriptive representation.

    But none of these factors (other than the resemblance factor) come into play with statistical representation (although, with time, the ascriptive function might evolve). Once you provide an analytic explanation of what form of hodgepodge representation is afforded by an AC with active functions we can have a sensible conversation about it — until that point the case must be considered unproven. I’m happy to move on from an older paradigm but you need to open up your toolbox first. Unfortunately people on this forum like to see sortition as a black box and are loath to give it the same analytic treatment as elective representation. By all means write the new sacred text on representation for the 21st century, but this will require a lot of painstaking analytical work.


  137. Keith: I understand what rational ignorance is but I don’t understand why you have introduced it as an objection to the system I described.

    I didn’t say the ACs would be arranged in a hierarchical structure, you assumed that for some reason. I explained that the ACs would be arranged into sets and I explained the basis of the sets. It follows that the structure is flat. It doesn’t need to be hierarchical. No AC ranks higher than any other. The members would have plenty of time to consider the information presented by opposing advocates before voting on a yes/no basis. Rational ignorance shouldn’t be an issue.

    There are around 900 courts of different types in England and Wales. There are more than 8,500 elected community and parish councils plus town councils, county councils, unitary authorities and metropolitan councils, arranged in a hierarchical structure. So why would it be impractical to organise hundreds of ACs and allotted policy forums?

    If you’re going to change the system, you have to consider the whole system not just a single part.


  138. >”By all means write the new sacred text on representation for the 21st century, but this will require a lot of painstaking analytical work.” – I’m afraid I need your help on that. It would also benefit your model – having “teeth” means that you can bite, and someone who is able to bite behaves differently from someone who isn’t ;-)

    And you perfectly know that we have discussed this: “representation” of an AC would have different angles… statistical, descriptive, homonomous, citizen-by-citizen, representation. There are elements of descriptiveness and statistical representation (members of the AC will resemble the population at large, will bring in many or most of the interests present in the population); authorisation (as it would be an institution established constitutionally), and non-elective forms of accountability (horizontal accountability; responsiveness regarding citizen initiatives; responsiveness through the media, responsiveness through referenda which could be called by a qualified minority of the AC or by the elective chamber if in disagreement…); symbolic representation will accrue over time if people feel that the ACs stands for and acts for the people at large (will they call it some time in the future “the people´s chamber”?); and, finally, substantive representation will happen as the decisions taken by the AC will most possibly reflect the weighing of interests present in the population at large.

    The words and concepts will evolve along the development of such institutions… Monarchy was there well before the concept of identification between Crown and Country was developed, Democracy in Athens didn’t evolve as a concept but much later of its establishment by Cleisthenes, etc. –

    So, in a way, you are simply putting an impossible task upon me/us…How could you possibly provide “flash-bulb photographs of the structure taken from different angles” of something which is still nascent?

    You can aggregate votes, but an AC is more than “aggregating votes” – even in your model- it is listening to arguments, then deliberating (privately I understand) and finally voting: this is an exercise of power and people in the AC will understand this quite clearly. With power and information comes a new role, a new responsibility vis-à-vis the population at large, new incentives and new behaviour.

    You cannot replace “illocutionary forces” – but you can´t do that in any setting (elective, allotted, direct-democratic) – we have talked about this before – and there is no or just a minimal link between elections and the “illocutionary force” of elected reps – so where is the democratic equality or the democratic legitimacy of assymetric “illocutionary” force in elected representatives (call them advocates if you wish)? In the votes they have received? That’s pretty much weak – the votes, if anything, determine with “equality” who has more reps among those who are able to run for elections (an elite in practical terms), but the votes have no incidence on the degree of “illocutionary” force elected reps will have once they are working in the chamber.
    (That must be perhaps a UK-linked assumption perhaps tied to the majoritarian system and a romantic view of soap-box speakers?).

    So, what I’m saying is that in any setting, (elective, direct democratic, allotted) you need rules and procedures in order to avoid the predominance of those talented people who do no stop talking and/or writing all the time. Anyone who has worked in commissions, assemblies etc knows that such rules exist and that they can be strengthened to different degrees. For instance, if we had such rules in this forum, we would have been granted one shot each and one reply each – with a maximum of e.g. 200 words each time, and then others would have been called upon to continue the discussion or decide. I have even suggested to separate initiative from advocacy in an AC – which also would limit the impact of such “natural” assymetries.

    I think we should begin to stop, for the sake of mental sanity ;-)


  139. Perhaps I’ve misunderstood your argument, but it would appear that you are marooned somewhere in between Scylla and Charybdis. If the statistical representation argument is correct (actually it’s not an argument, it’s just a tautology [statistical representation is . . . um err . . . statistical]), then you will require a huge number of active ACs in order to approximate the only other legitimising principle available (rotation). Although your vote has a significant weight in your own AC, this will only be turned into a policy outcome once the votes for all the different ACs have been aggregated. How is this different from the current operation of rational ignorance? If I vote in a general election in my village (pop. 160) polling station then my vote will have significant causal power in terms of the local (ward) result, but will become insignificant when aggregated with all the other rural and urban wards in the constituency; even more so then the constituency vote is aggregated with the other constituencies to form a governing party. Seeing as this gives me in the end only one vote out of around 40 million there’s no point taking the trouble to study the candidates and issues involved. The same principle applies for an assembly with deciding powers, whether it is constituted by election or sortition.

    What is the relevance of the example that you give for (elected) local government? No-one is disputing the mandate of elected parish council officers to make policies within their own limited domain. But their mandate would be entirely different is selected by lot, for the reasons given in this post. If you accept the argument/tautology then multiplying the number of sortive bodies doesn’t affect the principle involved. And that’s before you get to the practical difficulties of everyone having to go to committee meetings all the time.

    Have I missed something? Are you suggesting an alternative legitimising principle to statistical representation or rotation?


  140. Sorry, my last response was to Martin

    Jorge, I agree this is getting a little time-consuming, so I’ll be brief. In terms of your list of representative functions, you need to decide which of them are appropriate for democratic politics. Pitkin eliminates formal and symbolic (fascist) representation and dismisses authorisation theories as the product of a Hobbesian sleight of hand. Anyone can be responsive (even a king or an benign oligarchy), and citizen-by-citizen representation requires rotation (impossible in large states). I agree that substantive representation would occur if the decisions taken by the AC reflect the weighing of interests in the population at large, but that’s incompatible with an active role for individual reps for the reasons already given ad nauseam. I do think we need an analytic approach rather than a shopping list.

    I do share your hope that individual members will exercise their judgment dispassionately rather than just in their own interests, but this is little more than a moral exhortation. Prudent institutional designers hope for the best, but plan for the worst.

    Regarding speech act theory, voters can at least choose their representative — if they opt for a lazy and inarticulate bozo then they have only themselves to blame; not so for allotted representatives.



  141. Keith: I’m talking about the votes of 30 people for community issues, a few hundred for regional issues and a few thousand for national issues. At what number of votes, in your opinion, does rational ignorance become a fatal problem?


  142. >”you need to decide which of them are appropriate for democratic politics” – all of them – they are different angles. The kernel is of course the thought of them “being” and “acting” “like” the population at large – but that is more than mere static statistical representation.

    BTW – Now in your view Pitkin’s book is not only conceptual and descriptive, but also normative – wow – that’s great, really.

    >”but that’s incompatible with an active role for individual reps for the reasons already given ad nauseam. I do think we need an analytic approach rather than a shopping list.” Thanks for the term “shopping-list” – I thought I was giving reasons and arguments. And your voting will never be passive – forget about that…

    >”and citizen-by-citizen representation requires rotation (impossible in large states)” Is this a new article of faith?

    >”Regarding speech act theory, voters can at least choose their representative — if they opt for a lazy and inarticulate bozo then they have only themselves to blame; not so for allotted representatives.” What sort of argument is this? Who chooses reps according to their speech talent? (I insist: that must be sth of UK soap-box romanticism)


  143. Martin, there’s clearly a trade-off between fine-grained representative inclusiveness and rational ignorance, so a compromise is necessary. Most legislative chambers are a few hundred strong, so I would imagine this would be the right sort of figure. One vote in a few thousand runs the risk of impotence.

    Jorge, we’re not really making any progress here — if you are committed to sortition as a sufficient principle for democratic politics, then I don’t think anything that I say (to the effect that it’s only a necessary condition) will make much difference. I could reply to each and every of your points, but am happy to let you have the last word. But I think the sheer lack of progress in the deliberations on this forum must make one at least a little sceptical to the notion that free discursive exchange is the best way of resolving political issues. Whilst its true that loudmouths like us can be reined in by rules, I think you need something a little more structured than that.

    best wishes



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 )

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: