Edip Yuksel: Lotteries elections: Disinfecting democracy from lobbies

In 1998, Edip Yuskel, “an Islamic reformer”, wrote an article proposing selecting Congress using sortition:

Every citizen who meets the qualifications enumerated in Article I, sections 2 and 3 of the Constitution could become a candidate by filling out a simple application form. This application can be automatically done during voter registration. Every registered person will have an equal chance of becoming a member of Congress. The election or selection can be conducted by mechanical devises or computers with sufficient security and supervision.

Yuskel’s paper is interesting for going beyond complaints about the ills of the existing electoral system and beyond proposing sortition as an alternative. The paper first makes a brief but fairly insightful and principled analysis of the causes of the problems and the likely ineffectiveness of commonly offered reforms. Then, the advantages of sortition over the current system are enumerated, and some objections are considered.

Yuskel points at the inherent weakness of campaign finance reform proposals:

[The] proposed model, which argues for a mandatory campaign finance system in which each voter supports his or her favorite candidates by vouchers, ignores the fact that those candidates with more money or celebrity will eventually have more media coverage and name recognition and thus, will most likely receive more “coupons.”

Similarly, he makes the point that the clichéd proposals for techno-electoral systems and techno-participative systems do not address fundamental cognitive barriers:

[F]ew citizens have the luxury of time and talent to find useful political information in the jungle of Internet. If elections were heavily dependent on this source, the clutter in related web sites would discourage even me, an experienced netsurfer[,]


Even if we ignore the economic and social cost of direct participation, however, it is susceptible to becoming a power base for only those who have time on their hands, such as the elderly.

Among the objections considered are the corrupting effects of empowerment, statistical inaccuracies, and the reliability of the sampling procedure. Yuskel makes concise presentations of both the objections and some counter-arguments.

Interestingly, Yuskel mentions Amar’s “lottery voting” papers, but seems unaware of proposals that are closer to his own, such as that of Callenbach and Phillips, which I believe is essentially identical to the one proposed in the paper.

45 Responses

  1. THREE related items:

    What and where is the discussion regarding WHO should be allowed to be in the allotment pool? Common Lot Productions puts forward the proposal that the pool be of the ‘willing and the able’. Namely: those who register (indicating willingness to serve) and those who can pass the citizenship test required of immigrants (indicating a basic understanding of the system). The test is assumed to be no more difficult, with study, than the current driver’s licensing test.

    Under that regimen — and also, say a salary at the 90th percentile (thus eliminating almost all economic sacrifice) — the argument for sortition producing representation proportional to the populace would likely skew towards what demographic?

    Does any one know precisely what the Law of Large Numbers margin of error would be in choosing, say, 500 representatives to represent the 310 million people in the U.S. That would be a ratio of 1 to 650,000.

    Thank you.


  2. Yuskel addresses the issue of “manipulation of chances” – as he puts it – under the “some details” section. Personally, as I have written here before, I think any deviation from the “one person, one chance” rule is both problematic in principle and dangerous in practice.

    Regarding the statistical deviations issue: the size of the population being sampled is immaterial – the population could be infinite. It is only the size of the sample that matters. The question of “margin of error” has no precise answer unless you define precisely what you mean by “margin of error”. As a rule of thumb, on each particular issue the sampling error is about 1 / (2 sqrt(n)), where n is the size of the sample. Yuskel has a short discussion of this issue under the second point in the “arguments” section.


  3. A voluntary allotment pool of the “willing and able” may not skew towards a specific demographic (although the Athenians found it biased towards the poor and the old) but it will not be descriptively representative. I’m not aware of any modern jury pool that is based on volunteering, so if that’s thought to be necessary for justice, then surely the same principle must apply to democracy?


  4. Mandatory service makes no sense. The idea that forcing people to show up increases representativity is silly. You can force people to show up, but you cannot force them to participate, and mandatory service would likely result in reduced motivation to participate.

    Regarding the analogy to jury service: happily, in the US, at least, mandatory jury service is nominal. It is easy to get excused, and even if you simply do not show up there are no serious consequences. The last thing you want on your jury (if you are interested in justice) is someone who would prefer to go home over getting justice done.

    Of course, juries are not representative. If you really want representative juries, all you need to do is to pay reasonably well.


  5. If it was clear that allotted members had serious political power then compulsion would probably not be necessary and a public culture would develop that reflected the importance of participation — it would become a privilege rather than a tiresome duty. Reasonable pay is also necessary in order not to dissuade the more affluent. Fishkin cannot force his DP members to attend, but goes out of his way to ensure that as many as possible do so, as there is no “if” prefixing his desire for accurate representation. Opting in, however, would produce an entirely unrepresentative body and does not really merit serious discussion on this forum (assuming prior agreement on democratic norms).

    As far as regards the depth of the participation, as you know my proposal is rather minimal in that allotted members would only be expected to a) turn up b) stay awake and c) vote. That doesn’t require a very high level of motivation.


  6. I am still undecided on the mandatory/voluntary issue.

    As one scenario, I am imagining sortition panels that each deal with a single defined issue. One angle I have considered, to avoid the “volunteer bias” based on individual issues (you don’t want only those eager to deal with the issue of legalization of pot to volunteer to tackle that issue), is to establish a pool of people willing to serve in general, with no reference to WHICH issue an individual will be dealing with. This pool would not necessarily be fully descriptively representative of the entire population, but it would be vastly MORE representative than any current legislative body on Earth.

    In a mandatory system, and individual who would never have volunteered may (when “forced” into the situation) suddenly discover a joy of communal decision making that she would never have known otherwise. She may then grudgingly, or eagerly participate, rather than sit with arms-folded listening to an iPod. A danger of a purely voluntary system is that those with low self-esteem, for example, may not step up and actively volunteer, even though they might do a wonderful job, and would serve if semi-forced. So another possibility (as others have stated) is to follow the American jury model and make it mandatory (with good pay) but also have a relatively easy escape route – — Sort of like the mandatory voting law in Australia (you just need to write a note explaining why you couldn’t vote to avoid the modest fine).

    Terry Bouricius


  7. That strikes me as a sensible compromise. A lot would depend on the amount of power involved and the background cultural expectations. The Athenians got by with a voluntary system for the jury pool but it was a small homogeneous community with considerable pressure to homonoia (same mindedness). I don’t think volunteering would work in large modern states due to potentially substantial differences (temperament etc) between volunteers and everyone else. Agree that low self-esteem is the biggest problem — “I wouldn’t want to join a club that would accept me as a member”.


  8. Terry,

    The system you propose is not so much “mandatory” as “opt out”. This is fine, but I think it should be made clear that this is a voluntary system, i.e., that those not interested (and only those not interested) are expected to opt out. Showing up should be seen as indicating an interest in participating. “I showed up only because I have too” should not be seen as an acceptable attitude by a delegate.

    I don’t think, by the way, that the matter of low self-esteem is addressed by compulsion in any way. Low self-esteem is handled by making sure that everybody in the chamber is well rewarded for their time and effort, is treated equally and is endowed with ample resources and authority. At a deeper level, low self-esteem should be reduced across society by moving away from elitist systems and ideologies.


  9. Yoram, I’m grappling unsuccessfully with your formula in response to my ‘margin of error’ query. I evidently stated my question poorly.
    I want to know, for instance, if there were a sortitional selection of 500 people made from all the eligible voters in the US — about 210 million — what would be the probability of getting 250 women and 250 men? Maybe the term I’m looking for is ‘deviation’?
    I want to complete this sentence: Making a random selection of 500 people from a pool of 210 million (for simplicity’s sake, assume they are perfectly 50-50 men-women), we can expect that no more than ___ will be men or women.
    OR: The chance that there would be 249 of one and 251 of the other would be ____. Or of 248/252 the chance would be ___.
    It surprised me when someone responded to me by saying “Well, it might turn out to be all women, for instance!” That drove me to read up (Wikipedia) about the Law of Large Numbers … where I was surprised to read how difficult it was to mathematically prove in the early 18th Century what seems now to be common sense (at least to some of us).


  10. David,

    The first sentence should read
    “… we can expect no more than 250 + 3 x 1/2 sqrt(500) ~= 285 will be men or women.”
    or, changing the sentence a little bit, it could read,
    “… we can expect that there would be about 250 +- 1/2 sqrt(500) men and women.”

    The missing numbers in the second sentence are both trans-astronomically tiny – less than one googolth (10^-100). To bring things closer to reality (but still very far from it), the chance that either there would be more than 350 men or more than 350 women in the group of 500 is less than 0.2 millionth of a millionth of a millionth (2 x 10^-19).

    We can make analogous (but more interesting) statements for the numbers in the first sentence: The chance that there will be either more than 285 men or more than 285 women is about two tenths of one percent. The chance that there will be no more than 250 + 1/2 sqrt(500) = 261 men or women is about 70%.


  11. Many thanks, Yoram. This is very useful.

    {Technical question: Is there a way to reply only to a single person?}


  12. I think replying to a single person would need to be done through email.

    By the way, this issue of chance variations was one of the two objections raised by Dahl to relying on sortition for selection of decision making bodies (he suggested using them for advisory bodies). As I wrote before, I think he greatly overestimated the chance of large variations.


  13. Yoram,

    I think what Common Lot S. is looking for (and others could use as well) is some pithy statements that would make people have an “ah-ha” experience about how unlikely that an “extremist” group could ever be empaneled using random selection. Rather than assuring that women and men will be balanced, I think the fear is that a dangerous fringe group might “by chance” end up with a majority.

    As an example, of a pithy large numbers statement, I heard a scientist explain that there were about a trillion unique sequences to place just fifteen books in order on a bookshelf. That stuck with me because it was so contrary to most people’s assumptions.

    Is there a comparable statement you can devise reassuring people about the safety of random sampling…Something like:
    If a particular minority makes up 20 % of the population, the likelihood that they would end up with a majority in a sortition group of 500 people is one chance in X. (Or … They would be likely to be randomly selected as a majority of a 500 member group once every X times.)


  14. Terry,

    In one of the posts I linked to above I offered the statement (somewhat revised here):

    If a group makes less than 40% of the population, then the chance that it will form a majority in a group of 500 randomly selected people is less than 3 in a million.

    Does that serve the purpose you intended?


  15. My impression is that people are so accustomed to public opinion polls, (electoral) exit polls etc that they would need little mathematical evidence to back up the intuition of statistical representation. After all medieval cooks were quite content to assume that a sample of soup would represent the whole cauldron, so long as it had been stirred well and the spoon was big enough. That’s why I’m sceptical regarding Dowlen’s claim that the Greeks did not connect sortition with representation as they had no mathematical concept of proportionality. Of course if the assembly were to take on non-aggregate functions that would be an altogether different matter — it would be comparatively easy for a minority group with charismatic leadership to dominate an allotted assembly of ordinary Joes.

    The concern I hear more often about sortition is that it is too representative, the assumption being that most people are ignorant regarding public affairs so how can a representative sample of most people be competent to rule?


  16. Yoram,

    Yes that is close. Better would be a sort of story line. People grasp and retain information better through anecdotes…This isn’t polished, but a story might go something like this
    “In an election system with single seat districts a minority group that was 40% of the population, might well fail to elect even a single member to a legislature, depending on the distribution of voters and the skill of those gerrymandering the districts. With a random selection process this minority group could expect to hold around 40% of the seats every time. Some people might worry that this minority group could “by chance” end up with a majority and defeat the will of the majority of the population. This concern is unfounded, as those who appreciate the law of large numbers understand. If a new random panel of 500 were to be appointed once each month, a minority group that was 40% of the population would be unlikely to gain a majority of seats after X centuries.”

    By the way, can you “show your work” (the calculation needed)? I know a tiny bit about statistics and probability, but have forgotten most of what I learned in college.


  17. Some real examples of the bias of electoral systems can also be added to the story. Not only do some minorities get under-represented (Blacks and Hispanics, the young, the poor, etc.), but some tiny minorities get to be hugely over-represented (lawyers, the rich, white old men).

    The calculation is simply an evaluation of the Binomial cumulative distribution function. If B is distributed Binomially with parameters n = 500 and p = 0.4 then P(B > 250) ~= 2.5e-6.

    You can refer to Wikipedia for the definition and some properties of the Binomial distribution: http://en.wikipedia.org/wiki/Binomial_distribution

    An online calculator of the Binomial CDF is at: http://stattrek.com/tables/binomial.aspx


  18. Keith, I thought as you did … that common sense made it obvious that random selection of a large sample would tend towards a proportional number. So when an intelligent woman asked me “Might not such a sortition end up with, for instance, all women?” I was shocked that anyone would think such a thing. Of course, there is a chance, however astronomical, that 500 women would be sortitionally selected out of 200+ million.

    Yoram, I’m surprised that the divergence is as great as your calculation indicates. What I thought I had read somewhere was that it wouldn’t be likely to be more than 2 1/2 percent. Which, out of a group of 500 would mean no more than ~237/263. You said 285/215. I trust that in deciding what is ‘likely’, the question is: “Is 1 out of 10 chances that it will be one way or the other the definition of ‘likely’? Or 1 out of a 100? Or 1 out of a million? et cetera.” Is that the crux of this question about what is ‘likely’?


  19. Yes. As I wrote, +- 1/2 sqrt(n) covers about 70% of the drawings. This is the measure of uncertainty that is usually cited. In percentage terms this is +- 1/(2 sqrt(n)), which for n = 500 is 2.2%.


  20. >So when an intelligent woman asked me “Might not such a sortition end up with, for instance, all women?”

    I’ll resist the temptation to make a cheap sexist jibe about oxymorons, but I believe there is good evidence that mathematical ability is not equally distributed across the sexes. Tell her not to worry her pretty little head about math (leave it to us chaps), go back to her natural place (in the kitchen), cook up a soup, stir it and then sample the ingredients with a spoon!


  21. > I’ll resist the temptation

    Obviously, you did not manage to resist.

    Rather silly as well as obnoxious. I heard plenty of intelligent men expressing concern about the possibility that, say, the neo-Nazis or the insane would, by chance, gain a majority in an allotted legislature. We also have Dahl – ostensibly a male specimen of the homo sapiens species – expressing similar concerns.

    As for the soup analogy, I heard it before and I find that it overstates the case for sampling on several points. For one thing, the cook keeps mixing the soup until it is observably uniform – there is no analogous process with opinion sampling.


  22. I agree it was silly (in England we call it “joking”, something that is still permissible amongst friends). As for the disanalogy, soup (unlike statistical sampling) is subject to the effects of gravity, hence the need for continued stirring. This was intended to be a serious point — Richard Tuck notes that the ‘estimation of probabilities’ long predates Leibniz and Huygens’s mathematical studies – appearing, for example, in the writings of Grotius and the members of the Tew Circle, thereby casting doubt on Ian Hacking’s claim that the concept of probability is exclusively modern (Tuck, 1979, pp.104-5). Hacking’s (erroneous) claim is the foundation for Ollie Dowlen’s argument that sortition has nothing to do with descriptive representation. As for the risks of the domination of a random assembly by a small group, everyone would intuitively understand that the risk this can be overcome by expanding the size of the group; all that statisticians need to do is to quantify the risk and specify the necessary sample size in order to reduce it to a negligible level. The surprising thing is that the sample can be quite small.

    However the risk of a minority dominating an assembly empowered with active functions (other than voting) is much greater, and entirely impossible to model mathematically — even in principle. (If I’m wrong about that, then show me the mathematical proof). This is why those who argue the case for descriptive representation need to err on the side of caution with respect to what powers a randomly-selected assembly has.


  23. It’s in any case silly to say it can’t be modelled, Keith Sutherland. You have a model already, just an intuitive rather than formal one – and quite possibly, a flawed one. It could nonetheless be formalized, perhaps to cast some light on what your assumptions are.

    Asking those who disagree with you to present a formal model, when you don’t have one yourself, is unreasonable.

    –Harald Korneliussen


  24. OK, but the informal model that I am working on presupposes a level of complexity that would not be amenable to modelling in mathematical terms. As Yoram has demonstrated, it is a simple matter to model how well a sample represents a target population on factors such as age, socio-economic class, gender etc. But how would you represent mathematically the distribution within a small allotted group of factors such as persuasiveness, esteem and the illocutionary force of individual speech acts? These are the assumptions that underlie my model and I would defy anyone to formalize them in a way that would generate meaningful probabilities. And if the representativity of an allotted legislature that sets its own agenda cannot be demonstrated mathematically then the legislature would lack a democratic mandate.

    More importantly, how would you test if the priorities of the active agents within the allotted sample coincided with those of the target population? And why would you want to try, given that there are proven ways of sampling raw public opinion? I have a new post, pending with Yoram, which claims that the UK government e-petitions system is beginning to look like the most democratic way of providing an agenda for an allotted legislature and that it would appear not to be particularly exposed to the influence of money and power that has led to the claim that elections and plebiscites are necessarily open to corruption on this count.


  25. > More importantly, how would you test if the priorities of the active agents within the allotted sample coincided with those of the target population? And why would you want to try, given that there are proven ways of sampling raw public opinion?

    I know that trying to explain the the difference between day and night to a vase only makes one look like a fool, and that doing so repeatedly makes one’s peers suspect that one is insane. I’ll answer anyway: “raw public opinion”, as expressed in mass politics or in opinion surveys, is easily susceptible to manipulation, and, at best, is a set of general ideas – it cannot provide policy proposals. It is exactly the falsehood that this kind of construct is a useful tool for policy making that underlies electoralism and plebiscite-based systems.


  26. Take a look at http://epetitions.direct.gov.uk/petitions?order=desc&sort=count&state=open

    These all look more like policy proposals than general ideas, albeit that they would need operationalizing if they were to become parliamentary bills. Take the lead one, for example — it’s more of a policy proposal than a general philosophical meditation on the nature of human rights:

    Convicted London rioters should loose (sic) all benefits.

    Responsible department: Department for Work and Pensions

    Any persons convicted of criminal acts during the current London riots should have all financial benefits removed. No tax payer should have to contribute to those who have destroyed property, stolen from their community and shown a disregard for the country that provides for them.

    This is intended to be the subject for a parliamentary debate (not a legislative bill). If, at the end of the debate, there was a division in favour of the petition, it would be up to the responsible ministry to turn it into a parliamentary bill. If an allotted assembly were to make policy proposals they would adopt a very similar trajectory (from the general to the specific), the only difference being that the proposals would have no initial democratic mandate. All opinions are potentially subject to manipulation — this is the case with an elective/plebiscite system but also the internal deliberations of a small group — hence the ongoing need for trained moderators in deliberative forums. The safest way to avoid manipulation is to bring everything into the public sphere (along with taking every possible step to ensure diverse and pluralistic media ownership). You also refuse to consider the possibility (inevitable IMHO) that rich and powerful lobbyists would target allotted members of a legislative assembly in order to influence the genesis and advocacy of their pet projects. Manipulation of the e-petitions site would pale into insignificance in comparison with this very real opportunity for corruption.

    I certainly don’t think you are insane or speaking in a language that is not intelligible to a vase [me, presumably] but I’m struggling to follow the logic of your argument. I’m also struck by the manichean nature of your analogy (“explaining the difference between day and night to a vase”); a more commonsense perspective would be that political processes are a matter of degree rather than simple binaries.


  27. First, the snippet you quote is not policy (it is merely a tweak of existing public policy – and an under-defined tweak at that – what does “financial benefits” mean? It is inevitable that the details will be filled in by a small group of policy makers). It is also not “raw public opinion”, it is at most the opinion of a small fraction of the entire UK population. Even for that set of people it is clearly uninformed and unconsidered opinion – it is the knee-jerk reaction to partial and biased information. It is anti-democratic to set the public agenda based on such foundations.

    Regarding the possibility of corruption of allotted delegates: it is certain that the rich and powerful would try to do that. I think it is unlikely they would be generally successful. But whatever your opinion on the likelihood of such a situation, clearly introducing institutional corruption into the system – in the form of granting special powers to certain elite groups – can only increase corruption. The overwhelming influence of powerful interests of mass politics (elections, signature gathering, plebiscites, etc.) is not only theoretically well established it is a well established and widely recognized empirical fact.


  28. Well, that’s a curious choice of proposal you mention there, Keith Sutherland. Because as you hopefully know, this would be an ex post facto law, which is illegal according to the European Convention on Human Rights (as well as centuries of legal tradition). Do you think the proposers understand what they are trying to throw on the boat?

    I thought such proposals were exactly what you were worried about allotted agenda-setters coming up with.


  29. Yoram: The financial benefits referred to are social security and housing benefits. There is a very tight word limit on e-petitions. I agree that the petitions only reflect a small fraction of the UK population, that’s why I propose a second step in which the e-petitions that pass the signature threshold are then put to the whole electorate in a referendum. Those that score the most votes are then put to the allotted legislative assembly. My use of the phrase “raw public opinion” acknowledges that the views are unconsidered; that is the role of the (allotted) deliberative phase, during which knee-jerk proposals can be rejected. What I am suggesting is undoubtedly populist, but in what sense do you claim that it’s undemocratic? It’s a pretty good analogue for Athenian practice (raw votes at the ekklesia then subject to deliberative scrutiny in the nomothetai) and the Athenians invented democracy.

    Regarding the corruption of allotted delegates, I think it’s highly likely and you think it’s unlikely, so there’s not much more to be said on that matter. But what is the basis for your argument that e-petitions grant special powers to elite groups? Which elite groups are being represented in the petitions that have so far gained the most votes? The very fact that the author of the most popular petition cannot even spell would tend to indicate otherwise.

    I’m really trying to understand your general concerns — you’ve mentioned in the past that the public are brainwashed and indoctrinated by the mainstream media. I acknowledge that there is some truth in this claim, but if it’s true for the public in general then it will also be true for the individual members of the public who are allotted to the assembly. Are you claiming that the discursive process of assembly debate will somehow undo all this indoctrination? Do you think that the force of deliberative reasoning will enable the participants in the debate to transcend their conditioning, throw off their shackles and emerge blinking into the bright sunlight of pure thought? I think it would be much more effective to bring in laws ensuring a wide plurality of media ownership, so then at least indoctrination can be evenhanded. You have shown a preference for very small alternative media websites (that most of us have never even heard of), and you clearly believe that they are impartial (as opposed to the mainstream media, which are just a bunch of capitalist lackeys); others might claim that the sort of media sources you prefer are equally biased by left-wing prejudices. As we’ll never agree on this it would be much safer just to break up media monopolies and leave it to the free market (although you have previously argued that the latter is oxymoronic).

    Harald: It’s perfectly possible for successful e-petitions to be total gobbledegook and the fact that it’s ex post facto and illegal would again indicate that it was not the work of a rich and powerful lobby group. My *only* concern with agenda-setting is that it should be a fully democratic procedure; it’s up to the allotted chamber to decide if the proposal is desirable and up to government lawyers to decide if it’s feasible (or if there would be resulting entailments, such as modifying or repealing the Human Rights Act). It really isn’t up to you, me, Conan, Yoram or the Queen of Sheba to pronounce on the practicality or desirability of specific legislative proposals, however distasteful they may be to our refined liberal sensibilities.


  30. “… the fact that it’s ex post facto and illegal would again indicate that it was not the work of a rich and powerful lobby group.”

    This is nonsense. Many rich and powerful lobby groups (most notably, Rupert Murdoch) don’t give a hoot about human rights or laws. In fact, if you wanted to advance a more brutal society, wrapping it in this sort of populism and claiming it is an “honest expression of what the people feel”, would be the obvious thing to do. (Not that I think the elites in question are necessarily dishonest about this – they quite possibly believe the people really agree with them.)

    It’s election time in Norway, and we’ve just had a massacre by a right-wing nut who decided to be more than a forum warrior. This puts a certain fact into sharp relief: The people who currently totally dominate Norwegian online forums on newspapers and similar, are an insignificant force in (PR) elections.

    They are extreme loudmouths, exploiting anonymity and lack of accountability for their statements to give their “id”-impulses totally free rein. And they’re so extreme, and so persistent, that they eventually drive the more moderate away – even though, elections show, 90-95% of voters are more moderate.


  31. > This puts a certain fact into sharp relief: The people who currently totally dominate Norwegian online forums on newspapers and similar, are an insignificant force in (PR) elections.

    Online forums in mass readership sites (unlike intimate forums such as Equality-by-Lot) are another form of mass politics and are accordingly incapable of producing a discussion that is representative of informed and considered public opinion.

    This phenomenon naturally leads to considering the possibility of democratic media. That is, media where exposure is allocated by lot rather than handed off to an elite or to a self-selected group of loudmouths.


  32. Harald: Why would a well-resourced lobby group introduce an illegal ex post facto proposal in the full knowledge that it would be ruled out of order? And how do you explain the spelling error in the title? (unless this is a brilliant Machiavellian ploy by Murdoch’s copy editors to make it look like a grass-roots petition). This is really taking conspiracy theory to absurd limits; it’s much more plausible to think that this is a genuine grass-roots petition and that the popular press are struggling to keep up.

    I’m not sure what purpose your reference to Anders Breivik is intended to serve. Are you suggesting that there is some equivalence between the actions of a mass murderer and a popular petition to deprive rioters of welfare payments? And if this is the case, how do you square that with the earlier claim that it’s all a conspiracy by rich and powerful lobby groups?


  33. Yoram: I agree the debate on intimate forums is (sometimes) better informed and considered, but why do you use the adjective “public”? Although such forums are open in principle to anybody, they certainly don’t reflect public opinion — indeed our own forum, most of the time, reflects the views of a tiny handful of self-selected loudmouths (you, me and Harald being the principal culprits). Although we usually adhere to the rules of civilized discursive debate, nevertheless we are self-selected loudmouths, so I’m puzzled why you would use this as a template for democratic media. And if exposure is allocated by lot then why would the resultant debate be better informed and why would anyone want to follow it? I’m afraid the best we can do is to break up media monopolies and then leave it to the market, allowing readers to vote with their feet.


  34. Murdoch’s goal isn’t to pass any crazy proposal. Murdoch’s goal is what it has been for a long time, and which no sane analyst can disagree with: moving the Overton window to the right.

    For this purpose, a mass politics site like e-petitions is manna from heaven, because it will predictably be dominated by right-wing loudmouths. You can bet he will promote it, directly and indirectly.

    Thissite was never a template for democratic media, but it’s small enough that it isn’t needed. If there were thousands of posters posting day and night, the ad-hoc, live and let live attitude we have to visibility would have to be reconsidered.

    > And if exposure is allocated by lot then why would the resultant debate be better informed and why would anyone want to follow it?

    I don’t know if you’re familiar with the old tech site slashdot. It used some principles of sortition: any user having been active on the site for a certain period (I think it used to be one year), was a candidate to get 5 mod points. These he could assign, one at a time, on comments on stories he read the next day. Posts with 4 or 5 points (5 was the max) would be automatically expanded, posts at 1 or less would be hidden under a “more posts” link. There was also a system of meta-moderation, based on volunteerism, but here randomness was applied in a different way: you got assigned a handful of moderated posts, and were supposed to judge whether they had been moderated fairly.

    Now this is hardly an ideal use of sortition: The economy of attention wasn’t really handled well, so you would still expect an early post to get more upvotes than a late one, and an already upmodded post to get more upvotes. Metamoderation was open to a good deal of abuse, and I think it eventually was (in any case they seem to have abandoned it now). Since you could have mulitple accounts, you could probably game the system somewhat, as long as you had the patience to let them all “mature” for a year.

    Still, the discussion on technical matters (the site’s focus, after all, and a topic where there is little incentive for gaming the system), if you just looked at the high-ranked comments, was of a very high quality. Far better than any comparable site.

    So I have good reason to think a reflected, sortition based approach to moderation can foster informed debate. I and Yoram toyed with the idea of making a customized variant of reddit (or one of the other open source news aggregator sites) way back, but it didn’t get anywhere – it’s not as if attracting the thousands of readers such a site would need to be effective is trivial, either.


  35. Harald: “For this purpose, a mass politics site like e-petitions is manna from heaven, because it will predictably be dominated by right-wing loudmouths.”

    I can understand that e-petition sites will be likely to attract the attention of opinionated loudmouths, but why right-wing loudmouths? Currently the petition to retain the ban on capital punishment (22,245) outnumbers the petition to restore it (14,917), so this would tend to suggest that left-wing loudmouths are slightly more keen to participate. If, nevertheless, you are correct in your contention, then the only plausible explanation is that the public is naturally more right-wing than the media-political class, (unless you are suggesting that right-wingers are more computer savvy). This would tend to be be supported by data on UK newspaper circulation where readership of the left-wing media (Guardian, Independent, Mirror) is considerably less than the right-wing press. If so then it would be natural that right-wing voices would predominate on an e-petitions website, without the need to resort to (frankly incredible) conspiracy theories.

    Incidentally the reason that I argue for the role of an allotted assembly to be limited to aggregate functions like voting is because loudmouths will always dominate over the silent majority. That’s why democracy is based on the principal of one person one vote — everyone has only one token, however big their mouth.

    Moving on to the issue of democratic media, I’m not familiar with slashdot but the procedures you are suggesting sound a lot like voting and peer-review, I don’t see that sortition was a significant element.
    I believe Yoram was suggesting the general abolition of private media ownership and its replacement by a sortive-based system. So my question still stands — why would allotted debate be better informed and who would want to follow it?


  36. > I can understand that e-petition sites will be likely to attract the attention of opinionated loudmouths, but why right-wing loudmouths?

    Right wing opinion is friendly to the rich, and thus it is likely that right wing opinion would be backed by resources that left-wing opinion would be much less likely to garner.


  37. > (unless you are suggesting that right-wingers are more computer savvy).

    More properly, that the computer savvy are more likely to be right wing. But right-wingers appear also more persistent in online fora. Even when they aren’t the majority, they tend to set the agenda (in connection with the recent terrorist attack in Norway, many pointed out what the computer savvy have known for a long time: that no matter what the topic is in a Norwegian newspaper’s forums, after 8 comment it will be about the evil leftist government traitors, and after 12 it will be about muslims.)

    That the death penalty is up for debate at all (in a country where it is illegal), is a victory for right wing activists.

    > Incidentally the reason that I argue for the role of an allotted assembly to be limited to aggregate functions like voting is because loudmouths will always dominate over the silent majority. That’s why democracy is based on the principal of one person one vote — everyone has only one token, however big their mouth.

    For proposals, you have to use your mouth for more than saying yea or nay anyway; whoever is allowed to do it can take advantage of a big mouth. Anyway, the bigger the assembly, the larger the possibility for one truly big mouth to be in it – and when the “assembly” is the entire electorate, that becomes a so much worse problem.

    Anonymity an unaccountability are the would-be mob director’s best friend. You have plenty of both online, but the smaller an assembly you have, the less you have it, due to social pressures. In small enough assemblies (maybe 8-12 people), I think loudmouths are a non-problem, and in slightly larger ones (maybe 100-200), they are still just a modest problem.


  38. So does computer use just correlate with right-wingery or are you suggesting a causal link — perhaps something to do with the radiation from screens? If so then the move from CRT to LCD should lead to a political realignment. Seriously though, I can only repeat my plea for parsimony in explanation — if a sample group of whatever size has a consistent prevalence of certain characteristics (political or otherwise) then it’s likely that that reflects a prevalence in the general population. This, after all, is the prime argument for statistical representation and sortition. As for Yoram’s claim:

    >Right wing opinion is friendly to the rich, and thus it is likely that right wing opinion would be backed by resources that left-wing opinion would be much less likely to garner.

    Although this may be true in general, in the case of the e-petitions website, the only resources backing it are provided by the government. The petition in question mushroomed support virally, and cannot be attributed to sponsorship by a daily newspaper. And how do you explain the fact that the (left-wing) petition for retaining the ban on capital punishment is far more popular than the (right-wing) campaign for its re-introduction? Of course capital punishment is illegal — all laws are illegal until they have been introduced !

    Quite frankly I’m astonished at your mutual hostility to democracy and the explanatory contortions you have to adopt to explain the prevalence in the general public of attitudes that you find distasteful.


  39. Harald,

    > In small enough assemblies (maybe 8-12 people), I think loudmouths are a non-problem, and in slightly larger ones (maybe 100-200), they are still just a modest problem.

    The setting is important. As your Jante Law story indicates, a whole village – hundreds of people, I presume – could easily turn against a loudmouth. This is because in a village everybody is familiar with everybody and have a long time to form opinions and coalitions, and the stakes are high. So, while an ad-hoc, short-term, 200 strong assembly would likely be dominated by mass political effects, a long-term assembly with appropriate resources and authority could function reasonably well even if it was made of several hundred people.

    It is interesting that good democratic design turns out to be a balancing act between concentrating decision-making power too much and spreading it too thinly. (This is contrary to conventional democratic wisdom that asserts that decision-making power needs to be spread as thinly as possible.)


  40. Whilst it’s true that a whole village could turn against a loudmouth, it’s also the case that it could fall under her sway, depending on the subtlety of her persuasive charms. I live in a village and there is an informal but pervasive structure of power and influence which reflects the perceived status of the inhabitants. The village used to be run by the farmers and landowners but they have been displaced by a small group of doctors, academics and other professionals who run the council and dominate the parish plan committee. If we are going to be honest we have to be agnostic regarding the loudmouth effect, as there are too many “coulds” involved. Much better therefore to design political institutions in a way that do not allow the possibility of nefarious influences.

    I agree with Yoram’s argument for the democratic balancing act, but this is best achieved via functional differentiation. The micro-deliberations of a small statistically-representative body are the most democratic way of arriving at well-considered decisions, but the policy options and advocacy need to be supplied using macro-democratic procedures, in order to avoid the (hypothetical) danger of the loudmouth.

    The attempt by elective democracy to spread the decision-making power thinly (by the extension of the franchise) has failed, as in modern democracies all power is in the hands of party leaders and a small cabal of advisors. But we should resist the temptation to throw the baby out with the bathwater and acknowledge that some form of public opinion sampling is required in order to arrive at policy options. In my second book, A People’s Parliament, I argued that this called for an ongoing role for political parties, but I’m now more persuaded by the e-petition/referendum model as this cuts out the middle-man and all the resulting factional problems. Although we may choose to label certain petitions “right-wing” or “left-wing”, these are adjectives, not partisan groupings. I strongly encourage everyone to take a good look at the range of e-petitions involved and then ask themselves whether or not the site reflects true diversity:


    There is, of course, a further problem as to which petitions receive sufficient publicity to reach the 100,000-signature threshold (mine currently has 6, whereas the lead petition has nearly 200,000) but would argue that this is best served by breaking up media monopolies. I would also agree with Terry’s argument for a nominal signing fee (he suggested $1) in order to distinguish genuine support from crowd effects. The organisers should also concatenate the large number of nearly-identical proposals and consider imposing incremental signature targets (any proposal that did not achieve, say 100 signatures after a month, 1,000 after 2 months etc, should be dropped) in order to create a manageable number of options. At the moment it’s a bit like choosing from a menu with 7,000 main courses.


  41. > So does computer use just correlate with right-wingery or are you suggesting a causal link

    Since you’re so insistent on trying to ridicule, I’ll explain some more.
    People on the left are better organized in the out in the old analog world (the one with the scenery). There are ten genuine left-oriented grassroots groups to each right-oriented one. You’re not so eager at looking for new ways to communicate when you are so comfortable with the ones you have.

    Also, relevant to most countries: They suddenly came in contact with the US political scene, which is far to the right of even the UK. Far right-inclined people were overjoyed to find an environment where their views were considered normal, mainstream. Few people like to think of themselves as extremists, so they consequently got a lot more emotional payoff from being online. So they became more tech savvy.

    But the how is less important. I’m very surprised you haven’t noticed the right-wing, libertarian, atheist streak in the tech savvy.

    > Although this may be true in general, in the case of the e-petitions website, the only resources backing it are provided by the government.

    The most important resource is inbound links. Those are not provided by the government.


  42. Then why is it that the online signatures for the petition to retain the ban on the death penalty far outnumber those for the petition to re-introduce it? However I do agree with you about the inbound links, this is why I argue that e-petitions should be the first stage, followed by 2) a general referendum and 3) allotted deliberation.


  43. > Then why is it that the online signatures for the petition to retain the ban on the death penalty far outnumber those for the petition to re-introduce it?

    For the same reason your proposal sits on 6: inbound links.

    Obviously, the right-wing forces aren’t going to win every battle. If the Guardian links to the petition, and a handful of UKIP bloggers do so, the Guardian’s views will probably dominate (but not nearly as much as you’d think, because those right wing blogs have readership and mobilization potential way out of proportion to their share of the electorate. See also: US presidental candidate Ron Paul.)

    Why on earth you want the crudest of mass politics to be the first filter, I don’t understand. For one thing, it obviously doesn’t work for you. More importantly, the reasons it does not work for you applies to lots of other people.


  44. Because what works for me is not the criterion — in a democracy the deciding factor is what works for the majority. I may be able to persuade a small group of the merits of my proposal but would not have a democratic mandate for so doing, it would be just because I had a persuasive manner. This is why only those proposals that get through the first two filters should be eligible for deliberative scrutiny. No doubt this will preclude many worthwhile projects but that’s the price that you have to pay for majority rule.


  45. In this thread above is an exchange between me and Yoram regarding statistical probabilities in using sortition. This email correspondence fully clarifies the issue for me.

    =======first my query to Yoram:

    From: David Grant
    To: Yoram Gat
    Sent: Thursday, September 29, 2011 4:11 AM
    Subject: Margin of error, briefly

    Hello, Yoram,

    [regarding]…. that thread on Equality-by-Lot between you and me regarding the margin of error in sortitioning 500 seats out of … I think I reckoned 200 million?
    You said the probability would range, as I recall, between — for instance, for the dyad of women and men — 225/275.
    Which means, I take it, that men and women would sometimes be 225/275, sometimes 250/250, sometimes 240/260 … but unlikely to be 220/280 or worse.
    But I presume as well that ‘unlikely’ is a judgment call. A percentage, right? Would it be a 90%, for instance, likelihood for 225/275; but an 85% for 220/280?

    What I want to be able to say is: If 200 million adults were in the pool for 500 seats, the margin of error (correct term?) for choosing an accurate descriptive proportion of that population would not be any greater than plus-or-minus __%.
    Is it possible to make such a statement?

    +++++++++++Yoram’s reply

    From: Yoram Gat
    Subject: Re: Margin of error, briefly
    Date: 29September, 2011 2:58:13 AM EDT
    To: David Grant
    Reply-To: Yoram Gat

    Margin of error is not exactly the right term, I think – that is used when using a proportion in a sample to estimate the proportion in the population. In our case we assume that the proportion in the population is known, and we wish to bound the proportion in a sample. I would use something like “random fluctuation”.

    The size of the population doesn’t matter unless it is tiny – the statements are as true for a city of 100,000 as they are for a country of hundreds of millions.

    The proportion in the population +/-sqrt{n}/2 covers about 70% of the samples. Increasing the range to +/-sqrt{n} increases the coverage to about 95%, and increasing it to +/-3sqrt{n}/2 increases the coverage to about 99.5%. So in the case of a sample of 500 and an even split in the population, you will have 239–261 about 70% of the time, 227–273 about 95% of the time, and 216–284 about 99.5% of the time. The chance of having a split that is worse than 200/300, by the way, is about 1:100,000.


Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Connecting to %s

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

%d bloggers like this: