Somin: Sortition won’t solve political ignorance


Ilya Somin, Professor of Law at George Mason University, is the author of the book Democracy and Political Ignorance: Why Smaller Government is Smarter. Somin opens an opinion piece in the Washington Post thus:

Widespread political ignorance is a serious problem for modern democracy. In recent years, many scholars have argued that we can overcome it by relying on “sortition”: delegating various political decisions to jury-like bodies selected at random from the general population. In this post, I explain why such proposals are unlikely to succeed.

Before going into the substance of Somin’s arguments about sortition, it is important to realize that ignorance is actually only the second most important problem with the current system, the first being the difficulty of mass scale agenda setting. More details here. The emphasis on ignorance rather than agenda setting is typical of the “rational choice” line of political argument which ignores the complexity of organization largely for ideological reasons. That said, the ability of decision makers to become informed about their subject matters is important and worth discussing. Some of the considerations that are discussed below apply also to the matter of agenda setting.

Somin starts by explaining the obvious argument for why allotted body members would be likely be dramatically better informed than voters: the decision-making power of voters is infinitesimally small while that of allotted body members is noticeable. That improves motivation to become informed. Combining that with more time and more resources can in fact create an expectation for substantially reduced ignorance. Now Somin has to explain why this argument is invalid. He offers:

Unless the participants study for an extremely long time, they are unlikely to become knowledgeable about more than a small fraction of the many issues addressed by the modern state.[…]

This problem might be alleviated by by having each body selected by sortition address only a narrow range of issues. But then there would be serious problems of coordination between them.

So: irreducible complexity. The strength of this argument depends on the definition of “extremely long”. It is clear that no person can have a deep understanding of the many areas of public policy even if they devote their entire life to learning them. Yet, obviously coherent public policy not only exists, but is also quite effective at promoting the interests of policy makers in the existing system. So it is clear that the tenure of present-system policy makers – which is admittedly measured in years, often many years – is sufficient to allow them to become knowledgeable enough to craft policy effectively. But Somin may very well be right that shorter tenures may not allow allotted bodies to fashion policy democratically – i.e., in a way that reflects an independent understanding the issues.

The next paragraph makes it clear that “extremely long” was meant to include a tenure of years and that he concedes that a service term of years should be enough to overcome political ignorance.

Another possible way to make the participants better-informed would be to have them serve for long periods of time, perhaps even years on end. But in that scenario, the participants would gradually become a kind of professional governing class and would no longer be just randomly selected ordinary people.

According to this argument, knowledge is gained at the same time as representativity is lost. To some extent this is exactly the point: the allotted are meant to be more knowledgeable than the general population, and in this sense non-representative. See point #12 here. So the real question is whether knowledge is gained faster or slower than core values and interests are corrupted by power. I think the answer is there is a point of optimum, and a span of a few years – say 4, as is the typical term for elected officials – should sit comfortably close to this point. Much shorter terms (a year or shorter) and indeed no sufficient knowledge is gained to allow democratic decision making while much longer terms (10 years or longer) and decision makers can be expected to develop the self-serving worldviews that are typical of political elites today.

Somin continues:

Sortition systems are also vulnerable to manipulation in a variety of ways. The government could potentially skew the selection procedure in order to ensure that more of its supporters get selected. If, as in most proposals, the participants are expected to hear presentations about policy issues and engage in deliberation about them, there are many ways to bias the choice of presenters and the selection and framing of issues.

Skewing the sortition can be dismissed. It should be relatively easy to set up a system that would make skewing very difficult. The matter of controlling the information presented and the “framing” of the issues, on the other hand, is important. The way to handle this is to have allotted bodies set their own agendas, or have them set by other allotted bodies rather than by elite-dominated bodies, and have the allotted bodies responsible for gathering their own information rather than have it dictated to them by external forces.

Somin is also arguing that a bodies that are large enough to be representative would not be able overcome the rational ignorance effect:

[S]ortition systems will face difficult trade-offs between representativeness and minimizing incentives for rational ignorance. If the group selected is small, rational ignorance is unlikely to be a problem, since each vote will have a high chance of decisiveness.

Somin’s argument is quite unclear about the what he would consider as too-small-for-representativeness. There is certainly a trade-off involved in body size, although it is not exactly the one Somin describes. Very small bodies breed a sense of individual omnipotence and thus corruption while a mass-body leaves us with the same problems that sortition aims to solve. The optimum size would depend on the setting, but for high-powered long-termed bodies a size of about 200 people should in the right order of magnitude to avoid the problems of the extremes.

Finally, Somin mentions the Athenian precedent. According to Somin it is not relevant because modern government is much more complex. This is an issue worth considering separately.

In summary, while Somin’s arguments are not really effective in proving his sweeping conclusion against sortition, he does raise valid points that are important in guiding the design of the system under which sortition can foster democracy and in explaining why quite a few of the sortition-based proposals being made today cannot be expected to produce desirable results. It seems that a debate between Somin and a sortition advocate could be interesting.


9 Responses

  1. Sortition won’t cure ignorance but unlike the Washington Post, it won’t institutionalize it.

    Liked by 2 people

  2. Yoram< Ilya Somin

    The trouble with Somin's criticism of sortition is due in part to the failure of kelroterians to get beyond the traditional question: Who rules? In the contest of traditional centralisation of power. So sortition is often presented just as a way of replacing politicians by a sample of citizens in the frame of modern government structures.

    You rightly say that the prior question is setting the agenda, getting clear about how various practices and institutions are working and what practical alternatives we could try. Instead we naturally tend to focus on symptoms, just scratching where it is itchy.

    The only way to get beyond this sort of superficial reaction is to have a very thorough and clearly focussed public debate about specific problems. So the sortition body has the task of drawing its conclusions from as good a basis as we can get.

    Closely connected with this point is our need to bet away from the facile assumption that a nation-state is like a single business with a single aim. On the contrary it is involved in a huge range of diverse operations that have almost nothing to do with each other. It has picked up most of these responsibilities because they concern public goods which, it seems, can be financed only by taxation backed by the punitive power of the state. Who pays the piper calls the tune. I won't attempt to address that question here. Most of the sate's budget is precomitted. So just assume that each of the problem areas gets a traditional allocation of funds

    The basic question is: What needs to be done about each of the host of specific problems in the light of thorough discussion of each. I have argued that in each instance the people most directly affected by the problem are the best judges of what they need, in the light of a comprehensive discussion. Sortition becomes a matter of giving power to those most affected and getting the particular problems right.

    Of course, there are questions of coordination with other enterprises. But those problems only have clear and sensible solutions when the needs of each of the proposed coordinates are clear. The idea of planning as a centralised activity is almost purely ideological. Even in business conglomerates almost always fail, in spite of alleged savings of scale. They come to be run by bean counters who are out of touch with the problems of each of the component businesses. A great deal of what is wrong with the bureaucracies that discredit the quality of what they are charged with delivering is their distance from the problems and reliance on monetary assessments.

    Please read The Demarchy Manifesto for more details!

    Liked by 1 person

  3. John,

    > Sortition becomes a matter of giving power to those most affected and getting the particular problems right.

    This seems to largely assume away Somin’s points regarding ignorance, but regardless of this, can you specify the mechanism for determining who is “most affected”? (Having some body of experts making this decision, is a non-starter.) And what about the interests and points of view of those who are judged not to be “most affected”?

    Liked by 1 person

  4. Somin raises some good points:

    >Unless the participants study for an extremely long time, they are unlikely to become knowledgeable about more than a small fraction of the many issues addressed by the modern state.

    True, that’s why all a randomly-selected body can hope to do is adjudicate between opposing evidence/arguments from (elected) persons who know what they are talking about.

    >This [above] problem might be alleviated by by having each body selected by sortition address only a narrow range of issues. But then there would be serious problems of coordination between them. Moreover, groups addressing one area of policy might neglect important trade-offs between that issue and others.

    Yes, and that’s why sortition could only ever be a part of a mixed constitution, with the coordination role performed by a body appointed by a different mechanism. Naomi has argued that this should be the elected (parliamentary) government, whereas I prefer to arrogate the role to the bureaucracy, as a result of which the finance secretary would become de facto ‘prime’ minister. The increased power of the unelected bureaucracy would be offset by the fact that ministers would in turn be accountable to the democratic body (as was the case with magistrates in the Athenian democracy) and subject to removal from office by censure motion.

    >If, as in most proposals, the participants are expected to hear presentations about policy issues and engage in deliberation about them, there are many ways to bias the choice of presenters and the selection and framing of issues.

    Yes, that’s why (sensible) kleroterians argue that advocacy rights should be reserved for those selected by election (or direct-democratic initiative).

    >But the Athenian experience is unlikely to be a good model for us, because there are many crucial differences between ancient Athenian government and the modern state. Most notably, the former was a lot less complex than the latter, and handled a much narrower set of issues.

    Yes, the right-wing case for sortition does presuppose a reduction in the purview of the state. I think ​Somin’s​ post indicates the danger of proposals for “pure” sortition discrediting those of us who argue that it should be used more selectively — as a supplement to existing electoral systems.


  5. Keith,

    By limiting sortition to a final yes/no voting jury, you forfeit several other important benefits of sortition. Two that seem most critical are control of the agenda (what topics merit having legislation drafted and presented to a jury), and the benefits of diversity in deliberation gathering diffuse civic knowledge and insights in formulating optimal proposals for the juries to vote on. However, these benefits tread on each other’s toes and become ineffective if combined within a single body (whether elected or selected by lot), which is why I favor multi-body sortition and eschew the mean-spirited contentiousness of electoralism that promotes elite domination with negative “us vs. them” ripple effects throughout the rest of society. The point is that elections are not a neutral method of selecting “advocates.” Electoralism is as harmful as sortition is beneficial.

    Liked by 2 people

  6. Terry:> The point is that elections are not a neutral method of selecting “advocates.” [my emphasis]

    Of course not, elections are intrinsically partisan. If you hire a lawyer to represent you in court you would be rightly disappointed to find that her advocacy was neutral, ditto with the politicians you select to represent your political beliefs. This may be distasteful to the sensibilities of those offended by anything “mean-spirited and contentious” but we have yet to find a better way of arriving at the truth than adversarial combat between partisan advocates.

    The best explanation as to why this is the case can be found (ironically) in Helene Landemore’s “argumentative theory of reasoning”. According to Landemore and the evolutionary theorist Hugo Mercier, we don’t engage in reasoning and deliberation in order to update our own knowledge. There are two distinct mechanisms involved, convincing others (as jaw-jaw is less costly than war-war) and evaluating the arguments of others (rather than extracting the truth via torture). The confirmation bias and similar cognitive mechanisms are invaluable in the former case and dysfunctional in the latter, hence the need for the rigid separation of personnel in the court room and legislative jury (that we both endorse).

    Your agenda committees, however, rely on the classical (and discredited) theory of reasoning (updating one’s own beliefs). The examples of the value of cognitive diversity that Helene provides in Democratic Reason are unpersuasive. In the case of the New Haven neighbourhood committee (p.101), it only took two persons to find the optimum solution to the problem of mugging (solar lighting) and this could have been achieved by crowd-sourcing, competitive public competition etc. In the case of the three French députés (p.99), it’s harder to think of a less diverse group than elected politicians. So all that leaves is the Twelve Angry Men (p.98) and that’s just a Hollywood movie. Advocates of the benefits of deliberation for “gathering diffuse civic knowledge and insights in formulating optimal proposals” have a whole lot of work to do, especially now that the much-toted Hong and Page “Diversity Trumps Ability” theorem has been dismissed as an empty truism “that has neither mathematical content nor real-world applications”. (Abigail Thompson, “Does Diversity Trump Ability? An Example of the Misuse of Mathematics in the Social Sciences”, Notices of the American Mathematical Society, 61, 9, pp. 1024-30)

    P.S. It’s also worth noting the somewhat unworldy assumption underlying Helene’s case for democratic reason:

    “[T]he epistemic framework of the argument presented in this book assumes that people are voting what they think is right for the common good, no matter how unpleasant it is for them, whether ideologically or economically.” (Landemore, 2013, pp. 196-7, my emphasis).

    So if you can’t rely on Hong, Page and Landemore, what does that leave you — Aristotle’s pot-luck meal? Even the relevance of that for deliberative democracy has been called into question (Cammack, 2013).


  7. PS Helene’s evo-psych based theory would suggest that the adversarial conflict that characterises the political domain is a product of human psychology as opposed to “electoralism”. I’ve always been puzzled by her advocacy of an innate and dualistic mechanism like the argumentative theory of reasoning as it weakens her case for a collaborative approach to problem solving (which would suggest the classical model of reasoning as a way of updating one’s own beliefs in the light of the contribution of others). It’s also of interest that her “selective genealogy” for the talking (rather than counting) approach to collective deliberation struggles to find any advocates between Aristotle and J.S. Mill (neither of whom would pass muster as democrats). All the other sources — ancient, medieval and modern — fall into the wisdom of crowds (counting) rather than collective wisdom (talking) tradition, so “deliberative democracy” appears to be little more than a normative aspiration by a handful of very recent authors. I’ve sent Helene the chapter of my thesis dealing with this and I’m awaiting her response.


  8. Keith,

    I subscribe to the theory that reason evolved with most argument (as a form of deliberation) focused on trying to win, rather than revise one’s OWN opinions and seek truth. However, that theory also asserts that the other side of the coin is that reason evolved to become a better bullshit detector and EVALUATE the arguments of others (for the purpose of making best decisions and finding the “truth.”) So a key factor for effective deliberation to formulate optimal policies is to be sure that most of the participants do NOT enter the deliberation with preconceived outcomes. Thus ELECTED deliberators who have SOUGHT legislative office are LEAST suited to formulation of new syntheses as proposals. Yes, this indicates the importance of having non-author jurors listening to pro-con arguments for a final decision, but ALSO the value of having random citizens (rather than partisans) deliberating (and listening to conflicting experts) to help formulate proposals.


    Liked by 2 people

  9. Terry,

    I think we agree on the two different cognitive mechanisms involved but the first (persuasion) pertains to talking and the second (bullshit detection) pertains to listening. The collaborative give-and-take/brainstorming that pertains to your agenda councils presupposes a cognitive mechanism for truth seeking. Unfortunately such a mechanism doesn’t appear to exist, hence my claim that deliberative/epistemic democracy is just a set of ideal procedural norms that fly in the face of human nature. The fact that the small group sessions of the DP are so reliant on trained moderators bears this out as it would appear that ordinary people are just as prone to the confirmation bias (as soon as they open their mouths) as partisans. However if you sit in silence to judge the speech acts of others prior to voting in secret you are less likely to be enslaved by Forster’s dictum (how do I know what I think until I hear what I say?). Once you’ve said what you think, it’s a lot harder to change your mind.

    I would be curious to know your views on the Thompson paper, as you rely heavily on Hong and Page (although I think you said you haven’t actually read The Difference).

    Liked by 1 person

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: