The Luck of the Draw: The Role of Lotteries in Decision Making

I’ve been commissioned to write an ‘in brief’ review of Peter Stone’s new book (OUP, 2011) for Times Higher Education but wanted to bring up a couple of points here that I can’t squeeze into their measly 60-word limit. The book is an attempt at a theoretical clarification of lotteries as an equitable method for the ‘allocation of [scarce] goods’ and ‘assignment of responsibilities’ (both wanted and unwanted) (p.13), Peter’s thesis being that the distinguishing feature of the lottery is its ‘sanitizing effect’ (p.16). This is on account of the essentially arational nature of the lottery – it serves an entirely negative function by shielding the decision process from reasons of any kind (good or bad), therefore protecting it from partiality and corruption.

A lottery is a process capable of generating a set of outcomes, in which the particular outcome to be expected whenever the process occurs is unpredictable given available information (p.20).

Much of the book deals with allocative justice and covers similar ground to Barbara Goodwin’s Justice by Lottery but from a rigorous theoretical perspective that is hard to disagree with (presupposing certain Rawlsian assumptions).

The chapter on sortition, however, sits less easily with this essentially negative thesis. Peter clearly agrees with Oliver Dowlen (who he cites at length) that the prime purpose of sortition is to insulate the political process from factionalism and corruption. Ollie’s book (The Political Potential of Sortition) denies that sortition has anything to do with representative government, for reasons of anachronism and definition. In the first case sortition as a means of descriptive representation is ruled out because the Greeks (who invented sortition) had no (mathematical) concept of proportionality; in the second case representativity is ruled out by the (somewhat circular) argument that as the lottery is based on arationality then it cannot subserve any rational goal (such as democratic representation).

Peter, however, has a harder task, as he acknowledges that sortition has three normative goals (justice, incentive alignment and descriptive representation). Although these are all positive functions, nevertheless he feels constrained (for theoretical consistency?) to subsume all three functions under one (negative) explanatory principle: ‘all [three] ultimately appeal to the sanitizing effect of lotteries’ (p.124). This is unconvincing: taking the last function first, Peter acknowledges (following Dowlen) that it would be better to specify the particular characteristics that are felt to be necessary for an assembly to be descriptively-representative (age, sex, occupation, education or whatever) and then to ensure proportionality by stratified sampling. There is no need for the sampling method to be random, volunteering or even an auction would suffice, we only randomise in order to ensure that each stratified sample will be ‘sanitized’. This is, at best, a ‘weak’ example of the lottery principle. However the call for a legislature to be a ‘portrait in miniature’ is an end in itself – both a normative democratic ideal and also on account of the growing evidence that the (epistemically) best decisions require as much diversity as is possible. But (as Peter acknowledges) stratified sampling cannot achieve this as the list of relevant ‘diversity’ characteristics is arbitrary and combinatorially infinite so the only way of arriving at such an assembly is by sortition. He acknowledges in a footnote (p.176, n.27) that stratified sampling would not be necessary in a legislative-size assembly with mandatory service.

The descriptive-representation function is an entirely positive role for the lottery – it is a reason, so cannot have anything to do with arationality and resists being shoehorned into a single-factor scheme just for the sake of explanatory consistency. The (negative) sanitizing effect has to do with the second benefit: ‘incentive alignment’. This is the role of the lottery in preventing the domination or corruption of a legislature by partisan forces (the ‘rich and powerful’ in the parlance on this forum): ‘By preventing the “wrong” reasons from influencing either those who hold office or what officeholders do, the integrity of the ultimate decisions is protected’ (p.129). If only – as Peter acknowledges: ‘This protection does not last once the officials are identified and reachable’ (p.128); ‘A lottery can only sanitize the stages of a decision that have been undertaken up to the point at which the lottery’s outcome becomes known’ (p.154). If this is the case then the prime role of sortition (according to the Downlen-Stone thesis) in preventing corruption isn’t going to work very well, given that legislative members would normally hold office a lot longer than trial jury members and would be easily identified (and corrupted). Far from ‘sanitizing’ the political process, sortition has the potential to open the overflow sluices at the sewage works, as allotted members would not be constrained from offering their services to the highest bidder by manifesto commitments, party discipline or the need to be re-elected. As I’ve argued before (following Pitkin, 1967), this effectively limits the role of an allotted assembly to voting in secret, a process that is not open to corruption.

In sum, it would appear to the present reviewer that the attempt to shoehorn sortition into a theoretical framework designed around considerations of social justice is a doomed project.

Advertisements

27 Responses

  1. By way of clarification: sortition is a hybrid creature based on two completely distinct principles:

    1) The (Dowlen-Stone) Lottery Principle
    —————————————————–
    This is a negative principle which applies to the appointment process only. Because of the arationality and unpredictability of the lottery, the appointment process is “sanitized” from factional influences and corruption. As Peter acknowledges, this only applies until the results of the lottery are published, from that time onwards members of an allotted assembly are open to corruption (and are unconstrained by manifesto commitments, party discipline or the need to be re-elected).

    2) The Sampling Principle
    ———————————-
    This is a positive principle, which applies (after the lottery draw) to the day-to-day workings of a sortive body. There are good normative and epistemic reasons for wanting a legislature to be a “portrait in miniature” of the whole electorate. In principle this could be achieved by (non-random) stratified sampling, were it not for the combinatorial complexity involved — the necessity for the sortive body to contain the proportionate number of 30-40-year-old male, graduate (2:1, liberal arts, redbrick university), second-generation EU immigrant, dental-technician, Liberal Democrat-voting Guardian readers with two children (one of which is privately educated), living in a mid-size North Eastern town, no longer churchgoing but with a residual Catholic concern for social justice in the developing world. The ONLY (practical) way of ensuring statistical representation is via sortition, but this presupposes mandatory service.

    Postscript: The Rotation Principle
    ——————————————–
    The Sampling Principle only applies to a collective body — a legislature, advisory council or citizen jury. The (archaic) practice of appointing individual government officers by allotment is justified by the Rotation Principle (govern and be governed in turn) and is only applicable to small communities in which no specific executive skills are involved in the allotted task. As such it is really only of interest to historians (or the inhabitants of communes).

    Much of the confusion in the sortition debate is caused by failing to differentiate between these principles. Dowlen denies the Sampling Principle; Stone acknowledges it but (wrongly) attempts to subsume it under the Lottery Principle; many commentators on this forum overlook the temporal factor — the Lottery Principle only applies before the draw; the Sampling Principle after it; those who advocate the Rotation Principle fail to acknowledge the validity of the principle of the separation of powers.

    Like

  2. > Because of the arationality and unpredictability of the lottery, the appointment process is “sanitized” from factional influences and corruption. As Peter acknowledges, this only applies until the results of the lottery are published, from that time onwards members of an allotted assembly are open to corruption

    You are conflating two things here – corruption-free process and uncorrupted delegates. Any process that is completely mechanical is corruption-free, in the sense that it is not affected by the political influence. At the same time, a corruption-free process can easily produce corrupt delegates. For example, if the background level of corruption (i.e., the average level of corruption in the population) is high, then sortition (ideally, a corruption-free process) would likely produce corrupt delegates.

    Like

  3. I agree that descriptive representation can only mirror society as it is, hence the need for a “sanitizing” appointments process, coupled with checks and balances (separation of powers), as it would be naive to rely on some notion of uncorruptable delegates (a la Robespierre). The background level of corruption in society is an issue for another forum (political, religious or whatever). I’m not sure what you mean by “a corruption-free process can produce corrupt delegates” — allotted delegates are, by hypothesis, uncorrupt (in aggregate) at the time of appointment, but are then dangerously open to corruption (individually) as unconstrained by manifesto commitments, party discipline and the need to gain re-election. So the status of the allotted delegate can change from uncorrupted to corrupted over time. This is why I argue that allotted delegates should have no individual role, only a collective one (secret voting). In fact the concept of an individual descriptively-representative delegate is oxymoronic; there are only allotted delegates (plural). Perhaps, in order to avoid confusion, we should stop talking about allotted delegates and only refer in future to an allotted chamber, in order to avoid the temptation to believe that every plural entity necessarily entails a corresponding singularity.

    Or are we using the word “corruption” in two different senses? My meaning is the standard republican one of sacrificing the general interest for the sake of personal or factional ones. In this sense the secret votes of an allotted chamber are by definition uncorrupted because the aggregate will (as expressed by votes) of the microcosm is the same as that of the whole electorate. This is true in the analytical a priori sense — if the will of the microcosm wasn’t the same as that of the whole there would be something wrong with the sampling process.

    Like

  4. Sorry, that was me (Keith). My browser seems to have gone all anonymous on me (Mozilla upgrade downloaded yesterday).

    Like

  5. > allotted delegates are, by hypothesis, uncorrupt (in aggregate) at the time of appointment

    As I already explained above – this is untrue.

    The allotted delegates may – at the time of selection – be committed to promote special interests at the expense of the general interest. For example, their own interests, their families’, or those of some elite that they expect to collaborate with. The fact that this commitment is prevalent in the population doesn’t mean it is not corruption.

    (They may also be uncorrupted – in the sense of intending to promote the general interest – but still promote special interests due to incompetence.)

    Like

  6. What you are requiring of delegates is that they achieve the superhuman feat of transcending their own interests in order to divine the general will. This strikes me as both unrealistic and unnecessary — even Rousseau acknowledged that statistical representation was a way of aligning the general will with the (aggregate) will of all:

    “The [general will] looks only to private interest and is only the sum of particular wills: but take away from these same wills the pluses and minuses which cancel each other out and the general will remains as the sum of the differences.” (Rousseau, Social Contract, 1984, 76).

    If citizens were individually capable of divining and promoting the general interest there would be no need for descriptive representation. But in the postlapsarian world nobody can take the god’s eye view, the best you can hope for is a statistical aggregation of individual views, beliefs and interests, hence the need for sortition.

    The difference between your position and mine is the difference between classical and modern notions of republicanism. The former is dependent on the Machiavellian notion of virtu as a moral foundation for the collective good (which he romantically believed to be the “humor” of the populo); the latter presupposes a Harringtonian conception of conflicting interests as the dominant force and the resultant need to construct a mechanism to establish equilibrium (if you divide the cake then you don’t choose the first slice). Classical republicans deride the modern view as mechanistic, to which we would simply retort “get real”.

    Like

  7. > What you are requiring of delegates is that they achieve the superhuman feat of transcending their own interests in order to divine the general will.

    For an individual, representing the general interest is indeed difficult, but for a group of a few hundred people, all it takes is a commitment not to abuse the group’s power. I think such a commitment is quite common.

    Furthermore, I am not aware of any other basis for a representative government. In a society where such commitment is generally absent, there is no reason to assume that representative government is possible.

    Like

  8. This presupposes the negative (Dowlen-Stone) argument for sortition as a way of sanitizing the appointments process, but I don’t understand in what sense you claim such an assembly to be representative. Rousseau argued that, in principle, it only needs one person to divine the general will, although in practice it would be better to have many or (ideally) everybody. But why do you claim that the general will is an emergent property of group deliberation? How many people are required and what is the operative principle — some Kantian notion of rationality, or are you satisfied with proceduralism?

    Leaving aside the necessity for sortition to sanitise the appointments process there is no reason in principle to use sortition to establish the sort of deliberative chamber that you are advocating. It would be perfectly possible to simple exclude the corrupting influence (the rich and powerful) and then ask for volunteers or use some other non-randomised method. The value of sortition from the point of view of descriptive representation is to ensure a microcosm that is a reasonable match for the whole electorate in terms of interests, beliefs and attitudes (age, sex, income, location, occupation etc. in effect being proxies for interests/beliefs/attitudes). The sort of deliberative forum that you are advocating does not require what deliberative theorists call a “mini-populus”, hence the fact that they tend to be more relaxed about volunteering than those of us who privilege representative democracy.

    What you appear to be advocating in your last paragraph is classical republican government, but I fail to understand why you want to call this “representative”. “Representation” (in the descriptive sense) means “to make present again” (re-present), so a legislative assembly established on this principle would just be a portrait in miniature, warts and all’ of the larger society. You may not find that idea particularly appealing, but I’m afraid that’s what the word means. There are many other forums for the discussion of republican ideals, but it’s confusing to use the word “representative” when you mean “republican” (this was also Madison’s sleight of hand, although he did it the other way round and he was referring to elective rather than statistical representation).

    Like

  9. > I don’t understand in what sense you claim such an assembly to be representative.

    As always, I refer to representation of interests. It is the only sense that matters.

    > How many people are required

    As I already indicated, there are two opposing considerations here, requiring finding some sort of a happy middle. The right value would of course depend on the circumstances. For high power bodies, a few hundred people seems right.

    Like

  10. >As always, I refer to representation of interests. It is the only sense that matters.

    But you’ve just denied that, by positing the construct of the “general interest” and arguing that the task of an allotted body is to rise above the plurality of individual interests — “their own interests, their families’, or those of some elite that they expect to collaborate with”. I don’t deny the validity of the concept of the general interest as a normative goal that deliberators should aspire to, however in practice we usually have to make do with the statistical aggregation of interests (albeit tempered by appeals to the “forceless force of the better argument”). The representation of interests (plural) is a goal of modern (liberal) republicanism and is achievable by sortition, whereas the general interest is a mystical emergent property, beloved of classical republicans, and does not require either sortition or a large deliberative body, merely sufficient civic virtue in the representatives (who may be chosen by any means). The requirement for “a few hundred people” is derived from statistical sampling and is the order of magnitude for a microcosm to be a reasonably accurate reflection of the salient interests in the whole population. I would imagine that a figure of between 12 and 24 civic-minded, non-elite persons would be more suited to the discursive process that you have in mind — a process that has nothing to do with democracy or representation in the normal sense of the word.

    Like

  11. > the task of an allotted body is to rise above the plurality of individual interests

    No. The task (in fact, of course, one of the tasks) of the allotted body is to avoid serving its own interests as a group. This is a simple act of basic solidarity or fairness, not an act of transcendence or ennoblement.

    Like

  12. I have never quite understood how Sortition would deliver better government, but perhaps the present Corruption might have something to do with the Process of becoming a prone to corruption?

    Failing that, we could always draw on the example of the Doge of Venice, who once selected, in part by a lottery, was forced to live in a closed chamber. All his communications were monitored and recorded, just to ensure no undue influence.

    Makes you wonder why anyone wanted the job of Doge!
    delegate?

    Elective democracy, as we well know, leads to corruption, because of the need to raise funds in order to be elected.

    Sortition *may* lead to less corrupt, more public-spirited delegates. It seems to (mostly) work with juries. There is also evidence that people can be cajoled into cheating less. This comes from Experimental Economics, where the usual bunch of US students were observed: They were given an easy cheating opportunity. Those who were asked to read the 10 Commandments (!) beforehand were less prone to cheat. (I read about this in Ariely: Predictibly Irrational)

    So maybe we can hypothesise that the Proces of randomly selecting delegates (unlike ED) is not in essence corrupting, and with the right conditioning the sortitional delegates should be much less

    Like

  13. Sorry, that last post got garbled: It should have read

    I have never quite understood how Sortition would deliver better government, but perhaps the present Corruption might have something to do with the Process of becoming a prone to corruption?

    Elective democracy, as we well know, leads to corruption, because of the need to raise funds in order to be elected.

    Sortition *may* lead to less corrupt, more public-spirited delegates. It seems to (mostly) work with juries. There is also evidence that people can be cajoled into cheating less. This comes from Experimental Economics, where the usual bunch of US students were observed: They were given an easy cheating opportunity. Those who were asked to read the 10 Commandments (!) beforehand were less prone to cheat. (I read about this in Ariely: Predictibly Irrational)

    So maybe we can hypothesise that the Proces of randomly selecting delegates (unlike ED) is not in essence corrupting, and with the right conditioning the sortitional delegates should be much less

    Failing that, we could always draw on the example of the Doge of Venice, who once selected, in part by a lottery, was forced to live in a closed chamber. All his communications were monitored and recorded, just to ensure no undue influence.

    Makes you wonder why anyone wanted the job of Doge!
    delegate?

    Like

  14. >The task (in fact, of course, one of the tasks) of the allotted body is to avoid serving its own interests as a group.

    A noble republican sentiment, but it would be true whatever the appointment mechanism. The goal of sortition (as a mechanism of representation) is to align the interests of the group and the interests of the larger population. If so then a successful allotted body will be most effective (and uncorrupted) when it serves its own group interests.

    Like

  15. Conall: I agree with you (and Ollie and Peter) that sortition can only possibly sanitise the appointments process; to keep pure an allotted assembly with an active role you would need to lock up the members (and all their friends and relatives) and throw away the key. It’s interesting to hear the lengths that the Venetians had to go to maintain purity as this certainly gives weight to this argument.

    As for the epistemic benefits of an allotted chamber this is standard wisdom of crowds stuff (Surowiecki, Landemore, Page, Estlund etc). Nothing to do with sortition per se, only diversity. But note the wording “wisdom of crowds” — this is an aggregate claim, rather than a claim for the wisdom for any of the individuals that go to make up the crowd. Most of the people in an average crowd are of an average level of stupidity and I’m at a loss to understand why anyone would wish to privilege the ideation of an individual member of an allotted chamber just because she happened to draw the golden ticket.

    Like

  16. Conall,

    > I have never quite understood how Sortition would deliver better government

    Isn’t it reasonable to expect that a government that is made of average people would represent the interests of the average person better than a government that is made of an elite (electoral or otherwise)?

    Like

  17. > The goal of sortition (as a mechanism of representation) is to align the interests of the group and the interests of the larger population.

    Yes – it is a goal of the sortition mechanism, not an automatic result. In a society in which the background corruption level is high, this goal will not be achieved.

    Like

  18. Yes we would both agree that a mere mechanism cannot turn a pig’s ear into a silk purse, so sortition cannot achieve the aims of classical (virtue-based) republicanism — for that lofty goal you would need the sort of social revolution that you have consistently advocated. That’s why I’m aiming for the modest liberal goal of ensuring that the legislature should reflect the overall balance of interests and opinion in society in general. This is what is meant by the term “representative government”; we would both agree that Madison was being disingenuous when he described representative government as “republican”.

    What still puzzles me, though, is why you feel that a site dedicated to sortition is the right place to debate republican ideals. Perhaps we have different perspectives of the meaning of the word “equality” in the title of this blog — to me it’s a reference to equality of opportunity (we all have an equal chance of drawing the golden ticket), whereas your position seems more akin to equality of outcome, which would require something like Barbara Goodwin’s utopian vision (dystopian in the eyes of liberals) of the Total Social Lottery. There’s little hope of agreement regarding such fundamental normative issues, but at least if we clarify our use of words then we won’t have to waste such a lot of time in fruitless exchanges.

    Like

  19. “Politicians are born innocent, free from undue influence, but are everywhere chained to the self-serving vested interests”. (to parody Rouseau).

    Sortitional delegates would be easy meat compared to our MPs, who at least have to prove themselves in the ordeal of selection and election. And not just from the outsiders – consider the power of the civil servants who ‘advise’. They will be steeped in conventional wisdom.

    Like

  20. > Sortitional delegates would be easy meat compared to our MPs

    But the MPs are the predators, not the prey. That is, they are not manipulated by special interests, they are the special interests.

    The relevant question is: “who is easier meat, the allotted delegates, or the public at large?”

    Like

  21. The public at least has a method of redress and throws the rascals out every few years. And you are being a little uncharitable to backbench MPs who are often well motivated, until they find that they are part of the “dignified” constitution, without efficient power. This then obliges them to climb the greasy poll, with all the attendant moral hazards. It’s hard to imagine why anyone with half a brain would want to stand for parliament, given it’s current status, unless they were motivated by a sense of duty and service. I’m not very familiar with the US system but imagine there are very significant differences involved in the status and motivation of elected politicians on either side of the pond (in Conall’s home country I believe that seats for the Dail are in effect hereditary).

    Like

  22. The Athenian mechanisms for reducing corruption (appointment by lot as late as possible, .e.g., for the chairman of the prytaneis only on the morning of the day served; euthyne audits upon leaving office) seem to have worked reasonably well.

    Like

  23. If Ireland’s economic situation continues to deteriorate, I wonder for how long Dáil seats will continue to be effectively hereiditary. Already, in the election last February, Sinn Féin did remarkably well. I expect that trend to continue.

    Like

  24. I’m sure the mechanism worked well, but the need to leave the appointment until the last minute confirms that sortition is only an effective sanitisation method ex ante. This is why I’m puzzled that Stone and Dowlen argue that there is only one lottery principle (guarding the integrity of the political process). At the recent sortitition conference in Paris, Yves Sintomer argued that there were at least five — divination, sanitzation, rotation, descriptive representation (and another that I forget). These mechanisms are entirely independent and cannot be subsumed under one over-arching principle. To the Athenians, rotation was the principal merit of the lot.

    Like

  25. A lot of people, estimating independently and unemotionally about a quantitative matter of which they have direct experience can usually come near to the right answer. But in very many cases of great importance, especially where fear is involved or they have no sound experience, they get it very wrong. A poll in Australia showed that people thought about 30% of asylum seekers were arriving by boat. The actual figure is less than 2%. A lot of support for Hitler was based on similar errors about Jews in high places.
    The wisdom of crowds is a complete misnomer. It is not a question of wisdom but of estimates about quantities. Klerotarians should have nothing to do with it. What is true is that ordinary people are capable of reflective and critical judgement, given a task they are equipped by their experience to handle.
    The key to sortition is the need for trust between the public at large and the group who make decisions on their behalf. That trust is based on the identity of interests between a genuine sample of the population and the population as a whole. Anybody who is honest with herself must recognise that her ill-informed and untested opinions on most political matters are not worth much. What one would want is a thoroughly discussed decision by people who share the range of interests involved and are motivated to find a fair compromise between them.

    Like

  26. John, I don’t think any of us are advocating legislation by uninformed opinion: we all agree that ordinary people are capable of reflective and critical judgment. The only thing we differ on is as to what motivation and experience is required — your proposal (for volunteering) suggests that individuals are the best judge of their own suitability whereas others would argue for jury-style selection. Agree that the wisdom of crowds is a misnomer, what we all want is that ordinary people should be empowered to exercise informed judgment.

    Like

  27. I’ve found my notes on Yves Sintomer’s argument that there are five lottery principles:

    1) Revelation of divine will
    2) Impartiality — sanitizing the political process
    3) Rotation (anybody can govern)
    4) The jury principle (anybody can judge)
    5) Descriptive representation

    Yves’ claim is that these five principles are both exhaustive and independent — both logically and historically. The Greeks relied on 2, 3 and 4 (recent scholarship has discounted 1); whereas modern advocates of sortition are mostly concerned with 2, 4 and 5.

    1, 3, 4 and 5 are positive functions whereas 2 is negative so I’m still puzzled as to why Peter wishes to privilege 2 and to argue that all principles can be subsumed under this single “lottery principle”. Whilst it’s true that 1, 3, 4 and 5 can be realized using other methods (examining chicken entrails, stratified sampling etc), that doesn’t preclude the use of sortition to achieve the same effect (and the same can be said for 2, especially on account of its ineffectiveness ex-post). Yves’ conclusion was that the lottery was simply a selection mechanism that could be put to a variety of uses.

    Like

Leave a Reply

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

WordPress.com Logo

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

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

%d bloggers like this: