DOWN WITH ELECTIONS!

Part 1 Part 2 Part 3 Part 4 Part 5 Part 6

PART 5

The “Loose-Leaf” Approach

I have touched on the fact that elected politicians may become “locked into” a bad policy. (see §20 above) It is not simply a question of damage to the ego, or the embarrassment of publicly changing their stance after several TV or newspaper interviews. There may be the inertia of a whole political and publicity machine brought into being to push a policy which was once thought to be important for gaining office, and which later turns out to be doubtful or disastrous. (Examples include climate change denial, the “war on drugs”, and the search for weapons of mass destruction in Iraq.)

Changing policy means that the whole team of bullocks: tame journalists and editorialists, other politicians of the same party, public relations staff, and grass-roots party members have to change direction as well. A large part of the public may need to be convinced, not only of the need for the change, but also of the politicians’ sincerity in advocating first one policy, then another. Some sources of party finance may dry up overnight and other sources may have to be found. To make the change, some heads may have to roll, and there will not be many volunteers to rush in to make the sacrifice. All this is risky politically, and time-consuming.

An Assembly chosen by lot will not suffer from these disadvantages, of course. Members will be quite free to change their minds, and in any case, with a secret ballot, there is no need for anyone else even to be aware of a change of mind. This will permit errors of judgement to be quickly corrected, and policies which are no longer desirable because of changed circumstances to be altered.

This would facilitate a “loose-leaf” approach to legislature. In this, the entire corpus of law could be thought of as being similar to a loose-leaf folder. Laws would be drafted in such form and language that amendments could be made in a way that parallels the removal and replacement of one or two pages from a folder, without invalidating the rest of the law.

Accountability with Elections and Sortition

Accountability is often held to be crucial in an elected parliament. The problem is that it just doesn’t happen. As Terry Bouricius and David Schecter, discussing the elected government of the US, say:

“Holding elected legislators accountable is crucial because elected legislators generally have different interests than the people they represent. Having one opportunity every 2-6 years to “throw the bums out,” when legislators may vote on hundreds of bills each year, is an accountability mechanism that cannot possibly work.”ⁱ

It is sometimes assumed, by those who have not sufficiently thought about sortition, that accountability is always necessary, and that either some means must be found to hold members chosen by lot accountable for their actions, or else that it is a grave and irremediable failing of an allotted body that it does not provide accountability.

Étienne Chouard, an advocate of sortition, says in his proposalⁱⁱ:

les tirés au sort . . . sont surveillés en cours de mandat et révocables à tout moment,

(“those chosen by lot . . . are supervised during their mandate and may be removed at any time”)

and

ils sont évalués en fin de mandat, et éventuellement sanctionnés ou récompensés.

(“they are judged at the end of their term, and possibly punished or rewarded”)

We must ask who would have the moral right to supervise them, or to judge them afterwards? On what basis? What would be the criteria? Who would have any legitimate claim to find fault with their voting?

Rewarded or punished how? And for what?

Chouard does not answer these questions.

Clearly, since the vote of the Assembly will be that of the population, if the population had the time to intensively study and debate every issue, that vote must be legitimate.

So any person or body finding fault with it will have special interests that are not those of the ensemble of the people. Such a person or body would have no moral right to overrule the Assembly.

Every member of an allotted parliament should be entirely free to vote as he or she wishes, without regard to the opinions of others. For this, a secret ballot in the parliament is necessary.

As noted above it should be illegal for a member to reveal how he or she voted, at least while still a member. This might perhaps be hard to enforce, but it has the advantage that the member would have a valid reason for refusing to answer questions by a domineering spouse – or by the pushy reporter in the street with the microphone.

One can only guess how “accountability” might be implemented, though we can be sure that it would find favour with those seeking to destroy the independence of Assembly members.

A tribunal, perhaps nominated, perhaps elected, of “experts in politics” might be set up to judge members. The subjects of its hearings would of course be known in advance, and media campaigns would be launched to influence tribunal members, who would have every reason to fear for their reputations and safety if they did not follow the wisdom of the talking heads on TV, the shock jocks of pop radio, and the editorials and articles of the tabloids.

Members of the Assembly likewise would soon learn that they had no independence, that an honest vote in accordance with their own principles could sentence them to a campaign of harassment and vituperation by these same worthies, in addition to any sanction inflicted by the tribunal. Furthermore, since the voting in the Assembly could not then be by secret ballot, members would be liable to threats and intimidation by extremists.

In short, then, “accountability” is neither necessary nor desirable with sortition. It is in fact completely incompatible with it.

The Knowledge and Skills of an Assembly Chosen by Lot

We have seen above that the justification for an “allotted” Assembly is that, even though there will be a slight difference between sample and population, it is fair for the sample – the Assembly – to “represent” the population (using the word “represent” this time in the sense of “act for”) because the resemblance will be far closer than it would if the sample were chosen by non-random means, which is to say by any other means. A legislative body chosen by lot will vote as would the population as a whole, all other things being equal.

All other things, however, will not be equal. Assembly membership will after all be a full-time occupation. Members will have much more time to study the issues in question than the rest of us do. They will not have to spend most of their time earning their living in another occupation – or looking for a job – with the worries which that often brings, and they will have access to the best information available. Debating the issues, and listening to arguments for and against put forward by other members will help them to form an educated opinion. Most importantly, Assembly members will have the benefit of the work done beforehand by the members of the Single-Issue Policy Committees, who will work full-time on their issue, with the ability to call for expert opinions. Consequently, the Assembly will vote as the population would if everyone had an enormous amount of time available to study each proposal, and the best knowledge possible of the issues.

One of the criticisms of sortition that is sometimes advanced is that “ordinary” people do not have the necessary knowledge and skills to make decisions. It is true that many professional politicians have some formal training: law school in the USA, the École Normale d’Administration in France, or the schools of Politics, Philosophy, or Economics in the UK.

But what are these skills? What advantage do they bring?

There are two types of knowledge and skills involved: first, the knowledge and skills needed get a policy or a proposal made into law, and second, such technical knowledge as is needed to understand the issues involved in a particular proposal.

First, then, what skills are necessary to get laws passed? “Negotiating skills”, you might answer. But in an Assembly chosen by lot negotiating will not be required, since members will not be making deals. And really, what skills do successful elected politicians have? It is not overly cynical to reply “A glib tongue, the knack of avoiding awkward questions, a talent for raising money by fair means or foul, the readiness to make behind-the-scenes deals, and a good working knowledge of recent sporting events.”

In other words, the skills necessary to get elected and to get their way amongst colleagues, not those that are necessary for making responsible decisions, or that are beneficial to the nation after the election.

With sortition the need for these skills essentially disappears. The technical skills and legal knowledge required for putting a proposal into legal form, for making sure that laws do not contradict each other, will be held (much as at present with elected parliaments) by the parliamentary draftsmen, working under the direction of the Proposals Committee, and ultimately responsible to the Assembly.

The remaining knowledge necessary, a working knowledge of parliamentary procedure, is not particularly onerous to acquire, as the successful functioning of innumerable social and sporting clubs proves. The only person who needs a particularly good grounding here is the Speaker, and he or she will have four and a half years of experience in the Assembly in which to gain it.

As for the technical (engineering, scientific, medical, economic or whatever) knowledge needed for particular issues, that is what the SIPCs and the permanent Policy Committees are for. Those who say that “ordinary” people are incapable seem to forget that before gaining their expertise, the experts were ordinary mortals, like the rest of us. It is surely possible, then, for this knowledge to be acquired by other ordinary mortals.

“Ah, but that takes time”, reply the critics. “Professor Bloggs has spent forty years in his field.”

Certainly, and all of Bloggs’ knowledge and experience, or as much of it as is relevant, will be available to the Policy Committees, the SIPCs, and the Assembly unless Bloggs deliberately refuses to share it, which is unlikely, as most academics consider it both a pleasure and a duty to share their knowledge. Even if he does, there will always be other experts. In working with the policy committees, it may at times be frustrating for the experts to constantly have to explain to laymen things which are considered elementary in their field. This has the advantage, however, that they will be induced to re-examine their premises, and justify assumptions which are taken for granted by general consensus. Experts can be wrong, and sometimes a naive question will uncover a flaw in the accepted wisdom.

When technical matters are debated in parliament, elected politicians do not usually have specialised knowledge, any more than members of an Assembly chosen by lot. When the politicians honestly want to get the best legislation possible, and are not acting in accordance with some ideology, or in the service of special interests, they set up an inquiry, and call for submissions. That is exactly what will happen in the case of a new issue with a single-issue policy committee, with the difference that the SIPC members, being chosen by lot, will be impartial, which is very often not true with commissions appointed by politicians.

For matters which are of a general, persistent or recurring nature, it makes sense to have permanent Policy Committees, continually keeping abreast of the literature in their field, studying policies adopted or proposed overseas, and seeking opportunities to improve things generally.

Aptitude Tests and the Village Idiot

It is frequently suggested that members of an assembly chosen by lot should be subjected to a test of aptitude. From Chouard againⁱⁱⁱ:

“les tirés au sort sont soumis à un examen d’aptitude”

“those chosen will be subjected to a test of competence”

One wonders whether the advocates of aptitude tests have really thought the matter out.

The immediate consequence of a test of “aptitude” is to make an elite. If the test eliminates anyone – and after all that must be its purpose, to eliminate some people – it violates their civic rights, and makes the group remaining unrepresentative of the people as a whole. So such an assembly would lack legitimacy.

Certain questions must be asked:

Who decides what criteria to judge by?

Do we allow only those who have good character? Then does “good character” mean going regularly to Mass? Those who are good Marxists? Or good Muslims? Good patriots? Or what?

Should it be based on the opinions of those who know them? Then which people? The boss, the bank manager, the headmaster of the last school they attended? The head of the Taxation Department?

Do we go by some more objective criterion? Criminal records, for instance?

In the US it is estimated that, of those born in 2001, 6.6% of the total population will be incarcerated at some time in their life. For black males, the figure is 32% ^iv

Whatever their crimes (remembering that “justice” in the US is less than perfect, and many will be innocent) these people have a right to be heard, a right to have their interests taken into account. And what of those who are never punished for their crimes? Their interests would be taken into account with sortition, their votes are counted in elections, why should the unlucky ones who get caught be deprived of their rights when the unpunished are not?

Should it be a medical examination? Blood pressure, blood sugar and cholesterol levels, and all the rest? It would certainly be inelegant to have members indulging in epileptic fits or heart attacks in the Assembly instead of listening and debating. But removing epileptics or those at risk of heart attacks or strokes would make the Assembly unrepresentative, and in any case it is very simple to replace by lot a member no longer able to serve.

Is it to be an IQ test, or a test of “civic knowledge”? If it is the former, you bias not only towards the more intelligent, but also towards the wealthy; if the latter, towards those who have had a particular form of education, which may mean excluding immigrants, and may also exclude those who do not share the prejudices of the designers of the test.

The real question, though, is what is the purpose of eliminating the “less apt”? I suspect that there is a vague distrust of “ordinary” people, a feeling that “better” people will give (somehow) “better” decisions. One must ask “Better by whose standards?” The answer, unspoken, is probably by the standards of the intellectual elite to which the person proposing the test feels he belongs.

There may a fear that the lot could fall on the the morally depraved, or the “village idiot”. As far as the depraved are concerned, there is no objective standard for depravity, and no reliable way to test for it, so any idea of weeding them out is a fantasy. There is no shortage of them in elected governments

What about the so-called village idiot? In the first place, those who are severely mentally handicapped would be unable to take part for practical reasons. Secondly, those with mild intellectual disability, which we may take, as a rough approximation, as those with an IQ^v below 70 make up about 2.2% of the population. They will perhaps not contribute a great deal to debate, but there is no reason to deny them the right to vote in the Assembly. If we take the lowest, perhaps unfair, view of their abilities, we would expect their vote to be random “noise”, in the vocabulary of electrical engineers. If that is so, it would at least do no harm, as the “signal” – the choices made with intelligence – from the rest of the Assembly would be so much stronger.

“But what about those whose IQ is between 70 and 100 (the median IQ)? They can’t be too bright, can they?” Well, clearly the IQ test tells us that their IQ is lower than that of the rest of the Assembly. But as Bouricius and Schecter point out:

“It is not necessary for every member of a body to be intelligent and competent, in order for the group to function effectively. The intelligence of the group is magnified beyond that of any individual due to the cognitive diversity allowed by random selection.”^vi

This is discussed in the next section.

Incidentally, those who object to the idea of the mentally less-endowed being present in a legislature usually do not give much thought to the presence of alcoholics. Drunks are not unknown in elected parliaments, so we must expect them in any body chosen by lot. If their behaviour is disruptive, presumably the Speaker will have them ejected, which is exactly what happens in an elected parliament.

“Opt-in” and “opt-out “

Some proponents of sortition have suggested selection by lot from a pool of volunteers (those who opt-in), as happened in Athens. In effect, this is another form of aptitude test, of selecting an “elite”, (albeit rather inexact) by eliminating those who did not consider themselves competent. Hansen writes:

“the lot was based on voluntary candidature, which helps to eliminate those who had neither the talent nor the taste for administration. However, there must always have been some who had the taste but not the talent . . .”^vii

Obviously, too, there must have been some who had the talent, but not the taste, and who were left out, and these, the “upright persons who do not care to govern”^viii might well be those most desirable in a government. Their elimination would make nonsense of any elitist idea of getting “better” decisions by choosing from the self-selected.

Moreover, we could expect a much smaller, and thus unrepresentative pool if participants are obliged to opt-in rather than opt-out.

It is necessary, then, to make participation an “opt-out” rather than an “opt-in” choice.

The Jury Theorems

In “The Wisdom of Crowds” James Surowiecki says:

“If you put together a big enough and diverse enough group of people and ask them to make decisions affecting the general interest, that group’s decisions will, over time, be intellectually superior to the isolated individual, no matter how smart or well informed he is.”^ix

To back this up Surowiecki provides a number of examples where the success of a group was uncanny to the point of improbability, as in any given case one cannot completely rule out chance as a factor in this success. However, there is some good mathematical support for his ideas in the Jury Theorem(s).

The Jury Theorem, originally published by Condorcet in 1785, has been extended in the 20th century by several mathematicians, so that now we speak of jury theorems in the plural. These theorems have particularly interested political theorists seeking a robust justification of representative (electoral) democracy on “epistemic” grounds, that is to say based on a superior ability to make good decisions.

The argument for sortition made here so far is that it is preferable to elections because it better fulfills the criteria for democracy. It is an argument based on fairness and justice, not on efficiency. We are better off suffering from our own bad choices than from unjust decisions imposed on us by others. Nevertheless, the jury theorems help refute the notion of the supposed incompetence of the “lambda citizens”, the sometimes despised ordinary people.

A generalised form^x states :

“When a group votes, if the mean competence (propensity of a member to get the right answer) is greater than 0.5, the probability of the group verdict being correct approaches 1 as the number of members increases.

Conversely, if the mean competence is less than 0.5, the probability of the group verdict being correct approaches 0.”

In each case, the approach is quite rapid even if the competence is near 0.5, and gets faster the more it differs from 0.5.

The assumptions made in deriving this theorem are:

• Decision is by simple majority.
• The choice is binary (two options, as opposed to multiple choices), one of which is “correct” or better than the other.
• Independent decisions by the members of the group: no copying your neighbour’s answers, no blindly following the party line, though of course voters may agree, and still be independent.
• Voters do not necessarily have the same competence.

Some commentators, pointing to the existence of parties and ideologies, have considered that independent voting is impossible, and that this nullifies the epistemic justification of democracy.

James Hawthorne^xi has investigated the lack of independence in voting. It reduces the effective size of the assembly. In the extreme case where everyone follows some “guru”, the effective group size is one, and its competence is just that of the guru. In less extreme cases the effect is not disastrous, though independence is always desirable. For instance, if the amount of dependence is small, about 450 voters are necessary get the “right” answer with 80% probability if the mean competence is 0.52; if the mean competence is 0.53, the probability rises to 90%; for a mean competence of 0.6, the probability rises to over 99.9%. Hawthorne’s tabulated results show the dramatic rise in group competence as voter competence increases.

So far we have considered voter competence over 0.5. When voter competence is below 0.5, the probability of the group verdict being correct approaches 0. To put this in perspective, 0.5 represents the competence of voters who take a random guess. It is the result we would expect if we asked a tankful of tadpoles a question, and took the answer to be “yes” for those which swam right and “no” for those that swam left. So 0.5 is not a very high competence level. However, it is quite possible for competence to be below this. As an example, what would be the verdict if we had assembled a group of leading earth scientists in 1922 and asked them if the continents moved? That question can be answered with certainty because when Alfred Wegener presented his theory in America the academic audience was almost unanimous in ridiculing the idea, which wasn’t accepted until half-way through the 20th century. So the mean competence of the experts, in this case, was below that of our tadpoles. To be fair to the American scientists, the evidence in support of the theory was, of course, much less convincing than the evidence available to us now.

As an aside, some reasons for this may be easily guessed:

• Common background: All scientists present had received a similar education, with similar prejudices.
• Pressure to conform: Academic reputations – very important for job prospects and personal self esteem – would be damaged by accepting an “outlandish” idea.
• Personal factors: Wegener’s imperfect command of the English language, and perhaps anti-German prejudice in the aftermath of the Great War.

In the Assembly, secret voting means there can be no pressure to conform. We also have ensured diversity by the method of selection. What remains is to increase the competence (likelihood of making the best decision) of the members by giving them the best information available. This is the role of the policy committees.

List and Goodin^xii have extended the theorem of Condorcet to “plurality” voting in cases of more than two options. Their version may be expressed:

Suppose that there are k options and each voter has an independent probability of voting for the “correct” option which exceeds that of voting for any other option. Then the “correct” option is more likely than any other to receive the most votes.

(In the multi-option case, the expression “competence” is replaced by “probability of voting for the correct option”. It is not necessary for the voters’ probability of voting for the “correct” option to exceed 50%, it need only exceed 1/k for k options.)

List and Goodin tabulate some illustrative results, which show that the likelihood of reaching the correct decision rises very steeply as the probability of a voter choosing one option rises, even when the size of the group is as small as 51.

The importance of this is not as a justification for “plurality”, ie accepting the option that receives the most votes in the Assembly even when it is a minority of the total votes cast. Rather, it would be useful to the Assembly when evaluating a number of differing recommendations from a Policy Committee to know that the recommendation that received the most support was very likely the best.

Opinion Polls and Sortition

Critics have sometimes conflated sortition and opinion polls or surveys, and said that sortition is not to be trusted, since surveys often do not accurately measure what they are supposed to. For instance surveys of voting intentions often give a result which is not borne out by the results of the election. To answer this we should consider the difference between sortition, in which the members of a body are chosen as a random sample taken from the whole population, and the sort of sampling used in opinion polls.

Both are subject to “sampling error” or variability between sample and population. As noted above, in sortition the representation, although close, will not be perfect unless sample and population are identical, that is, until all the members of a population are included in the sample.

This error is certain to be less than it is in a survey, since it is very difficult in surveys to get a true random sample. To overcome this so-called “stratified sampling” is sometimes used. In this, those conducting the survey divide the population into “strata” (for instance “black” “white” “male” “female” “under 30” and so forth), and attempt to make sure that each stratum is present in the sample in the same proportion that one finds in the population. This is at best a palliative, since it is impossible to reduce people or their opinions – on perhaps hundreds of issues – to such crude measures.

In addition “non-sampling errors” occur in surveys but not in sortition. Such errors have been classified as follows:

Specification Error: for instance, in a survey the question doesn’t match what the researchers are trying to find.

Survey: “Will you vote for a candidate opposed to legal abortion?”

Response: “No”.

If the researcher assumes that the respondent is in favour of legal abortion, that might not be the case: the respondent might consider that all the candidates opposed to it were unacceptable for other reasons, or may intend to be overseas on the day of the vote, or may not wish to vote on such a sensitive issue, or may be ineligible to vote . . .

Specification error can also occur with referenda if the voters do not fully understand the implications of the question they are being asked to vote on. As an aside, this is sometimes the intention of those who formulate the question.

Frame Error or Coverage Bias: (some people are not counted, or counted more than once, or ineligible people are counted). Telephone surveys suffer from this because they don’t include those without a telephone, or those who do not answer when called. Internet surveys only include those who have access to internet, who happen to see the survey, and who can be bothered to answer. Street corner surveys only include those who are in that location at that time, and so on.

Non-response Error and Response Bias: Some people don’t answer the survey, or don’t completely answer. To this should be added the deliberately false responses given for any number of reasons: by those who are fed up with answering too many questions, or are grumpy about being woken up, or who just don’t like surveys, or who fear that a survey will be used in some way to their disadvantage . . .

Measurement Error: The classic case in surveying opinions occurs when the question influences the response. Compare:

“Would you vote for Bill Buggins?” with “Would you vote for a man who has been divorced twice, has been a bankrupt, and has a criminal record?” Of course, this is an extreme and grotesque example, but it is possible for researchers to influence the response quite unintentionally; or deliberately, but in a more subtle fashion.

It also appears likely that voters are often affected by the published results of an opinion poll, and change their vote accordingly.

Processing Error: This includes errors arising in processing the data, eg arithmetic, encoding, assigning weights in stratified sampling, and so on.

Again, since polls of voting intentions are always taken some time before an election, the results may differ from the votes cast in the election because voters change their minds between the survey and the election, for reasons that have nothing to do with any influence of the published results of the poll.

Finally, when the survey and the election results are different, there is always the possibility that irregularities in the election have caused the difference.

It should be clear that the method of choosing an assembly proposed here – a true random sample from the whole population – does not suffer from these non-sampling errors, nor from changes of mind by the public, nor, (above all!), from electoral fraud. Therefore it is quite incorrect to suggest that sortition will suffer from the defects of survey polls.

Filed under: Academia, Elections, Sortition, Theory |