In 2010 Iceland elected a constitutional assembly applying single transferable vote (STV). I analysed the election in a newspaper article (Election of persons—two elected). Later a periodical published by Institute of Public Administration and Politics, University of Iceland, published an analysis by Þorkell Helgason (ÞH) of the election results. I have commented the analysis. In the following two points of the comment are retold (notes 2 and 3).
1. Election of persons—two elected
3. Constitution made by sequential choice and fund voting
1. Election of persons—two elected
This article on election of persons is based on an actual state of affairs. The purpose is to highlight characteristics of the single transferable vote (STV), which have not been dealt with in voting theory. The uninitiated needs to know the following background facts before reading on.
An amendment of the Icelandic constitution is brought about in the following way: After the parliament has adopted an amendment it is dissolved at once and new elections are held. As soon as the new parliament has confirmed the amendment it comes into force.
As a consequence of the financial turmoil in Iceland since 2008 an interest for new procedure for bringing about amendments to the constitution emerged. In 2010 an act was passed on for electing a body for presenting to the parliament a proposition for a new constitution. This body was to be composed of 25 members who were to be elected by single transferable vote, a procedure without precedence in Iceland.
To become a candidate was relatively easy and the number was astonishingly high, 522 in all. Voters were to elect – and rank – up to 25 candidates.
The election of persons on 27 November was a thought promoter. On election day I was present at the largest polling-station in Reykjavik as supervisor. There a Scottish gentleman, who had been invited by the Ministry of Justice in connection with the elections, contacted me. He had been in Iceland on an earlier occasion, i.e. in September, invited by the parliament, then in connection with the handling of the proposed act. He had, as he told me, rewritten the Act for the parliament.
The Scotchman told me the following: This method is in use in Scotland, Ireland, Northern Ireland, Australia, New Zealand and Canada – the British world, he interposed – as well as in Cambridge in Massachusetts, New England. The number of persons to be elected could very well be ten to twelve, seventy being the highest number so far since the method was first applied. Then the voter was to elect – and rank – all the candidates, e.g. 1, 2, etc. all the way up to 69, 70. For the election of Members to the body that shall propose amendments to Iceland´s constitution each voter shall elect – and rank – only 25 candidates.
Here this method will be evaluated with reference to the lessons of the November elections.
But first we would like to introduce a tentative example: One person is to be elected. There are 21 voters. To determine the quota. i.e the number of votes a candidate must have to be elected, a divisor must be set. The divisor equals the number of candidates to be elected plus the number 1. The number of voters, i.e. 21, is then divided by this number, i.e. 21/2 = 10.5.
The quota is found by eliminating the decimal and adding 1. Consequently the quota is 11.
There are 5 candidates: A, B, C, D and E. Eleven voters rank them as follows:
A
B
C
D
E
while 10 voters rank them:
B
C
D
E
A
The number of votes spent on A equal the quota. The 11 voters who place him at top put B right below him. The ones who place B at top put A at the bottom. With this voting method it does not change anything.
This example is presented to compare single tranferable vote with sequential choice in Democracy with sequential choice and fund voting, chapter II.B.2. In sequential choice B would come out first: He receives 10x4 points for the top position, and 11x3 for position 2, i.e. 73 points in all. A receives 11x4 points for position 1 but none for the bottom position.
Now let us examine what the outcome of such an evaluation by the voters, i.e. wishing to counteract the election of specific candidates, would have been if the method applied at the election of the Constitution Assembly last November had been used. In our example 25 are to be elected, i.e the divisor is 25+1. Let us also assume the following: The number of voters is 100,000. The quota, i.e. the number of votes needed for a candidate to be elected is 3,847. A is placed at the top by 3,900 voters so he is safely in. That would also have been the case even if no other voters had placed him among the 25 ranked. That could also have been the fate of B, who was put at top by 3,850 voters, even if no other voter had included him in their selection. Therefore it is conceivable that 25 candidates are elected, each backed by only a small number but rejected by a vast majority. On the other hand there is a possibility that candidates, who are placed number two or three by most voters, have no chance of being elected because so few of them are placed first in the list.
Of the 25 who were elected for the Constitution Assembly 22 reside in the capital area, 2 in the principal town of Northern Iceland, but only 1 elsewhere. This was the outcome but it is to be presumed that no voter actually wanted things to turn out that way. It is a fact that there is no co-ordination in single transferable vote, it is a coincidence what the final combination will be like. The same is even true of an election of only two persons. That I shall now demonstrate by a near-authentic example.
At a given school the teachers are to elect two teachers for the School Board. The school has premises at two locations, in Ladies' Street – with subjects taught mostly by women – and in Gents' Street with subjects taught mostly by men. Everybody agrees that both places shall be represented – as well as both sexes. In an unbound election of persons or in single transferable vote where every voter places teachers of both sexes and from both locations at top it is not at all sure that the result will match the opinion held by all; it can easily happen that two men will be elected from Gents' Street or two women from Ladies' Street. By applying sequential choice of pairs this can be avoided. A list of all possible combinations will be established and each pair given an identification. The voters rank the pairs and the points will be calculated according to the rules for sequential choice.
This is quite easy when only two shall be elected and there are only two attributes to be taken into consideration. When 25 shall be elected and there is a number of attributes to be taken into consideration, sequential choice is not so easy to apply. The discussion of sequential choice for electing board members in Democracy with sequential choice and fund voting, chapter II.C.2, presents an approach that is without the detriments which are so characteristic of the single transferable vote, as we have seen.
Morgunbladid 2 February 2011 [translated from the Icelandic with minor amendments]
In the section «Advantages and disadvantages of different rules» ÞH compares the election method used for the Constitution Assembly´s election with some other rules. Of these Borda count will be discussed here; it certainly is the one most discussed in the literature. By Borda count the best alternative is chosen and nothing else. A Borda election of course shows which alternative comes next to be viewed as best, but that doesn´t mean that some voter has voted for it thinking that it should be number 2 in the group´s conclusion. Assume there are four alternatives, A, B, C og D. A voter ranks A,B,C,D. This doesn´t mean, if two are to be elected, that the voter would rank B after A. He may have the opinion that A and B present nearly the same qualities, so B could substitute A, if A get small support. On the contrary, the voter could mean that A and C constitute the best pair. Therefore, there is no reason to handle election sheets in the Constitutional Assembly´s election as if they were Borda rankings where only one is elected. Nonetheless, this is done in the section «What if other methods had been applied».
3. Constitution made by sequential choice and fund voting
Jon Elster was invited to Iceland concerning a new constitution which was to be made. In an interview in the newspaper Morgunbladid 17th april 2011 he was asked if separate articles could be submitted to plebiscite. In his opinion it would be unfortunate and even dangerous; that kind of procedure could result in patchwork, while a constitution ought to be consistent. He found it to be a really good idea to submit the Constitution Assembly´s motion to plebiscite before submitting to the Parliament.
Submitting only one version to plebiscite involves strain. Applying sequential choice, it is no problem to submit more versions of a constitution. One of the versions could include no sovereignty reduction; that would be in accordance with an unanimous standpoint taken by the People´s Parliament, which the Constitution Assembly ought to take into account. The opposite version, a reduction of sovereignty, could be implemented in various ways. That kind of procedure is not a patchwork as, by it, the electorate gives priority to whole versions of a constitution.
When a constitution is shaped, according to Elster in the inverview, the main intention should be to give interests, passions, prejudices and subjectivity the least possible significance; this was the newspaper´s main headline. My opinion is the opposite: democracy should take care of interests, such as national interests towards other states and other general interests. There will not be one opinion about how special interests may converge into general interests; in this respect, people divide in political parties. Moreover, passions should rule, such as the passion for freedom, democracy and humanism. Different opinions in this respect may not be limited.
A constitution can easily be shaped through fund voting. It is a procedure which can be applied whenever a society shall shape a convention, or a political party or a society shall form a program. The procedure is described in chapter III.D, «Large issues», in the book Democracy with sequential choice and fund voting. There, a master city plan is an example. Forming a plan starts with an outline. Subsequently, proposals for further forming and changes are handled, the one after the other. When somebody finds that the outline, through decisions and changes, has not developed as expected, it can be dealt with again. By fund voting it is possible to balance ideas (proposals) which only few strive for but more are against, each of them with less intensity. I presented this procedure for the leadership of Reykjavik town in 2008. I mentioned the possibility that the 30 which had been on the slate of candidates participated and that each of them received votes in their fund proportionally to the votes their slate got in the election.
With regard to this, a constitutional assembly of 25 members could be organized as follows. We assume that there are 522 candidates and that each voter votes for up to 25 of them, without ranking. The assembly will consist of the 25 candidates having received the most votes, each candidate receiving in his fund a number of votes proportional to the number of votes received in the election. The assembly forms proposals and submits them to all candidates. As all the candidates participate in the fund voting, candidates which are generally liked — they may be many — can let themselves be heard in proportion to their support; together, this support can be much stronger than the support allotted to the members of the assembly. A constitutional assembly of this kind cannot work fast.