Universal Workshop

books etc. by
Guy Ottewell

Guy's blog Follow @guyottewell

The Other AV
    The 2011 referendum on electoral reform was a long-awaited chance for Britain to grow out of its primitive system of voting. That system seems simple and obvious; yet whenever there are more than two candidates it causes serious unfairness.
    But the only method we were allowed to choose instead was not the best, and it was rejected, because it was complicated enough to cause confusions, suspicions, misunderstandings, and ourright misrepresentations.
    There is another system, Approval Voting, that has been shut out of the discussion, yet is far simpler and does as good a job and in some respects a better. If this had been the “alternative” offered in the referendum, it would have had a better chance of being understood and therefore accepted. It should be the system offered in any future effort to make voting fairer.

Terminology (dammit!)
    First — though I hate to have to do it — I must say something about the names of these three voting systems.
    The ancient system, used not only in Britain but in the U.S.A. and elsewhere, is commonly called First Past the Post. This, besides being clumsy enough to stumble over (how many times have people said or written FTPT instead of FPTP?), is a metaphor and an inaccurate one. It doesn't mean that whoever first reaches a certain total wins, it means that whoever gets the most votes wins. It used to be known as One Man One Vote, so I call it One Person One Vote, OPOV.
    The best and second simplest system is called Approval Voting. That characterises it well enough. It means that the winner is the one that the largest number of voters could live with.
    The system that was on offer was being called Alternative Voting. There are four objections to that term: It does nothing to describe the system. It (deliberately?) suggests that the system is the only alternative — which is far from true. It could hardly have continued to be used if the system had become the established one. And it steals the AV acronym from Approval Voting, whose name dates from 1977.
    “AV” is one of the many forms of Preferential Voting. It is used in Australia, Papua New Guinea, and Fiji; actually the form of it used in Australia is even more complicated. It is called in the USA, and was previously called in Britain, Instant Runoff Voting — which does not mean that it produces an instant result, only that the “runoffs” are not separate elections (as they are in France). It has also been called Ware's Method, and by other names.

What's wrong with One Person One Vote?
    It forces you to choose between the candidates you consider good.
    Suppose A or B could win and C has only an outside chance; you don't like A, you could live with B, but you like C as well or better.
    (Many people have been in this situation at every election of their lives.)
    If you give C your “heartfelt” vote, you make it more likely for B to lose to A. If you give B your “realistic” vote, the true extent of support for C will not be known. And even if B and C together get more votes than A, A typically wins.
    From the point of view of the individual, this is the agonising “voter's dilemma” or the problem of whether to “vote tactically”. From the point of view of a party or philosophy, it is the problem of the “split vote” and of “spoilers.” — people who weaken a side, sometimes deliberately, by offering themselves as candidates.
    The system often allows a candidate to win who is approved by only a minority, indeed by fewer than another candidate is.
    These unfair results have happened in countless real elections. Solving these problems — the split vote and the unrepresentative win — is the urgent purpose of electoral reform.

What is Preferential Voting (“AV”)?
    The rule is that you may vote for as many of the candidates as you wish, and you must number them in order of preference. If no candidate gets at least 50% of first preferences, then the candidate with the smallest number of those is eliminated and that candidate's ballots are reallocated to the remaining candidates according to the next preference on each ballot. (The “next” preference is the second, unless that has already been used, in which case it may be the third, fourth, or lower.) This process is repeated until one of the remaining candidates gets at least 50% of the remaining votes.
    This system would have been an improvement in that you could give a form of your “heartfelt” as well as your “realistic” vote, and the winner would usually (not, as proponents claimed, always) have been voted for by more than 50% of those voting, whether as first or other preferences.

Confusions, suspicions, misunderstandings
    Sure, it isn't as complicated as astrophysics. But it's complicated enough to go wrong.
    Most people need to read that rule at least twice, and many never really grasp it. Easy enough to understand that you must write “1,” “2,” “3,” etc. But it is not easy to feel sure of what will happen to your vote afterwards, as it travels that process of reallocations. If voters cannot clearly imagine the consequences of what they do, is the system truly democratic?
    I've talked with well-educated people who did not realize that you don't have to rank all the candidates. Even the Yes to Fairer Votes campaign, which championed this system, sometimes got it wrong: I went to a meeting at which mock ballots were handed out, and the instructions on the “AV Ballot Paper” were: “Put a number next to each candidate . . .  until all candidates have been given a number”!
    The instructions went on: “The higher the number, the less you favour them, much like a top 5 list of favourites.” For there was some danger of voters getting that wrong, and thus doing the exact opposite of what they intended. (If any did, it could never be known.)
    And there was some risk of voters spoiling their ballots by inadvertently writing two 2s, or a 3 and a 5 but no 4.
    The Electoral Commission's booklet, distributed a month before the referendum, looked like a desperate effort to deconfuse the public. It took four pages to explain the “Alternative Vote,” with pictures of piles of votes.
    And were the reallocatings to be done in a flash inside computers, or by manual re-counting as those pictures suggested, and how long would it take? It was not easy to get answers to these questions, in fact it took me weeks of requests sent to officials. The answer was that the Electoral Commission did “not plan to introduce any voting machines”; and that the Bill decreeing the referendum required that, after each round of reallocating and re-counting (in each constituency), the detailed numbers must be made public.
    All this enabled the No To AV campaign to spread stories that reform would cost millions of pounds. Fictitious, in that there would have been no new machines; but the counting process might well have required more people.
    A conspicuous example of a mistaken idea came from no less a statesman than David Owen, leader of the former Social Democratic Party, in an article in The Independent on Sunday (13 March 2011): “ . . . the second preferences of the least popular candidates have the most influence. Because the first candidate to get eliminated under AV — typically a fringe or extremist candidate — gets their votes redistributed first, they have the best chance of determining the final result.” Wow, that sounded like a penetrating criticism that we hadn't thought of! But think it through. It doesn't matter that the 14th strongest party's votes are the first to get redistributed, since the 13th, of which there are more, will get redistributed next, and so on up to the 3rd.
    If Lord Owen could get lost in the logic, anyone could.

Two sharper questions about Preferential Voting (“AV”)
    What if you write “1” for a candidate who gets eliminated, and no “2” for anybody else? To whom will your vote be reassigned? Answer: To nobody. It is thrown away, or as the Electoral Commission said “is not used.” More broadly, this is what happens if none of the candidates who survive to the last round is one to whom you've given a number. Your vote will not be part of the final totals. If, say, the final reallocation results in 51% for A and 49% for B, those numbers exclude everyone who gave preferences only to C and D. In most constituencies, for example, people who can bring themselves to vote only Ukip or only Green will not be counted in the final totals. So we should more truly say that the winner will have been voted for by at least 50% of those whose votes count.
    And: What effect will it have if you rank the candidates you approve — B and C, say — as 1 and 2, or the other way around? Will you still disadvantage the one to whom you don't give your first preference? It took me quite a bit of figuring with different scenarios to realize that it makes no difference. Whether you rank C as 1 or 2, your vote will eventually go to B. More generally, for all of the names that you approve, it doesn't matter which number you put beside any of them. It's a decision you are required to make, sometimes a dilemma, but a dilemma whose resolution doesn't matter.
    So why rank at all? Why not just tick?

Approval Voting
    That is what is done in Approval Voting, which has been described by me and other authors since 1977, but has been used, without being called by a specific name, in various elections since ancient times. It is essentially the system used in almost all elections for members of boards and other councils.
    In Approval Voting, as in “AV,” you can vote for as many candidates as you wish. The difference is that you don't have to rank them. If there are four, you can vote for one or two or three. Of course voting for all would have the same non-effect as voting for none. As in One Person One Vote, whoever gets the most votes wins. That's all.
    Suppose that roughly 40% approve Conservative only, 30% Labour only, 15% Lib Dem only, and 15% approve both Labour and Lib Dem, in either order. Then the approval totals are Conservative 40, Labour 45, Lib Dem 30.
    Given the distribution of the voters' real wishes, there is — unlike under One Person One Vote, “AV,” or any other system — only one possible outcome. That “distribution of the voters' real wishes” is made manifest in those simple figures.
    “AV” records it either too simply (A and B get X% and Y%, and all the votes for the eliminated candidates go unstated) or with an excess of detail that varies according to how voters resolve their unnecessary dilemmas, and is impractical to publish. It might take up a whole page to report that in a certain constituency “X voted for Green 1, Conservative 2, Lib Dem 3, BNP 4 . . . , Official Monster Raving Loony Party 15” and all other combinations that voters happened to make.
    Various other situations — kinds of distribution of the voters' wishes — are explored at www.universalworkshop.com/ApprovalVoting.

Surprising simplicity
    Approval Voting has been described as “a surprisingly simple costless reform.” It is the next simplest after One Person One Vote, in that nobody has to do any ranking. It is simple and costless in that the ballots are counted only once, there is no reallocating, no re-counting. It sounds surprising, in that you can vote for more than one and the votes add up to more than 100%. But, just as in the elections for boards, where you may vote for up to six people out of twenty, nobody who votes for two or more candidates has more power over the result than those who vote for only one.
    A pleasant advantage of Approval Voting over “AV” is that people can have real-life experiences of it at any time, thus becoming comfortable with it. A situation arises in which a group of people has to select three out of nine options — say, as the short-list for further action, or recipients of an award. They each write down as many of the options as they think fit. They soon have the result. “Alternative Voting” would be no alternative, because it would be impracticable to do all that reallocating.

The future
    What about Proportional Representation? Many regard it as the real reform beyond “AV.” But it is not a method of casting votes; it has to do with the number of seats being voted for. If a region is to have 10 seats, then those can be shared among parties in proportion to the numbers of votes cast under One Person One Vote, or (much more fairly) under Approval Voting. But not under the intricate results of “AV,” because only the first preferences could be used (think, for instance, of the false result for voters who rank not some but all of the options). So the voter's dilemma would be reintroduced; in fact, the system would reduce merely to a needlessly complex version of OPOV.
    If “AV” had been instituted and we had then gone on to get Proportional Representation, we would then have had to go back to using OPOV as the voting method. Unless . . .  I say no more.
    “AV” lost. Its needless complexity — slight, but sufficient to cause misunderstanding — was the underlying reason for its failure. Has it discredited the whole idea of electoral reform? Do we have to go on chafing under the maddening unfairness of OPOV?
    Those who have not given up on reform may be well advised to begin listening to the few small voices that have been trying to call attention to Approval Voting. This “surprisingly simple costless reform” has been consistently shut out of discussion, presumably because its surprisingness has prevented the pundits who control our press and our politics from noticing its elegant soundness.

BACK