When I was at University the Students Union used AV, and one year a man stood for the post of Womens' Officer under the name of "Captain Kirk". Against him were two more conventional candidates standing on the Conservative and Labour platforms.
When the votes were counted Captain Kirk got 49% of the first preference votes, with the Conservative getting 26% and Labour 25%. Under FPTP Kirk would have won, but now the Labour candidate was eliminated and it turned out that all of the people who voted for her first had voted for the other woman as second preference. So now the Conservative candidate had 26%+25%=51% of the vote, and Kirk was defeated.
I'm well aware of the Arrow Impossibility Theorem, which shows that given an election with three or more candidates and three or more voters it is impossible to have a voting system that always delivers the right result, but it seems to me that this is rather like the fact that any programming language is logically equivalent to a Turing Machine; its true, but it doesn't mean that all (voting systems | programming languages) are equally good. FPTP seems particularly prone to perverse outcomes, such as the election of a male Womens' Officer when the majority of the electorate wants a female. Pretty much every election in the UK is a 2-horse-1-pony race, with the Liberal Democrats as the pony, and every election pundits discuss the impact of "tactical voting" as people who would like to elect the Lib Dems instead vote Labour for fear of otherwise "letting the Conservatives in". Under AV they can simply vote Lib Dem first and Labour second.
The anti-AV campaign's arguments seem to come down to:
- It will create more hung parliaments. Possibly. I don't see this as a bad thing. The current coalition seems to be doing OK, and a coalition means that more of the electorate's views get represented in government. Having a government elected by 40% of the people get 100% of the power seems to me rather undemocratic.
- It will empower the National Front. For those who don't know, the National Front (NF) is an extreme right wing racist party, which occasionally does well in poor inner-city areas where the "immigrants are taking your jobs and houses" line gets a sympathetic hearing. But equally, if most people object to the NF on principle then AV makes it much easier to vote against them; just put the NF at the bottom of your list. That ensures that no matter how other people vote, your vote will count against the NF.
- The most popular person doesn't always win. Well yes, if you define "most popular" as "winner under FPTP" then this is true; if AV didn't give a different result sometimes then there wouldn't be any difference. The point about AV is that the candidates with the broadest support tend to win much more often, whereas FPTP is prone to producing winners who most of the electorate actively dislike.
5 comments:
I fully agree that AV is preferable in theory.
One big issue, though, is that if there is n candidates and less than n! voters (in a given district), you can coerce voting: http://hal.archives-ouvertes.fr/hal-00142440
It is not directly related to AV, but to any system where you have many choices on a single ballot, thus leaking information. (And, contrary to the Condorcet method, I don't think you can split AV in several ballots since you gather information from eliminated ballots for the second round.)
Maybe not an issue for elections in the UK (I don't know the typical value of n), but elections in the US are prone to such attacks because they have many questions on a single ballot: http://www.electronic-voting.org/GIF/ballotusa2004.jpg
I'll try again, my first attempt was rather rubbish.
You write: "just put the NF at the bottom of your list. That ensures that no matter how other people vote, your vote will count against the NF."
The problem is, this isn't really true.
In the first round, only the first preference votes are counted.
If you vote for the most popular candidate, the order of your votes for other candidates are not relevant, because only those ballot papers for the eliminated candidates get re-examined after the first round.
Suppose the procedure comes to a conclusion after the second. The fact that you put one candidate 2 and another 5 has not influenced the vote at all.
This means that if you happen to prefer the more popular candidates, your preferences against the other candidates are not counted, but if you prefer the least popular candidates, your other preferences are counted.
As you say, no voting system is fair, but AV is biased towards the minority opinions.
This is why many people think that it is AV, not FPTP, that produces the perverse results.
Hi Paul, just came across your blog from Planet Haskell. Here in the antipodes (i.e. NZ) we have had MMP since 1995. We will also be voting this year to decide whether to keep it/go back to FPP/get something else.
I was interested in your comments about hung parliaments. We haven't had a simple majority in parliament since MMP was introduced, but we've had reasonably effective government anyway. The main problem that people who support FPP seem to worry about is how a government is formed in these situations, and I thought it might be interesting to share our experience with this.
The first year that MMP was introduced in NZ the leader of the party with the balance of power (Winston Peters) dicked around something terrible in choosing which of the major parties he would support in a coalition. Voters remembered this and his party was pretty much removed from parliament at the next election. Since then most coalitions have been on the table for voters to assess before voting even takes place, a coalition is formed within about a week of voting and government continues as usual.
So there might be some teething problems to start with but these will sort themselves out. I suspect that MMP will remain in NZ after the vote this year with a bit of luck.
In Australia they've had the first hung parliament in 70 years or something, and despite all the wailing about ineffective government, they've actually got a lot of good things done. And the independent MPs who are backing up the government have forced them to take account of one or two issues which would otherwise have been ignored (e.g. gambling reform/carbon pricing). So I would argue that a hung parliament has led to better government in Australia as well.
Canada on the other hand is a bit of a problem. I don't know enough about how they select their parliaments but they haven't had a stable government for the last decade it seems. Maybe their politicians are just dicks.
With regard to the National Front getting into parliament, well I would argue that if they can get people to vote for them then they should be there. Parliament is not about restricting membership only to people that are acceptable to everyone in society. Proportional representation is exactly that, proportional. At any rate bigot parties like the NF tend to self-destruct with the higher scrutiny that having an MP brings.
Anyway, sorry about the blog-length comment - I just think that better proportional representation of society in parliament is a _demonstrably_ good thing, and I hope the UK moves in this direction too.
Luke,
In your example, the only reason that the vote doesn't count against the NF is that they've already lost, so its irrelevant.
"Bias" is a loaded term; it seems to me that FPTP is biased towards the big parties. The Lib Dems need far more votes to get an MP than either the Conservatives or Labour.
With AV, you need to rank candidates on a single ballot: this is prone to the ballot-as-signature attack (coercing someone's vote).
http://hal.archives-ouvertes.fr/hal-00142440
Post a Comment