BC Electoral Reform: Single Transferable Vote?

[cowgirl]

So the British Colombia (western Canadian province) government created a Citizens' Assembly on Electoral Reform (http://www.citizensassembly.bc.ca/public), where two people were selected out of each riding in the province to create an assembly of people to research and decide on an alternative to the existing First Past the Post electoral system.

They came up with the Single Transferable Vote (http://en.wikipedia.org/wiki/Single_transferable_vote), or STV, which I understand will be voted on in a referendum this year.

BC Dopers, or anyone else with an opinion: What do you think of it?

On the one hand, just about anything is better than FPTP in a system with more than two parties. It results in massively skewed majorities (in one notable example, in 1987 the Liberals won 100% of the seats in New Brunswick with only 60% of the popular vote) and massively inhibits emerging new parties.

On the other hand, I've seen a number of commentators with whom I usually disagree, as well as some of those with whom I normally agree, criticize it. I confess I don't really understand it enough to form an opinion. Please help.

[ultrafilter]

In a single-seat election, this is known as instant runoff voting, and it's the only system I've seen that's worse than first past the post. I don't know for certain that all the problems carry over, but the potential is there, and that alone is enough that I'd be seriously worried were I a citizen of BC.

You can read all about the problems of IRV over at www.electionmethods.org.

[BrainGlutton]

From http://www.fairvote.org/choice/:

Choice Voting

Choice voting is a full representation system in which voters maximize their vote’s effectiveness by ranking candidates and the threshold of support necessary to win is lower than in winner-take-all elections. Full representation systems are ones where as many voters as possible in a given constituency elect a preferred candidate. Choice voting (also known as “preference voting”, the “Hare system” and the “single transferable vote”) is the fairest method of full representation that can be used in non-partisan elections and also has a well-established history in partisan elections.

Choice voting’s advantages include greater minority representation, minimization of wasted votes, elimination of vote-splitting and encouragement of coalition-building among minority groups or parties. Choice voting has been used primarily in the English-speaking nations, in large part because of John Stuart Mill’s strong advocacy. As of January 2004, choice voting is used for electing such legislatures as: the parliaments of Malta and the Republic of Ireland, the federal senate in Australia; the regional assembly and most cities in Northern Ireland; all local health boards in New Zealand and for city council elections in the capital city of Wellington; and the city council and school committee in Cambridge, Massachusetts. It is also frequently used to elect the boards of non-governmental organizations.

Approximately two dozen cities in the United States have used choice voting, mostly in the first half of the 20th century when it was highlighted in the model city charter of the National Municipal League. New York City used it for five city council elections during the era of Mayor Fiorello La Guardia. Cincinnati used it for council elections from 1925 to 1955. Others using choice voting included Cleveland, Sacramento (CA), Toledo (OH) and Worcester (MA). Generally adopted to reform “machine” governments, choice voting typically faced persistent opposition that won despite voters typically opposing initial repeal efforts. The need for hand-counts and the fact that it represented racial minorities well were the main political problems for choice voting in the United States in this era.

Choice voting has won recent support from charter commissions in cities such as Kalamazoo (MI) and Pasadena (CA). It won 45% of the vote in stand-alone ballot measures in Cincinnati in 1988 and 1991 and in San Francisco in 1996.

From http://www.fairvote.org/pr/choiceintro.htm:

Introduction to Choice Voting

Choice voting (e.g, "single transferable vote" or "preference voting") is a form of limited voting in which voters maximize their one vote's effectiveness through ranking choices. Choice voting is very likely to provide fair results, can be used in both partisan and non-partisan elections and does not require primaries. It is recommended as the best system for local government elections.

To vote, voters simply rank candidates in order of preference, putting a "1" by their first choice, a "2" by their second choice and so on. Voters can rank as few or as many candidates as they wish, knowing that a lower choice will never count against the chances of a higher choice.

To determine winners, the number of votes necessary for a candidate to earn office is established based on a formula using the numbers of seats and ballots: one more than

1/(# of seats + 1). In a race to elect three seats, the winning threshold would be one vote more than 25% of the vote -- a total that would be mathemetically impossible for four candidates to reach.

After counting first choices, candidates with the winning threshold are elected. To maximize the number of voters who help elect someone, "surplus" ballots beyond the threshold are transferred to remaining candidates according to voters' next-choice preferences: in the most precise method, every ballot is transferred at an equally reduced value. After transferring surplus ballots until no remaining candidate has obtained the winning threshold, the candidate with the fewest votes is eliminated. All of his/her ballots are distributed among remaining candidates according to voters' next-choice preferences. This process continues until all seats are filled. Computer programs have been developed to conduct the count, although the ballot count often is done by hand.

From http://www.fairvote.org/pr/perfectsystem.htm:

A brief, non-technical response to mathematical critiques of choice voting

Some people criticize the choice voting form of proportional representation because in certain very limited circumstances it is at least theoretically possible for choice voting to lead to the election of a candidate that common sense suggests should not be elected.

Why is this not a fatal problem for choice voting?

Because these kinds of situations can occur with all voting systems, but in the real world, these situations never actually occur in choice voting elections. For examples of the types of situations in which different voting systems suffer flaws, check out these examples and judge for yourself which ones would be likely to occur in actual public elections.

Kenneth Arrow won a Nobel Prize in 1972 for showing, in effect, that no voting system is perfect. If you choose a few reasonable criteria, such as the majority should rule and voting for your favorite candidate should not help your least favorite candidate, Arrow showed that all voting systems violate some of these criteria some of the time. A voting system that meets all of the criteria all of the time simply does not exist.

Thus, when someone points out that a particular voting system in a particular situation leads to an undesired response, we must ask, "Does this problem only occur in theory, or does it actually occur in real word elections?"

Choice voting actually works

Choice voting has been used in public elections for over 100 years in the US and around the world, and we have not uncovered even a single example where any of the theoretical defects have occurred.

This includes tens of thousands of elections with tens of millions of voters. If the theoretical problems with choice voting occurred even as frequently as 0.1% of the time, there would be many such examples, but there are none.

Choice voting allows voters to vote sincerely for the candidates they actually support, most voters end up being represented by their preferred candidates, and the outcome is broadly representative of the entire electorate.

Kenneth Arrow showed that all voting systems can be "gamed" in certain situations. The situations in which choice voting can be "gamed" are so infrequent as to be virtually undetectable, and if a voter ever tries to game the system, rather than voting sincerely for her preferred candidates, she is far more likely to shoot herself in the foot than help her cause.

See also: http://www.fairvote.org/consulting/stvtechnical.pdf

And http://www.fairvote.org/pr/global/alaskahighwaynews.htm

[BrainGlutton]

In a single-seat election, this is known as instant runoff voting, and it's the only system I've seen that's worse than first past the post.

Why?

[Grey]

BrainGlutton I really do appreciate the cites, but could you maybe sum them up? Not everyone (ok maybe just me) is able to pour over the quotes and links.

[BrainGlutton]

BrainGlutton I really do appreciate the cites, but could you maybe sum them up? Not everyone (ok maybe just me) is able to pour over the quotes and links.

OK. here's how it works in most places it's been tried. In Canada you elect provincial and national M.P.s by the single-member-district system, don't you? That is, the territory is divided into districts -- I believe you call them "ridings" -- of equal population, and you elect one member to represent each, by the winner-take-all, first-past-the-post plurality system. If there are three candidates in the riding, the one with the most votes wins, even if that's less than 50%.

Naturally this tends to squeeze out the minor parties. E.g., if the New Democrats (my personal fave) have a 20% level of support throughout British Columbia -- they still might not elect a single member to the provincial parliemant, because there might not be enough NDP supporters in any one riding to form a plurality.

Under a multi-member-district system, you would merge five existing ridings into one big riding electing a five-member delegation, and each member, ideally, is the choice of roughly one-fifth of the voters in the new megariding.

Single-transferable-voting is the mechanism for achieving that desired outcome. You, as a voter, needn't worry about the mechanics of it, about how the votes are tallied; that's the election office's problem. (But you can read a good, concise explanation here: [ur]http://www.fairvote.org/factshts/choice1.htm[/url].)

All parties contesting the election would nominate five candidates; there also would be room on the ballot for independent candidates affiliated with no party. When you vote, you would be given a rather long ballot listing all candidates standing for that riding (sorted and labelled by party, presumably); and you would pick your top five favorite names, rank-ordering them by preference. (See sample ballot at http://www.fairvote.org/consulting/choiceex.pdf.) E.g., if you are an NDP supporter you likely would pick the five NDP candidates, ranking them in the order the party leadership proposed them for listing on the ballot, or in some other order if you have a different opinion about their relative merits. (If the NDP has only 20% or so support in the riding it will get to elect only one member, so which one is an important decision -- and under this system, the party's grassroots supporters as well as the local party officials get to participate in that decision.)

Practical results:

1. At present, "your" single M.P. might well be somebody you didn't vote for or even very enthusiastically voted against. Under the system described above, you can vote with some confidence that at least one of your five choices will end up in your riding's elected delegation. Somebody you can really talk to, you know?

2. The elected parliament will be more diverse -- that is, it will represent a wider range of viewpoints, from (by Canadian standards) far left to far right. The minor parties still won't have much power in the way of votes, but they will have the right to participate in the debates, on the floor and in committees, and everybody else will have to listen to them.

3. There probably will never again be a "majority" party in parliament. All government will have to be by coalitions or other power-sharing arrangements. No proposal ever will pass a vote unless it can drum up support from several very different political camps, supporting it for their own reasons or as the result of a logrolling bargain.

So that's how it works. Still want it?

[BrainGlutton]

Sorry, flubbed that link --

"You, as a voter, needn't worry about the mechanics of it, about how the votes are tallied; that's the election office's problem. (But you can read a good, concise explanation here: http://www.fairvote.org/factshts/choice1.htm.)"

[ultrafilter]

Why?

For all the same reasons I detailed in this thread (http://boards.straightdope.com/sdmb/showthread.php?t=284577).

[BrainGlutton]

For all the same reasons I detailed in this thread (http://boards.straightdope.com/sdmb/showthread.php?t=284577).

None of your arguments in that thread demonstrated that IRV* is worse than first-past-the-post, only that it is arguably a less than optimal improvement when compared with, say, the Condorcet method. But it's still better than what we've got now.

And I advise you to read "A brief, non-technical response to mathematical critiques of choice voting," excerpted in post #3 above.

*Instant runoff voting is essentially the STV system adapted to fill just one post, rather than seats in a multimember policymaking body. E.g., if we had had IRV in place for the 2000 presidential election, then a Nader supporter might have had the option to rank-order his/her vote as follows:

First choice: Nader (Green)
Second choice: Gore (Democrat)
Third choice: Browne (Libertarian)
Fourth choice: Bush (Republican)
Fifth choice: Buchanan (Reform)

In the first round of counting, all the first-choice votes would be tallied. If nobody got more than 50% of those, then the candidate with the fewest first-choice votes would be eliminated from the counting, and in the second round, every ballot naming that candidate as first choice would be counted toward the voter's second choice; the process repeats until a candidate with majority support emerges. This eliminates the "opposition-splitting" problem presented by third-party candidacies under the present plurality system. I.e., Nader could have run without fear of "taking away" votes for Gore -- in fact, he would have attracted votes for Gore, by drawing to the polls people who might have stayed home if Nader were not in the race, but who, once at the polls, would name Gore as their second choice.

[ultrafilter]

None of your arguments in that thread demonstrated that IRV* is worse than first-past-the-post....

Well, except the monotonicity criterion. Why don't you think that that's fundamentally important?

And I advise you to read "A brief, non-technical response to mathematical critiques of choice voting," excerpted in post #3 above.

Well, I did, and most of the arguments boil down to "Yeah, they're right, but nothing's perfect, so let's do this anyway". As a mathematician, I'm not particularly interested in a non-technical argument in the first place.

[BrainGlutton]

Well, I did, and most of the arguments boil down to "Yeah, they're right, but nothing's perfect, so let's do this anyway". As a mathematician, I'm not particularly interested in a non-technical argument in the first place.

But this is a political argument, not a mathematical or technical one. Our concern should be whether a proposed reform will actually further its underlying political goals -- in this case, giving each voter a more effective choice, and producing a political system open to a more diverse range of parties and viewpoints. Why does IRV fail by that standard?

[ultrafilter]

But this is a political argument, not a mathematical or technical one. Our concern should be whether a proposed reform will actually further its underlying political goals -- in this case, giving each voter a more effective choice, and producing a political system open to a more diverse range of parties and viewpoints. Why does IRV fail by that standard?

It is a technical argument, because there's a whole theory of voting mechanisms that exists just to deal with stuff like this. You've already made it technical by asking for a more "effective" choice? What does that mean? How do you measure whether a given system provides effective choices? How do you show that a particular system gives voice to a wider range of political opinions than some other system? Those are all technical points that need to be addressed.

I've already linked (in here or the other thread) to at least one analysis showing that IRV behaves counterintuitively in the case where there are three or more viable parties. Isn't that the sort of situation where you want a system that's intuitive?

[cowgirl]

Can anybody shed some light on how this is going down in BC?

[Cunctator]

There's a bit more information here (http://www.aec.gov.au/_content/what/voting/count_senate.htm) about how votes are counted for Australian senate elections, where STV is used.

[Barbarian]

I was living in BC during the last election, and I still hold a driver's license there. It's important to note that the current legislature is incredibly lopsided: In 2001 the ruling NDP dropped down to 2 seats, and the Liberals swept the remaining 77 seats with 57% of the popular vote. in 1996, the NDP had a majority government with 39.5% of the popular vote, 39 out of 75 seats, or 52% of seats in the House.

(The current roster (http://www.legis.gov.bc.ca/mla/3-1-5.htm) has changed, since a bunch of Libs jumped parties, some involuntarily.)

While that type of lopsidedness seems unusual, it's not that rare for BC. The Social Credit party ruled virtually unopposed from the '50s to the '70s, and frequent flip-flops are common since people in BC like to have clowns as politicians and often come out to the polls with no urge other than to "throw the bastards out".

The current house is about as uneven as it gets, and doesn't really reflect the population, especially since the Green party has polled somewhere around 10% in the past two elections and won nary a seat. There's also the SoCreds doing a bit less than that, not the mention the dozen or so fringe parties whose votes could be transferred either way.

If you want to know what's going on in Vancouver, check out the Georgia Straight (http://www.straight.com/section.cfm?id=172), but bear in mind that Bill Tieleman is an NDP attacklapdog. Just about every other paper in the province is owned by GlobalCanWest, so you can read the Vancouver Sun (motto: our business writer is brother to the Premer!).

Or ping Larry Mudd, Flamsterette, and our other BC dopers.

[ruadh]

We use it in Ireland and I almost never hear any complaints about it. The only criticism is that it tends to hurt parties that get a significant number of first preference votes, but few transfers, but some would argue that's exactly how it's supposed to work.

No way would we want to go to FPTP.

[BrainGlutton]

We use it in Ireland and I almost never hear any complaints about it. The only criticism is that it tends to hurt parties that get a significant number of first preference votes, but few transfers, but some would argue that's exactly how it's supposed to work.

No way would we want to go to FPTP.

How does your system work in terms of stability from one election to the next? Barbarian's post addresses something I've been thinking about a long time: Some people fear a proportional-representation, multiparty system would lead to "instability," but my thinking is that it is actually a first-past-the-post system which produces stability -- by artificially inflating the power of the "swing voters" in the middle. Under such a system, a shift in one-half of one percent of the aggregate votes cast for one party or another can lead to a total shift in power in the legislature -- that happened in the U.S. in the 1994 Congressional election. And, according to Barbarian, it happens regularly in elections to the BC Parliament. Under a PR system, on the other hand, each party would have a "market share" of a certain percentage, and there would be no more "electoral revolutions" -- an election would be about one party stealing a few percentage points' worth of voters from the fringes of its nearest neighbors. Things would be a lot more predictable, at least in their broad outlines. Is that how it works in Ireland? Or do you have dramatic "lopsidedness" and "flip-flops" in every election?

[Mr2001]

*Instant runoff voting is essentially the STV system adapted to fill just one post, rather than seats in a multimember policymaking body. E.g., if we had had IRV in place for the 2000 presidential election, then a Nader supporter might have had the option to rank-order his/her vote as follows:

First choice: Nader (Green)
Second choice: Gore (Democrat)
Third choice: Browne (Libertarian)
Fourth choice: Bush (Republican)
Fifth choice: Buchanan (Reform)

[...]
I.e., Nader could have run without fear of "taking away" votes for Gore -- in fact, he would have attracted votes for Gore, by drawing to the polls people who might have stayed home if Nader were not in the race, but who, once at the polls, would name Gore as their second choice.
In fact, because of the flaws in IRV that have been discussed at length in previous threads, someone who votes that particular ballot might help Gore even more than someone who ranks Gore first and Nader second. But then, he might also help Bush more than someone who ranks Bush first and Nader last. IRV just isn't trustworthy, because ranking someone higher can end up hurting him instead of helping.

[ruadh]

How does your system work in terms of stability from one election to the next?

Judge for yourself:

Irish general election results since 1923 (http://en.wikipedia.org/wiki/List_of_Irish_general_elections)