Bell Curve The Law Talking Guy Raised by Republicans U.S. West
Well, he's kind of had it in for me ever since I accidentally ran over his dog. Actually, replace "accidentally" with "repeatedly," and replace "dog" with "son."

Monday, August 09, 2004

San Francisco to Use STV Electoral System

Hi Everyone,

The LA Times (and other papers I'm sure) are reporting that San Francisco will use the Single Transferable Vote (STV) in this November's city elections.

In this system, voters rank order the candidates (San Francisco will allow 1st, 2nd and 3rd rankings). The votes are counted several times. After the first round of counting, the candidate with the fewest 1st rankings is eliminated. All the people who ranked the eliminated candidate 1st will then have their votes transferred to the candidate they ranked 2nd. The count process is repeated until one candidate gets over 50% of the vote.

This system has a number of advantages over the single member district plurality system (SMD) used in most US elections. The LA Times article discusses these and I've posted about this before.

This system would most likely benefit moderate candidates. Consider a three way mayoral race between a Green (G), a Democrat (D) and Republican (R). Let's suppose there are several kinds of voters with the following rank orderings for candidates:

Greens: 1 = G; 2 = D; 3 = R (15% of voters)
Republicans: 1 = R; 2 = D; 3 = G (43% of voters)
Liberal Democrats: 1 = D; 2 = G; 3 = R (32% of voters)
Moderate Democrats: 1 = D; 2 = R; 3 = G (10% of voters)

Under the current US electoral system if everyone voted for their first choice, the Republican would win with 43%, the Democrat would come in second with 42%, and the Green would lose with 15%. However, this hypothetical city has a 57% majority against the Republican. What is more, 47% of the voters, more than actually voted for her, consider the Republican the worst possible candidate! Would a Republican mayor really be the best, most democratic expression of the voters' preferred policies in this city??

Now consider the same city with the same candidates but this time, the votes will be counted using STV and each group of voters will rank order the candidates as indicated above. After the first round, the Green candidate would be eliminated and his votes transferred to the Democrat. The Democrat would then win with 57% of the votes to the Republican's 43%. What's more the Democrat would be 1st choice of 42% of the voters and the second choice of the other 58%. So no one would feel like the election produced their worst case scenario.

Obviously this example makes some assumptions about the candidates involved and voters' preferences. You could play around with it on your own and figure out how such a system would work with 4 parties by adding a party on the far right too. Also, this kind of example assumes one dimensional politics (left versus right) that assumption works most of the time.

Comments? Discussion?


BearPatrol said...

Nice to see this system of voting catching on. We used it for student body elections where I went for undergrad.

It's important because it allows a third-party candidate to be viable and reduces the need for strategic voting (e.g., "I want Nader to win, but I'll vote for Kerry since he has a chance at beating Bush.")

My main concern is that it may be too complicated for some people to understand. It seems pretty simple, but then again, people had trouble with the "butterfly ballots". Also, I seem to recall that some problems with Condorcet cycles can arise with this system; perhaps one of the resident political scientists can enlighten us?

Raised By Republicans said...

First a definition of a "Condorcet Cycle." This is when you have three options that people are voting for and you get A gets more votes than B. B gets more votes than C. But C gets more votes than A. You can see that in this situation, you there would be a constant cycling through the options as new coalitions form to defeat the most recent winner. This is why I think the California recall law is disasterous. It allows a minority of voters (the losers in the last election) to determine the timing and circumstances of the next election. That is how you get condorcet cycles.

OK, now STV actually avoids this problem by imposing institutional structure on the preferences. Note that the candidate with the fewest 1st rank votes is dropped and is not included in the subsequent rounds of counting. That takes care of the condorcet cycling problem.

A "condorcet winner" is a candidate who defeats all other candidates when compared to them one at a time (as opposed to all at once as in an election). When no condorcet winner exists, you get cycling. STV solves that problem but does not gaurantee a condorcet winning winner of the election.

STV can also be applied to multi-seat elections. It could be used for City Council elections for example and people would vote city wide for the same list of candidates which they would then rank order. But instead of eliminating and tranfering until someone gets over 50% of the votes, you would eliminate and transfer until you have the same number of candidates as seats on the council. This would allow multiple candidates from the same party to more or less compete with each other in the general election.

Anonymous said...

I personally find this amusing. Growing up in a country (yes, Australia) where preference voting and compulsory voting are the norm, to think that the average American voter cannot handle numbering candidates 1 to 3 just confirms ugly stereotyping of you seppos.

It is also amusing to me that it is limited to the top three preferences: maybe three preferences to result in a valid ballot I could understand, but so what if there are six candidates to be counted? By limiting it to three you ensure situations where some votes do not count, where the voter still has to think about who is likely to get in rather than who they would prefer.

A couple of examples of elections I can remember from Australia: Victorian elections for national parliament senators involved 62 candidates, which you could number or simply tick your preferred party. Ticking the preferred party meant the parties preferences were allocated, giving minor parties some real negotiating power with the majors. It also allowed me the pleasure of placing the real nutters last and knowing where my vote would end up.

Another election would have to be Tasmanian state elections done under the Hare-Clarke system: preference voting for multiple candidates (7 per electorate). It made election nights very entertaining. A former premier was constantly getting 3 quotas in his electorate due to his popularity, mainly due to his incredible ability to remember everyone's names and jobs. I digress.

Ignoring compulsory voting as a separate issue, preference voting gives a great deal more flexibility with little risk of deadlock. Distribute votes, the candidate with the least number is eliminated and their votes redistributed. Yes, it results in centralised candidates in one dimensional politics but is that a bad thing? I'm sure the French would have been happier with that in their last presidential primaries. Hell, I would be much happier if you guys could get direct presidential elections with full preference voting.

Raised By Republicans said...

Re: the complexity of STV voting. I don't think it would pose as much of a problem as its critics suppose. And Florida demonstrated that you can have screwed up ballots in even the simplist electoral sytems. However, it is also true that Australia (which allows the longest lists to be ranked) has one of, if not THE highest rates of spoiled ballots in any democracy. There are a number of explanations for this, protest voting in a country where not voting is illegal is one. Another explanation might be that a small but noticable minority of Australians can't properly mark their complicated ballots. But such voter incompetance is likely to be more or less randomly distributed (I would hope) and so it will all come out in the wash.

I think San Francisco picked three rankings because there are currently three parties in that city: Republicans, Democrats and Greens. If there were a lot of Libertarians in that city, they might have gone for ranking 4 candidates. But I'm just speculating.

Now, as for gridlock. There is no way that increasing the role of minor parties can reduce gridlock. Adding parties to the system can only increase gridlock or leave it the same as before. Think about a group of friends deciding on what movie to go to. Adding more people with different movie preferences to the mix will only make the decision more difficult.

I have mixed feelings about gridlock. I find it maddening when politicians I like are in power but I thank the Founding Fathers for it every day someone like Bush is in power.

Anonymous said...

I am not sure I agree with your comments about the number of spoilt ballots. Voter turn-out in Australia runs at around 95% (monetary fines for not voting), and informal votes are at less than 4%. Even assuming a slight bias in informal voting, that is not going to swing the majority.

It is interesting to me that SF is considering implementing a system that is not generic -- ie requires only 1-3 -- but still I am used to a national electoral commission, with set rules for defining electoral boundaries and the like. Still everyone can improve from examining their electoral systems: Australia could possibly value from a partial preferential system, where once your preference are run through your ballot is removed and the number of counted ballots reduced.

To gridlock: maybe you are using it in a sense I do not understand. How does adding more parties or candidates increase gridlock? Under a preferential system, unless the number of votes for the lowest number of vote getters are exactly the same, there is no gridlock. Even then tossing a coin to select one is unlikely to influence the highest vote getter.

Raised By Republicans said...

Usually "Gridlock" is used in American politics (and increasingly in the study of politics globally) to refer to an inability of legislators to change the policy status quo. In other words, gridlock means nothing gets done. It comes from a term used first, I believe, in New York city to describe what can happen in the grid like street pattern of Manhattan when there is heavy trafic.

In practical terms, the multicameral legislature in the US combines with the Presidential veto and the 2/3 cloture rule in the Senate to make big policy changes very unlikely. This is especially true when one party is not in firm control of all three institutions. Adding a third party to that mix won't make change more likely. On the contrary.

Regarding 4% spoiled ballot levels. I think I heard somewhere that they used to be much higher. But I have to admit, Australia is about 10,000 miles outside of my area of expertise (Europe).

Dr. Strangelove said...

Let me pose an interesting situation. Imagine there are three candidates for President:

1. Republican
2. Independent (or a minor third party)
3. Democrat

One can easily imagine the case where about half of the country picks the Republican as first choice and the Independent second, while the other half picks the Democrat as first choice and the Independent second.

So the Republican and Democrat are close to a draw, and even though few people picked the Independent as their first choice, almost everyone finds him an acceptable alternative (and would certainly prefer the Independent to their nemesis!) In this case, who ought to win?

I am not asking who *would* win, but who you think *should* win. I figure most voting systems would result in an effective coin-toss between the Democrat and Republican, determined by a slender difference of 50.1% to 49.9% in first place votes, or maybe by a similar slight difference in second or third place preferences. And yet the Independent *is* an acceptable alternative to nearly everyone... I'm not sure what I think should happen--what do you guys think?

Raised By Republicans said...

I think that the winner should be determined by who the Independent's supporters pick most as their second choice.

If we make the Independent the winner, we would essentially be saying that 2nd choice votes are more important than 1st choice votes. And of course there is strong reason to believe that the opposite is true. After all if the overwhelming majority pick the Independent as 2nd, they had reasons to do so.

Dr. Strangelove said...

Test... previous post did not appear.

Anonymous said...

Not sure about your assumption, but the figures I quoted were for four elections before the most recent polls.

I understand what you mean by "gridlock" now. My original comment was about election "deadlock" and not passing legislation.

The Law Talking Guy said...

The practical effect of SF's system is to give third parties a huge boost, with the spillover effect in areas still using the traditional winner-take-all system. This might hurt Democrats, since it will (in SF) favor the Greens. If Orange County adopted this system, by contrast, it would favor the libertarians.

Raised By Republicans said...

I'm not sure the Greens will be the major beneficiaries of STV voting in SF. They would be ONLY if the Greens are able to get close to or over 50% of the 1st choice votes. I'm not sure they are, even in SF. They might be almost there in some neighborhoods but I doubt they are so dominant citywide. Also, in order to benefit from coming in second under STV's first round, they would also need to be the second choice of the Democratic voters. I would imagine Democrats would be split on whether they thought the Greens or a moderate Republican would be their 2nd choice.

I would be shocked if the Democrats came out losers in this. I think its far more likely that the Republican party in SF change positions with the Greens as the local fringe party (if it hasn't already). If the Greens are the biggest vote getters I still think the Democrats could beat the Republicans in the first round and would be the 2nd choice of most Republicans and so would still beat the Greens after the second round of voting. Alterantively, if the Greens were the 2nd biggest vote getters in the 1st round, the Democrats would likely be the biggest vote getters and again, it would come down to who would the Republicans pick as their 2nd choice.

The Greens would benefit in that the number of people openly identifying with their party would go up. But I doubt they would win any more Council seats than they already have.

Ultimately, I think Green hopes for victory resulting from changing from one majoritarian electoral system to a slightly less majoritarian electoral system miss the point. The biggest single reason people don't vote for Green candidates isn't because the electoral system forces them not to. It's because the policies advocated by Green party candidates are unpopular with the overwhelming majority of voters. Even in very left wing cities like San Francisco, there is not likely to be anything like a Green majority or even a Green plurality regardless of electoral system.