Discussion of alternate election methods for the U.S.

I, like many Americans, think it is a shame that third-party candidates often act as ‘spoilers’ in elections. In the 2000 election, Nader voters were told that their votes were more of a vote against Gore than they were a vote for Nader. In the coming election, the same dilemma will apparently exist.

This problem comes from the way American elections are run. We have what is known as a plurality system, in which each voter must pick only one candidate to support. In a lot of cases, the voter is forced to vote strategically. An example would be a Nader supporter that voted for Gore in order to block Bush, even though that wasn’t his favorite candidate.

There are several other election systems available that might alleviate this problem. Three major ones are Approval Voting, Instant Runoff Voting, and Condorcet Voting.

In Approval Voting, voters can vote for as many candidates as they please. In 2000, someone could vote for both Nader and Gore, which would show they approve both of those, and don’t approve of Bush and Buchannan (and others). No ranking among candidates is recorded (i.e. the tally doesn’t show whether this is a Nader voter who thinks Gore is merely good enough, or vice versa).

In Instant Runoff Voting and Condorcet Voting, voters rank some number of candidates. The specifics can vary from numbering all candidates (giving each a number–1 for most-preferred, 2 for next most-preferred, etc.) to picking only the favorite and second-favorite. The two methods only differ in how the winner is determined when the votes are counted.

In Instant Runoff, all the favorite votes are totalled, then the loser’s votes are converted to those voter’s second-favorites. Things are added up again, and the new loser’s votes are converted. Eventually, a clear winner emerges. In this case, a person who voted Nader first and Gore second would have their vote converted to a Gore vote after it was determined that Nader wasn’t winning in one of the rounds of totalling.

In Condorcet, all candidates are paired in as many 1-on-1 races as necessary to compare every candidate with every other one. If a voter ranks one candidate above another, that candidate gets the vote in that pairing. A winner comes when a candidate wins against all other candidates in every pairing. If no candidate wins all pairings, there are a few methods to resolve the ambiguity.

I’ve only read a few things on this subject. If I’ve mis-described any of the methods above, please correct me. The best site I’ve found in my few searches is www.electionmethods.org, though it is not impartial. (it favors adopting Approval Voting as soon as possible, then transistioning to Condorcet. It is strongly against Instant Runoff Voting.) It also features a few other methods.

Personally, I think it would be great if my state were to go straight into Condorcet. However, I know the major parties would be against any change, as each of these changes would strengthen third (and fourth, fifth, etc.) parties.

Would you support a change in American voting methods? Which method would you support?

Do you think the American political landscape would ever allow such a change? What would have to happen?

By the way, this debate is on a change in how elections are run in the U.S., not on changing our political system entirely. I know some people will want to suggest changing to proportional representation, but that’s not what were going to discuss. That would require a Constitutional amendment and would totally alter the way Congress operates after the election. Changing election methods could be done at the state level, and Congress would function as usual once the seats were determined.

Also, keep in mind we’re talking about the unspecified future. The changes we’re discussing will not be a factor in the 2004 presidential election, by any stretch of the imagination.

If we had used the Condorcet method for the California recall, we’d still be counting.

No, that’s not right. We’d still be voting.

Approval voting is probably the best idea, but the problem there is that you end up with hundreds of parties and thousands of candidates. Americans barely care about the mudslinging between two of them - I don’t think adding more would be a good idea. However, it would lend more power to “alternative” parties like the Greens. Unfortunately, if they were going to be popular enough to outvote the Democrats, they would already have done so. It WOULD help an eventual situation where the Greens could get, say, 20% of the vote, and the Dems 30%, and the Reps 50%, though. It would certainly throw a wrench into the two-party system we have now. However, it also allows one party to run multiple candidates, which is an iffy situation.

The others are actually more muddled than what we have now. Instant Runoff would assign someone’s votes (eventually) to someone they didn’t want to vote for. In your example, if they ranked Nader 1 and Gore 6, eventually their vote would be folded into Gore’s party. Concordant is just too confusing. We could barely keep track of the direct democratic popular vote in the last Presidential election, and The Governator won here in California largely because no one really gave a crap and thought it would be funny. No, seriously, that is why most people I talked to voted for him. It was kinda another Anybody But Bush campaign.

We use a preferential voting system in both federal and state elections here in Australia.

Voters mark their ballot papers 1, 2, 3 etc in order of preference for the candidates. Then the counting proceeds as follows:

(a) each candidate’s total number of first preference votes is determined;
(b) if a candidate has an absolute majority (ie 50% +1) of the total votes then he is elected;
© otherwise, if no candidate has an absolute majority, the candidate who gained the smallest number of first preference votes is eliminated from the count and all of that candidate’s votes are re-distributed to whichever candidate was given the second preference;
(d) then you keep going back to steps (b) and © until one candidate has an absolute majority.

So that means that there’s no such thing as a “wasted” vote if you vote for a minority party. There’s no need for “tactical” voting either.

Well that is the whole idea of the voting method. It ensures that your vote eventually does not assist the person you don’t want. So if the election finished Bush 45%, Kerry 44% and Nader 11% and the Nader voters all preferred Kerry to Bush you don’t get a President who is last choice of the majority of voters

Voting with the Condorcet method isn’t any more complicated than voting with instant runoff. You just list the candidates in your preferred order, and computers take care of the “virtual matchups”. It’s only slightly more complicated than voting with the approval method - instead of just checking the candidates you like, you rank them in order.

You could say the same thing about the system we have now. A Nader voter is effectively helping the candidate he likes least.

With IRV or Condorcet, he can express his preference for all the candidates, instead of having to choose between the candidate he really prefers, and the one he thinks has a chance of winning. He can say “I’d rather pick Nader than anyone else, but if Nader loses, I’d rather pick these other guys instead of Gore.”

His vote will only be assigned to Gore if Nader and his other choices #2-#5 are already sure losers, so if Gore is the candidate he likes least, he can just rank Gore last. With the Condorcet system, he will never hurt his preferred candidates’ chances by ranking them higher (which isn’t true of IRV, according to examples at the OP’s link).

Does anyone know what specific steps are needed to change voting methods? I know that it is a state-level issue (right now–there could be federal standards of some sort coming down the pike).

Here in Washington, we just had our presidential primary changed through a lawsuit, the state legislature, and the governor. So it seems like it wouldn’t require a vote of the populace. Of course, if it’s entirely up to elected officials, they’ll probably be against ti. Anybody know anything more specific?

Under our current system could someone explain what would happen if shortly before the election we had polls showing Bush 45%, Kerry 44% and Nader 11% and then Nader dropped out of the race and publicly asked his supporters to vote for Kerry?

I am puzzled as to why this doesn’t happen (Perot?).

Cunctator, that’s the best system I ever heard of! But how long does it take to count all the ballots?

Under the current system, nothing happens if a candidate quits the race. The votes aren’t his to give away. Candidates withdraw from all sorts of races. Usually, some (usually small) group of supporters insist on voting for them anyway. Hell, here in Missouri, we elected a candidate who died before the election.

In fact, even in the American politcal party system, when a candidate withdraws, all he can do is release the delegates he has from their pledge to support him. He can’t (at least officially) bind them to any other candidate.

As a voter, you might not mind if you voted for Nader and he “assigned” your vote to Kerry, but suppose Nader decided to throw his support to Bush, or Lyndon LaRouche?

I like this preferential sytem.

The system Cunctator described is Instant Runoff, and unfortunately it suffers from some problems that aren’t there in the Condorcet method:

The Problem with Instant Runoff Voting

Very interesting link Mr2001. I guess this is why we’re appropriately reluctant to reform the current system. Everything seems to have unintended consequences.

I don’t think the argument of **Mr2001 ** is valid. In his example, once the Republican candidate is eliminated, his preferences have to go somewhere. If they flow to the Libertarian candidate, who is thus elected, then Mr2001 will be happy. If instead the preferences result in the election of the Democrat candidate, then Mr2001 won’t like the outcome, even though it represents the will of the majority. This problem of disenfranchisement of a (possibly significant) minority of voters is always going to be an issue whenever the electoral system is based on single-member electorates. The preferential voting system isn’t perfect, but it’s definitely better than any system requiring merely a *relative * majority.

Check out this page at the site I linked to above:

Technical Evaluation of Election Methods

It shows a lot of the weaknesses of Instant Runoff Voting, which seems to be what you’re using in Australia. It also shows how Condorcet Voting does better in most criteria.

On the other hand, I don’t think IRV is as flawed as that site argues. I think the main thing that the preferential systems do is relieve the voter of the feeling that any vote for a third-party candidate is a wasted vote, or a vote for one’s least favorite candidate. The very complexity of these voting systems makes that sort of strategizing difficult.

Again, the site Mr2001 and I have used is definitely biased against IRV and for Condorcet. Perhaps tomorrow I’ll go looking for another site that’s a tad more balanced.

Also, in answer to those who’ve commented on how long it must take to count votes in a preferential system: Keep in mind that we now use computers to tabulate with our very simple Plurality system. Computers could count preferential ballots just as easily and as fast.

I’ll respond by paraphrasing an example from the same site (found here).

Suppose there are three candidates in an Instant Runoff election: a Republican, a Democrat, and a Libertarian. The votes are as follows:

8 people vote #1 Republican, #2 Libertarian, #3 Democrat.
5 people vote #1 Democrat, #2 Republican, #3 Libertarian.
4 people vote #1 Libertarian, #2 Democrat, #3 Republican.

The Libertarian has the fewest #1 votes, so he is eliminated. His four votes are given to the Democrat, who now wins, 9 votes to 8.

Now suppose two of the Republican voters decide to rank the Libertarian in first place instead. The votes are now as follows:

6 people vote #1 Republican, #2 Libertarian, #3 Democrat.
2 people vote #1 Libertarian, #2 Republican, #3 Democrat.
5 people vote #1 Democrat, #2 Republican, #3 Libertarian.
4 people vote #1 Libertarian, #2 Democrat, #3 Republican.

This time, the Democrat has the fewest #1 votes, so he is eliminated. His five votes are given to the Republican, who now wins, 11 votes to 6.

See what just happened? By ranking the Republican lower, those two voters caused him to win! This means that under IRV, you still can’t vote your true preference; you must try to predict how everyone else will vote, and vote strategically, otherwise you’ll end up hurting the candidates you want to help (and vice versa).

Perhaps. But as long as you’re asking voters to rank candidates in order, there’s no reason to count their votes using IRV when you could be using the Condorcet method instead, avoiding unintuitive results like those above.

The site does have a preference for Condorcet, but I think it’s justified. The criteria on that evaluation page are objective and sound, and it certainly seems that Condorcet is the best option (and IRV the worst) when you consider those criteria.

I’m afraid that I don’t see. In the second example the two voters made their preferences very clear. They wanted the Libertarian candidate to win. In default of a Libertarian candidate they preferred a Republican win to a Democrat win - which is precisely what they got.

Try the website of the Center for Voting and Democracy – www.fairvote.org. Not exactly “balanced,” it’s an organization in favor of IRV. I’ve never seen the Condorcet method mentioned in its literature, but in principle CV&D would have no reason to favor one over the other – the point is to support any reform which will make individual votes more effective and open the ballots to realistic third-party contenders.

CV&D’s other main issue is proportional representation (PR), which unlike IRV is a system for electing a multimember policymaking body (Congress, your state legislature, your local school board, etc.). What we use now to elect the House of Representatives and both houses of every state legislature is the “single-member district” system, inspired by Britain’s system for electing Parliament. In fact, we;ve been using it since colonial times. The territory to be represented is divided into districts of (roughly) equal population, and the people of each district elect one representative by majority or, more often, plurality vote (whoever gets the most votes wins, even if he or she does not get a majority of votes cast). One result of this is that very few Congressional districts are really competitive; most are “safe seats” for whichever party predominates in that district. Another result is that the system is naturally bipolar – minor parties are squeezed out. If 20% of all Americans decided they wanted to vote Green (or pick your own favorite third party) in the next Congressional election, how man Greens get into Congress? None, because there wouldn’t be enough Green voters in any one district to make up a plurality of the vote. But under a PR system, if the Greens get 20% of the vote, they get 20% of the seats.

See the following GD threads:

“Should the U.S. adopt alternative, pro-multipartisan voting systems?” –

“What would a multipartisan America be like?” –

The point of Mr2001’s example is that those two voters wanted the Republican to win. Furthermore, since they were savvy enough to predict others’ votes, they turned a Democratic win into a Republican win by pretending to prefer the Libertarian.

As I said earlier, however, I don’t think a real election would have this sort of activity (at least not often), because it would be almost impossible to predict everyone’s vote well enough to ‘target’ your strategy in the way the example was done. Still, it seems clear that Condorcet avoids the problem while not making things more difficult on the voter.